diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,192521 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 27497,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 3.6367603738589666e-05,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4242424242424243e-08,
+      "loss": 2.0117,
+      "step": 1
+    },
+    {
+      "epoch": 7.273520747717933e-05,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8484848484848486e-08,
+      "loss": 2.1145,
+      "step": 2
+    },
+    {
+      "epoch": 0.000109102811215769,
+      "grad_norm": 0.0,
+      "learning_rate": 7.272727272727274e-08,
+      "loss": 2.1519,
+      "step": 3
+    },
+    {
+      "epoch": 0.00014547041495435866,
+      "grad_norm": 0.0,
+      "learning_rate": 9.696969696969697e-08,
+      "loss": 1.9951,
+      "step": 4
+    },
+    {
+      "epoch": 0.00018183801869294832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2121212121212122e-07,
+      "loss": 2.0078,
+      "step": 5
+    },
+    {
+      "epoch": 0.000218205622431538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4545454545454548e-07,
+      "loss": 2.0085,
+      "step": 6
+    },
+    {
+      "epoch": 0.0002545732261701277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6969696969696974e-07,
+      "loss": 2.1094,
+      "step": 7
+    },
+    {
+      "epoch": 0.0002909408299087173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9393939393939395e-07,
+      "loss": 2.1797,
+      "step": 8
+    },
+    {
+      "epoch": 0.000327308433647307,
+      "grad_norm": 0.0,
+      "learning_rate": 2.181818181818182e-07,
+      "loss": 1.9698,
+      "step": 9
+    },
+    {
+      "epoch": 0.00036367603738589663,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4242424242424244e-07,
+      "loss": 2.1689,
+      "step": 10
+    },
+    {
+      "epoch": 0.0004000436411244863,
+      "grad_norm": 0.0,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 2.0037,
+      "step": 11
+    },
+    {
+      "epoch": 0.000436411244863076,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9090909090909096e-07,
+      "loss": 2.0271,
+      "step": 12
+    },
+    {
+      "epoch": 0.00047277884860166564,
+      "grad_norm": 0.0,
+      "learning_rate": 3.151515151515152e-07,
+      "loss": 1.9519,
+      "step": 13
+    },
+    {
+      "epoch": 0.0005091464523402553,
+      "grad_norm": 0.0,
+      "learning_rate": 3.393939393939395e-07,
+      "loss": 1.8694,
+      "step": 14
+    },
+    {
+      "epoch": 0.000545514056078845,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6363636363636366e-07,
+      "loss": 1.8102,
+      "step": 15
+    },
+    {
+      "epoch": 0.0005818816598174347,
+      "grad_norm": 0.0,
+      "learning_rate": 3.878787878787879e-07,
+      "loss": 1.8485,
+      "step": 16
+    },
+    {
+      "epoch": 0.0006182492635560243,
+      "grad_norm": 0.0,
+      "learning_rate": 4.121212121212122e-07,
+      "loss": 1.7985,
+      "step": 17
+    },
+    {
+      "epoch": 0.000654616867294614,
+      "grad_norm": 0.0,
+      "learning_rate": 4.363636363636364e-07,
+      "loss": 1.689,
+      "step": 18
+    },
+    {
+      "epoch": 0.0006909844710332036,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6060606060606064e-07,
+      "loss": 1.7521,
+      "step": 19
+    },
+    {
+      "epoch": 0.0007273520747717933,
+      "grad_norm": 0.0,
+      "learning_rate": 4.848484848484849e-07,
+      "loss": 1.7284,
+      "step": 20
+    },
+    {
+      "epoch": 0.0007637196785103829,
+      "grad_norm": 0.0,
+      "learning_rate": 5.090909090909092e-07,
+      "loss": 1.6492,
+      "step": 21
+    },
+    {
+      "epoch": 0.0008000872822489726,
+      "grad_norm": 0.0,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 1.5232,
+      "step": 22
+    },
+    {
+      "epoch": 0.0008364548859875623,
+      "grad_norm": 0.0,
+      "learning_rate": 5.575757575757576e-07,
+      "loss": 1.5225,
+      "step": 23
+    },
+    {
+      "epoch": 0.000872822489726152,
+      "grad_norm": 0.0,
+      "learning_rate": 5.818181818181819e-07,
+      "loss": 1.4633,
+      "step": 24
+    },
+    {
+      "epoch": 0.0009091900934647416,
+      "grad_norm": 0.0,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 1.6095,
+      "step": 25
+    },
+    {
+      "epoch": 0.0009455576972033313,
+      "grad_norm": 0.0,
+      "learning_rate": 6.303030303030304e-07,
+      "loss": 1.6414,
+      "step": 26
+    },
+    {
+      "epoch": 0.000981925300941921,
+      "grad_norm": 0.0,
+      "learning_rate": 6.545454545454547e-07,
+      "loss": 1.506,
+      "step": 27
+    },
+    {
+      "epoch": 0.0010182929046805107,
+      "grad_norm": 0.0,
+      "learning_rate": 6.78787878787879e-07,
+      "loss": 1.7639,
+      "step": 28
+    },
+    {
+      "epoch": 0.0010546605084191002,
+      "grad_norm": 0.0,
+      "learning_rate": 7.03030303030303e-07,
+      "loss": 1.5128,
+      "step": 29
+    },
+    {
+      "epoch": 0.00109102811215769,
+      "grad_norm": 0.0,
+      "learning_rate": 7.272727272727273e-07,
+      "loss": 1.5779,
+      "step": 30
+    },
+    {
+      "epoch": 0.0011273957158962795,
+      "grad_norm": 0.0,
+      "learning_rate": 7.515151515151516e-07,
+      "loss": 1.4362,
+      "step": 31
+    },
+    {
+      "epoch": 0.0011637633196348693,
+      "grad_norm": 0.0,
+      "learning_rate": 7.757575757575758e-07,
+      "loss": 1.4406,
+      "step": 32
+    },
+    {
+      "epoch": 0.0012001309233734588,
+      "grad_norm": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.5795,
+      "step": 33
+    },
+    {
+      "epoch": 0.0012364985271120486,
+      "grad_norm": 0.0,
+      "learning_rate": 8.242424242424244e-07,
+      "loss": 1.392,
+      "step": 34
+    },
+    {
+      "epoch": 0.0012728661308506382,
+      "grad_norm": 0.0,
+      "learning_rate": 8.484848484848486e-07,
+      "loss": 1.489,
+      "step": 35
+    },
+    {
+      "epoch": 0.001309233734589228,
+      "grad_norm": 0.0,
+      "learning_rate": 8.727272727272728e-07,
+      "loss": 1.5786,
+      "step": 36
+    },
+    {
+      "epoch": 0.0013456013383278177,
+      "grad_norm": 0.0,
+      "learning_rate": 8.96969696969697e-07,
+      "loss": 1.6243,
+      "step": 37
+    },
+    {
+      "epoch": 0.0013819689420664072,
+      "grad_norm": 0.0,
+      "learning_rate": 9.212121212121213e-07,
+      "loss": 1.6385,
+      "step": 38
+    },
+    {
+      "epoch": 0.001418336545804997,
+      "grad_norm": 0.0,
+      "learning_rate": 9.454545454545455e-07,
+      "loss": 1.4865,
+      "step": 39
+    },
+    {
+      "epoch": 0.0014547041495435865,
+      "grad_norm": 0.0,
+      "learning_rate": 9.696969696969698e-07,
+      "loss": 1.5259,
+      "step": 40
+    },
+    {
+      "epoch": 0.0014910717532821763,
+      "grad_norm": 0.0,
+      "learning_rate": 9.93939393939394e-07,
+      "loss": 1.4402,
+      "step": 41
+    },
+    {
+      "epoch": 0.0015274393570207658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0181818181818183e-06,
+      "loss": 1.4863,
+      "step": 42
+    },
+    {
+      "epoch": 0.0015638069607593556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0424242424242426e-06,
+      "loss": 1.4957,
+      "step": 43
+    },
+    {
+      "epoch": 0.0016001745644979451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 1.455,
+      "step": 44
+    },
+    {
+      "epoch": 0.0016365421682365349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.090909090909091e-06,
+      "loss": 1.3968,
+      "step": 45
+    },
+    {
+      "epoch": 0.0016729097719751247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1151515151515153e-06,
+      "loss": 1.507,
+      "step": 46
+    },
+    {
+      "epoch": 0.0017092773757137142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1393939393939395e-06,
+      "loss": 1.4561,
+      "step": 47
+    },
+    {
+      "epoch": 0.001745644979452304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1636363636363638e-06,
+      "loss": 1.4909,
+      "step": 48
+    },
+    {
+      "epoch": 0.0017820125831908935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.187878787878788e-06,
+      "loss": 1.5271,
+      "step": 49
+    },
+    {
+      "epoch": 0.0018183801869294833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 1.4473,
+      "step": 50
+    },
+    {
+      "epoch": 0.0018547477906680728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2363636363636365e-06,
+      "loss": 1.4314,
+      "step": 51
+    },
+    {
+      "epoch": 0.0018911153944066626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2606060606060608e-06,
+      "loss": 1.5528,
+      "step": 52
+    },
+    {
+      "epoch": 0.001927482998145252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2848484848484848e-06,
+      "loss": 1.4731,
+      "step": 53
+    },
+    {
+      "epoch": 0.001963850601883842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3090909090909093e-06,
+      "loss": 1.6626,
+      "step": 54
+    },
+    {
+      "epoch": 0.0020002182056224316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.5104,
+      "step": 55
+    },
+    {
+      "epoch": 0.0020365858093610214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.357575757575758e-06,
+      "loss": 1.4947,
+      "step": 56
+    },
+    {
+      "epoch": 0.0020729534130996107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.381818181818182e-06,
+      "loss": 1.4957,
+      "step": 57
+    },
+    {
+      "epoch": 0.0021093210168382005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.406060606060606e-06,
+      "loss": 1.5498,
+      "step": 58
+    },
+    {
+      "epoch": 0.0021456886205767902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4303030303030306e-06,
+      "loss": 1.4562,
+      "step": 59
+    },
+    {
+      "epoch": 0.00218205622431538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4545454545454546e-06,
+      "loss": 1.329,
+      "step": 60
+    },
+    {
+      "epoch": 0.0022184238280539693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.478787878787879e-06,
+      "loss": 1.5116,
+      "step": 61
+    },
+    {
+      "epoch": 0.002254791431792559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5030303030303032e-06,
+      "loss": 1.3334,
+      "step": 62
+    },
+    {
+      "epoch": 0.002291159035531149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5272727272727275e-06,
+      "loss": 1.3886,
+      "step": 63
+    },
+    {
+      "epoch": 0.0023275266392697386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5515151515151516e-06,
+      "loss": 1.3915,
+      "step": 64
+    },
+    {
+      "epoch": 0.0023638942430083284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5757575757575759e-06,
+      "loss": 1.4551,
+      "step": 65
+    },
+    {
+      "epoch": 0.0024002618467469177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.4904,
+      "step": 66
+    },
+    {
+      "epoch": 0.0024366294504855075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6242424242424242e-06,
+      "loss": 1.4669,
+      "step": 67
+    },
+    {
+      "epoch": 0.002472997054224097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6484848484848487e-06,
+      "loss": 1.5348,
+      "step": 68
+    },
+    {
+      "epoch": 0.002509364657962687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6727272727272728e-06,
+      "loss": 1.3731,
+      "step": 69
+    },
+    {
+      "epoch": 0.0025457322617012763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6969696969696973e-06,
+      "loss": 1.4272,
+      "step": 70
+    },
+    {
+      "epoch": 0.002582099865439866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7212121212121214e-06,
+      "loss": 1.4087,
+      "step": 71
+    },
+    {
+      "epoch": 0.002618467469178456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7454545454545456e-06,
+      "loss": 1.3931,
+      "step": 72
+    },
+    {
+      "epoch": 0.0026548350729170456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.76969696969697e-06,
+      "loss": 1.5868,
+      "step": 73
+    },
+    {
+      "epoch": 0.0026912026766556353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.793939393939394e-06,
+      "loss": 1.2712,
+      "step": 74
+    },
+    {
+      "epoch": 0.0027275702803942247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 1.3785,
+      "step": 75
+    },
+    {
+      "epoch": 0.0027639378841328144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8424242424242426e-06,
+      "loss": 1.3925,
+      "step": 76
+    },
+    {
+      "epoch": 0.002800305487871404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 1.3066,
+      "step": 77
+    },
+    {
+      "epoch": 0.002836673091609994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890909090909091e-06,
+      "loss": 1.4801,
+      "step": 78
+    },
+    {
+      "epoch": 0.0028730406953485833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9151515151515154e-06,
+      "loss": 1.3511,
+      "step": 79
+    },
+    {
+      "epoch": 0.002909408299087173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9393939393939395e-06,
+      "loss": 1.4011,
+      "step": 80
+    },
+    {
+      "epoch": 0.002945775902825763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9636363636363636e-06,
+      "loss": 1.4792,
+      "step": 81
+    },
+    {
+      "epoch": 0.0029821435065643526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987878787878788e-06,
+      "loss": 1.5513,
+      "step": 82
+    },
+    {
+      "epoch": 0.0030185111103029423,
+      "grad_norm": 0.0,
+      "learning_rate": 2.012121212121212e-06,
+      "loss": 1.3022,
+      "step": 83
+    },
+    {
+      "epoch": 0.0030548787140415317,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0363636363636367e-06,
+      "loss": 1.4351,
+      "step": 84
+    },
+    {
+      "epoch": 0.0030912463177801214,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0606060606060607e-06,
+      "loss": 1.4193,
+      "step": 85
+    },
+    {
+      "epoch": 0.003127613921518711,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0848484848484852e-06,
+      "loss": 1.3557,
+      "step": 86
+    },
+    {
+      "epoch": 0.003163981525257301,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1090909090909093e-06,
+      "loss": 1.3598,
+      "step": 87
+    },
+    {
+      "epoch": 0.0032003491289958903,
+      "grad_norm": 0.0,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 1.3892,
+      "step": 88
+    },
+    {
+      "epoch": 0.00323671673273448,
+      "grad_norm": 0.0,
+      "learning_rate": 2.157575757575758e-06,
+      "loss": 1.4038,
+      "step": 89
+    },
+    {
+      "epoch": 0.0032730843364730698,
+      "grad_norm": 0.0,
+      "learning_rate": 2.181818181818182e-06,
+      "loss": 1.4088,
+      "step": 90
+    },
+    {
+      "epoch": 0.0033094519402116595,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2060606060606064e-06,
+      "loss": 1.2576,
+      "step": 91
+    },
+    {
+      "epoch": 0.0033458195439502493,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2303030303030305e-06,
+      "loss": 1.3773,
+      "step": 92
+    },
+    {
+      "epoch": 0.0033821871476888386,
+      "grad_norm": 0.0,
+      "learning_rate": 2.254545454545455e-06,
+      "loss": 1.3254,
+      "step": 93
+    },
+    {
+      "epoch": 0.0034185547514274284,
+      "grad_norm": 0.0,
+      "learning_rate": 2.278787878787879e-06,
+      "loss": 1.2699,
+      "step": 94
+    },
+    {
+      "epoch": 0.003454922355166018,
+      "grad_norm": 0.0,
+      "learning_rate": 2.303030303030303e-06,
+      "loss": 1.2252,
+      "step": 95
+    },
+    {
+      "epoch": 0.003491289958904608,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3272727272727277e-06,
+      "loss": 1.3873,
+      "step": 96
+    },
+    {
+      "epoch": 0.0035276575626431972,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3515151515151517e-06,
+      "loss": 1.4008,
+      "step": 97
+    },
+    {
+      "epoch": 0.003564025166381787,
+      "grad_norm": 0.0,
+      "learning_rate": 2.375757575757576e-06,
+      "loss": 1.3918,
+      "step": 98
+    },
+    {
+      "epoch": 0.0036003927701203768,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.3557,
+      "step": 99
+    },
+    {
+      "epoch": 0.0036367603738589665,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 1.3966,
+      "step": 100
+    },
+    {
+      "epoch": 0.0036731279775975563,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4484848484848485e-06,
+      "loss": 1.319,
+      "step": 101
+    },
+    {
+      "epoch": 0.0037094955813361456,
+      "grad_norm": 0.0,
+      "learning_rate": 2.472727272727273e-06,
+      "loss": 1.3094,
+      "step": 102
+    },
+    {
+      "epoch": 0.0037458631850747354,
+      "grad_norm": 0.0,
+      "learning_rate": 2.496969696969697e-06,
+      "loss": 1.4945,
+      "step": 103
+    },
+    {
+      "epoch": 0.003782230788813325,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5212121212121215e-06,
+      "loss": 1.4531,
+      "step": 104
+    },
+    {
+      "epoch": 0.003818598392551915,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5454545454545456e-06,
+      "loss": 1.4244,
+      "step": 105
+    },
+    {
+      "epoch": 0.003854965996290504,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5696969696969697e-06,
+      "loss": 1.3542,
+      "step": 106
+    },
+    {
+      "epoch": 0.003891333600029094,
+      "grad_norm": 0.0,
+      "learning_rate": 2.593939393939394e-06,
+      "loss": 1.4354,
+      "step": 107
+    },
+    {
+      "epoch": 0.003927701203767684,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6181818181818187e-06,
+      "loss": 1.3257,
+      "step": 108
+    },
+    {
+      "epoch": 0.003964068807506273,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6424242424242423e-06,
+      "loss": 1.2194,
+      "step": 109
+    },
+    {
+      "epoch": 0.004000436411244863,
+      "grad_norm": 0.0,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.3506,
+      "step": 110
+    },
+    {
+      "epoch": 0.004036804014983453,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6909090909090913e-06,
+      "loss": 1.3224,
+      "step": 111
+    },
+    {
+      "epoch": 0.004073171618722043,
+      "grad_norm": 0.0,
+      "learning_rate": 2.715151515151516e-06,
+      "loss": 1.2516,
+      "step": 112
+    },
+    {
+      "epoch": 0.004109539222460632,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7393939393939395e-06,
+      "loss": 1.3091,
+      "step": 113
+    },
+    {
+      "epoch": 0.0041459068261992214,
+      "grad_norm": 0.0,
+      "learning_rate": 2.763636363636364e-06,
+      "loss": 1.2928,
+      "step": 114
+    },
+    {
+      "epoch": 0.004182274429937812,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7878787878787885e-06,
+      "loss": 1.3373,
+      "step": 115
+    },
+    {
+      "epoch": 0.004218642033676401,
+      "grad_norm": 0.0,
+      "learning_rate": 2.812121212121212e-06,
+      "loss": 1.3739,
+      "step": 116
+    },
+    {
+      "epoch": 0.004255009637414991,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8363636363636366e-06,
+      "loss": 1.5255,
+      "step": 117
+    },
+    {
+      "epoch": 0.0042913772411535805,
+      "grad_norm": 0.0,
+      "learning_rate": 2.860606060606061e-06,
+      "loss": 1.295,
+      "step": 118
+    },
+    {
+      "epoch": 0.00432774484489217,
+      "grad_norm": 0.0,
+      "learning_rate": 2.884848484848485e-06,
+      "loss": 1.1973,
+      "step": 119
+    },
+    {
+      "epoch": 0.00436411244863076,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9090909090909093e-06,
+      "loss": 1.4072,
+      "step": 120
+    },
+    {
+      "epoch": 0.004400480052369349,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 1.3391,
+      "step": 121
+    },
+    {
+      "epoch": 0.004436847656107939,
+      "grad_norm": 0.0,
+      "learning_rate": 2.957575757575758e-06,
+      "loss": 1.4307,
+      "step": 122
+    },
+    {
+      "epoch": 0.004473215259846529,
+      "grad_norm": 0.0,
+      "learning_rate": 2.981818181818182e-06,
+      "loss": 1.412,
+      "step": 123
+    },
+    {
+      "epoch": 0.004509582863585118,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0060606060606064e-06,
+      "loss": 1.4076,
+      "step": 124
+    },
+    {
+      "epoch": 0.004545950467323708,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 1.2626,
+      "step": 125
+    },
+    {
+      "epoch": 0.004582318071062298,
+      "grad_norm": 0.0,
+      "learning_rate": 3.054545454545455e-06,
+      "loss": 1.3144,
+      "step": 126
+    },
+    {
+      "epoch": 0.004618685674800887,
+      "grad_norm": 0.0,
+      "learning_rate": 3.078787878787879e-06,
+      "loss": 1.5006,
+      "step": 127
+    },
+    {
+      "epoch": 0.004655053278539477,
+      "grad_norm": 0.0,
+      "learning_rate": 3.103030303030303e-06,
+      "loss": 1.3877,
+      "step": 128
+    },
+    {
+      "epoch": 0.0046914208822780665,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1272727272727276e-06,
+      "loss": 1.382,
+      "step": 129
+    },
+    {
+      "epoch": 0.004727788486016657,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1515151515151517e-06,
+      "loss": 1.4066,
+      "step": 130
+    },
+    {
+      "epoch": 0.004764156089755246,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1757575757575758e-06,
+      "loss": 1.2379,
+      "step": 131
+    },
+    {
+      "epoch": 0.004800523693493835,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.3102,
+      "step": 132
+    },
+    {
+      "epoch": 0.004836891297232426,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2242424242424248e-06,
+      "loss": 1.3011,
+      "step": 133
+    },
+    {
+      "epoch": 0.004873258900971015,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2484848484848484e-06,
+      "loss": 1.2786,
+      "step": 134
+    },
+    {
+      "epoch": 0.004909626504709605,
+      "grad_norm": 0.0,
+      "learning_rate": 3.272727272727273e-06,
+      "loss": 1.4695,
+      "step": 135
+    },
+    {
+      "epoch": 0.004945994108448194,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2969696969696974e-06,
+      "loss": 1.2533,
+      "step": 136
+    },
+    {
+      "epoch": 0.004982361712186784,
+      "grad_norm": 0.0,
+      "learning_rate": 3.321212121212121e-06,
+      "loss": 1.3934,
+      "step": 137
+    },
+    {
+      "epoch": 0.005018729315925374,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3454545454545456e-06,
+      "loss": 1.2443,
+      "step": 138
+    },
+    {
+      "epoch": 0.005055096919663963,
+      "grad_norm": 0.0,
+      "learning_rate": 3.36969696969697e-06,
+      "loss": 1.3268,
+      "step": 139
+    },
+    {
+      "epoch": 0.005091464523402553,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3939393939393946e-06,
+      "loss": 1.2601,
+      "step": 140
+    },
+    {
+      "epoch": 0.005127832127141143,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4181818181818182e-06,
+      "loss": 1.3925,
+      "step": 141
+    },
+    {
+      "epoch": 0.005164199730879732,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4424242424242427e-06,
+      "loss": 1.2889,
+      "step": 142
+    },
+    {
+      "epoch": 0.005200567334618322,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 1.3899,
+      "step": 143
+    },
+    {
+      "epoch": 0.005236934938356912,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4909090909090913e-06,
+      "loss": 1.3905,
+      "step": 144
+    },
+    {
+      "epoch": 0.005273302542095501,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5151515151515154e-06,
+      "loss": 1.3458,
+      "step": 145
+    },
+    {
+      "epoch": 0.005309670145834091,
+      "grad_norm": 0.0,
+      "learning_rate": 3.53939393939394e-06,
+      "loss": 1.2813,
+      "step": 146
+    },
+    {
+      "epoch": 0.0053460377495726805,
+      "grad_norm": 0.0,
+      "learning_rate": 3.563636363636364e-06,
+      "loss": 1.2688,
+      "step": 147
+    },
+    {
+      "epoch": 0.005382405353311271,
+      "grad_norm": 0.0,
+      "learning_rate": 3.587878787878788e-06,
+      "loss": 1.3318,
+      "step": 148
+    },
+    {
+      "epoch": 0.00541877295704986,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6121212121212125e-06,
+      "loss": 1.3444,
+      "step": 149
+    },
+    {
+      "epoch": 0.005455140560788449,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 1.3625,
+      "step": 150
+    },
+    {
+      "epoch": 0.0054915081645270395,
+      "grad_norm": 0.0,
+      "learning_rate": 3.660606060606061e-06,
+      "loss": 1.1973,
+      "step": 151
+    },
+    {
+      "epoch": 0.005527875768265629,
+      "grad_norm": 0.0,
+      "learning_rate": 3.684848484848485e-06,
+      "loss": 1.2345,
+      "step": 152
+    },
+    {
+      "epoch": 0.005564243372004219,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7090909090909092e-06,
+      "loss": 1.3761,
+      "step": 153
+    },
+    {
+      "epoch": 0.005600610975742808,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 1.1974,
+      "step": 154
+    },
+    {
+      "epoch": 0.005636978579481398,
+      "grad_norm": 0.0,
+      "learning_rate": 3.757575757575758e-06,
+      "loss": 1.391,
+      "step": 155
+    },
+    {
+      "epoch": 0.005673346183219988,
+      "grad_norm": 0.0,
+      "learning_rate": 3.781818181818182e-06,
+      "loss": 1.2848,
+      "step": 156
+    },
+    {
+      "epoch": 0.005709713786958577,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8060606060606064e-06,
+      "loss": 1.3304,
+      "step": 157
+    },
+    {
+      "epoch": 0.0057460813906971666,
+      "grad_norm": 0.0,
+      "learning_rate": 3.830303030303031e-06,
+      "loss": 1.4203,
+      "step": 158
+    },
+    {
+      "epoch": 0.005782448994435757,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8545454545454545e-06,
+      "loss": 1.3425,
+      "step": 159
+    },
+    {
+      "epoch": 0.005818816598174346,
+      "grad_norm": 0.0,
+      "learning_rate": 3.878787878787879e-06,
+      "loss": 1.3171,
+      "step": 160
+    },
+    {
+      "epoch": 0.005855184201912936,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9030303030303035e-06,
+      "loss": 1.4928,
+      "step": 161
+    },
+    {
+      "epoch": 0.005891551805651526,
+      "grad_norm": 0.0,
+      "learning_rate": 3.927272727272727e-06,
+      "loss": 1.406,
+      "step": 162
+    },
+    {
+      "epoch": 0.005927919409390115,
+      "grad_norm": 0.0,
+      "learning_rate": 3.951515151515152e-06,
+      "loss": 1.3874,
+      "step": 163
+    },
+    {
+      "epoch": 0.005964287013128705,
+      "grad_norm": 0.0,
+      "learning_rate": 3.975757575757576e-06,
+      "loss": 1.3611,
+      "step": 164
+    },
+    {
+      "epoch": 0.0060006546168672945,
+      "grad_norm": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.2985,
+      "step": 165
+    },
+    {
+      "epoch": 0.006037022220605885,
+      "grad_norm": 0.0,
+      "learning_rate": 4.024242424242424e-06,
+      "loss": 1.3567,
+      "step": 166
+    },
+    {
+      "epoch": 0.006073389824344474,
+      "grad_norm": 0.0,
+      "learning_rate": 4.048484848484849e-06,
+      "loss": 1.2971,
+      "step": 167
+    },
+    {
+      "epoch": 0.006109757428083063,
+      "grad_norm": 0.0,
+      "learning_rate": 4.072727272727273e-06,
+      "loss": 1.3406,
+      "step": 168
+    },
+    {
+      "epoch": 0.0061461250318216535,
+      "grad_norm": 0.0,
+      "learning_rate": 4.096969696969697e-06,
+      "loss": 1.3558,
+      "step": 169
+    },
+    {
+      "epoch": 0.006182492635560243,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1212121212121215e-06,
+      "loss": 1.2321,
+      "step": 170
+    },
+    {
+      "epoch": 0.006218860239298833,
+      "grad_norm": 0.0,
+      "learning_rate": 4.145454545454546e-06,
+      "loss": 1.1345,
+      "step": 171
+    },
+    {
+      "epoch": 0.006255227843037422,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1696969696969705e-06,
+      "loss": 1.2773,
+      "step": 172
+    },
+    {
+      "epoch": 0.006291595446776012,
+      "grad_norm": 0.0,
+      "learning_rate": 4.193939393939394e-06,
+      "loss": 1.3777,
+      "step": 173
+    },
+    {
+      "epoch": 0.006327963050514602,
+      "grad_norm": 0.0,
+      "learning_rate": 4.218181818181819e-06,
+      "loss": 1.4131,
+      "step": 174
+    },
+    {
+      "epoch": 0.006364330654253191,
+      "grad_norm": 0.0,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 1.2722,
+      "step": 175
+    },
+    {
+      "epoch": 0.0064006982579917805,
+      "grad_norm": 0.0,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 1.3842,
+      "step": 176
+    },
+    {
+      "epoch": 0.006437065861730371,
+      "grad_norm": 0.0,
+      "learning_rate": 4.290909090909091e-06,
+      "loss": 1.2546,
+      "step": 177
+    },
+    {
+      "epoch": 0.00647343346546896,
+      "grad_norm": 0.0,
+      "learning_rate": 4.315151515151516e-06,
+      "loss": 1.3432,
+      "step": 178
+    },
+    {
+      "epoch": 0.00650980106920755,
+      "grad_norm": 0.0,
+      "learning_rate": 4.33939393939394e-06,
+      "loss": 1.2925,
+      "step": 179
+    },
+    {
+      "epoch": 0.0065461686729461396,
+      "grad_norm": 0.0,
+      "learning_rate": 4.363636363636364e-06,
+      "loss": 1.4094,
+      "step": 180
+    },
+    {
+      "epoch": 0.006582536276684729,
+      "grad_norm": 0.0,
+      "learning_rate": 4.387878787878788e-06,
+      "loss": 1.2784,
+      "step": 181
+    },
+    {
+      "epoch": 0.006618903880423319,
+      "grad_norm": 0.0,
+      "learning_rate": 4.412121212121213e-06,
+      "loss": 1.3638,
+      "step": 182
+    },
+    {
+      "epoch": 0.006655271484161908,
+      "grad_norm": 0.0,
+      "learning_rate": 4.436363636363637e-06,
+      "loss": 1.3947,
+      "step": 183
+    },
+    {
+      "epoch": 0.006691639087900499,
+      "grad_norm": 0.0,
+      "learning_rate": 4.460606060606061e-06,
+      "loss": 1.3953,
+      "step": 184
+    },
+    {
+      "epoch": 0.006728006691639088,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4848484848484855e-06,
+      "loss": 1.2687,
+      "step": 185
+    },
+    {
+      "epoch": 0.006764374295377677,
+      "grad_norm": 0.0,
+      "learning_rate": 4.50909090909091e-06,
+      "loss": 1.3275,
+      "step": 186
+    },
+    {
+      "epoch": 0.0068007418991162675,
+      "grad_norm": 0.0,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 1.2107,
+      "step": 187
+    },
+    {
+      "epoch": 0.006837109502854857,
+      "grad_norm": 0.0,
+      "learning_rate": 4.557575757575758e-06,
+      "loss": 1.4572,
+      "step": 188
+    },
+    {
+      "epoch": 0.006873477106593447,
+      "grad_norm": 0.0,
+      "learning_rate": 4.581818181818183e-06,
+      "loss": 1.3351,
+      "step": 189
+    },
+    {
+      "epoch": 0.006909844710332036,
+      "grad_norm": 0.0,
+      "learning_rate": 4.606060606060606e-06,
+      "loss": 1.3547,
+      "step": 190
+    },
+    {
+      "epoch": 0.006946212314070626,
+      "grad_norm": 0.0,
+      "learning_rate": 4.630303030303031e-06,
+      "loss": 1.0996,
+      "step": 191
+    },
+    {
+      "epoch": 0.006982579917809216,
+      "grad_norm": 0.0,
+      "learning_rate": 4.654545454545455e-06,
+      "loss": 1.3695,
+      "step": 192
+    },
+    {
+      "epoch": 0.007018947521547805,
+      "grad_norm": 0.0,
+      "learning_rate": 4.678787878787879e-06,
+      "loss": 1.4422,
+      "step": 193
+    },
+    {
+      "epoch": 0.0070553151252863945,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7030303030303035e-06,
+      "loss": 1.2711,
+      "step": 194
+    },
+    {
+      "epoch": 0.007091682729024985,
+      "grad_norm": 0.0,
+      "learning_rate": 4.727272727272728e-06,
+      "loss": 1.274,
+      "step": 195
+    },
+    {
+      "epoch": 0.007128050332763574,
+      "grad_norm": 0.0,
+      "learning_rate": 4.751515151515152e-06,
+      "loss": 1.3708,
+      "step": 196
+    },
+    {
+      "epoch": 0.007164417936502164,
+      "grad_norm": 0.0,
+      "learning_rate": 4.775757575757576e-06,
+      "loss": 1.3057,
+      "step": 197
+    },
+    {
+      "epoch": 0.0072007855402407535,
+      "grad_norm": 0.0,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.2299,
+      "step": 198
+    },
+    {
+      "epoch": 0.007237153143979343,
+      "grad_norm": 0.0,
+      "learning_rate": 4.824242424242424e-06,
+      "loss": 1.2327,
+      "step": 199
+    },
+    {
+      "epoch": 0.007273520747717933,
+      "grad_norm": 0.0,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 1.2998,
+      "step": 200
+    },
+    {
+      "epoch": 0.007309888351456522,
+      "grad_norm": 0.0,
+      "learning_rate": 4.872727272727273e-06,
+      "loss": 1.3274,
+      "step": 201
+    },
+    {
+      "epoch": 0.0073462559551951126,
+      "grad_norm": 0.0,
+      "learning_rate": 4.896969696969697e-06,
+      "loss": 1.2706,
+      "step": 202
+    },
+    {
+      "epoch": 0.007382623558933702,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9212121212121214e-06,
+      "loss": 1.2106,
+      "step": 203
+    },
+    {
+      "epoch": 0.007418991162672291,
+      "grad_norm": 0.0,
+      "learning_rate": 4.945454545454546e-06,
+      "loss": 1.2982,
+      "step": 204
+    },
+    {
+      "epoch": 0.007455358766410881,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9696969696969696e-06,
+      "loss": 1.319,
+      "step": 205
+    },
+    {
+      "epoch": 0.007491726370149471,
+      "grad_norm": 0.0,
+      "learning_rate": 4.993939393939394e-06,
+      "loss": 1.2953,
+      "step": 206
+    },
+    {
+      "epoch": 0.007528093973888061,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0181818181818186e-06,
+      "loss": 1.4233,
+      "step": 207
+    },
+    {
+      "epoch": 0.00756446157762665,
+      "grad_norm": 0.0,
+      "learning_rate": 5.042424242424243e-06,
+      "loss": 1.1942,
+      "step": 208
+    },
+    {
+      "epoch": 0.00760082918136524,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 1.2477,
+      "step": 209
+    },
+    {
+      "epoch": 0.00763719678510383,
+      "grad_norm": 0.0,
+      "learning_rate": 5.090909090909091e-06,
+      "loss": 1.2482,
+      "step": 210
+    },
+    {
+      "epoch": 0.007673564388842419,
+      "grad_norm": 0.0,
+      "learning_rate": 5.115151515151515e-06,
+      "loss": 1.2304,
+      "step": 211
+    },
+    {
+      "epoch": 0.007709931992581008,
+      "grad_norm": 0.0,
+      "learning_rate": 5.139393939393939e-06,
+      "loss": 1.2847,
+      "step": 212
+    },
+    {
+      "epoch": 0.007746299596319599,
+      "grad_norm": 0.0,
+      "learning_rate": 5.163636363636364e-06,
+      "loss": 1.3267,
+      "step": 213
+    },
+    {
+      "epoch": 0.007782667200058188,
+      "grad_norm": 0.0,
+      "learning_rate": 5.187878787878788e-06,
+      "loss": 1.1304,
+      "step": 214
+    },
+    {
+      "epoch": 0.007819034803796778,
+      "grad_norm": 0.0,
+      "learning_rate": 5.212121212121213e-06,
+      "loss": 1.301,
+      "step": 215
+    },
+    {
+      "epoch": 0.007855402407535367,
+      "grad_norm": 0.0,
+      "learning_rate": 5.236363636363637e-06,
+      "loss": 1.3692,
+      "step": 216
+    },
+    {
+      "epoch": 0.007891770011273957,
+      "grad_norm": 0.0,
+      "learning_rate": 5.26060606060606e-06,
+      "loss": 1.3538,
+      "step": 217
+    },
+    {
+      "epoch": 0.007928137615012546,
+      "grad_norm": 0.0,
+      "learning_rate": 5.284848484848485e-06,
+      "loss": 1.3509,
+      "step": 218
+    },
+    {
+      "epoch": 0.007964505218751137,
+      "grad_norm": 0.0,
+      "learning_rate": 5.309090909090909e-06,
+      "loss": 1.351,
+      "step": 219
+    },
+    {
+      "epoch": 0.008000872822489727,
+      "grad_norm": 0.0,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.3427,
+      "step": 220
+    },
+    {
+      "epoch": 0.008037240426228316,
+      "grad_norm": 0.0,
+      "learning_rate": 5.357575757575758e-06,
+      "loss": 1.2725,
+      "step": 221
+    },
+    {
+      "epoch": 0.008073608029966905,
+      "grad_norm": 0.0,
+      "learning_rate": 5.381818181818183e-06,
+      "loss": 1.1678,
+      "step": 222
+    },
+    {
+      "epoch": 0.008109975633705494,
+      "grad_norm": 0.0,
+      "learning_rate": 5.406060606060607e-06,
+      "loss": 1.3605,
+      "step": 223
+    },
+    {
+      "epoch": 0.008146343237444086,
+      "grad_norm": 0.0,
+      "learning_rate": 5.430303030303032e-06,
+      "loss": 1.2948,
+      "step": 224
+    },
+    {
+      "epoch": 0.008182710841182675,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 1.378,
+      "step": 225
+    },
+    {
+      "epoch": 0.008219078444921264,
+      "grad_norm": 0.0,
+      "learning_rate": 5.478787878787879e-06,
+      "loss": 1.3796,
+      "step": 226
+    },
+    {
+      "epoch": 0.008255446048659854,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5030303030303034e-06,
+      "loss": 1.4288,
+      "step": 227
+    },
+    {
+      "epoch": 0.008291813652398443,
+      "grad_norm": 0.0,
+      "learning_rate": 5.527272727272728e-06,
+      "loss": 1.2742,
+      "step": 228
+    },
+    {
+      "epoch": 0.008328181256137034,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5515151515151524e-06,
+      "loss": 1.2874,
+      "step": 229
+    },
+    {
+      "epoch": 0.008364548859875623,
+      "grad_norm": 0.0,
+      "learning_rate": 5.575757575757577e-06,
+      "loss": 1.1699,
+      "step": 230
+    },
+    {
+      "epoch": 0.008400916463614213,
+      "grad_norm": 0.0,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.3976,
+      "step": 231
+    },
+    {
+      "epoch": 0.008437284067352802,
+      "grad_norm": 0.0,
+      "learning_rate": 5.624242424242424e-06,
+      "loss": 1.3611,
+      "step": 232
+    },
+    {
+      "epoch": 0.008473651671091391,
+      "grad_norm": 0.0,
+      "learning_rate": 5.648484848484849e-06,
+      "loss": 1.2936,
+      "step": 233
+    },
+    {
+      "epoch": 0.008510019274829982,
+      "grad_norm": 0.0,
+      "learning_rate": 5.672727272727273e-06,
+      "loss": 1.3268,
+      "step": 234
+    },
+    {
+      "epoch": 0.008546386878568572,
+      "grad_norm": 0.0,
+      "learning_rate": 5.696969696969698e-06,
+      "loss": 1.373,
+      "step": 235
+    },
+    {
+      "epoch": 0.008582754482307161,
+      "grad_norm": 0.0,
+      "learning_rate": 5.721212121212122e-06,
+      "loss": 1.293,
+      "step": 236
+    },
+    {
+      "epoch": 0.00861912208604575,
+      "grad_norm": 0.0,
+      "learning_rate": 5.745454545454546e-06,
+      "loss": 1.3135,
+      "step": 237
+    },
+    {
+      "epoch": 0.00865548968978434,
+      "grad_norm": 0.0,
+      "learning_rate": 5.76969696969697e-06,
+      "loss": 1.2324,
+      "step": 238
+    },
+    {
+      "epoch": 0.008691857293522929,
+      "grad_norm": 0.0,
+      "learning_rate": 5.793939393939394e-06,
+      "loss": 1.2873,
+      "step": 239
+    },
+    {
+      "epoch": 0.00872822489726152,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8181818181818185e-06,
+      "loss": 1.3966,
+      "step": 240
+    },
+    {
+      "epoch": 0.00876459250100011,
+      "grad_norm": 0.0,
+      "learning_rate": 5.842424242424243e-06,
+      "loss": 1.2852,
+      "step": 241
+    },
+    {
+      "epoch": 0.008800960104738699,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 1.3105,
+      "step": 242
+    },
+    {
+      "epoch": 0.008837327708477288,
+      "grad_norm": 0.0,
+      "learning_rate": 5.890909090909091e-06,
+      "loss": 1.1254,
+      "step": 243
+    },
+    {
+      "epoch": 0.008873695312215877,
+      "grad_norm": 0.0,
+      "learning_rate": 5.915151515151516e-06,
+      "loss": 1.3091,
+      "step": 244
+    },
+    {
+      "epoch": 0.008910062915954468,
+      "grad_norm": 0.0,
+      "learning_rate": 5.93939393939394e-06,
+      "loss": 1.3094,
+      "step": 245
+    },
+    {
+      "epoch": 0.008946430519693058,
+      "grad_norm": 0.0,
+      "learning_rate": 5.963636363636364e-06,
+      "loss": 1.2481,
+      "step": 246
+    },
+    {
+      "epoch": 0.008982798123431647,
+      "grad_norm": 0.0,
+      "learning_rate": 5.987878787878788e-06,
+      "loss": 1.3497,
+      "step": 247
+    },
+    {
+      "epoch": 0.009019165727170236,
+      "grad_norm": 0.0,
+      "learning_rate": 6.012121212121213e-06,
+      "loss": 1.174,
+      "step": 248
+    },
+    {
+      "epoch": 0.009055533330908826,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0363636363636365e-06,
+      "loss": 1.3183,
+      "step": 249
+    },
+    {
+      "epoch": 0.009091900934647417,
+      "grad_norm": 0.0,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 1.3206,
+      "step": 250
+    },
+    {
+      "epoch": 0.009128268538386006,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0848484848484855e-06,
+      "loss": 1.3763,
+      "step": 251
+    },
+    {
+      "epoch": 0.009164636142124595,
+      "grad_norm": 0.0,
+      "learning_rate": 6.10909090909091e-06,
+      "loss": 1.3643,
+      "step": 252
+    },
+    {
+      "epoch": 0.009201003745863185,
+      "grad_norm": 0.0,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 1.3313,
+      "step": 253
+    },
+    {
+      "epoch": 0.009237371349601774,
+      "grad_norm": 0.0,
+      "learning_rate": 6.157575757575758e-06,
+      "loss": 1.25,
+      "step": 254
+    },
+    {
+      "epoch": 0.009273738953340365,
+      "grad_norm": 0.0,
+      "learning_rate": 6.181818181818182e-06,
+      "loss": 1.3075,
+      "step": 255
+    },
+    {
+      "epoch": 0.009310106557078954,
+      "grad_norm": 0.0,
+      "learning_rate": 6.206060606060606e-06,
+      "loss": 1.3536,
+      "step": 256
+    },
+    {
+      "epoch": 0.009346474160817544,
+      "grad_norm": 0.0,
+      "learning_rate": 6.230303030303031e-06,
+      "loss": 1.3041,
+      "step": 257
+    },
+    {
+      "epoch": 0.009382841764556133,
+      "grad_norm": 0.0,
+      "learning_rate": 6.254545454545455e-06,
+      "loss": 1.3945,
+      "step": 258
+    },
+    {
+      "epoch": 0.009419209368294722,
+      "grad_norm": 0.0,
+      "learning_rate": 6.27878787878788e-06,
+      "loss": 1.3204,
+      "step": 259
+    },
+    {
+      "epoch": 0.009455576972033313,
+      "grad_norm": 0.0,
+      "learning_rate": 6.303030303030303e-06,
+      "loss": 1.3526,
+      "step": 260
+    },
+    {
+      "epoch": 0.009491944575771903,
+      "grad_norm": 0.0,
+      "learning_rate": 6.327272727272727e-06,
+      "loss": 1.3036,
+      "step": 261
+    },
+    {
+      "epoch": 0.009528312179510492,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3515151515151516e-06,
+      "loss": 1.2926,
+      "step": 262
+    },
+    {
+      "epoch": 0.009564679783249081,
+      "grad_norm": 0.0,
+      "learning_rate": 6.375757575757576e-06,
+      "loss": 1.2267,
+      "step": 263
+    },
+    {
+      "epoch": 0.00960104738698767,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.3182,
+      "step": 264
+    },
+    {
+      "epoch": 0.009637414990726262,
+      "grad_norm": 0.0,
+      "learning_rate": 6.424242424242425e-06,
+      "loss": 1.2764,
+      "step": 265
+    },
+    {
+      "epoch": 0.009673782594464851,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4484848484848496e-06,
+      "loss": 1.2298,
+      "step": 266
+    },
+    {
+      "epoch": 0.00971015019820344,
+      "grad_norm": 0.0,
+      "learning_rate": 6.472727272727272e-06,
+      "loss": 1.3278,
+      "step": 267
+    },
+    {
+      "epoch": 0.00974651780194203,
+      "grad_norm": 0.0,
+      "learning_rate": 6.496969696969697e-06,
+      "loss": 1.3215,
+      "step": 268
+    },
+    {
+      "epoch": 0.00978288540568062,
+      "grad_norm": 0.0,
+      "learning_rate": 6.521212121212121e-06,
+      "loss": 1.1003,
+      "step": 269
+    },
+    {
+      "epoch": 0.00981925300941921,
+      "grad_norm": 0.0,
+      "learning_rate": 6.545454545454546e-06,
+      "loss": 1.2266,
+      "step": 270
+    },
+    {
+      "epoch": 0.0098556206131578,
+      "grad_norm": 0.0,
+      "learning_rate": 6.56969696969697e-06,
+      "loss": 1.3698,
+      "step": 271
+    },
+    {
+      "epoch": 0.009891988216896389,
+      "grad_norm": 0.0,
+      "learning_rate": 6.593939393939395e-06,
+      "loss": 1.2462,
+      "step": 272
+    },
+    {
+      "epoch": 0.009928355820634978,
+      "grad_norm": 0.0,
+      "learning_rate": 6.618181818181819e-06,
+      "loss": 1.2917,
+      "step": 273
+    },
+    {
+      "epoch": 0.009964723424373568,
+      "grad_norm": 0.0,
+      "learning_rate": 6.642424242424242e-06,
+      "loss": 1.3555,
+      "step": 274
+    },
+    {
+      "epoch": 0.010001091028112157,
+      "grad_norm": 0.0,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.3026,
+      "step": 275
+    },
+    {
+      "epoch": 0.010037458631850748,
+      "grad_norm": 0.0,
+      "learning_rate": 6.690909090909091e-06,
+      "loss": 1.2011,
+      "step": 276
+    },
+    {
+      "epoch": 0.010073826235589337,
+      "grad_norm": 0.0,
+      "learning_rate": 6.715151515151516e-06,
+      "loss": 1.2745,
+      "step": 277
+    },
+    {
+      "epoch": 0.010110193839327927,
+      "grad_norm": 0.0,
+      "learning_rate": 6.73939393939394e-06,
+      "loss": 1.4165,
+      "step": 278
+    },
+    {
+      "epoch": 0.010146561443066516,
+      "grad_norm": 0.0,
+      "learning_rate": 6.763636363636365e-06,
+      "loss": 1.3008,
+      "step": 279
+    },
+    {
+      "epoch": 0.010182929046805105,
+      "grad_norm": 0.0,
+      "learning_rate": 6.787878787878789e-06,
+      "loss": 1.221,
+      "step": 280
+    },
+    {
+      "epoch": 0.010219296650543696,
+      "grad_norm": 0.0,
+      "learning_rate": 6.812121212121212e-06,
+      "loss": 1.2923,
+      "step": 281
+    },
+    {
+      "epoch": 0.010255664254282286,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8363636363636364e-06,
+      "loss": 1.2681,
+      "step": 282
+    },
+    {
+      "epoch": 0.010292031858020875,
+      "grad_norm": 0.0,
+      "learning_rate": 6.860606060606061e-06,
+      "loss": 1.181,
+      "step": 283
+    },
+    {
+      "epoch": 0.010328399461759464,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8848484848484854e-06,
+      "loss": 1.2939,
+      "step": 284
+    },
+    {
+      "epoch": 0.010364767065498054,
+      "grad_norm": 0.0,
+      "learning_rate": 6.90909090909091e-06,
+      "loss": 1.3181,
+      "step": 285
+    },
+    {
+      "epoch": 0.010401134669236645,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 1.2789,
+      "step": 286
+    },
+    {
+      "epoch": 0.010437502272975234,
+      "grad_norm": 0.0,
+      "learning_rate": 6.957575757575759e-06,
+      "loss": 1.2008,
+      "step": 287
+    },
+    {
+      "epoch": 0.010473869876713823,
+      "grad_norm": 0.0,
+      "learning_rate": 6.981818181818183e-06,
+      "loss": 1.2657,
+      "step": 288
+    },
+    {
+      "epoch": 0.010510237480452413,
+      "grad_norm": 0.0,
+      "learning_rate": 7.006060606060606e-06,
+      "loss": 1.2665,
+      "step": 289
+    },
+    {
+      "epoch": 0.010546605084191002,
+      "grad_norm": 0.0,
+      "learning_rate": 7.030303030303031e-06,
+      "loss": 1.3638,
+      "step": 290
+    },
+    {
+      "epoch": 0.010582972687929593,
+      "grad_norm": 0.0,
+      "learning_rate": 7.054545454545455e-06,
+      "loss": 1.382,
+      "step": 291
+    },
+    {
+      "epoch": 0.010619340291668182,
+      "grad_norm": 0.0,
+      "learning_rate": 7.07878787878788e-06,
+      "loss": 1.3742,
+      "step": 292
+    },
+    {
+      "epoch": 0.010655707895406772,
+      "grad_norm": 0.0,
+      "learning_rate": 7.103030303030304e-06,
+      "loss": 1.1795,
+      "step": 293
+    },
+    {
+      "epoch": 0.010692075499145361,
+      "grad_norm": 0.0,
+      "learning_rate": 7.127272727272728e-06,
+      "loss": 1.2091,
+      "step": 294
+    },
+    {
+      "epoch": 0.01072844310288395,
+      "grad_norm": 0.0,
+      "learning_rate": 7.151515151515152e-06,
+      "loss": 1.188,
+      "step": 295
+    },
+    {
+      "epoch": 0.010764810706622541,
+      "grad_norm": 0.0,
+      "learning_rate": 7.175757575757576e-06,
+      "loss": 1.2866,
+      "step": 296
+    },
+    {
+      "epoch": 0.01080117831036113,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.2762,
+      "step": 297
+    },
+    {
+      "epoch": 0.01083754591409972,
+      "grad_norm": 0.0,
+      "learning_rate": 7.224242424242425e-06,
+      "loss": 1.3645,
+      "step": 298
+    },
+    {
+      "epoch": 0.01087391351783831,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2484848484848495e-06,
+      "loss": 1.2509,
+      "step": 299
+    },
+    {
+      "epoch": 0.010910281121576899,
+      "grad_norm": 0.0,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.2419,
+      "step": 300
+    },
+    {
+      "epoch": 0.01094664872531549,
+      "grad_norm": 0.0,
+      "learning_rate": 7.296969696969698e-06,
+      "loss": 1.2255,
+      "step": 301
+    },
+    {
+      "epoch": 0.010983016329054079,
+      "grad_norm": 0.0,
+      "learning_rate": 7.321212121212122e-06,
+      "loss": 1.2828,
+      "step": 302
+    },
+    {
+      "epoch": 0.011019383932792668,
+      "grad_norm": 0.0,
+      "learning_rate": 7.345454545454546e-06,
+      "loss": 1.2332,
+      "step": 303
+    },
+    {
+      "epoch": 0.011055751536531258,
+      "grad_norm": 0.0,
+      "learning_rate": 7.36969696969697e-06,
+      "loss": 1.4247,
+      "step": 304
+    },
+    {
+      "epoch": 0.011092119140269847,
+      "grad_norm": 0.0,
+      "learning_rate": 7.393939393939395e-06,
+      "loss": 1.2299,
+      "step": 305
+    },
+    {
+      "epoch": 0.011128486744008438,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4181818181818185e-06,
+      "loss": 1.2297,
+      "step": 306
+    },
+    {
+      "epoch": 0.011164854347747027,
+      "grad_norm": 0.0,
+      "learning_rate": 7.442424242424243e-06,
+      "loss": 1.2023,
+      "step": 307
+    },
+    {
+      "epoch": 0.011201221951485617,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 1.2966,
+      "step": 308
+    },
+    {
+      "epoch": 0.011237589555224206,
+      "grad_norm": 0.0,
+      "learning_rate": 7.490909090909092e-06,
+      "loss": 1.3006,
+      "step": 309
+    },
+    {
+      "epoch": 0.011273957158962795,
+      "grad_norm": 0.0,
+      "learning_rate": 7.515151515151516e-06,
+      "loss": 1.3301,
+      "step": 310
+    },
+    {
+      "epoch": 0.011310324762701385,
+      "grad_norm": 0.0,
+      "learning_rate": 7.53939393939394e-06,
+      "loss": 1.293,
+      "step": 311
+    },
+    {
+      "epoch": 0.011346692366439976,
+      "grad_norm": 0.0,
+      "learning_rate": 7.563636363636364e-06,
+      "loss": 1.3842,
+      "step": 312
+    },
+    {
+      "epoch": 0.011383059970178565,
+      "grad_norm": 0.0,
+      "learning_rate": 7.587878787878788e-06,
+      "loss": 1.21,
+      "step": 313
+    },
+    {
+      "epoch": 0.011419427573917154,
+      "grad_norm": 0.0,
+      "learning_rate": 7.612121212121213e-06,
+      "loss": 1.2355,
+      "step": 314
+    },
+    {
+      "epoch": 0.011455795177655744,
+      "grad_norm": 0.0,
+      "learning_rate": 7.636363636363638e-06,
+      "loss": 1.2337,
+      "step": 315
+    },
+    {
+      "epoch": 0.011492162781394333,
+      "grad_norm": 0.0,
+      "learning_rate": 7.660606060606062e-06,
+      "loss": 1.2568,
+      "step": 316
+    },
+    {
+      "epoch": 0.011528530385132924,
+      "grad_norm": 0.0,
+      "learning_rate": 7.684848484848485e-06,
+      "loss": 1.395,
+      "step": 317
+    },
+    {
+      "epoch": 0.011564897988871514,
+      "grad_norm": 0.0,
+      "learning_rate": 7.709090909090909e-06,
+      "loss": 1.1822,
+      "step": 318
+    },
+    {
+      "epoch": 0.011601265592610103,
+      "grad_norm": 0.0,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 1.179,
+      "step": 319
+    },
+    {
+      "epoch": 0.011637633196348692,
+      "grad_norm": 0.0,
+      "learning_rate": 7.757575757575758e-06,
+      "loss": 1.264,
+      "step": 320
+    },
+    {
+      "epoch": 0.011674000800087281,
+      "grad_norm": 0.0,
+      "learning_rate": 7.781818181818183e-06,
+      "loss": 1.3413,
+      "step": 321
+    },
+    {
+      "epoch": 0.011710368403825873,
+      "grad_norm": 0.0,
+      "learning_rate": 7.806060606060607e-06,
+      "loss": 1.159,
+      "step": 322
+    },
+    {
+      "epoch": 0.011746736007564462,
+      "grad_norm": 0.0,
+      "learning_rate": 7.83030303030303e-06,
+      "loss": 1.3473,
+      "step": 323
+    },
+    {
+      "epoch": 0.011783103611303051,
+      "grad_norm": 0.0,
+      "learning_rate": 7.854545454545454e-06,
+      "loss": 1.2646,
+      "step": 324
+    },
+    {
+      "epoch": 0.01181947121504164,
+      "grad_norm": 0.0,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 1.3365,
+      "step": 325
+    },
+    {
+      "epoch": 0.01185583881878023,
+      "grad_norm": 0.0,
+      "learning_rate": 7.903030303030303e-06,
+      "loss": 1.3524,
+      "step": 326
+    },
+    {
+      "epoch": 0.011892206422518821,
+      "grad_norm": 0.0,
+      "learning_rate": 7.927272727272729e-06,
+      "loss": 1.1953,
+      "step": 327
+    },
+    {
+      "epoch": 0.01192857402625741,
+      "grad_norm": 0.0,
+      "learning_rate": 7.951515151515152e-06,
+      "loss": 1.282,
+      "step": 328
+    },
+    {
+      "epoch": 0.011964941629996,
+      "grad_norm": 0.0,
+      "learning_rate": 7.975757575757576e-06,
+      "loss": 1.3365,
+      "step": 329
+    },
+    {
+      "epoch": 0.012001309233734589,
+      "grad_norm": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.4064,
+      "step": 330
+    },
+    {
+      "epoch": 0.012037676837473178,
+      "grad_norm": 0.0,
+      "learning_rate": 8.024242424242425e-06,
+      "loss": 1.2696,
+      "step": 331
+    },
+    {
+      "epoch": 0.01207404444121177,
+      "grad_norm": 0.0,
+      "learning_rate": 8.048484848484849e-06,
+      "loss": 1.2266,
+      "step": 332
+    },
+    {
+      "epoch": 0.012110412044950359,
+      "grad_norm": 0.0,
+      "learning_rate": 8.072727272727274e-06,
+      "loss": 1.3081,
+      "step": 333
+    },
+    {
+      "epoch": 0.012146779648688948,
+      "grad_norm": 0.0,
+      "learning_rate": 8.096969696969698e-06,
+      "loss": 1.2759,
+      "step": 334
+    },
+    {
+      "epoch": 0.012183147252427537,
+      "grad_norm": 0.0,
+      "learning_rate": 8.121212121212121e-06,
+      "loss": 1.2281,
+      "step": 335
+    },
+    {
+      "epoch": 0.012219514856166127,
+      "grad_norm": 0.0,
+      "learning_rate": 8.145454545454547e-06,
+      "loss": 1.3265,
+      "step": 336
+    },
+    {
+      "epoch": 0.012255882459904718,
+      "grad_norm": 0.0,
+      "learning_rate": 8.16969696969697e-06,
+      "loss": 1.3663,
+      "step": 337
+    },
+    {
+      "epoch": 0.012292250063643307,
+      "grad_norm": 0.0,
+      "learning_rate": 8.193939393939394e-06,
+      "loss": 1.2424,
+      "step": 338
+    },
+    {
+      "epoch": 0.012328617667381896,
+      "grad_norm": 0.0,
+      "learning_rate": 8.21818181818182e-06,
+      "loss": 1.1833,
+      "step": 339
+    },
+    {
+      "epoch": 0.012364985271120486,
+      "grad_norm": 0.0,
+      "learning_rate": 8.242424242424243e-06,
+      "loss": 1.2877,
+      "step": 340
+    },
+    {
+      "epoch": 0.012401352874859075,
+      "grad_norm": 0.0,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 1.2456,
+      "step": 341
+    },
+    {
+      "epoch": 0.012437720478597666,
+      "grad_norm": 0.0,
+      "learning_rate": 8.290909090909092e-06,
+      "loss": 1.2897,
+      "step": 342
+    },
+    {
+      "epoch": 0.012474088082336255,
+      "grad_norm": 0.0,
+      "learning_rate": 8.315151515151516e-06,
+      "loss": 1.3253,
+      "step": 343
+    },
+    {
+      "epoch": 0.012510455686074845,
+      "grad_norm": 0.0,
+      "learning_rate": 8.339393939393941e-06,
+      "loss": 1.3652,
+      "step": 344
+    },
+    {
+      "epoch": 0.012546823289813434,
+      "grad_norm": 0.0,
+      "learning_rate": 8.363636363636365e-06,
+      "loss": 1.2935,
+      "step": 345
+    },
+    {
+      "epoch": 0.012583190893552023,
+      "grad_norm": 0.0,
+      "learning_rate": 8.387878787878788e-06,
+      "loss": 1.2552,
+      "step": 346
+    },
+    {
+      "epoch": 0.012619558497290613,
+      "grad_norm": 0.0,
+      "learning_rate": 8.412121212121212e-06,
+      "loss": 1.2818,
+      "step": 347
+    },
+    {
+      "epoch": 0.012655926101029204,
+      "grad_norm": 0.0,
+      "learning_rate": 8.436363636363637e-06,
+      "loss": 1.2886,
+      "step": 348
+    },
+    {
+      "epoch": 0.012692293704767793,
+      "grad_norm": 0.0,
+      "learning_rate": 8.460606060606061e-06,
+      "loss": 1.3738,
+      "step": 349
+    },
+    {
+      "epoch": 0.012728661308506382,
+      "grad_norm": 0.0,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 1.2515,
+      "step": 350
+    },
+    {
+      "epoch": 0.012765028912244972,
+      "grad_norm": 0.0,
+      "learning_rate": 8.50909090909091e-06,
+      "loss": 1.2025,
+      "step": 351
+    },
+    {
+      "epoch": 0.012801396515983561,
+      "grad_norm": 0.0,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 1.1987,
+      "step": 352
+    },
+    {
+      "epoch": 0.012837764119722152,
+      "grad_norm": 0.0,
+      "learning_rate": 8.557575757575757e-06,
+      "loss": 1.2668,
+      "step": 353
+    },
+    {
+      "epoch": 0.012874131723460741,
+      "grad_norm": 0.0,
+      "learning_rate": 8.581818181818183e-06,
+      "loss": 1.3541,
+      "step": 354
+    },
+    {
+      "epoch": 0.01291049932719933,
+      "grad_norm": 0.0,
+      "learning_rate": 8.606060606060606e-06,
+      "loss": 1.2358,
+      "step": 355
+    },
+    {
+      "epoch": 0.01294686693093792,
+      "grad_norm": 0.0,
+      "learning_rate": 8.630303030303032e-06,
+      "loss": 1.2645,
+      "step": 356
+    },
+    {
+      "epoch": 0.01298323453467651,
+      "grad_norm": 0.0,
+      "learning_rate": 8.654545454545455e-06,
+      "loss": 1.2999,
+      "step": 357
+    },
+    {
+      "epoch": 0.0130196021384151,
+      "grad_norm": 0.0,
+      "learning_rate": 8.67878787878788e-06,
+      "loss": 1.3091,
+      "step": 358
+    },
+    {
+      "epoch": 0.01305596974215369,
+      "grad_norm": 0.0,
+      "learning_rate": 8.703030303030304e-06,
+      "loss": 1.2921,
+      "step": 359
+    },
+    {
+      "epoch": 0.013092337345892279,
+      "grad_norm": 0.0,
+      "learning_rate": 8.727272727272728e-06,
+      "loss": 1.3084,
+      "step": 360
+    },
+    {
+      "epoch": 0.013128704949630868,
+      "grad_norm": 0.0,
+      "learning_rate": 8.751515151515151e-06,
+      "loss": 1.2242,
+      "step": 361
+    },
+    {
+      "epoch": 0.013165072553369458,
+      "grad_norm": 0.0,
+      "learning_rate": 8.775757575757577e-06,
+      "loss": 1.3338,
+      "step": 362
+    },
+    {
+      "epoch": 0.013201440157108049,
+      "grad_norm": 0.0,
+      "learning_rate": 8.8e-06,
+      "loss": 1.2266,
+      "step": 363
+    },
+    {
+      "epoch": 0.013237807760846638,
+      "grad_norm": 0.0,
+      "learning_rate": 8.824242424242426e-06,
+      "loss": 1.3994,
+      "step": 364
+    },
+    {
+      "epoch": 0.013274175364585227,
+      "grad_norm": 0.0,
+      "learning_rate": 8.84848484848485e-06,
+      "loss": 1.3351,
+      "step": 365
+    },
+    {
+      "epoch": 0.013310542968323817,
+      "grad_norm": 0.0,
+      "learning_rate": 8.872727272727275e-06,
+      "loss": 1.2402,
+      "step": 366
+    },
+    {
+      "epoch": 0.013346910572062406,
+      "grad_norm": 0.0,
+      "learning_rate": 8.896969696969697e-06,
+      "loss": 1.3299,
+      "step": 367
+    },
+    {
+      "epoch": 0.013383278175800997,
+      "grad_norm": 0.0,
+      "learning_rate": 8.921212121212122e-06,
+      "loss": 1.2988,
+      "step": 368
+    },
+    {
+      "epoch": 0.013419645779539587,
+      "grad_norm": 0.0,
+      "learning_rate": 8.945454545454546e-06,
+      "loss": 1.2095,
+      "step": 369
+    },
+    {
+      "epoch": 0.013456013383278176,
+      "grad_norm": 0.0,
+      "learning_rate": 8.969696969696971e-06,
+      "loss": 1.3851,
+      "step": 370
+    },
+    {
+      "epoch": 0.013492380987016765,
+      "grad_norm": 0.0,
+      "learning_rate": 8.993939393939395e-06,
+      "loss": 1.281,
+      "step": 371
+    },
+    {
+      "epoch": 0.013528748590755355,
+      "grad_norm": 0.0,
+      "learning_rate": 9.01818181818182e-06,
+      "loss": 1.2776,
+      "step": 372
+    },
+    {
+      "epoch": 0.013565116194493946,
+      "grad_norm": 0.0,
+      "learning_rate": 9.042424242424244e-06,
+      "loss": 1.1395,
+      "step": 373
+    },
+    {
+      "epoch": 0.013601483798232535,
+      "grad_norm": 0.0,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 1.3719,
+      "step": 374
+    },
+    {
+      "epoch": 0.013637851401971124,
+      "grad_norm": 0.0,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 1.3182,
+      "step": 375
+    },
+    {
+      "epoch": 0.013674219005709714,
+      "grad_norm": 0.0,
+      "learning_rate": 9.115151515151516e-06,
+      "loss": 1.2819,
+      "step": 376
+    },
+    {
+      "epoch": 0.013710586609448303,
+      "grad_norm": 0.0,
+      "learning_rate": 9.13939393939394e-06,
+      "loss": 1.2092,
+      "step": 377
+    },
+    {
+      "epoch": 0.013746954213186894,
+      "grad_norm": 0.0,
+      "learning_rate": 9.163636363636365e-06,
+      "loss": 1.3202,
+      "step": 378
+    },
+    {
+      "epoch": 0.013783321816925483,
+      "grad_norm": 0.0,
+      "learning_rate": 9.187878787878789e-06,
+      "loss": 1.3485,
+      "step": 379
+    },
+    {
+      "epoch": 0.013819689420664073,
+      "grad_norm": 0.0,
+      "learning_rate": 9.212121212121213e-06,
+      "loss": 1.2341,
+      "step": 380
+    },
+    {
+      "epoch": 0.013856057024402662,
+      "grad_norm": 0.0,
+      "learning_rate": 9.236363636363636e-06,
+      "loss": 1.2904,
+      "step": 381
+    },
+    {
+      "epoch": 0.013892424628141251,
+      "grad_norm": 0.0,
+      "learning_rate": 9.260606060606062e-06,
+      "loss": 1.2327,
+      "step": 382
+    },
+    {
+      "epoch": 0.01392879223187984,
+      "grad_norm": 0.0,
+      "learning_rate": 9.284848484848485e-06,
+      "loss": 1.2468,
+      "step": 383
+    },
+    {
+      "epoch": 0.013965159835618432,
+      "grad_norm": 0.0,
+      "learning_rate": 9.30909090909091e-06,
+      "loss": 1.2682,
+      "step": 384
+    },
+    {
+      "epoch": 0.014001527439357021,
+      "grad_norm": 0.0,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.2292,
+      "step": 385
+    },
+    {
+      "epoch": 0.01403789504309561,
+      "grad_norm": 0.0,
+      "learning_rate": 9.357575757575758e-06,
+      "loss": 1.2025,
+      "step": 386
+    },
+    {
+      "epoch": 0.0140742626468342,
+      "grad_norm": 0.0,
+      "learning_rate": 9.381818181818183e-06,
+      "loss": 1.251,
+      "step": 387
+    },
+    {
+      "epoch": 0.014110630250572789,
+      "grad_norm": 0.0,
+      "learning_rate": 9.406060606060607e-06,
+      "loss": 1.285,
+      "step": 388
+    },
+    {
+      "epoch": 0.01414699785431138,
+      "grad_norm": 0.0,
+      "learning_rate": 9.43030303030303e-06,
+      "loss": 1.307,
+      "step": 389
+    },
+    {
+      "epoch": 0.01418336545804997,
+      "grad_norm": 0.0,
+      "learning_rate": 9.454545454545456e-06,
+      "loss": 1.332,
+      "step": 390
+    },
+    {
+      "epoch": 0.014219733061788559,
+      "grad_norm": 0.0,
+      "learning_rate": 9.47878787878788e-06,
+      "loss": 1.1412,
+      "step": 391
+    },
+    {
+      "epoch": 0.014256100665527148,
+      "grad_norm": 0.0,
+      "learning_rate": 9.503030303030303e-06,
+      "loss": 1.3536,
+      "step": 392
+    },
+    {
+      "epoch": 0.014292468269265737,
+      "grad_norm": 0.0,
+      "learning_rate": 9.527272727272729e-06,
+      "loss": 1.2491,
+      "step": 393
+    },
+    {
+      "epoch": 0.014328835873004328,
+      "grad_norm": 0.0,
+      "learning_rate": 9.551515151515152e-06,
+      "loss": 1.3567,
+      "step": 394
+    },
+    {
+      "epoch": 0.014365203476742918,
+      "grad_norm": 0.0,
+      "learning_rate": 9.575757575757576e-06,
+      "loss": 1.183,
+      "step": 395
+    },
+    {
+      "epoch": 0.014401571080481507,
+      "grad_norm": 0.0,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.2445,
+      "step": 396
+    },
+    {
+      "epoch": 0.014437938684220096,
+      "grad_norm": 0.0,
+      "learning_rate": 9.624242424242425e-06,
+      "loss": 1.3254,
+      "step": 397
+    },
+    {
+      "epoch": 0.014474306287958686,
+      "grad_norm": 0.0,
+      "learning_rate": 9.648484848484849e-06,
+      "loss": 1.205,
+      "step": 398
+    },
+    {
+      "epoch": 0.014510673891697277,
+      "grad_norm": 0.0,
+      "learning_rate": 9.672727272727274e-06,
+      "loss": 1.3961,
+      "step": 399
+    },
+    {
+      "epoch": 0.014547041495435866,
+      "grad_norm": 0.0,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 1.376,
+      "step": 400
+    },
+    {
+      "epoch": 0.014583409099174455,
+      "grad_norm": 0.0,
+      "learning_rate": 9.721212121212123e-06,
+      "loss": 1.1164,
+      "step": 401
+    },
+    {
+      "epoch": 0.014619776702913045,
+      "grad_norm": 0.0,
+      "learning_rate": 9.745454545454547e-06,
+      "loss": 1.2793,
+      "step": 402
+    },
+    {
+      "epoch": 0.014656144306651634,
+      "grad_norm": 0.0,
+      "learning_rate": 9.76969696969697e-06,
+      "loss": 1.288,
+      "step": 403
+    },
+    {
+      "epoch": 0.014692511910390225,
+      "grad_norm": 0.0,
+      "learning_rate": 9.793939393939394e-06,
+      "loss": 1.2759,
+      "step": 404
+    },
+    {
+      "epoch": 0.014728879514128814,
+      "grad_norm": 0.0,
+      "learning_rate": 9.81818181818182e-06,
+      "loss": 1.4071,
+      "step": 405
+    },
+    {
+      "epoch": 0.014765247117867404,
+      "grad_norm": 0.0,
+      "learning_rate": 9.842424242424243e-06,
+      "loss": 1.3276,
+      "step": 406
+    },
+    {
+      "epoch": 0.014801614721605993,
+      "grad_norm": 0.0,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 1.1772,
+      "step": 407
+    },
+    {
+      "epoch": 0.014837982325344582,
+      "grad_norm": 0.0,
+      "learning_rate": 9.890909090909092e-06,
+      "loss": 1.2216,
+      "step": 408
+    },
+    {
+      "epoch": 0.014874349929083173,
+      "grad_norm": 0.0,
+      "learning_rate": 9.915151515151515e-06,
+      "loss": 1.223,
+      "step": 409
+    },
+    {
+      "epoch": 0.014910717532821763,
+      "grad_norm": 0.0,
+      "learning_rate": 9.939393939393939e-06,
+      "loss": 1.352,
+      "step": 410
+    },
+    {
+      "epoch": 0.014947085136560352,
+      "grad_norm": 0.0,
+      "learning_rate": 9.963636363636364e-06,
+      "loss": 1.369,
+      "step": 411
+    },
+    {
+      "epoch": 0.014983452740298941,
+      "grad_norm": 0.0,
+      "learning_rate": 9.987878787878788e-06,
+      "loss": 1.29,
+      "step": 412
+    },
+    {
+      "epoch": 0.01501982034403753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0012121212121213e-05,
+      "loss": 1.253,
+      "step": 413
+    },
+    {
+      "epoch": 0.015056187947776122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0036363636363637e-05,
+      "loss": 1.3641,
+      "step": 414
+    },
+    {
+      "epoch": 0.015092555551514711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0060606060606062e-05,
+      "loss": 1.2062,
+      "step": 415
+    },
+    {
+      "epoch": 0.0151289231552533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0084848484848486e-05,
+      "loss": 1.2538,
+      "step": 416
+    },
+    {
+      "epoch": 0.01516529075899189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.010909090909091e-05,
+      "loss": 1.2617,
+      "step": 417
+    },
+    {
+      "epoch": 0.01520165836273048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 1.2076,
+      "step": 418
+    },
+    {
+      "epoch": 0.015238025966469068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0157575757575759e-05,
+      "loss": 1.1562,
+      "step": 419
+    },
+    {
+      "epoch": 0.01527439357020766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0181818181818182e-05,
+      "loss": 1.0605,
+      "step": 420
+    },
+    {
+      "epoch": 0.015310761173946249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0206060606060606e-05,
+      "loss": 1.1892,
+      "step": 421
+    },
+    {
+      "epoch": 0.015347128777684838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.023030303030303e-05,
+      "loss": 1.3706,
+      "step": 422
+    },
+    {
+      "epoch": 0.015383496381423428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0254545454545455e-05,
+      "loss": 1.1965,
+      "step": 423
+    },
+    {
+      "epoch": 0.015419863985162017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0278787878787879e-05,
+      "loss": 1.4166,
+      "step": 424
+    },
+    {
+      "epoch": 0.015456231588900608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0303030303030304e-05,
+      "loss": 1.2863,
+      "step": 425
+    },
+    {
+      "epoch": 0.015492599192639197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0327272727272728e-05,
+      "loss": 1.2488,
+      "step": 426
+    },
+    {
+      "epoch": 0.015528966796377787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0351515151515153e-05,
+      "loss": 1.2823,
+      "step": 427
+    },
+    {
+      "epoch": 0.015565334400116376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0375757575757577e-05,
+      "loss": 1.2809,
+      "step": 428
+    },
+    {
+      "epoch": 0.015601702003854965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.04e-05,
+      "loss": 1.259,
+      "step": 429
+    },
+    {
+      "epoch": 0.015638069607593556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0424242424242426e-05,
+      "loss": 1.2817,
+      "step": 430
+    },
+    {
+      "epoch": 0.015674437211332146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.044848484848485e-05,
+      "loss": 1.3707,
+      "step": 431
+    },
+    {
+      "epoch": 0.015710804815070735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0472727272727275e-05,
+      "loss": 1.2503,
+      "step": 432
+    },
+    {
+      "epoch": 0.015747172418809324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0496969696969698e-05,
+      "loss": 1.2139,
+      "step": 433
+    },
+    {
+      "epoch": 0.015783540022547914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.052121212121212e-05,
+      "loss": 1.2943,
+      "step": 434
+    },
+    {
+      "epoch": 0.015819907626286503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0545454545454546e-05,
+      "loss": 1.2576,
+      "step": 435
+    },
+    {
+      "epoch": 0.015856275230025092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.056969696969697e-05,
+      "loss": 1.2505,
+      "step": 436
+    },
+    {
+      "epoch": 0.01589264283376368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0593939393939395e-05,
+      "loss": 1.2325,
+      "step": 437
+    },
+    {
+      "epoch": 0.015929010437502274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0618181818181818e-05,
+      "loss": 1.1708,
+      "step": 438
+    },
+    {
+      "epoch": 0.015965378041240864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0642424242424244e-05,
+      "loss": 1.2472,
+      "step": 439
+    },
+    {
+      "epoch": 0.016001745644979453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 1.3326,
+      "step": 440
+    },
+    {
+      "epoch": 0.016038113248718042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0690909090909091e-05,
+      "loss": 1.2577,
+      "step": 441
+    },
+    {
+      "epoch": 0.01607448085245663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0715151515151516e-05,
+      "loss": 1.2756,
+      "step": 442
+    },
+    {
+      "epoch": 0.01611084845619522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.073939393939394e-05,
+      "loss": 1.2823,
+      "step": 443
+    },
+    {
+      "epoch": 0.01614721605993381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0763636363636365e-05,
+      "loss": 1.2764,
+      "step": 444
+    },
+    {
+      "epoch": 0.0161835836636724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0787878787878789e-05,
+      "loss": 1.2994,
+      "step": 445
+    },
+    {
+      "epoch": 0.01621995126741099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0812121212121214e-05,
+      "loss": 1.2753,
+      "step": 446
+    },
+    {
+      "epoch": 0.01625631887114958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0836363636363638e-05,
+      "loss": 1.3279,
+      "step": 447
+    },
+    {
+      "epoch": 0.01629268647488817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0860606060606063e-05,
+      "loss": 1.3711,
+      "step": 448
+    },
+    {
+      "epoch": 0.01632905407862676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0884848484848485e-05,
+      "loss": 1.4347,
+      "step": 449
+    },
+    {
+      "epoch": 0.01636542168236535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 1.328,
+      "step": 450
+    },
+    {
+      "epoch": 0.01640178928610394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 1.341,
+      "step": 451
+    },
+    {
+      "epoch": 0.01643815688984253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0957575757575758e-05,
+      "loss": 1.329,
+      "step": 452
+    },
+    {
+      "epoch": 0.016474524493581118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0981818181818182e-05,
+      "loss": 1.1932,
+      "step": 453
+    },
+    {
+      "epoch": 0.016510892097319707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1006060606060607e-05,
+      "loss": 1.3445,
+      "step": 454
+    },
+    {
+      "epoch": 0.016547259701058296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.103030303030303e-05,
+      "loss": 1.3911,
+      "step": 455
+    },
+    {
+      "epoch": 0.016583627304796886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1054545454545456e-05,
+      "loss": 1.188,
+      "step": 456
+    },
+    {
+      "epoch": 0.016619994908535475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.107878787878788e-05,
+      "loss": 1.1836,
+      "step": 457
+    },
+    {
+      "epoch": 0.016656362512274068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1103030303030305e-05,
+      "loss": 1.3312,
+      "step": 458
+    },
+    {
+      "epoch": 0.016692730116012657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1127272727272729e-05,
+      "loss": 1.2747,
+      "step": 459
+    },
+    {
+      "epoch": 0.016729097719751247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1151515151515154e-05,
+      "loss": 1.1972,
+      "step": 460
+    },
+    {
+      "epoch": 0.016765465323489836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1175757575757578e-05,
+      "loss": 1.1483,
+      "step": 461
+    },
+    {
+      "epoch": 0.016801832927228425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.2277,
+      "step": 462
+    },
+    {
+      "epoch": 0.016838200530967014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1224242424242425e-05,
+      "loss": 1.3436,
+      "step": 463
+    },
+    {
+      "epoch": 0.016874568134705604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1248484848484848e-05,
+      "loss": 1.2126,
+      "step": 464
+    },
+    {
+      "epoch": 0.016910935738444193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1272727272727272e-05,
+      "loss": 1.2437,
+      "step": 465
+    },
+    {
+      "epoch": 0.016947303342182782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1296969696969697e-05,
+      "loss": 1.3087,
+      "step": 466
+    },
+    {
+      "epoch": 0.016983670945921372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1321212121212121e-05,
+      "loss": 1.3416,
+      "step": 467
+    },
+    {
+      "epoch": 0.017020038549659965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1345454545454546e-05,
+      "loss": 1.2062,
+      "step": 468
+    },
+    {
+      "epoch": 0.017056406153398554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.136969696969697e-05,
+      "loss": 1.3575,
+      "step": 469
+    },
+    {
+      "epoch": 0.017092773757137143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1393939393939395e-05,
+      "loss": 1.231,
+      "step": 470
+    },
+    {
+      "epoch": 0.017129141360875733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1418181818181819e-05,
+      "loss": 1.3608,
+      "step": 471
+    },
+    {
+      "epoch": 0.017165508964614322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1442424242424244e-05,
+      "loss": 1.3125,
+      "step": 472
+    },
+    {
+      "epoch": 0.01720187656835291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1466666666666668e-05,
+      "loss": 1.4034,
+      "step": 473
+    },
+    {
+      "epoch": 0.0172382441720915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1490909090909092e-05,
+      "loss": 1.2243,
+      "step": 474
+    },
+    {
+      "epoch": 0.01727461177583009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1515151515151517e-05,
+      "loss": 1.2938,
+      "step": 475
+    },
+    {
+      "epoch": 0.01731097937956868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.153939393939394e-05,
+      "loss": 1.2006,
+      "step": 476
+    },
+    {
+      "epoch": 0.01734734698330727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1563636363636363e-05,
+      "loss": 1.2771,
+      "step": 477
+    },
+    {
+      "epoch": 0.017383714587045858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1587878787878788e-05,
+      "loss": 1.2138,
+      "step": 478
+    },
+    {
+      "epoch": 0.01742008219078445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1612121212121212e-05,
+      "loss": 1.239,
+      "step": 479
+    },
+    {
+      "epoch": 0.01745644979452304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1636363636363637e-05,
+      "loss": 1.3982,
+      "step": 480
+    },
+    {
+      "epoch": 0.01749281739826163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.166060606060606e-05,
+      "loss": 1.4167,
+      "step": 481
+    },
+    {
+      "epoch": 0.01752918500200022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1684848484848486e-05,
+      "loss": 1.2051,
+      "step": 482
+    },
+    {
+      "epoch": 0.017565552605738808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.170909090909091e-05,
+      "loss": 1.081,
+      "step": 483
+    },
+    {
+      "epoch": 0.017601920209477397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 1.3057,
+      "step": 484
+    },
+    {
+      "epoch": 0.017638287813215987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1757575757575759e-05,
+      "loss": 1.3824,
+      "step": 485
+    },
+    {
+      "epoch": 0.017674655416954576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1781818181818182e-05,
+      "loss": 1.277,
+      "step": 486
+    },
+    {
+      "epoch": 0.017711023020693165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1806060606060608e-05,
+      "loss": 1.1995,
+      "step": 487
+    },
+    {
+      "epoch": 0.017747390624431755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1830303030303031e-05,
+      "loss": 1.3004,
+      "step": 488
+    },
+    {
+      "epoch": 0.017783758228170347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1854545454545457e-05,
+      "loss": 1.4205,
+      "step": 489
+    },
+    {
+      "epoch": 0.017820125831908937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.187878787878788e-05,
+      "loss": 1.374,
+      "step": 490
+    },
+    {
+      "epoch": 0.017856493435647526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1903030303030302e-05,
+      "loss": 1.4194,
+      "step": 491
+    },
+    {
+      "epoch": 0.017892861039386115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1927272727272728e-05,
+      "loss": 1.1522,
+      "step": 492
+    },
+    {
+      "epoch": 0.017929228643124705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1951515151515151e-05,
+      "loss": 1.1833,
+      "step": 493
+    },
+    {
+      "epoch": 0.017965596246863294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1975757575757577e-05,
+      "loss": 1.2334,
+      "step": 494
+    },
+    {
+      "epoch": 0.018001963850601883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.2918,
+      "step": 495
+    },
+    {
+      "epoch": 0.018038331454340473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2024242424242426e-05,
+      "loss": 1.34,
+      "step": 496
+    },
+    {
+      "epoch": 0.018074699058079062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.204848484848485e-05,
+      "loss": 1.3194,
+      "step": 497
+    },
+    {
+      "epoch": 0.01811106666181765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2072727272727273e-05,
+      "loss": 1.2363,
+      "step": 498
+    },
+    {
+      "epoch": 0.018147434265556244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2096969696969698e-05,
+      "loss": 1.1693,
+      "step": 499
+    },
+    {
+      "epoch": 0.018183801869294833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 1.2618,
+      "step": 500
+    },
+    {
+      "epoch": 0.018220169473033423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2145454545454547e-05,
+      "loss": 1.2824,
+      "step": 501
+    },
+    {
+      "epoch": 0.018256537076772012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2169696969696971e-05,
+      "loss": 1.3588,
+      "step": 502
+    },
+    {
+      "epoch": 0.0182929046805106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2193939393939396e-05,
+      "loss": 1.1961,
+      "step": 503
+    },
+    {
+      "epoch": 0.01832927228424919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.221818181818182e-05,
+      "loss": 1.2827,
+      "step": 504
+    },
+    {
+      "epoch": 0.01836563988798778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2242424242424242e-05,
+      "loss": 1.2812,
+      "step": 505
+    },
+    {
+      "epoch": 0.01840200749172637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 1.3618,
+      "step": 506
+    },
+    {
+      "epoch": 0.01843837509546496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2290909090909091e-05,
+      "loss": 1.2849,
+      "step": 507
+    },
+    {
+      "epoch": 0.018474742699203548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2315151515151516e-05,
+      "loss": 1.3597,
+      "step": 508
+    },
+    {
+      "epoch": 0.018511110302942137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.233939393939394e-05,
+      "loss": 1.187,
+      "step": 509
+    },
+    {
+      "epoch": 0.01854747790668073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2363636363636364e-05,
+      "loss": 1.4229,
+      "step": 510
+    },
+    {
+      "epoch": 0.01858384551041932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2387878787878789e-05,
+      "loss": 1.3484,
+      "step": 511
+    },
+    {
+      "epoch": 0.01862021311415791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2412121212121213e-05,
+      "loss": 1.2581,
+      "step": 512
+    },
+    {
+      "epoch": 0.018656580717896498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2436363636363638e-05,
+      "loss": 1.2253,
+      "step": 513
+    },
+    {
+      "epoch": 0.018692948321635088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2460606060606062e-05,
+      "loss": 1.2898,
+      "step": 514
+    },
+    {
+      "epoch": 0.018729315925373677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2484848484848487e-05,
+      "loss": 1.3224,
+      "step": 515
+    },
+    {
+      "epoch": 0.018765683529112266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.250909090909091e-05,
+      "loss": 1.2151,
+      "step": 516
+    },
+    {
+      "epoch": 0.018802051132850855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2533333333333336e-05,
+      "loss": 1.3374,
+      "step": 517
+    },
+    {
+      "epoch": 0.018838418736589445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.255757575757576e-05,
+      "loss": 1.3527,
+      "step": 518
+    },
+    {
+      "epoch": 0.018874786340328034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2581818181818183e-05,
+      "loss": 1.4056,
+      "step": 519
+    },
+    {
+      "epoch": 0.018911153944066627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2606060606060607e-05,
+      "loss": 1.2683,
+      "step": 520
+    },
+    {
+      "epoch": 0.018947521547805216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.263030303030303e-05,
+      "loss": 1.3289,
+      "step": 521
+    },
+    {
+      "epoch": 0.018983889151543806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2654545454545454e-05,
+      "loss": 1.3569,
+      "step": 522
+    },
+    {
+      "epoch": 0.019020256755282395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.267878787878788e-05,
+      "loss": 1.2594,
+      "step": 523
+    },
+    {
+      "epoch": 0.019056624359020984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2703030303030303e-05,
+      "loss": 1.3134,
+      "step": 524
+    },
+    {
+      "epoch": 0.019092991962759574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 1.1496,
+      "step": 525
+    },
+    {
+      "epoch": 0.019129359566498163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2751515151515152e-05,
+      "loss": 1.2329,
+      "step": 526
+    },
+    {
+      "epoch": 0.019165727170236752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2775757575757577e-05,
+      "loss": 1.2197,
+      "step": 527
+    },
+    {
+      "epoch": 0.01920209477397534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.3134,
+      "step": 528
+    },
+    {
+      "epoch": 0.01923846237771393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2824242424242426e-05,
+      "loss": 1.2729,
+      "step": 529
+    },
+    {
+      "epoch": 0.019274829981452524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.284848484848485e-05,
+      "loss": 1.3225,
+      "step": 530
+    },
+    {
+      "epoch": 0.019311197585191113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2872727272727274e-05,
+      "loss": 1.2096,
+      "step": 531
+    },
+    {
+      "epoch": 0.019347565188929702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2896969696969699e-05,
+      "loss": 1.2195,
+      "step": 532
+    },
+    {
+      "epoch": 0.01938393279266829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2921212121212123e-05,
+      "loss": 1.2389,
+      "step": 533
+    },
+    {
+      "epoch": 0.01942030039640688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2945454545454545e-05,
+      "loss": 1.2756,
+      "step": 534
+    },
+    {
+      "epoch": 0.01945666800014547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.296969696969697e-05,
+      "loss": 1.2173,
+      "step": 535
+    },
+    {
+      "epoch": 0.01949303560388406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2993939393939394e-05,
+      "loss": 1.4097,
+      "step": 536
+    },
+    {
+      "epoch": 0.01952940320762265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3018181818181819e-05,
+      "loss": 1.2989,
+      "step": 537
+    },
+    {
+      "epoch": 0.01956577081136124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3042424242424243e-05,
+      "loss": 1.3265,
+      "step": 538
+    },
+    {
+      "epoch": 0.019602138415099828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3066666666666668e-05,
+      "loss": 1.308,
+      "step": 539
+    },
+    {
+      "epoch": 0.01963850601883842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3090909090909092e-05,
+      "loss": 1.3691,
+      "step": 540
+    },
+    {
+      "epoch": 0.01967487362257701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3115151515151517e-05,
+      "loss": 1.1305,
+      "step": 541
+    },
+    {
+      "epoch": 0.0197112412263156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.313939393939394e-05,
+      "loss": 1.2813,
+      "step": 542
+    },
+    {
+      "epoch": 0.01974760883005419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3163636363636364e-05,
+      "loss": 1.2375,
+      "step": 543
+    },
+    {
+      "epoch": 0.019783976433792778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.318787878787879e-05,
+      "loss": 1.2184,
+      "step": 544
+    },
+    {
+      "epoch": 0.019820344037531367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3212121212121213e-05,
+      "loss": 1.2287,
+      "step": 545
+    },
+    {
+      "epoch": 0.019856711641269956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3236363636363639e-05,
+      "loss": 1.1804,
+      "step": 546
+    },
+    {
+      "epoch": 0.019893079245008546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3260606060606062e-05,
+      "loss": 1.3185,
+      "step": 547
+    },
+    {
+      "epoch": 0.019929446848747135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3284848484848484e-05,
+      "loss": 1.2403,
+      "step": 548
+    },
+    {
+      "epoch": 0.019965814452485724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.330909090909091e-05,
+      "loss": 1.1691,
+      "step": 549
+    },
+    {
+      "epoch": 0.020002182056224314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.3455,
+      "step": 550
+    },
+    {
+      "epoch": 0.020038549659962906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3357575757575759e-05,
+      "loss": 1.254,
+      "step": 551
+    },
+    {
+      "epoch": 0.020074917263701496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3381818181818182e-05,
+      "loss": 1.2773,
+      "step": 552
+    },
+    {
+      "epoch": 0.020111284867440085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3406060606060608e-05,
+      "loss": 1.2976,
+      "step": 553
+    },
+    {
+      "epoch": 0.020147652471178674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3430303030303031e-05,
+      "loss": 1.3046,
+      "step": 554
+    },
+    {
+      "epoch": 0.020184020074917264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3454545454545455e-05,
+      "loss": 1.2788,
+      "step": 555
+    },
+    {
+      "epoch": 0.020220387678655853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.347878787878788e-05,
+      "loss": 1.3125,
+      "step": 556
+    },
+    {
+      "epoch": 0.020256755282394442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3503030303030304e-05,
+      "loss": 1.2975,
+      "step": 557
+    },
+    {
+      "epoch": 0.020293122886133032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.352727272727273e-05,
+      "loss": 1.2094,
+      "step": 558
+    },
+    {
+      "epoch": 0.02032949048987162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3551515151515153e-05,
+      "loss": 1.1942,
+      "step": 559
+    },
+    {
+      "epoch": 0.02036585809361021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3575757575757578e-05,
+      "loss": 1.1126,
+      "step": 560
+    },
+    {
+      "epoch": 0.020402225697348803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 1.1928,
+      "step": 561
+    },
+    {
+      "epoch": 0.020438593301087393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3624242424242424e-05,
+      "loss": 1.2512,
+      "step": 562
+    },
+    {
+      "epoch": 0.020474960904825982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.364848484848485e-05,
+      "loss": 1.2344,
+      "step": 563
+    },
+    {
+      "epoch": 0.02051132850856457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3672727272727273e-05,
+      "loss": 1.3596,
+      "step": 564
+    },
+    {
+      "epoch": 0.02054769611230316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3696969696969698e-05,
+      "loss": 1.1975,
+      "step": 565
+    },
+    {
+      "epoch": 0.02058406371604175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3721212121212122e-05,
+      "loss": 1.4313,
+      "step": 566
+    },
+    {
+      "epoch": 0.02062043131978034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3745454545454546e-05,
+      "loss": 1.2395,
+      "step": 567
+    },
+    {
+      "epoch": 0.02065679892351893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3769696969696971e-05,
+      "loss": 1.2803,
+      "step": 568
+    },
+    {
+      "epoch": 0.020693166527257518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3793939393939395e-05,
+      "loss": 1.267,
+      "step": 569
+    },
+    {
+      "epoch": 0.020729534130996107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.381818181818182e-05,
+      "loss": 1.3618,
+      "step": 570
+    },
+    {
+      "epoch": 0.0207659017347347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3842424242424244e-05,
+      "loss": 1.2041,
+      "step": 571
+    },
+    {
+      "epoch": 0.02080226933847329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 1.1918,
+      "step": 572
+    },
+    {
+      "epoch": 0.02083863694221188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3890909090909093e-05,
+      "loss": 1.2347,
+      "step": 573
+    },
+    {
+      "epoch": 0.020875004545950468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3915151515151518e-05,
+      "loss": 1.2585,
+      "step": 574
+    },
+    {
+      "epoch": 0.020911372149689057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3939393939393942e-05,
+      "loss": 1.3073,
+      "step": 575
+    },
+    {
+      "epoch": 0.020947739753427647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3963636363636365e-05,
+      "loss": 1.2078,
+      "step": 576
+    },
+    {
+      "epoch": 0.020984107357166236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3987878787878789e-05,
+      "loss": 1.2471,
+      "step": 577
+    },
+    {
+      "epoch": 0.021020474960904825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4012121212121212e-05,
+      "loss": 1.3461,
+      "step": 578
+    },
+    {
+      "epoch": 0.021056842564643415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4036363636363636e-05,
+      "loss": 1.231,
+      "step": 579
+    },
+    {
+      "epoch": 0.021093210168382004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4060606060606061e-05,
+      "loss": 1.2773,
+      "step": 580
+    },
+    {
+      "epoch": 0.021129577772120597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4084848484848485e-05,
+      "loss": 1.2749,
+      "step": 581
+    },
+    {
+      "epoch": 0.021165945375859186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.410909090909091e-05,
+      "loss": 1.2563,
+      "step": 582
+    },
+    {
+      "epoch": 0.021202312979597775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 1.2728,
+      "step": 583
+    },
+    {
+      "epoch": 0.021238680583336365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.415757575757576e-05,
+      "loss": 1.2852,
+      "step": 584
+    },
+    {
+      "epoch": 0.021275048187074954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4181818181818183e-05,
+      "loss": 1.2162,
+      "step": 585
+    },
+    {
+      "epoch": 0.021311415790813543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4206060606060608e-05,
+      "loss": 1.1967,
+      "step": 586
+    },
+    {
+      "epoch": 0.021347783394552133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4230303030303032e-05,
+      "loss": 1.2177,
+      "step": 587
+    },
+    {
+      "epoch": 0.021384150998290722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4254545454545456e-05,
+      "loss": 1.2236,
+      "step": 588
+    },
+    {
+      "epoch": 0.02142051860202931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4278787878787881e-05,
+      "loss": 1.2442,
+      "step": 589
+    },
+    {
+      "epoch": 0.0214568862057679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4303030303030305e-05,
+      "loss": 1.2899,
+      "step": 590
+    },
+    {
+      "epoch": 0.02149325380950649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4327272727272727e-05,
+      "loss": 1.207,
+      "step": 591
+    },
+    {
+      "epoch": 0.021529621413245083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4351515151515152e-05,
+      "loss": 1.322,
+      "step": 592
+    },
+    {
+      "epoch": 0.021565989016983672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4375757575757576e-05,
+      "loss": 1.2391,
+      "step": 593
+    },
+    {
+      "epoch": 0.02160235662072226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 1.3035,
+      "step": 594
+    },
+    {
+      "epoch": 0.02163872422446085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4424242424242425e-05,
+      "loss": 1.2458,
+      "step": 595
+    },
+    {
+      "epoch": 0.02167509182819944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.444848484848485e-05,
+      "loss": 1.3777,
+      "step": 596
+    },
+    {
+      "epoch": 0.02171145943193803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4472727272727274e-05,
+      "loss": 1.3392,
+      "step": 597
+    },
+    {
+      "epoch": 0.02174782703567662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4496969696969699e-05,
+      "loss": 1.303,
+      "step": 598
+    },
+    {
+      "epoch": 0.021784194639415208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4521212121212123e-05,
+      "loss": 1.2838,
+      "step": 599
+    },
+    {
+      "epoch": 0.021820562243153797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 1.3063,
+      "step": 600
+    },
+    {
+      "epoch": 0.021856929846892387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4569696969696972e-05,
+      "loss": 1.2248,
+      "step": 601
+    },
+    {
+      "epoch": 0.02189329745063098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4593939393939395e-05,
+      "loss": 1.3246,
+      "step": 602
+    },
+    {
+      "epoch": 0.02192966505436957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.461818181818182e-05,
+      "loss": 1.1393,
+      "step": 603
+    },
+    {
+      "epoch": 0.021966032658108158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4642424242424244e-05,
+      "loss": 1.3748,
+      "step": 604
+    },
+    {
+      "epoch": 0.022002400261846747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 1.3094,
+      "step": 605
+    },
+    {
+      "epoch": 0.022038767865585337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4690909090909092e-05,
+      "loss": 1.2373,
+      "step": 606
+    },
+    {
+      "epoch": 0.022075135469323926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4715151515151515e-05,
+      "loss": 1.2934,
+      "step": 607
+    },
+    {
+      "epoch": 0.022111503073062515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.473939393939394e-05,
+      "loss": 1.4495,
+      "step": 608
+    },
+    {
+      "epoch": 0.022147870676801105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4763636363636364e-05,
+      "loss": 1.1727,
+      "step": 609
+    },
+    {
+      "epoch": 0.022184238280539694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.478787878787879e-05,
+      "loss": 1.1969,
+      "step": 610
+    },
+    {
+      "epoch": 0.022220605884278283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4812121212121213e-05,
+      "loss": 1.3033,
+      "step": 611
+    },
+    {
+      "epoch": 0.022256973488016876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4836363636363637e-05,
+      "loss": 1.3209,
+      "step": 612
+    },
+    {
+      "epoch": 0.022293341091755466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4860606060606062e-05,
+      "loss": 1.3888,
+      "step": 613
+    },
+    {
+      "epoch": 0.022329708695494055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4884848484848486e-05,
+      "loss": 1.2355,
+      "step": 614
+    },
+    {
+      "epoch": 0.022366076299232644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4909090909090911e-05,
+      "loss": 1.2922,
+      "step": 615
+    },
+    {
+      "epoch": 0.022402443902971234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 1.3674,
+      "step": 616
+    },
+    {
+      "epoch": 0.022438811506709823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.495757575757576e-05,
+      "loss": 1.337,
+      "step": 617
+    },
+    {
+      "epoch": 0.022475179110448412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4981818181818184e-05,
+      "loss": 1.2685,
+      "step": 618
+    },
+    {
+      "epoch": 0.022511546714187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5006060606060606e-05,
+      "loss": 1.2027,
+      "step": 619
+    },
+    {
+      "epoch": 0.02254791431792559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5030303030303031e-05,
+      "loss": 1.2003,
+      "step": 620
+    },
+    {
+      "epoch": 0.02258428192166418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5054545454545455e-05,
+      "loss": 1.2126,
+      "step": 621
+    },
+    {
+      "epoch": 0.02262064952540277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.507878787878788e-05,
+      "loss": 1.2825,
+      "step": 622
+    },
+    {
+      "epoch": 0.022657017129141362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5103030303030304e-05,
+      "loss": 1.2944,
+      "step": 623
+    },
+    {
+      "epoch": 0.02269338473287995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5127272727272728e-05,
+      "loss": 1.3845,
+      "step": 624
+    },
+    {
+      "epoch": 0.02272975233661854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 1.2311,
+      "step": 625
+    },
+    {
+      "epoch": 0.02276611994035713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5175757575757577e-05,
+      "loss": 1.2612,
+      "step": 626
+    },
+    {
+      "epoch": 0.02280248754409572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 1.3226,
+      "step": 627
+    },
+    {
+      "epoch": 0.02283885514783431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5224242424242426e-05,
+      "loss": 1.2713,
+      "step": 628
+    },
+    {
+      "epoch": 0.0228752227515729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5248484848484851e-05,
+      "loss": 1.2483,
+      "step": 629
+    },
+    {
+      "epoch": 0.022911590355311488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5272727272727276e-05,
+      "loss": 1.1781,
+      "step": 630
+    },
+    {
+      "epoch": 0.022947957959050077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5296969696969698e-05,
+      "loss": 1.3387,
+      "step": 631
+    },
+    {
+      "epoch": 0.022984325562788666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5321212121212123e-05,
+      "loss": 1.1894,
+      "step": 632
+    },
+    {
+      "epoch": 0.02302069316652726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5345454545454545e-05,
+      "loss": 1.1714,
+      "step": 633
+    },
+    {
+      "epoch": 0.02305706077026585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.536969696969697e-05,
+      "loss": 1.2813,
+      "step": 634
+    },
+    {
+      "epoch": 0.023093428374004438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5393939393939393e-05,
+      "loss": 1.3838,
+      "step": 635
+    },
+    {
+      "epoch": 0.023129795977743027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5418181818181818e-05,
+      "loss": 1.2598,
+      "step": 636
+    },
+    {
+      "epoch": 0.023166163581481616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5442424242424243e-05,
+      "loss": 1.3787,
+      "step": 637
+    },
+    {
+      "epoch": 0.023202531185220206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 1.2157,
+      "step": 638
+    },
+    {
+      "epoch": 0.023238898788958795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.549090909090909e-05,
+      "loss": 1.2809,
+      "step": 639
+    },
+    {
+      "epoch": 0.023275266392697384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5515151515151516e-05,
+      "loss": 1.249,
+      "step": 640
+    },
+    {
+      "epoch": 0.023311633996435974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.553939393939394e-05,
+      "loss": 1.2673,
+      "step": 641
+    },
+    {
+      "epoch": 0.023348001600174563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5563636363636367e-05,
+      "loss": 1.3242,
+      "step": 642
+    },
+    {
+      "epoch": 0.023384369203913156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.558787878787879e-05,
+      "loss": 1.2599,
+      "step": 643
+    },
+    {
+      "epoch": 0.023420736807651745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5612121212121214e-05,
+      "loss": 1.3228,
+      "step": 644
+    },
+    {
+      "epoch": 0.023457104411390334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.563636363636364e-05,
+      "loss": 1.3143,
+      "step": 645
+    },
+    {
+      "epoch": 0.023493472015128924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.566060606060606e-05,
+      "loss": 1.2433,
+      "step": 646
+    },
+    {
+      "epoch": 0.023529839618867513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5684848484848487e-05,
+      "loss": 1.2521,
+      "step": 647
+    },
+    {
+      "epoch": 0.023566207222606102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.570909090909091e-05,
+      "loss": 1.2781,
+      "step": 648
+    },
+    {
+      "epoch": 0.023602574826344692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 1.3071,
+      "step": 649
+    },
+    {
+      "epoch": 0.02363894243008328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.575757575757576e-05,
+      "loss": 1.1893,
+      "step": 650
+    },
+    {
+      "epoch": 0.02367531003382187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.578181818181818e-05,
+      "loss": 1.2164,
+      "step": 651
+    },
+    {
+      "epoch": 0.02371167763756046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5806060606060607e-05,
+      "loss": 1.2007,
+      "step": 652
+    },
+    {
+      "epoch": 0.023748045241299053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5830303030303032e-05,
+      "loss": 1.3035,
+      "step": 653
+    },
+    {
+      "epoch": 0.023784412845037642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5854545454545457e-05,
+      "loss": 1.2692,
+      "step": 654
+    },
+    {
+      "epoch": 0.02382078044877623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.587878787878788e-05,
+      "loss": 1.2634,
+      "step": 655
+    },
+    {
+      "epoch": 0.02385714805251482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5903030303030305e-05,
+      "loss": 1.3218,
+      "step": 656
+    },
+    {
+      "epoch": 0.02389351565625341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.592727272727273e-05,
+      "loss": 1.2972,
+      "step": 657
+    },
+    {
+      "epoch": 0.023929883259992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5951515151515152e-05,
+      "loss": 1.2693,
+      "step": 658
+    },
+    {
+      "epoch": 0.02396625086373059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5975757575757577e-05,
+      "loss": 1.3866,
+      "step": 659
+    },
+    {
+      "epoch": 0.024002618467469178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.2901,
+      "step": 660
+    },
+    {
+      "epoch": 0.024038986071207767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6024242424242428e-05,
+      "loss": 1.225,
+      "step": 661
+    },
+    {
+      "epoch": 0.024075353674946356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.604848484848485e-05,
+      "loss": 1.2086,
+      "step": 662
+    },
+    {
+      "epoch": 0.024111721278684946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6072727272727272e-05,
+      "loss": 1.2927,
+      "step": 663
+    },
+    {
+      "epoch": 0.02414808888242354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6096969696969697e-05,
+      "loss": 1.2642,
+      "step": 664
+    },
+    {
+      "epoch": 0.024184456486162128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6121212121212123e-05,
+      "loss": 1.3831,
+      "step": 665
+    },
+    {
+      "epoch": 0.024220824089900717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6145454545454548e-05,
+      "loss": 1.2634,
+      "step": 666
+    },
+    {
+      "epoch": 0.024257191693639307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.616969696969697e-05,
+      "loss": 1.1989,
+      "step": 667
+    },
+    {
+      "epoch": 0.024293559297377896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6193939393939395e-05,
+      "loss": 1.1824,
+      "step": 668
+    },
+    {
+      "epoch": 0.024329926901116485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.621818181818182e-05,
+      "loss": 1.2726,
+      "step": 669
+    },
+    {
+      "epoch": 0.024366294504855075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6242424242424243e-05,
+      "loss": 1.2296,
+      "step": 670
+    },
+    {
+      "epoch": 0.024402662108593664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6266666666666668e-05,
+      "loss": 1.2009,
+      "step": 671
+    },
+    {
+      "epoch": 0.024439029712332253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6290909090909093e-05,
+      "loss": 1.1417,
+      "step": 672
+    },
+    {
+      "epoch": 0.024475397316070843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.631515151515152e-05,
+      "loss": 1.2377,
+      "step": 673
+    },
+    {
+      "epoch": 0.024511764919809435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.633939393939394e-05,
+      "loss": 1.2395,
+      "step": 674
+    },
+    {
+      "epoch": 0.024548132523548025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 1.2834,
+      "step": 675
+    },
+    {
+      "epoch": 0.024584500127286614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6387878787878788e-05,
+      "loss": 1.2848,
+      "step": 676
+    },
+    {
+      "epoch": 0.024620867731025203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6412121212121213e-05,
+      "loss": 1.2807,
+      "step": 677
+    },
+    {
+      "epoch": 0.024657235334763793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.643636363636364e-05,
+      "loss": 1.3081,
+      "step": 678
+    },
+    {
+      "epoch": 0.024693602938502382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646060606060606e-05,
+      "loss": 1.2659,
+      "step": 679
+    },
+    {
+      "epoch": 0.02472997054224097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6484848484848486e-05,
+      "loss": 1.2233,
+      "step": 680
+    },
+    {
+      "epoch": 0.02476633814597956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650909090909091e-05,
+      "loss": 1.2795,
+      "step": 681
+    },
+    {
+      "epoch": 0.02480270574971815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 1.3377,
+      "step": 682
+    },
+    {
+      "epoch": 0.02483907335345674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.655757575757576e-05,
+      "loss": 1.3348,
+      "step": 683
+    },
+    {
+      "epoch": 0.024875440957195332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6581818181818184e-05,
+      "loss": 1.3417,
+      "step": 684
+    },
+    {
+      "epoch": 0.02491180856093392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.660606060606061e-05,
+      "loss": 1.3126,
+      "step": 685
+    },
+    {
+      "epoch": 0.02494817616467251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663030303030303e-05,
+      "loss": 1.1704,
+      "step": 686
+    },
+    {
+      "epoch": 0.0249845437684111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6654545454545456e-05,
+      "loss": 1.2556,
+      "step": 687
+    },
+    {
+      "epoch": 0.02502091137214969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6678787878787882e-05,
+      "loss": 1.3587,
+      "step": 688
+    },
+    {
+      "epoch": 0.02505727897588828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6703030303030304e-05,
+      "loss": 1.2654,
+      "step": 689
+    },
+    {
+      "epoch": 0.025093646579626868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672727272727273e-05,
+      "loss": 1.2459,
+      "step": 690
+    },
+    {
+      "epoch": 0.025130014183365457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.675151515151515e-05,
+      "loss": 1.3003,
+      "step": 691
+    },
+    {
+      "epoch": 0.025166381787104047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6775757575757576e-05,
+      "loss": 1.2746,
+      "step": 692
+    },
+    {
+      "epoch": 0.025202749390842636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.1566,
+      "step": 693
+    },
+    {
+      "epoch": 0.025239116994581225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6824242424242424e-05,
+      "loss": 1.34,
+      "step": 694
+    },
+    {
+      "epoch": 0.025275484598319818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.684848484848485e-05,
+      "loss": 1.3008,
+      "step": 695
+    },
+    {
+      "epoch": 0.025311852202058407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6872727272727274e-05,
+      "loss": 1.1307,
+      "step": 696
+    },
+    {
+      "epoch": 0.025348219805796997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.68969696969697e-05,
+      "loss": 1.3574,
+      "step": 697
+    },
+    {
+      "epoch": 0.025384587409535586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6921212121212122e-05,
+      "loss": 1.2191,
+      "step": 698
+    },
+    {
+      "epoch": 0.025420955013274175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6945454545454547e-05,
+      "loss": 1.3024,
+      "step": 699
+    },
+    {
+      "epoch": 0.025457322617012765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6969696969696972e-05,
+      "loss": 1.3829,
+      "step": 700
+    },
+    {
+      "epoch": 0.025493690220751354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6993939393939394e-05,
+      "loss": 1.3354,
+      "step": 701
+    },
+    {
+      "epoch": 0.025530057824489943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.701818181818182e-05,
+      "loss": 1.2676,
+      "step": 702
+    },
+    {
+      "epoch": 0.025566425428228533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7042424242424245e-05,
+      "loss": 1.2055,
+      "step": 703
+    },
+    {
+      "epoch": 0.025602793031967122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 1.18,
+      "step": 704
+    },
+    {
+      "epoch": 0.025639160635705715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7090909090909092e-05,
+      "loss": 1.3596,
+      "step": 705
+    },
+    {
+      "epoch": 0.025675528239444304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7115151515151514e-05,
+      "loss": 1.2686,
+      "step": 706
+    },
+    {
+      "epoch": 0.025711895843182894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713939393939394e-05,
+      "loss": 1.2979,
+      "step": 707
+    },
+    {
+      "epoch": 0.025748263446921483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7163636363636365e-05,
+      "loss": 1.2415,
+      "step": 708
+    },
+    {
+      "epoch": 0.025784631050660072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718787878787879e-05,
+      "loss": 1.1884,
+      "step": 709
+    },
+    {
+      "epoch": 0.02582099865439866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7212121212121212e-05,
+      "loss": 1.2387,
+      "step": 710
+    },
+    {
+      "epoch": 0.02585736625813725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7236363636363638e-05,
+      "loss": 1.2866,
+      "step": 711
+    },
+    {
+      "epoch": 0.02589373386187584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7260606060606063e-05,
+      "loss": 1.3044,
+      "step": 712
+    },
+    {
+      "epoch": 0.02593010146561443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7284848484848485e-05,
+      "loss": 1.1721,
+      "step": 713
+    },
+    {
+      "epoch": 0.02596646906935302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730909090909091e-05,
+      "loss": 1.2346,
+      "step": 714
+    },
+    {
+      "epoch": 0.02600283667309161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 1.2125,
+      "step": 715
+    },
+    {
+      "epoch": 0.0260392042768302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735757575757576e-05,
+      "loss": 1.1098,
+      "step": 716
+    },
+    {
+      "epoch": 0.02607557188056879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7381818181818183e-05,
+      "loss": 1.1895,
+      "step": 717
+    },
+    {
+      "epoch": 0.02611193948430738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7406060606060608e-05,
+      "loss": 1.3655,
+      "step": 718
+    },
+    {
+      "epoch": 0.02614830708804597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743030303030303e-05,
+      "loss": 1.162,
+      "step": 719
+    },
+    {
+      "epoch": 0.026184674691784558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7454545454545456e-05,
+      "loss": 1.3556,
+      "step": 720
+    },
+    {
+      "epoch": 0.026221042295523148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.747878787878788e-05,
+      "loss": 1.3748,
+      "step": 721
+    },
+    {
+      "epoch": 0.026257409899261737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7503030303030303e-05,
+      "loss": 1.2855,
+      "step": 722
+    },
+    {
+      "epoch": 0.026293777503000326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7527272727272728e-05,
+      "loss": 1.3518,
+      "step": 723
+    },
+    {
+      "epoch": 0.026330145106738916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7551515151515154e-05,
+      "loss": 1.3061,
+      "step": 724
+    },
+    {
+      "epoch": 0.02636651271047751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7575757575757576e-05,
+      "loss": 1.4001,
+      "step": 725
+    },
+    {
+      "epoch": 0.026402880314216098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.76e-05,
+      "loss": 1.2136,
+      "step": 726
+    },
+    {
+      "epoch": 0.026439247917954687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7624242424242426e-05,
+      "loss": 1.3274,
+      "step": 727
+    },
+    {
+      "epoch": 0.026475615521693276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.764848484848485e-05,
+      "loss": 1.3422,
+      "step": 728
+    },
+    {
+      "epoch": 0.026511983125431866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7672727272727274e-05,
+      "loss": 1.3331,
+      "step": 729
+    },
+    {
+      "epoch": 0.026548350729170455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.76969696969697e-05,
+      "loss": 1.3055,
+      "step": 730
+    },
+    {
+      "epoch": 0.026584718332909044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7721212121212124e-05,
+      "loss": 1.07,
+      "step": 731
+    },
+    {
+      "epoch": 0.026621085936647634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774545454545455e-05,
+      "loss": 1.4048,
+      "step": 732
+    },
+    {
+      "epoch": 0.026657453540386223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.776969696969697e-05,
+      "loss": 1.2173,
+      "step": 733
+    },
+    {
+      "epoch": 0.026693821144124812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7793939393939393e-05,
+      "loss": 1.3009,
+      "step": 734
+    },
+    {
+      "epoch": 0.0267301887478634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.781818181818182e-05,
+      "loss": 1.2208,
+      "step": 735
+    },
+    {
+      "epoch": 0.026766556351601994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7842424242424244e-05,
+      "loss": 1.2675,
+      "step": 736
+    },
+    {
+      "epoch": 0.026802923955340584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 1.3207,
+      "step": 737
+    },
+    {
+      "epoch": 0.026839291559079173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.789090909090909e-05,
+      "loss": 1.1985,
+      "step": 738
+    },
+    {
+      "epoch": 0.026875659162817762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7915151515151517e-05,
+      "loss": 1.4114,
+      "step": 739
+    },
+    {
+      "epoch": 0.02691202676655635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7939393939393942e-05,
+      "loss": 1.1566,
+      "step": 740
+    },
+    {
+      "epoch": 0.02694839437029494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7963636363636364e-05,
+      "loss": 1.2816,
+      "step": 741
+    },
+    {
+      "epoch": 0.02698476197403353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.798787878787879e-05,
+      "loss": 1.2347,
+      "step": 742
+    },
+    {
+      "epoch": 0.02702112957777212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8012121212121215e-05,
+      "loss": 1.2062,
+      "step": 743
+    },
+    {
+      "epoch": 0.02705749718151071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803636363636364e-05,
+      "loss": 1.3381,
+      "step": 744
+    },
+    {
+      "epoch": 0.0270938647852493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8060606060606062e-05,
+      "loss": 1.2825,
+      "step": 745
+    },
+    {
+      "epoch": 0.02713023238898789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8084848484848487e-05,
+      "loss": 1.3056,
+      "step": 746
+    },
+    {
+      "epoch": 0.02716659999272648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810909090909091e-05,
+      "loss": 1.3213,
+      "step": 747
+    },
+    {
+      "epoch": 0.02720296759646507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 1.103,
+      "step": 748
+    },
+    {
+      "epoch": 0.02723933520020366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8157575757575757e-05,
+      "loss": 1.2829,
+      "step": 749
+    },
+    {
+      "epoch": 0.02727570280394225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 1.2565,
+      "step": 750
+    },
+    {
+      "epoch": 0.027312070407680838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8206060606060607e-05,
+      "loss": 1.28,
+      "step": 751
+    },
+    {
+      "epoch": 0.027348438011419427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8230303030303033e-05,
+      "loss": 1.1713,
+      "step": 752
+    },
+    {
+      "epoch": 0.027384805615158016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8254545454545455e-05,
+      "loss": 1.338,
+      "step": 753
+    },
+    {
+      "epoch": 0.027421173218896606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827878787878788e-05,
+      "loss": 1.2297,
+      "step": 754
+    },
+    {
+      "epoch": 0.027457540822635195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8303030303030305e-05,
+      "loss": 1.2649,
+      "step": 755
+    },
+    {
+      "epoch": 0.027493908426373788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832727272727273e-05,
+      "loss": 1.3187,
+      "step": 756
+    },
+    {
+      "epoch": 0.027530276030112377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8351515151515153e-05,
+      "loss": 1.1286,
+      "step": 757
+    },
+    {
+      "epoch": 0.027566643633850967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8375757575757578e-05,
+      "loss": 1.3,
+      "step": 758
+    },
+    {
+      "epoch": 0.027603011237589556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 1.1952,
+      "step": 759
+    },
+    {
+      "epoch": 0.027639378841328145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8424242424242425e-05,
+      "loss": 1.217,
+      "step": 760
+    },
+    {
+      "epoch": 0.027675746445066735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8448484848484847e-05,
+      "loss": 1.2283,
+      "step": 761
+    },
+    {
+      "epoch": 0.027712114048805324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8472727272727273e-05,
+      "loss": 1.3339,
+      "step": 762
+    },
+    {
+      "epoch": 0.027748481652543913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8496969696969698e-05,
+      "loss": 1.2623,
+      "step": 763
+    },
+    {
+      "epoch": 0.027784849256282503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8521212121212123e-05,
+      "loss": 1.3704,
+      "step": 764
+    },
+    {
+      "epoch": 0.027821216860021092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8545454545454545e-05,
+      "loss": 1.2155,
+      "step": 765
+    },
+    {
+      "epoch": 0.02785758446375968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856969696969697e-05,
+      "loss": 1.2833,
+      "step": 766
+    },
+    {
+      "epoch": 0.027893952067498274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8593939393939396e-05,
+      "loss": 1.3369,
+      "step": 767
+    },
+    {
+      "epoch": 0.027930319671236863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861818181818182e-05,
+      "loss": 1.2456,
+      "step": 768
+    },
+    {
+      "epoch": 0.027966687274975453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8642424242424243e-05,
+      "loss": 1.2933,
+      "step": 769
+    },
+    {
+      "epoch": 0.028003054878714042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 1.2463,
+      "step": 770
+    },
+    {
+      "epoch": 0.02803942248245263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8690909090909094e-05,
+      "loss": 1.2432,
+      "step": 771
+    },
+    {
+      "epoch": 0.02807579008619122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8715151515151516e-05,
+      "loss": 1.2286,
+      "step": 772
+    },
+    {
+      "epoch": 0.02811215768992981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873939393939394e-05,
+      "loss": 1.2789,
+      "step": 773
+    },
+    {
+      "epoch": 0.0281485252936684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8763636363636367e-05,
+      "loss": 1.3964,
+      "step": 774
+    },
+    {
+      "epoch": 0.02818489289740699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8787878787878792e-05,
+      "loss": 1.2203,
+      "step": 775
+    },
+    {
+      "epoch": 0.028221260501145578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8812121212121214e-05,
+      "loss": 1.2488,
+      "step": 776
+    },
+    {
+      "epoch": 0.02825762810488417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8836363636363636e-05,
+      "loss": 1.1246,
+      "step": 777
+    },
+    {
+      "epoch": 0.02829399570862276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886060606060606e-05,
+      "loss": 1.1865,
+      "step": 778
+    },
+    {
+      "epoch": 0.02833036331236135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8884848484848487e-05,
+      "loss": 1.3567,
+      "step": 779
+    },
+    {
+      "epoch": 0.02836673091609994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8909090909090912e-05,
+      "loss": 1.3915,
+      "step": 780
+    },
+    {
+      "epoch": 0.028403098519838528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 1.2252,
+      "step": 781
+    },
+    {
+      "epoch": 0.028439466123577117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895757575757576e-05,
+      "loss": 1.2074,
+      "step": 782
+    },
+    {
+      "epoch": 0.028475833727315707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8981818181818185e-05,
+      "loss": 1.2296,
+      "step": 783
+    },
+    {
+      "epoch": 0.028512201331054296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9006060606060607e-05,
+      "loss": 1.3525,
+      "step": 784
+    },
+    {
+      "epoch": 0.028548568934792885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9030303030303032e-05,
+      "loss": 1.318,
+      "step": 785
+    },
+    {
+      "epoch": 0.028584936538531475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9054545454545457e-05,
+      "loss": 1.2938,
+      "step": 786
+    },
+    {
+      "epoch": 0.028621304142270067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9078787878787883e-05,
+      "loss": 1.2703,
+      "step": 787
+    },
+    {
+      "epoch": 0.028657671746008657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9103030303030305e-05,
+      "loss": 1.1742,
+      "step": 788
+    },
+    {
+      "epoch": 0.028694039349747246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912727272727273e-05,
+      "loss": 1.3488,
+      "step": 789
+    },
+    {
+      "epoch": 0.028730406953485835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9151515151515152e-05,
+      "loss": 1.3009,
+      "step": 790
+    },
+    {
+      "epoch": 0.028766774557224425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9175757575757577e-05,
+      "loss": 1.1706,
+      "step": 791
+    },
+    {
+      "epoch": 0.028803142160963014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.2116,
+      "step": 792
+    },
+    {
+      "epoch": 0.028839509764701603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9224242424242424e-05,
+      "loss": 1.2829,
+      "step": 793
+    },
+    {
+      "epoch": 0.028875877368440193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924848484848485e-05,
+      "loss": 1.1851,
+      "step": 794
+    },
+    {
+      "epoch": 0.028912244972178782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9272727272727275e-05,
+      "loss": 1.2848,
+      "step": 795
+    },
+    {
+      "epoch": 0.02894861257591737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9296969696969697e-05,
+      "loss": 1.2443,
+      "step": 796
+    },
+    {
+      "epoch": 0.028984980179655964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9321212121212122e-05,
+      "loss": 1.3414,
+      "step": 797
+    },
+    {
+      "epoch": 0.029021347783394554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9345454545454548e-05,
+      "loss": 1.4141,
+      "step": 798
+    },
+    {
+      "epoch": 0.029057715387133143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9369696969696973e-05,
+      "loss": 1.2904,
+      "step": 799
+    },
+    {
+      "epoch": 0.029094082990871732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 1.234,
+      "step": 800
+    },
+    {
+      "epoch": 0.02913045059461032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941818181818182e-05,
+      "loss": 1.3546,
+      "step": 801
+    },
+    {
+      "epoch": 0.02916681819834891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9442424242424246e-05,
+      "loss": 1.105,
+      "step": 802
+    },
+    {
+      "epoch": 0.0292031858020875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 1.2448,
+      "step": 803
+    },
+    {
+      "epoch": 0.02923955340582609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9490909090909093e-05,
+      "loss": 1.2679,
+      "step": 804
+    },
+    {
+      "epoch": 0.02927592100956468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9515151515151515e-05,
+      "loss": 1.3535,
+      "step": 805
+    },
+    {
+      "epoch": 0.029312288613303268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953939393939394e-05,
+      "loss": 1.3729,
+      "step": 806
+    },
+    {
+      "epoch": 0.029348656217041857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9563636363636366e-05,
+      "loss": 1.2326,
+      "step": 807
+    },
+    {
+      "epoch": 0.02938502382078045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9587878787878788e-05,
+      "loss": 1.2333,
+      "step": 808
+    },
+    {
+      "epoch": 0.02942139142451904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9612121212121213e-05,
+      "loss": 1.1964,
+      "step": 809
+    },
+    {
+      "epoch": 0.02945775902825763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963636363636364e-05,
+      "loss": 1.3282,
+      "step": 810
+    },
+    {
+      "epoch": 0.029494126631996218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9660606060606064e-05,
+      "loss": 1.4754,
+      "step": 811
+    },
+    {
+      "epoch": 0.029530494235734808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9684848484848486e-05,
+      "loss": 1.3533,
+      "step": 812
+    },
+    {
+      "epoch": 0.029566861839473397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970909090909091e-05,
+      "loss": 1.3246,
+      "step": 813
+    },
+    {
+      "epoch": 0.029603229443211986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 1.2772,
+      "step": 814
+    },
+    {
+      "epoch": 0.029639597046950576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975757575757576e-05,
+      "loss": 1.3264,
+      "step": 815
+    },
+    {
+      "epoch": 0.029675964650689165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9781818181818184e-05,
+      "loss": 1.3032,
+      "step": 816
+    },
+    {
+      "epoch": 0.029712332254427754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980606060606061e-05,
+      "loss": 1.2356,
+      "step": 817
+    },
+    {
+      "epoch": 0.029748699858166347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983030303030303e-05,
+      "loss": 1.355,
+      "step": 818
+    },
+    {
+      "epoch": 0.029785067461904936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9854545454545456e-05,
+      "loss": 1.3586,
+      "step": 819
+    },
+    {
+      "epoch": 0.029821435065643526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9878787878787878e-05,
+      "loss": 1.296,
+      "step": 820
+    },
+    {
+      "epoch": 0.029857802669382115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9903030303030304e-05,
+      "loss": 1.2018,
+      "step": 821
+    },
+    {
+      "epoch": 0.029894170273120704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992727272727273e-05,
+      "loss": 1.2265,
+      "step": 822
+    },
+    {
+      "epoch": 0.029930537876859294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951515151515154e-05,
+      "loss": 1.174,
+      "step": 823
+    },
+    {
+      "epoch": 0.029966905480597883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975757575757576e-05,
+      "loss": 1.2844,
+      "step": 824
+    },
+    {
+      "epoch": 0.030003273084336472,
+      "grad_norm": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.2995,
+      "step": 825
+    },
+    {
+      "epoch": 0.03003964068807506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999999930632096e-05,
+      "loss": 1.3342,
+      "step": 826
+    },
+    {
+      "epoch": 0.03007600829181365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999999972252838e-05,
+      "loss": 1.2912,
+      "step": 827
+    },
+    {
+      "epoch": 0.030112375895552244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999999375688855e-05,
+      "loss": 1.3039,
+      "step": 828
+    },
+    {
+      "epoch": 0.030148743499290833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999998890113527e-05,
+      "loss": 1.1496,
+      "step": 829
+    },
+    {
+      "epoch": 0.030185111103029422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999998265802404e-05,
+      "loss": 1.2339,
+      "step": 830
+    },
+    {
+      "epoch": 0.03022147870676801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999997502755488e-05,
+      "loss": 1.3729,
+      "step": 831
+    },
+    {
+      "epoch": 0.0302578463105066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99999966009728e-05,
+      "loss": 1.1852,
+      "step": 832
+    },
+    {
+      "epoch": 0.03029421391424519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999995560454347e-05,
+      "loss": 1.4029,
+      "step": 833
+    },
+    {
+      "epoch": 0.03033058151798378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999999438120014e-05,
+      "loss": 1.1796,
+      "step": 834
+    },
+    {
+      "epoch": 0.03036694912172237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999993063210202e-05,
+      "loss": 1.2119,
+      "step": 835
+    },
+    {
+      "epoch": 0.03040331672546096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999999160648455e-05,
+      "loss": 1.3304,
+      "step": 836
+    },
+    {
+      "epoch": 0.030439684329199548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99999900110232e-05,
+      "loss": 1.2615,
+      "step": 837
+    },
+    {
+      "epoch": 0.030476051932938137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999988276826175e-05,
+      "loss": 1.2672,
+      "step": 838
+    },
+    {
+      "epoch": 0.03051241953667673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999986403893505e-05,
+      "loss": 1.2736,
+      "step": 839
+    },
+    {
+      "epoch": 0.03054878714041532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999984392225208e-05,
+      "loss": 1.2761,
+      "step": 840
+    },
+    {
+      "epoch": 0.03058515474415391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999998224182132e-05,
+      "loss": 1.3382,
+      "step": 841
+    },
+    {
+      "epoch": 0.030621522347892498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999979952681863e-05,
+      "loss": 1.2683,
+      "step": 842
+    },
+    {
+      "epoch": 0.030657889951631087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999977524806874e-05,
+      "loss": 1.3506,
+      "step": 843
+    },
+    {
+      "epoch": 0.030694257555369676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999974958196383e-05,
+      "loss": 1.1892,
+      "step": 844
+    },
+    {
+      "epoch": 0.030730625159108266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999997225285043e-05,
+      "loss": 1.3037,
+      "step": 845
+    },
+    {
+      "epoch": 0.030766992762846855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999969408769047e-05,
+      "loss": 1.3737,
+      "step": 846
+    },
+    {
+      "epoch": 0.030803360366585444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999996642595228e-05,
+      "loss": 1.3668,
+      "step": 847
+    },
+    {
+      "epoch": 0.030839727970324034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999996330440017e-05,
+      "loss": 1.2431,
+      "step": 848
+    },
+    {
+      "epoch": 0.030876095574062627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999996004411275e-05,
+      "loss": 1.3718,
+      "step": 849
+    },
+    {
+      "epoch": 0.030912463177801216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999956645090072e-05,
+      "loss": 1.3691,
+      "step": 850
+    },
+    {
+      "epoch": 0.030948830781539805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999995310733219e-05,
+      "loss": 1.2713,
+      "step": 851
+    },
+    {
+      "epoch": 0.030985198385278395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999949430839143e-05,
+      "loss": 1.2247,
+      "step": 852
+    },
+    {
+      "epoch": 0.031021565989016984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999945615610987e-05,
+      "loss": 1.3676,
+      "step": 853
+    },
+    {
+      "epoch": 0.031057933592755573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999941661647773e-05,
+      "loss": 1.2723,
+      "step": 854
+    },
+    {
+      "epoch": 0.031094301196494162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999937568949557e-05,
+      "loss": 1.2984,
+      "step": 855
+    },
+    {
+      "epoch": 0.031130668800232752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999933337516393e-05,
+      "loss": 1.3194,
+      "step": 856
+    },
+    {
+      "epoch": 0.03116703640397134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999928967348344e-05,
+      "loss": 1.2365,
+      "step": 857
+    },
+    {
+      "epoch": 0.03120340400770993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999924458445467e-05,
+      "loss": 1.2453,
+      "step": 858
+    },
+    {
+      "epoch": 0.031239771611448523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999919810807828e-05,
+      "loss": 1.299,
+      "step": 859
+    },
+    {
+      "epoch": 0.03127613921518711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999991502443549e-05,
+      "loss": 1.2195,
+      "step": 860
+    },
+    {
+      "epoch": 0.0313125068189257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999910099328517e-05,
+      "loss": 1.2017,
+      "step": 861
+    },
+    {
+      "epoch": 0.03134887442266429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999905035486982e-05,
+      "loss": 1.2121,
+      "step": 862
+    },
+    {
+      "epoch": 0.03138524202640288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999899832910952e-05,
+      "loss": 1.3163,
+      "step": 863
+    },
+    {
+      "epoch": 0.03142160963014147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99998944916005e-05,
+      "loss": 1.1982,
+      "step": 864
+    },
+    {
+      "epoch": 0.03145797723388006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999889011555696e-05,
+      "loss": 1.2965,
+      "step": 865
+    },
+    {
+      "epoch": 0.03149434483761865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999883392776623e-05,
+      "loss": 1.3543,
+      "step": 866
+    },
+    {
+      "epoch": 0.03153071244135724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999877635263355e-05,
+      "loss": 1.2632,
+      "step": 867
+    },
+    {
+      "epoch": 0.03156708004509583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999871739015974e-05,
+      "loss": 1.235,
+      "step": 868
+    },
+    {
+      "epoch": 0.03160344764883442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999986570403456e-05,
+      "loss": 1.2428,
+      "step": 869
+    },
+    {
+      "epoch": 0.031639815252573006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99998595303192e-05,
+      "loss": 1.2955,
+      "step": 870
+    },
+    {
+      "epoch": 0.031676182856311595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999853217869972e-05,
+      "loss": 1.2055,
+      "step": 871
+    },
+    {
+      "epoch": 0.031712550460050185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999846766686973e-05,
+      "loss": 1.3582,
+      "step": 872
+    },
+    {
+      "epoch": 0.031748918063788774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999840176770285e-05,
+      "loss": 1.3942,
+      "step": 873
+    },
+    {
+      "epoch": 0.03178528566752736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999833448120004e-05,
+      "loss": 1.2592,
+      "step": 874
+    },
+    {
+      "epoch": 0.03182165327126596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999982658073622e-05,
+      "loss": 1.2577,
+      "step": 875
+    },
+    {
+      "epoch": 0.03185802087500455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999819574619033e-05,
+      "loss": 1.2155,
+      "step": 876
+    },
+    {
+      "epoch": 0.03189438847874314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999812429768536e-05,
+      "loss": 1.3446,
+      "step": 877
+    },
+    {
+      "epoch": 0.03193075608248173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999980514618483e-05,
+      "loss": 1.2977,
+      "step": 878
+    },
+    {
+      "epoch": 0.03196712368622032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999797723868014e-05,
+      "loss": 1.3439,
+      "step": 879
+    },
+    {
+      "epoch": 0.032003491289958906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999790162818195e-05,
+      "loss": 1.2796,
+      "step": 880
+    },
+    {
+      "epoch": 0.032039858893697495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999782463035472e-05,
+      "loss": 1.1975,
+      "step": 881
+    },
+    {
+      "epoch": 0.032076226497436085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999774624519953e-05,
+      "loss": 1.3485,
+      "step": 882
+    },
+    {
+      "epoch": 0.032112594101174674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999766647271755e-05,
+      "loss": 1.302,
+      "step": 883
+    },
+    {
+      "epoch": 0.03214896170491326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999758531290977e-05,
+      "loss": 1.2125,
+      "step": 884
+    },
+    {
+      "epoch": 0.03218532930865185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999750276577743e-05,
+      "loss": 1.291,
+      "step": 885
+    },
+    {
+      "epoch": 0.03222169691239044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999741883132156e-05,
+      "loss": 1.3207,
+      "step": 886
+    },
+    {
+      "epoch": 0.03225806451612903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999733350954344e-05,
+      "loss": 1.2903,
+      "step": 887
+    },
+    {
+      "epoch": 0.03229443211986762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999724680044417e-05,
+      "loss": 1.1606,
+      "step": 888
+    },
+    {
+      "epoch": 0.03233079972360621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999715870402497e-05,
+      "loss": 1.3877,
+      "step": 889
+    },
+    {
+      "epoch": 0.0323671673273448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999706922028706e-05,
+      "loss": 1.3629,
+      "step": 890
+    },
+    {
+      "epoch": 0.03240353493108339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999969783492317e-05,
+      "loss": 1.252,
+      "step": 891
+    },
+    {
+      "epoch": 0.03243990253482198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999688609086015e-05,
+      "loss": 1.2674,
+      "step": 892
+    },
+    {
+      "epoch": 0.03247627013856057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999967924451737e-05,
+      "loss": 1.4077,
+      "step": 893
+    },
+    {
+      "epoch": 0.03251263774229916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999966974121736e-05,
+      "loss": 1.3108,
+      "step": 894
+    },
+    {
+      "epoch": 0.03254900534603775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999660099186123e-05,
+      "loss": 1.262,
+      "step": 895
+    },
+    {
+      "epoch": 0.03258537294977634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999650318423786e-05,
+      "loss": 1.2695,
+      "step": 896
+    },
+    {
+      "epoch": 0.03262174055351493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999640398930492e-05,
+      "loss": 1.4779,
+      "step": 897
+    },
+    {
+      "epoch": 0.03265810815725352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999963034070637e-05,
+      "loss": 1.3058,
+      "step": 898
+    },
+    {
+      "epoch": 0.03269447576099211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999620143751573e-05,
+      "loss": 1.22,
+      "step": 899
+    },
+    {
+      "epoch": 0.0327308433647307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999960980806623e-05,
+      "loss": 1.3358,
+      "step": 900
+    },
+    {
+      "epoch": 0.03276721096846929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999599333650487e-05,
+      "loss": 1.3583,
+      "step": 901
+    },
+    {
+      "epoch": 0.03280357857220788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999958872050449e-05,
+      "loss": 1.2688,
+      "step": 902
+    },
+    {
+      "epoch": 0.03283994617594647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999957796862839e-05,
+      "loss": 1.3054,
+      "step": 903
+    },
+    {
+      "epoch": 0.03287631377968506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999956707802233e-05,
+      "loss": 1.4522,
+      "step": 904
+    },
+    {
+      "epoch": 0.032912681383423646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999556048686468e-05,
+      "loss": 1.1689,
+      "step": 905
+    },
+    {
+      "epoch": 0.032949048987162236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999544880620952e-05,
+      "loss": 1.3289,
+      "step": 906
+    },
+    {
+      "epoch": 0.032985416590900825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999533573825937e-05,
+      "loss": 1.1806,
+      "step": 907
+    },
+    {
+      "epoch": 0.033021784194639414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999522128301586e-05,
+      "loss": 1.3126,
+      "step": 908
+    },
+    {
+      "epoch": 0.033058151798378003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999510544048047e-05,
+      "loss": 1.2891,
+      "step": 909
+    },
+    {
+      "epoch": 0.03309451940211659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999498821065487e-05,
+      "loss": 1.2663,
+      "step": 910
+    },
+    {
+      "epoch": 0.03313088700585518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999486959354072e-05,
+      "loss": 1.2873,
+      "step": 911
+    },
+    {
+      "epoch": 0.03316725460959377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999947495891396e-05,
+      "loss": 1.1666,
+      "step": 912
+    },
+    {
+      "epoch": 0.03320362221333236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999946281974532e-05,
+      "loss": 1.3486,
+      "step": 913
+    },
+    {
+      "epoch": 0.03323998981707095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999450541848322e-05,
+      "loss": 1.3246,
+      "step": 914
+    },
+    {
+      "epoch": 0.03327635742080954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999438125223136e-05,
+      "loss": 1.285,
+      "step": 915
+    },
+    {
+      "epoch": 0.033312725024548136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999425569869932e-05,
+      "loss": 1.3323,
+      "step": 916
+    },
+    {
+      "epoch": 0.033349092628286725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999412875788886e-05,
+      "loss": 1.3556,
+      "step": 917
+    },
+    {
+      "epoch": 0.033385460232025314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999400042980173e-05,
+      "loss": 1.2673,
+      "step": 918
+    },
+    {
+      "epoch": 0.033421827835763904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999387071443975e-05,
+      "loss": 1.4058,
+      "step": 919
+    },
+    {
+      "epoch": 0.03345819543950249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999373961180466e-05,
+      "loss": 1.281,
+      "step": 920
+    },
+    {
+      "epoch": 0.03349456304324108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999936071218983e-05,
+      "loss": 1.4579,
+      "step": 921
+    },
+    {
+      "epoch": 0.03353093064697967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999347324472252e-05,
+      "loss": 1.3358,
+      "step": 922
+    },
+    {
+      "epoch": 0.03356729825071826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999333798027914e-05,
+      "loss": 1.238,
+      "step": 923
+    },
+    {
+      "epoch": 0.03360366585445685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999932013285701e-05,
+      "loss": 1.2629,
+      "step": 924
+    },
+    {
+      "epoch": 0.03364003345819544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999306328959726e-05,
+      "loss": 1.3899,
+      "step": 925
+    },
+    {
+      "epoch": 0.03367640106193403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999929238633626e-05,
+      "loss": 1.2133,
+      "step": 926
+    },
+    {
+      "epoch": 0.03371276866567262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999927830498679e-05,
+      "loss": 1.3716,
+      "step": 927
+    },
+    {
+      "epoch": 0.03374913626941121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999264084911524e-05,
+      "loss": 1.1871,
+      "step": 928
+    },
+    {
+      "epoch": 0.0337855038731498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999249726110653e-05,
+      "loss": 1.3651,
+      "step": 929
+    },
+    {
+      "epoch": 0.033821871476888386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999235228584383e-05,
+      "loss": 1.1614,
+      "step": 930
+    },
+    {
+      "epoch": 0.033858239080626976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999922059233291e-05,
+      "loss": 1.255,
+      "step": 931
+    },
+    {
+      "epoch": 0.033894606684365565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999920581735644e-05,
+      "loss": 1.2641,
+      "step": 932
+    },
+    {
+      "epoch": 0.033930974288104154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999190903655175e-05,
+      "loss": 1.2383,
+      "step": 933
+    },
+    {
+      "epoch": 0.033967341891842744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999175851229323e-05,
+      "loss": 1.3434,
+      "step": 934
+    },
+    {
+      "epoch": 0.03400370949558133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999916066007909e-05,
+      "loss": 1.2653,
+      "step": 935
+    },
+    {
+      "epoch": 0.03404007709931993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999145330204694e-05,
+      "loss": 1.3231,
+      "step": 936
+    },
+    {
+      "epoch": 0.03407644470305852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999129861606345e-05,
+      "loss": 1.2404,
+      "step": 937
+    },
+    {
+      "epoch": 0.03411281230679711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999114254284252e-05,
+      "loss": 1.2287,
+      "step": 938
+    },
+    {
+      "epoch": 0.0341491799105357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999098508238638e-05,
+      "loss": 1.2379,
+      "step": 939
+    },
+    {
+      "epoch": 0.034185547514274287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999082623469718e-05,
+      "loss": 1.131,
+      "step": 940
+    },
+    {
+      "epoch": 0.034221915118012876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999066599977712e-05,
+      "loss": 1.3317,
+      "step": 941
+    },
+    {
+      "epoch": 0.034258282721751465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999050437762846e-05,
+      "loss": 1.3358,
+      "step": 942
+    },
+    {
+      "epoch": 0.034294650325490054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999903413682534e-05,
+      "loss": 1.2328,
+      "step": 943
+    },
+    {
+      "epoch": 0.034331017929228644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999017697165425e-05,
+      "loss": 1.3242,
+      "step": 944
+    },
+    {
+      "epoch": 0.03436738553296723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9999001118783322e-05,
+      "loss": 1.1305,
+      "step": 945
+    },
+    {
+      "epoch": 0.03440375313670582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999898440167927e-05,
+      "loss": 1.3146,
+      "step": 946
+    },
+    {
+      "epoch": 0.03444012074044441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998967545853492e-05,
+      "loss": 1.3064,
+      "step": 947
+    },
+    {
+      "epoch": 0.034476488344183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999895055130623e-05,
+      "loss": 1.3512,
+      "step": 948
+    },
+    {
+      "epoch": 0.03451285594792159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998933418037715e-05,
+      "loss": 1.2975,
+      "step": 949
+    },
+    {
+      "epoch": 0.03454922355166018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998916146048182e-05,
+      "loss": 1.3275,
+      "step": 950
+    },
+    {
+      "epoch": 0.03458559115539877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998898735337877e-05,
+      "loss": 1.4317,
+      "step": 951
+    },
+    {
+      "epoch": 0.03462195875913736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998881185907036e-05,
+      "loss": 1.2847,
+      "step": 952
+    },
+    {
+      "epoch": 0.03465832636287595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998863497755908e-05,
+      "loss": 1.2355,
+      "step": 953
+    },
+    {
+      "epoch": 0.03469469396661454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998845670884733e-05,
+      "loss": 1.257,
+      "step": 954
+    },
+    {
+      "epoch": 0.034731061570353126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998827705293764e-05,
+      "loss": 1.1496,
+      "step": 955
+    },
+    {
+      "epoch": 0.034767429174091716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998809600983246e-05,
+      "loss": 1.2953,
+      "step": 956
+    },
+    {
+      "epoch": 0.03480379677783031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998791357953428e-05,
+      "loss": 1.2606,
+      "step": 957
+    },
+    {
+      "epoch": 0.0348401643815689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999877297620457e-05,
+      "loss": 1.1718,
+      "step": 958
+    },
+    {
+      "epoch": 0.03487653198530749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998754455736922e-05,
+      "loss": 1.2664,
+      "step": 959
+    },
+    {
+      "epoch": 0.03491289958904608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998735796550744e-05,
+      "loss": 1.154,
+      "step": 960
+    },
+    {
+      "epoch": 0.03494926719278467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999871699864629e-05,
+      "loss": 1.3612,
+      "step": 961
+    },
+    {
+      "epoch": 0.03498563479652326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999869806202383e-05,
+      "loss": 1.3159,
+      "step": 962
+    },
+    {
+      "epoch": 0.03502200240026185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998678986683613e-05,
+      "loss": 1.261,
+      "step": 963
+    },
+    {
+      "epoch": 0.03505837000400044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998659772625914e-05,
+      "loss": 1.2426,
+      "step": 964
+    },
+    {
+      "epoch": 0.03509473760773903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998640419850998e-05,
+      "loss": 1.3936,
+      "step": 965
+    },
+    {
+      "epoch": 0.035131105211477616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998620928359132e-05,
+      "loss": 1.1825,
+      "step": 966
+    },
+    {
+      "epoch": 0.035167472815216205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998601298150588e-05,
+      "loss": 1.2384,
+      "step": 967
+    },
+    {
+      "epoch": 0.035203840418954795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998581529225633e-05,
+      "loss": 1.2039,
+      "step": 968
+    },
+    {
+      "epoch": 0.035240208022693384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999856162158455e-05,
+      "loss": 1.1518,
+      "step": 969
+    },
+    {
+      "epoch": 0.03527657562643197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998541575227607e-05,
+      "loss": 1.198,
+      "step": 970
+    },
+    {
+      "epoch": 0.03531294323017056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998521390155087e-05,
+      "loss": 1.3069,
+      "step": 971
+    },
+    {
+      "epoch": 0.03534931083390915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998501066367272e-05,
+      "loss": 1.2131,
+      "step": 972
+    },
+    {
+      "epoch": 0.03538567843764774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998480603864435e-05,
+      "loss": 1.231,
+      "step": 973
+    },
+    {
+      "epoch": 0.03542204604138633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998460002646872e-05,
+      "loss": 1.2413,
+      "step": 974
+    },
+    {
+      "epoch": 0.03545841364512492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998439262714856e-05,
+      "loss": 1.2611,
+      "step": 975
+    },
+    {
+      "epoch": 0.03549478124886351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998418384068686e-05,
+      "loss": 1.2872,
+      "step": 976
+    },
+    {
+      "epoch": 0.035531148852602105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998397366708647e-05,
+      "loss": 1.4122,
+      "step": 977
+    },
+    {
+      "epoch": 0.035567516456340695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999837621063503e-05,
+      "loss": 1.2941,
+      "step": 978
+    },
+    {
+      "epoch": 0.035603884060079284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998354915848125e-05,
+      "loss": 1.2728,
+      "step": 979
+    },
+    {
+      "epoch": 0.03564025166381787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998333482348233e-05,
+      "loss": 1.3126,
+      "step": 980
+    },
+    {
+      "epoch": 0.03567661926755646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999831191013565e-05,
+      "loss": 1.3036,
+      "step": 981
+    },
+    {
+      "epoch": 0.03571298687129505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998290199210678e-05,
+      "loss": 1.179,
+      "step": 982
+    },
+    {
+      "epoch": 0.03574935447503364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999826834957361e-05,
+      "loss": 1.1888,
+      "step": 983
+    },
+    {
+      "epoch": 0.03578572207877223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999824636122476e-05,
+      "loss": 1.2465,
+      "step": 984
+    },
+    {
+      "epoch": 0.03582208968251082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998224234164423e-05,
+      "loss": 1.1903,
+      "step": 985
+    },
+    {
+      "epoch": 0.03585845728624941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999820196839291e-05,
+      "loss": 1.2213,
+      "step": 986
+    },
+    {
+      "epoch": 0.035894824889988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998179563910535e-05,
+      "loss": 1.2347,
+      "step": 987
+    },
+    {
+      "epoch": 0.03593119249372659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99981570207176e-05,
+      "loss": 1.3309,
+      "step": 988
+    },
+    {
+      "epoch": 0.03596756009746518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998134338814423e-05,
+      "loss": 1.3048,
+      "step": 989
+    },
+    {
+      "epoch": 0.03600392770120377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998111518201317e-05,
+      "loss": 1.3403,
+      "step": 990
+    },
+    {
+      "epoch": 0.036040295304942356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99980885588786e-05,
+      "loss": 1.1875,
+      "step": 991
+    },
+    {
+      "epoch": 0.036076662908680945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999806546084659e-05,
+      "loss": 1.2372,
+      "step": 992
+    },
+    {
+      "epoch": 0.036113030512419535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998042224105608e-05,
+      "loss": 1.2509,
+      "step": 993
+    },
+    {
+      "epoch": 0.036149398116158124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9998018848655974e-05,
+      "loss": 1.4116,
+      "step": 994
+    },
+    {
+      "epoch": 0.03618576571989671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997995334498014e-05,
+      "loss": 1.1471,
+      "step": 995
+    },
+    {
+      "epoch": 0.0362221333236353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997971681632052e-05,
+      "loss": 1.198,
+      "step": 996
+    },
+    {
+      "epoch": 0.03625850092737389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997947890058422e-05,
+      "loss": 1.3519,
+      "step": 997
+    },
+    {
+      "epoch": 0.03629486853111249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997923959777447e-05,
+      "loss": 1.0715,
+      "step": 998
+    },
+    {
+      "epoch": 0.03633123613485108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997899890789464e-05,
+      "loss": 1.3289,
+      "step": 999
+    },
+    {
+      "epoch": 0.03636760373858967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997875683094805e-05,
+      "loss": 1.183,
+      "step": 1000
+    },
+    {
+      "epoch": 0.036403971342328256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997851336693804e-05,
+      "loss": 1.2652,
+      "step": 1001
+    },
+    {
+      "epoch": 0.036440338946066846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997826851586804e-05,
+      "loss": 1.2173,
+      "step": 1002
+    },
+    {
+      "epoch": 0.036476706549805435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997802227774138e-05,
+      "loss": 1.2531,
+      "step": 1003
+    },
+    {
+      "epoch": 0.036513074153544024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997777465256157e-05,
+      "loss": 1.2697,
+      "step": 1004
+    },
+    {
+      "epoch": 0.036549441757282614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997752564033194e-05,
+      "loss": 1.3204,
+      "step": 1005
+    },
+    {
+      "epoch": 0.0365858093610212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997727524105597e-05,
+      "loss": 1.3837,
+      "step": 1006
+    },
+    {
+      "epoch": 0.03662217696475979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997702345473723e-05,
+      "loss": 1.1874,
+      "step": 1007
+    },
+    {
+      "epoch": 0.03665854456849838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997677028137906e-05,
+      "loss": 1.2067,
+      "step": 1008
+    },
+    {
+      "epoch": 0.03669491217223697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999765157209851e-05,
+      "loss": 1.3507,
+      "step": 1009
+    },
+    {
+      "epoch": 0.03673127977597556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997625977355883e-05,
+      "loss": 1.2908,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03676764737971415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997600243910377e-05,
+      "loss": 1.259,
+      "step": 1011
+    },
+    {
+      "epoch": 0.03680401498345274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997574371762356e-05,
+      "loss": 1.2076,
+      "step": 1012
+    },
+    {
+      "epoch": 0.03684038258719133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997548360912174e-05,
+      "loss": 1.2952,
+      "step": 1013
+    },
+    {
+      "epoch": 0.03687675019092992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997522211360192e-05,
+      "loss": 1.1971,
+      "step": 1014
+    },
+    {
+      "epoch": 0.03691311779466851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997495923106775e-05,
+      "loss": 1.2614,
+      "step": 1015
+    },
+    {
+      "epoch": 0.036949485398407096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997469496152288e-05,
+      "loss": 1.2285,
+      "step": 1016
+    },
+    {
+      "epoch": 0.036985853002145685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999744293049709e-05,
+      "loss": 1.2527,
+      "step": 1017
+    },
+    {
+      "epoch": 0.037022220605884275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997416226141564e-05,
+      "loss": 1.2674,
+      "step": 1018
+    },
+    {
+      "epoch": 0.03705858820962287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999738938308607e-05,
+      "loss": 1.3461,
+      "step": 1019
+    },
+    {
+      "epoch": 0.03709495581336146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999736240133098e-05,
+      "loss": 1.2666,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03713132341710005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997335280876676e-05,
+      "loss": 1.2638,
+      "step": 1021
+    },
+    {
+      "epoch": 0.03716769102083864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997308021723523e-05,
+      "loss": 1.2614,
+      "step": 1022
+    },
+    {
+      "epoch": 0.03720405862457723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999728062387191e-05,
+      "loss": 1.201,
+      "step": 1023
+    },
+    {
+      "epoch": 0.03724042622831582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999725308732221e-05,
+      "loss": 1.2436,
+      "step": 1024
+    },
+    {
+      "epoch": 0.03727679383205441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999722541207481e-05,
+      "loss": 1.2739,
+      "step": 1025
+    },
+    {
+      "epoch": 0.037313161435792996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999719759813009e-05,
+      "loss": 1.2372,
+      "step": 1026
+    },
+    {
+      "epoch": 0.037349529039531586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997169645488438e-05,
+      "loss": 1.3914,
+      "step": 1027
+    },
+    {
+      "epoch": 0.037385896643270175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997141554150243e-05,
+      "loss": 1.2283,
+      "step": 1028
+    },
+    {
+      "epoch": 0.037422264247008764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999711332411589e-05,
+      "loss": 1.1849,
+      "step": 1029
+    },
+    {
+      "epoch": 0.037458631850747354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997084955385776e-05,
+      "loss": 1.3353,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03749499945448594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999705644796029e-05,
+      "loss": 1.2847,
+      "step": 1031
+    },
+    {
+      "epoch": 0.03753136705822453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9997027801839832e-05,
+      "loss": 1.264,
+      "step": 1032
+    },
+    {
+      "epoch": 0.03756773466196312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996999017024795e-05,
+      "loss": 1.3571,
+      "step": 1033
+    },
+    {
+      "epoch": 0.03760410226570171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999697009351558e-05,
+      "loss": 1.3717,
+      "step": 1034
+    },
+    {
+      "epoch": 0.0376404698694403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996941031312593e-05,
+      "loss": 1.2193,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03767683747317889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996911830416233e-05,
+      "loss": 1.131,
+      "step": 1036
+    },
+    {
+      "epoch": 0.03771320507691748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99968824908269e-05,
+      "loss": 1.2686,
+      "step": 1037
+    },
+    {
+      "epoch": 0.03774957268065607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999685301254501e-05,
+      "loss": 1.3186,
+      "step": 1038
+    },
+    {
+      "epoch": 0.037785940284394665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996823395570967e-05,
+      "loss": 1.2035,
+      "step": 1039
+    },
+    {
+      "epoch": 0.037822307888133254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996793639905184e-05,
+      "loss": 1.1081,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03785867549187184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999676374554807e-05,
+      "loss": 1.3172,
+      "step": 1041
+    },
+    {
+      "epoch": 0.03789504309561043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996733712500044e-05,
+      "loss": 1.1987,
+      "step": 1042
+    },
+    {
+      "epoch": 0.03793141069934902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996703540761523e-05,
+      "loss": 1.2234,
+      "step": 1043
+    },
+    {
+      "epoch": 0.03796777830308761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996673230332925e-05,
+      "loss": 1.224,
+      "step": 1044
+    },
+    {
+      "epoch": 0.0380041459068262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996642781214665e-05,
+      "loss": 1.2393,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03804051351056479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999661219340717e-05,
+      "loss": 1.3171,
+      "step": 1046
+    },
+    {
+      "epoch": 0.03807688111430338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999658146691087e-05,
+      "loss": 1.2401,
+      "step": 1047
+    },
+    {
+      "epoch": 0.03811324871804197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999655060172618e-05,
+      "loss": 1.3371,
+      "step": 1048
+    },
+    {
+      "epoch": 0.03814961632178056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996519597853533e-05,
+      "loss": 1.2261,
+      "step": 1049
+    },
+    {
+      "epoch": 0.03818598392551915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999648845529336e-05,
+      "loss": 1.2706,
+      "step": 1050
+    },
+    {
+      "epoch": 0.038222351529257736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996457174046097e-05,
+      "loss": 1.3964,
+      "step": 1051
+    },
+    {
+      "epoch": 0.038258719132996326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999642575411217e-05,
+      "loss": 1.2809,
+      "step": 1052
+    },
+    {
+      "epoch": 0.038295086736734915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999639419549202e-05,
+      "loss": 1.3287,
+      "step": 1053
+    },
+    {
+      "epoch": 0.038331454340473504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996362498186084e-05,
+      "loss": 1.2507,
+      "step": 1054
+    },
+    {
+      "epoch": 0.038367821944212094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996330662194802e-05,
+      "loss": 1.1802,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03840418954795068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996298687518613e-05,
+      "loss": 1.2247,
+      "step": 1056
+    },
+    {
+      "epoch": 0.03844055715168927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996266574157964e-05,
+      "loss": 1.3852,
+      "step": 1057
+    },
+    {
+      "epoch": 0.03847692475542786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99962343221133e-05,
+      "loss": 1.2747,
+      "step": 1058
+    },
+    {
+      "epoch": 0.03851329235916645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996201931385065e-05,
+      "loss": 1.2171,
+      "step": 1059
+    },
+    {
+      "epoch": 0.03854965996290505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996169401973714e-05,
+      "loss": 1.1225,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03858602756664364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999613673387969e-05,
+      "loss": 1.3381,
+      "step": 1061
+    },
+    {
+      "epoch": 0.038622395170382226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996103927103458e-05,
+      "loss": 1.269,
+      "step": 1062
+    },
+    {
+      "epoch": 0.038658762774120815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996070981645464e-05,
+      "loss": 1.1921,
+      "step": 1063
+    },
+    {
+      "epoch": 0.038695130377859405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996037897506168e-05,
+      "loss": 1.2224,
+      "step": 1064
+    },
+    {
+      "epoch": 0.038731497981597994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9996004674686026e-05,
+      "loss": 1.3069,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03876786558533658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995971313185506e-05,
+      "loss": 1.2673,
+      "step": 1066
+    },
+    {
+      "epoch": 0.03880423318907517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995937813005063e-05,
+      "loss": 1.2269,
+      "step": 1067
+    },
+    {
+      "epoch": 0.03884060079281376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995904174145163e-05,
+      "loss": 1.1976,
+      "step": 1068
+    },
+    {
+      "epoch": 0.03887696839655235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999587039660628e-05,
+      "loss": 1.2209,
+      "step": 1069
+    },
+    {
+      "epoch": 0.03891333600029094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995836480388876e-05,
+      "loss": 1.3664,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03894970360402953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995802425493426e-05,
+      "loss": 1.2996,
+      "step": 1071
+    },
+    {
+      "epoch": 0.03898607120776812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995768231920394e-05,
+      "loss": 1.37,
+      "step": 1072
+    },
+    {
+      "epoch": 0.03902243881150671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999573389967026e-05,
+      "loss": 1.2412,
+      "step": 1073
+    },
+    {
+      "epoch": 0.0390588064152453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995699428743503e-05,
+      "loss": 1.148,
+      "step": 1074
+    },
+    {
+      "epoch": 0.03909517401898389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99956648191406e-05,
+      "loss": 1.3099,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03913154162272248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995630070862025e-05,
+      "loss": 1.3552,
+      "step": 1076
+    },
+    {
+      "epoch": 0.039167909226461066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999559518390827e-05,
+      "loss": 1.277,
+      "step": 1077
+    },
+    {
+      "epoch": 0.039204276830199655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995560158279808e-05,
+      "loss": 1.2738,
+      "step": 1078
+    },
+    {
+      "epoch": 0.039240644433938245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995524993977135e-05,
+      "loss": 1.1999,
+      "step": 1079
+    },
+    {
+      "epoch": 0.03927701203767684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995489691000735e-05,
+      "loss": 1.1692,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03931337964141543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995454249351093e-05,
+      "loss": 1.2577,
+      "step": 1081
+    },
+    {
+      "epoch": 0.03934974724515402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999541866902871e-05,
+      "loss": 1.2474,
+      "step": 1082
+    },
+    {
+      "epoch": 0.03938611484889261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995382950034074e-05,
+      "loss": 1.264,
+      "step": 1083
+    },
+    {
+      "epoch": 0.0394224824526312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999534709236768e-05,
+      "loss": 1.3248,
+      "step": 1084
+    },
+    {
+      "epoch": 0.03945885005636979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995311096030024e-05,
+      "loss": 1.3386,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03949521766010838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999527496102161e-05,
+      "loss": 1.3646,
+      "step": 1086
+    },
+    {
+      "epoch": 0.039531585263846966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995238687342936e-05,
+      "loss": 1.2939,
+      "step": 1087
+    },
+    {
+      "epoch": 0.039567952867585555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995202274994508e-05,
+      "loss": 1.2758,
+      "step": 1088
+    },
+    {
+      "epoch": 0.039604320471324145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999516572397683e-05,
+      "loss": 1.4004,
+      "step": 1089
+    },
+    {
+      "epoch": 0.039640688075062734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995129034290408e-05,
+      "loss": 1.2296,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03967705567880132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995092205935752e-05,
+      "loss": 1.2576,
+      "step": 1091
+    },
+    {
+      "epoch": 0.03971342328253991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995055238913376e-05,
+      "loss": 1.1848,
+      "step": 1092
+    },
+    {
+      "epoch": 0.0397497908862785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9995018133223785e-05,
+      "loss": 1.2189,
+      "step": 1093
+    },
+    {
+      "epoch": 0.03978615849001709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99949808888675e-05,
+      "loss": 1.2295,
+      "step": 1094
+    },
+    {
+      "epoch": 0.03982252609375568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999494350584504e-05,
+      "loss": 1.234,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03985889369749427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994905984156913e-05,
+      "loss": 1.4423,
+      "step": 1096
+    },
+    {
+      "epoch": 0.03989526130123286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999486832380365e-05,
+      "loss": 1.2245,
+      "step": 1097
+    },
+    {
+      "epoch": 0.03993162890497145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999483052478577e-05,
+      "loss": 1.0827,
+      "step": 1098
+    },
+    {
+      "epoch": 0.03996799650871004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994792587103796e-05,
+      "loss": 1.2853,
+      "step": 1099
+    },
+    {
+      "epoch": 0.04000436411244863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994754510758256e-05,
+      "loss": 1.3356,
+      "step": 1100
+    },
+    {
+      "epoch": 0.040040731716187224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994716295749682e-05,
+      "loss": 1.2626,
+      "step": 1101
+    },
+    {
+      "epoch": 0.04007709931992581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994677942078593e-05,
+      "loss": 1.3022,
+      "step": 1102
+    },
+    {
+      "epoch": 0.0401134669236644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994639449745533e-05,
+      "loss": 1.1587,
+      "step": 1103
+    },
+    {
+      "epoch": 0.04014983452740299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999460081875103e-05,
+      "loss": 1.2281,
+      "step": 1104
+    },
+    {
+      "epoch": 0.04018620213114158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999456204909562e-05,
+      "loss": 1.1468,
+      "step": 1105
+    },
+    {
+      "epoch": 0.04022256973488017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994523140779843e-05,
+      "loss": 1.2212,
+      "step": 1106
+    },
+    {
+      "epoch": 0.04025893733861876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999448409380424e-05,
+      "loss": 1.286,
+      "step": 1107
+    },
+    {
+      "epoch": 0.04029530494235735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999444490816935e-05,
+      "loss": 1.2147,
+      "step": 1108
+    },
+    {
+      "epoch": 0.04033167254609594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994405583875714e-05,
+      "loss": 1.2578,
+      "step": 1109
+    },
+    {
+      "epoch": 0.04036804014983453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999436612092388e-05,
+      "loss": 1.2332,
+      "step": 1110
+    },
+    {
+      "epoch": 0.04040440775357312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994326519314396e-05,
+      "loss": 1.2883,
+      "step": 1111
+    },
+    {
+      "epoch": 0.040440775357311706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994286779047817e-05,
+      "loss": 1.2897,
+      "step": 1112
+    },
+    {
+      "epoch": 0.040477142961050296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994246900124686e-05,
+      "loss": 1.1298,
+      "step": 1113
+    },
+    {
+      "epoch": 0.040513510564788885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994206882545558e-05,
+      "loss": 1.1218,
+      "step": 1114
+    },
+    {
+      "epoch": 0.040549878168527474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994166726310992e-05,
+      "loss": 1.1562,
+      "step": 1115
+    },
+    {
+      "epoch": 0.040586245772266064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999412643142154e-05,
+      "loss": 1.3295,
+      "step": 1116
+    },
+    {
+      "epoch": 0.04062261337600465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994085997877766e-05,
+      "loss": 1.2051,
+      "step": 1117
+    },
+    {
+      "epoch": 0.04065898097974324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994045425680228e-05,
+      "loss": 1.3306,
+      "step": 1118
+    },
+    {
+      "epoch": 0.04069534858348183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9994004714829488e-05,
+      "loss": 1.304,
+      "step": 1119
+    },
+    {
+      "epoch": 0.04073171618722042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993963865326115e-05,
+      "loss": 1.2303,
+      "step": 1120
+    },
+    {
+      "epoch": 0.04076808379095902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999392287717067e-05,
+      "loss": 1.1799,
+      "step": 1121
+    },
+    {
+      "epoch": 0.040804451394697606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993881750363723e-05,
+      "loss": 1.2543,
+      "step": 1122
+    },
+    {
+      "epoch": 0.040840818998436196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999384048490585e-05,
+      "loss": 1.3766,
+      "step": 1123
+    },
+    {
+      "epoch": 0.040877186602174785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993799080797618e-05,
+      "loss": 1.1833,
+      "step": 1124
+    },
+    {
+      "epoch": 0.040913554205913374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993757538039605e-05,
+      "loss": 1.1502,
+      "step": 1125
+    },
+    {
+      "epoch": 0.040949921809651964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999371585663238e-05,
+      "loss": 1.2507,
+      "step": 1126
+    },
+    {
+      "epoch": 0.04098628941339055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993674036576533e-05,
+      "loss": 1.2197,
+      "step": 1127
+    },
+    {
+      "epoch": 0.04102265701712914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993632077872637e-05,
+      "loss": 1.3263,
+      "step": 1128
+    },
+    {
+      "epoch": 0.04105902462086773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999358998052127e-05,
+      "loss": 1.2342,
+      "step": 1129
+    },
+    {
+      "epoch": 0.04109539222460632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993547744523027e-05,
+      "loss": 1.2592,
+      "step": 1130
+    },
+    {
+      "epoch": 0.04113175982834491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993505369878486e-05,
+      "loss": 1.4176,
+      "step": 1131
+    },
+    {
+      "epoch": 0.0411681274320835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993462856588235e-05,
+      "loss": 1.3713,
+      "step": 1132
+    },
+    {
+      "epoch": 0.04120449503582209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999342020465287e-05,
+      "loss": 1.4033,
+      "step": 1133
+    },
+    {
+      "epoch": 0.04124086263956068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993377414072976e-05,
+      "loss": 1.2792,
+      "step": 1134
+    },
+    {
+      "epoch": 0.04127723024329927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993334484849147e-05,
+      "loss": 1.0812,
+      "step": 1135
+    },
+    {
+      "epoch": 0.04131359784703786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993291416981983e-05,
+      "loss": 1.3799,
+      "step": 1136
+    },
+    {
+      "epoch": 0.041349965450776446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999324821047208e-05,
+      "loss": 1.2953,
+      "step": 1137
+    },
+    {
+      "epoch": 0.041386333054515036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9993204865320034e-05,
+      "loss": 1.2979,
+      "step": 1138
+    },
+    {
+      "epoch": 0.041422700658253625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999316138152645e-05,
+      "loss": 1.3038,
+      "step": 1139
+    },
+    {
+      "epoch": 0.041459068261992214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999311775909193e-05,
+      "loss": 1.3765,
+      "step": 1140
+    },
+    {
+      "epoch": 0.041495435865730804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999307399801708e-05,
+      "loss": 1.269,
+      "step": 1141
+    },
+    {
+      "epoch": 0.0415318034694694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999303009830251e-05,
+      "loss": 1.2519,
+      "step": 1142
+    },
+    {
+      "epoch": 0.04156817107320799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999298605994882e-05,
+      "loss": 1.1407,
+      "step": 1143
+    },
+    {
+      "epoch": 0.04160453867694658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999294188295663e-05,
+      "loss": 1.3204,
+      "step": 1144
+    },
+    {
+      "epoch": 0.04164090628068517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999289756732655e-05,
+      "loss": 1.2698,
+      "step": 1145
+    },
+    {
+      "epoch": 0.04167727388442376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999285311305919e-05,
+      "loss": 1.2688,
+      "step": 1146
+    },
+    {
+      "epoch": 0.04171364148816235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992808520155175e-05,
+      "loss": 1.3765,
+      "step": 1147
+    },
+    {
+      "epoch": 0.041750009091900936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992763788615116e-05,
+      "loss": 1.232,
+      "step": 1148
+    },
+    {
+      "epoch": 0.041786376695639525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992718918439643e-05,
+      "loss": 1.2275,
+      "step": 1149
+    },
+    {
+      "epoch": 0.041822744299378115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999267390962937e-05,
+      "loss": 1.2033,
+      "step": 1150
+    },
+    {
+      "epoch": 0.041859111903116704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992628762184924e-05,
+      "loss": 1.2971,
+      "step": 1151
+    },
+    {
+      "epoch": 0.04189547950685529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999258347610693e-05,
+      "loss": 1.267,
+      "step": 1152
+    },
+    {
+      "epoch": 0.04193184711059388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999253805139602e-05,
+      "loss": 1.3147,
+      "step": 1153
+    },
+    {
+      "epoch": 0.04196821471433247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992492488052822e-05,
+      "loss": 1.1418,
+      "step": 1154
+    },
+    {
+      "epoch": 0.04200458231807106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992446786077967e-05,
+      "loss": 1.2946,
+      "step": 1155
+    },
+    {
+      "epoch": 0.04204094992180965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992400945472093e-05,
+      "loss": 1.303,
+      "step": 1156
+    },
+    {
+      "epoch": 0.04207731752554824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999235496623583e-05,
+      "loss": 1.1238,
+      "step": 1157
+    },
+    {
+      "epoch": 0.04211368512928683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992308848369823e-05,
+      "loss": 1.2306,
+      "step": 1158
+    },
+    {
+      "epoch": 0.04215005273302542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992262591874705e-05,
+      "loss": 1.209,
+      "step": 1159
+    },
+    {
+      "epoch": 0.04218642033676401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999221619675112e-05,
+      "loss": 1.1302,
+      "step": 1160
+    },
+    {
+      "epoch": 0.0422227879405026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992169662999715e-05,
+      "loss": 1.298,
+      "step": 1161
+    },
+    {
+      "epoch": 0.04225915554424119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999212299062113e-05,
+      "loss": 1.3236,
+      "step": 1162
+    },
+    {
+      "epoch": 0.04229552314797978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999207617961602e-05,
+      "loss": 1.3033,
+      "step": 1163
+    },
+    {
+      "epoch": 0.04233189075171837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9992029229985024e-05,
+      "loss": 1.2524,
+      "step": 1164
+    },
+    {
+      "epoch": 0.04236825835545696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991982141728804e-05,
+      "loss": 1.1606,
+      "step": 1165
+    },
+    {
+      "epoch": 0.04240462595919555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991934914848005e-05,
+      "loss": 1.1351,
+      "step": 1166
+    },
+    {
+      "epoch": 0.04244099356293414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991887549343288e-05,
+      "loss": 1.2097,
+      "step": 1167
+    },
+    {
+      "epoch": 0.04247736116667273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991840045215307e-05,
+      "loss": 1.2897,
+      "step": 1168
+    },
+    {
+      "epoch": 0.04251372877041132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999179240246472e-05,
+      "loss": 1.2806,
+      "step": 1169
+    },
+    {
+      "epoch": 0.04255009637414991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991744621092195e-05,
+      "loss": 1.2018,
+      "step": 1170
+    },
+    {
+      "epoch": 0.0425864639778885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991696701098386e-05,
+      "loss": 1.1829,
+      "step": 1171
+    },
+    {
+      "epoch": 0.04262283158162709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991648642483962e-05,
+      "loss": 1.3437,
+      "step": 1172
+    },
+    {
+      "epoch": 0.042659199185365676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999160044524959e-05,
+      "loss": 1.2885,
+      "step": 1173
+    },
+    {
+      "epoch": 0.042695566789104265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999155210939594e-05,
+      "loss": 1.2753,
+      "step": 1174
+    },
+    {
+      "epoch": 0.042731934392842855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991503634923678e-05,
+      "loss": 1.2192,
+      "step": 1175
+    },
+    {
+      "epoch": 0.042768301996581444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999145502183348e-05,
+      "loss": 1.1258,
+      "step": 1176
+    },
+    {
+      "epoch": 0.04280466960032003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999140627012602e-05,
+      "loss": 1.2104,
+      "step": 1177
+    },
+    {
+      "epoch": 0.04284103720405862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991357379801973e-05,
+      "loss": 1.3481,
+      "step": 1178
+    },
+    {
+      "epoch": 0.04287740480779721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991308350862017e-05,
+      "loss": 1.2913,
+      "step": 1179
+    },
+    {
+      "epoch": 0.0429137724115358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991259183306834e-05,
+      "loss": 1.1889,
+      "step": 1180
+    },
+    {
+      "epoch": 0.04295014001527439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999120987713711e-05,
+      "loss": 1.3386,
+      "step": 1181
+    },
+    {
+      "epoch": 0.04298650761901298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999116043235352e-05,
+      "loss": 1.2687,
+      "step": 1182
+    },
+    {
+      "epoch": 0.043022875222751576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991110848956755e-05,
+      "loss": 1.2831,
+      "step": 1183
+    },
+    {
+      "epoch": 0.043059242826490166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991061126947503e-05,
+      "loss": 1.1869,
+      "step": 1184
+    },
+    {
+      "epoch": 0.043095610430228755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9991011266326453e-05,
+      "loss": 1.1732,
+      "step": 1185
+    },
+    {
+      "epoch": 0.043131978033967344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99909612670943e-05,
+      "loss": 1.1619,
+      "step": 1186
+    },
+    {
+      "epoch": 0.043168345637705934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990911129251734e-05,
+      "loss": 1.4414,
+      "step": 1187
+    },
+    {
+      "epoch": 0.04320471324144452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999086085279945e-05,
+      "loss": 1.2616,
+      "step": 1188
+    },
+    {
+      "epoch": 0.04324108084518311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990810437738142e-05,
+      "loss": 1.1898,
+      "step": 1189
+    },
+    {
+      "epoch": 0.0432774484489217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990759884068524e-05,
+      "loss": 1.3367,
+      "step": 1190
+    },
+    {
+      "epoch": 0.04331381605266029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999070919179128e-05,
+      "loss": 1.2958,
+      "step": 1191
+    },
+    {
+      "epoch": 0.04335018365639888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990658360907125e-05,
+      "loss": 1.2739,
+      "step": 1192
+    },
+    {
+      "epoch": 0.04338655126013747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990607391416758e-05,
+      "loss": 1.242,
+      "step": 1193
+    },
+    {
+      "epoch": 0.04342291886387606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999055628332089e-05,
+      "loss": 1.2379,
+      "step": 1194
+    },
+    {
+      "epoch": 0.04345928646761465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990505036620228e-05,
+      "loss": 1.1399,
+      "step": 1195
+    },
+    {
+      "epoch": 0.04349565407135324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999045365131548e-05,
+      "loss": 1.3311,
+      "step": 1196
+    },
+    {
+      "epoch": 0.04353202167509183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990402127407365e-05,
+      "loss": 1.198,
+      "step": 1197
+    },
+    {
+      "epoch": 0.043568389278830416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990350464896595e-05,
+      "loss": 1.1966,
+      "step": 1198
+    },
+    {
+      "epoch": 0.043604756882569005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990298663783886e-05,
+      "loss": 1.2284,
+      "step": 1199
+    },
+    {
+      "epoch": 0.043641124486307595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999024672406996e-05,
+      "loss": 1.264,
+      "step": 1200
+    },
+    {
+      "epoch": 0.043677492090046184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999019464575553e-05,
+      "loss": 1.373,
+      "step": 1201
+    },
+    {
+      "epoch": 0.04371385969378477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990142428841328e-05,
+      "loss": 1.2325,
+      "step": 1202
+    },
+    {
+      "epoch": 0.04375022729752336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990090073328075e-05,
+      "loss": 1.1991,
+      "step": 1203
+    },
+    {
+      "epoch": 0.04378659490126196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990037579216492e-05,
+      "loss": 1.3705,
+      "step": 1204
+    },
+    {
+      "epoch": 0.04382296250500055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989984946507314e-05,
+      "loss": 1.3058,
+      "step": 1205
+    },
+    {
+      "epoch": 0.04385933010873914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989932175201268e-05,
+      "loss": 1.2592,
+      "step": 1206
+    },
+    {
+      "epoch": 0.04389569771247773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989879265299087e-05,
+      "loss": 1.3291,
+      "step": 1207
+    },
+    {
+      "epoch": 0.043932065316216316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989826216801506e-05,
+      "loss": 1.235,
+      "step": 1208
+    },
+    {
+      "epoch": 0.043968432919954906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998977302970926e-05,
+      "loss": 1.1989,
+      "step": 1209
+    },
+    {
+      "epoch": 0.044004800523693495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989719704023084e-05,
+      "loss": 1.1792,
+      "step": 1210
+    },
+    {
+      "epoch": 0.044041168127432084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998966623974372e-05,
+      "loss": 1.2342,
+      "step": 1211
+    },
+    {
+      "epoch": 0.044077535731170674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989612636871915e-05,
+      "loss": 1.2474,
+      "step": 1212
+    },
+    {
+      "epoch": 0.04411390333490926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989558895408406e-05,
+      "loss": 1.2288,
+      "step": 1213
+    },
+    {
+      "epoch": 0.04415027093864785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989505015353942e-05,
+      "loss": 1.3633,
+      "step": 1214
+    },
+    {
+      "epoch": 0.04418663854238644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998945099670927e-05,
+      "loss": 1.1485,
+      "step": 1215
+    },
+    {
+      "epoch": 0.04422300614612503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989396839475134e-05,
+      "loss": 1.1548,
+      "step": 1216
+    },
+    {
+      "epoch": 0.04425937374986362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998934254365229e-05,
+      "loss": 1.2226,
+      "step": 1217
+    },
+    {
+      "epoch": 0.04429574135360221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989288109241496e-05,
+      "loss": 1.1073,
+      "step": 1218
+    },
+    {
+      "epoch": 0.0443321089573408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99892335362435e-05,
+      "loss": 1.3723,
+      "step": 1219
+    },
+    {
+      "epoch": 0.04436847656107939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998917882465906e-05,
+      "loss": 1.2785,
+      "step": 1220
+    },
+    {
+      "epoch": 0.04440484416481798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989123974488937e-05,
+      "loss": 1.3295,
+      "step": 1221
+    },
+    {
+      "epoch": 0.04444121176855657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9989068985733895e-05,
+      "loss": 1.3174,
+      "step": 1222
+    },
+    {
+      "epoch": 0.044477579372295156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998901385839469e-05,
+      "loss": 1.3271,
+      "step": 1223
+    },
+    {
+      "epoch": 0.04451394697603375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988958592472092e-05,
+      "loss": 1.2592,
+      "step": 1224
+    },
+    {
+      "epoch": 0.04455031457977234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988903187966866e-05,
+      "loss": 1.3771,
+      "step": 1225
+    },
+    {
+      "epoch": 0.04458668218351093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998884764487978e-05,
+      "loss": 1.2056,
+      "step": 1226
+    },
+    {
+      "epoch": 0.04462304978724952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988791963211603e-05,
+      "loss": 1.2534,
+      "step": 1227
+    },
+    {
+      "epoch": 0.04465941739098811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988736142963114e-05,
+      "loss": 1.1585,
+      "step": 1228
+    },
+    {
+      "epoch": 0.0446957849947267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988680184135082e-05,
+      "loss": 1.3018,
+      "step": 1229
+    },
+    {
+      "epoch": 0.04473215259846529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988624086728286e-05,
+      "loss": 1.3752,
+      "step": 1230
+    },
+    {
+      "epoch": 0.04476852020220388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988567850743497e-05,
+      "loss": 1.3679,
+      "step": 1231
+    },
+    {
+      "epoch": 0.04480488780594247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988511476181506e-05,
+      "loss": 1.2551,
+      "step": 1232
+    },
+    {
+      "epoch": 0.044841255409681056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998845496304309e-05,
+      "loss": 1.1309,
+      "step": 1233
+    },
+    {
+      "epoch": 0.044877623013419646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998839831132903e-05,
+      "loss": 1.237,
+      "step": 1234
+    },
+    {
+      "epoch": 0.044913990617158235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998834152104012e-05,
+      "loss": 1.2098,
+      "step": 1235
+    },
+    {
+      "epoch": 0.044950358220896824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988284592177138e-05,
+      "loss": 1.2257,
+      "step": 1236
+    },
+    {
+      "epoch": 0.044986725824635414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988227524740882e-05,
+      "loss": 1.3123,
+      "step": 1237
+    },
+    {
+      "epoch": 0.045023093428374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998817031873214e-05,
+      "loss": 1.3047,
+      "step": 1238
+    },
+    {
+      "epoch": 0.04505946103211259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988112974151704e-05,
+      "loss": 1.1888,
+      "step": 1239
+    },
+    {
+      "epoch": 0.04509582863585118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9988055491000374e-05,
+      "loss": 1.2871,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04513219623958977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987997869278947e-05,
+      "loss": 1.3372,
+      "step": 1241
+    },
+    {
+      "epoch": 0.04516856384332836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987940108988215e-05,
+      "loss": 1.2451,
+      "step": 1242
+    },
+    {
+      "epoch": 0.04520493144706695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998788221012899e-05,
+      "loss": 1.3173,
+      "step": 1243
+    },
+    {
+      "epoch": 0.04524129905080554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987824172702067e-05,
+      "loss": 1.3357,
+      "step": 1244
+    },
+    {
+      "epoch": 0.045277666654544135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987765996708256e-05,
+      "loss": 1.2229,
+      "step": 1245
+    },
+    {
+      "epoch": 0.045314034258282725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987707682148363e-05,
+      "loss": 1.2471,
+      "step": 1246
+    },
+    {
+      "epoch": 0.045350401862021314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987649229023196e-05,
+      "loss": 1.1642,
+      "step": 1247
+    },
+    {
+      "epoch": 0.0453867694657599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987590637333567e-05,
+      "loss": 1.2005,
+      "step": 1248
+    },
+    {
+      "epoch": 0.04542313706949849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987531907080288e-05,
+      "loss": 1.3983,
+      "step": 1249
+    },
+    {
+      "epoch": 0.04545950467323708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987473038264172e-05,
+      "loss": 1.2331,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04549587227697567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987414030886043e-05,
+      "loss": 1.196,
+      "step": 1251
+    },
+    {
+      "epoch": 0.04553223988071426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987354884946713e-05,
+      "loss": 1.2016,
+      "step": 1252
+    },
+    {
+      "epoch": 0.04556860748445285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987295600447003e-05,
+      "loss": 1.1596,
+      "step": 1253
+    },
+    {
+      "epoch": 0.04560497508819144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987236177387736e-05,
+      "loss": 1.2153,
+      "step": 1254
+    },
+    {
+      "epoch": 0.04564134269193003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998717661576974e-05,
+      "loss": 1.2573,
+      "step": 1255
+    },
+    {
+      "epoch": 0.04567771029566862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987116915593835e-05,
+      "loss": 1.2924,
+      "step": 1256
+    },
+    {
+      "epoch": 0.04571407789940721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9987057076860853e-05,
+      "loss": 1.2231,
+      "step": 1257
+    },
+    {
+      "epoch": 0.0457504455031458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986997099571625e-05,
+      "loss": 1.1846,
+      "step": 1258
+    },
+    {
+      "epoch": 0.045786813106884386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986936983726984e-05,
+      "loss": 1.2183,
+      "step": 1259
+    },
+    {
+      "epoch": 0.045823180710622975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998687672932776e-05,
+      "loss": 1.1221,
+      "step": 1260
+    },
+    {
+      "epoch": 0.045859548314361565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998681633637479e-05,
+      "loss": 1.2349,
+      "step": 1261
+    },
+    {
+      "epoch": 0.045895915918100154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998675580486891e-05,
+      "loss": 1.2407,
+      "step": 1262
+    },
+    {
+      "epoch": 0.04593228352183874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986695134810968e-05,
+      "loss": 1.2502,
+      "step": 1263
+    },
+    {
+      "epoch": 0.04596865112557733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99866343262018e-05,
+      "loss": 1.3708,
+      "step": 1264
+    },
+    {
+      "epoch": 0.04600501872931593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998657337904225e-05,
+      "loss": 1.3246,
+      "step": 1265
+    },
+    {
+      "epoch": 0.04604138633305452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986512293333156e-05,
+      "loss": 1.2308,
+      "step": 1266
+    },
+    {
+      "epoch": 0.04607775393679311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986451069075383e-05,
+      "loss": 1.2924,
+      "step": 1267
+    },
+    {
+      "epoch": 0.0461141215405317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986389706269765e-05,
+      "loss": 1.4183,
+      "step": 1268
+    },
+    {
+      "epoch": 0.046150489144270286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986328204917157e-05,
+      "loss": 1.2557,
+      "step": 1269
+    },
+    {
+      "epoch": 0.046186856748008875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998626656501842e-05,
+      "loss": 1.2587,
+      "step": 1270
+    },
+    {
+      "epoch": 0.046223224351747465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986204786574398e-05,
+      "loss": 1.236,
+      "step": 1271
+    },
+    {
+      "epoch": 0.046259591955486054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986142869585955e-05,
+      "loss": 1.2849,
+      "step": 1272
+    },
+    {
+      "epoch": 0.04629595955922464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998608081405395e-05,
+      "loss": 1.1937,
+      "step": 1273
+    },
+    {
+      "epoch": 0.04633232716296323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9986018619979237e-05,
+      "loss": 1.3466,
+      "step": 1274
+    },
+    {
+      "epoch": 0.04636869476670182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985956287362688e-05,
+      "loss": 1.2524,
+      "step": 1275
+    },
+    {
+      "epoch": 0.04640506237044041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998589381620516e-05,
+      "loss": 1.3201,
+      "step": 1276
+    },
+    {
+      "epoch": 0.046441429974179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998583120650753e-05,
+      "loss": 1.2705,
+      "step": 1277
+    },
+    {
+      "epoch": 0.04647779757791759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985768458270656e-05,
+      "loss": 1.2985,
+      "step": 1278
+    },
+    {
+      "epoch": 0.04651416518165618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985705571495414e-05,
+      "loss": 1.2055,
+      "step": 1279
+    },
+    {
+      "epoch": 0.04655053278539477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985642546182672e-05,
+      "loss": 1.2697,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04658690038913336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985579382333308e-05,
+      "loss": 1.1599,
+      "step": 1281
+    },
+    {
+      "epoch": 0.04662326799287195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985516079948198e-05,
+      "loss": 1.3014,
+      "step": 1282
+    },
+    {
+      "epoch": 0.04665963559661054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985452639028224e-05,
+      "loss": 1.3188,
+      "step": 1283
+    },
+    {
+      "epoch": 0.046696003200349126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998538905957426e-05,
+      "loss": 1.2537,
+      "step": 1284
+    },
+    {
+      "epoch": 0.046732370804087715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985325341587186e-05,
+      "loss": 1.2832,
+      "step": 1285
+    },
+    {
+      "epoch": 0.04676873840782631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985261485067895e-05,
+      "loss": 1.2746,
+      "step": 1286
+    },
+    {
+      "epoch": 0.0468051060115649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998519749001727e-05,
+      "loss": 1.2227,
+      "step": 1287
+    },
+    {
+      "epoch": 0.04684147361530349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998513335643619e-05,
+      "loss": 1.2321,
+      "step": 1288
+    },
+    {
+      "epoch": 0.04687784121904208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985069084325557e-05,
+      "loss": 1.1952,
+      "step": 1289
+    },
+    {
+      "epoch": 0.04691420882278067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9985004673686255e-05,
+      "loss": 1.1818,
+      "step": 1290
+    },
+    {
+      "epoch": 0.04695057642651926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998494012451918e-05,
+      "loss": 1.12,
+      "step": 1291
+    },
+    {
+      "epoch": 0.04698694403025785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984875436825232e-05,
+      "loss": 1.201,
+      "step": 1292
+    },
+    {
+      "epoch": 0.04702331163399644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984810610605295e-05,
+      "loss": 1.217,
+      "step": 1293
+    },
+    {
+      "epoch": 0.047059679237735026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984745645860286e-05,
+      "loss": 1.2881,
+      "step": 1294
+    },
+    {
+      "epoch": 0.047096046841473616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984680542591092e-05,
+      "loss": 1.1884,
+      "step": 1295
+    },
+    {
+      "epoch": 0.047132414445212205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984615300798628e-05,
+      "loss": 1.301,
+      "step": 1296
+    },
+    {
+      "epoch": 0.047168782048950794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984549920483788e-05,
+      "loss": 1.2695,
+      "step": 1297
+    },
+    {
+      "epoch": 0.047205149652689384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984484401647487e-05,
+      "loss": 1.2908,
+      "step": 1298
+    },
+    {
+      "epoch": 0.04724151725642797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998441874429063e-05,
+      "loss": 1.3076,
+      "step": 1299
+    },
+    {
+      "epoch": 0.04727788486016656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984352948414124e-05,
+      "loss": 1.2014,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04731425246390515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984287014018892e-05,
+      "loss": 1.2319,
+      "step": 1301
+    },
+    {
+      "epoch": 0.04735062006764374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984220941105845e-05,
+      "loss": 1.3,
+      "step": 1302
+    },
+    {
+      "epoch": 0.04738698767138233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984154729675894e-05,
+      "loss": 1.2471,
+      "step": 1303
+    },
+    {
+      "epoch": 0.04742335527512092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984088379729964e-05,
+      "loss": 1.3668,
+      "step": 1304
+    },
+    {
+      "epoch": 0.04745972287885951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9984021891268973e-05,
+      "loss": 1.2837,
+      "step": 1305
+    },
+    {
+      "epoch": 0.047496090482598105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983955264293843e-05,
+      "loss": 1.2889,
+      "step": 1306
+    },
+    {
+      "epoch": 0.047532458086336694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983888498805503e-05,
+      "loss": 1.1856,
+      "step": 1307
+    },
+    {
+      "epoch": 0.047568825690075284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998382159480487e-05,
+      "loss": 1.3186,
+      "step": 1308
+    },
+    {
+      "epoch": 0.04760519329381387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998375455229288e-05,
+      "loss": 1.2626,
+      "step": 1309
+    },
+    {
+      "epoch": 0.04764156089755246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998368737127046e-05,
+      "loss": 1.2305,
+      "step": 1310
+    },
+    {
+      "epoch": 0.04767792850129105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983620051738545e-05,
+      "loss": 1.2377,
+      "step": 1311
+    },
+    {
+      "epoch": 0.04771429610502964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983552593698063e-05,
+      "loss": 1.234,
+      "step": 1312
+    },
+    {
+      "epoch": 0.04775066370876823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998348499714996e-05,
+      "loss": 1.2568,
+      "step": 1313
+    },
+    {
+      "epoch": 0.04778703131250682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998341726209516e-05,
+      "loss": 1.117,
+      "step": 1314
+    },
+    {
+      "epoch": 0.04782339891624541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983349388534612e-05,
+      "loss": 1.0989,
+      "step": 1315
+    },
+    {
+      "epoch": 0.047859766519984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983281376469255e-05,
+      "loss": 1.1865,
+      "step": 1316
+    },
+    {
+      "epoch": 0.04789613412372259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983213225900036e-05,
+      "loss": 1.2345,
+      "step": 1317
+    },
+    {
+      "epoch": 0.04793250172746118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983144936827895e-05,
+      "loss": 1.262,
+      "step": 1318
+    },
+    {
+      "epoch": 0.047968869331199766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998307650925379e-05,
+      "loss": 1.2346,
+      "step": 1319
+    },
+    {
+      "epoch": 0.048005236934938356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983007943178655e-05,
+      "loss": 1.266,
+      "step": 1320
+    },
+    {
+      "epoch": 0.048041604538676945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982939238603447e-05,
+      "loss": 1.2956,
+      "step": 1321
+    },
+    {
+      "epoch": 0.048077972142415534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982870395529125e-05,
+      "loss": 1.2099,
+      "step": 1322
+    },
+    {
+      "epoch": 0.048114339746154124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998280141395664e-05,
+      "loss": 1.2322,
+      "step": 1323
+    },
+    {
+      "epoch": 0.04815070734989271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998273229388695e-05,
+      "loss": 1.2724,
+      "step": 1324
+    },
+    {
+      "epoch": 0.0481870749536313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998266303532101e-05,
+      "loss": 1.2416,
+      "step": 1325
+    },
+    {
+      "epoch": 0.04822344255736989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982593638259785e-05,
+      "loss": 1.1628,
+      "step": 1326
+    },
+    {
+      "epoch": 0.04825981016110849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998252410270424e-05,
+      "loss": 1.2451,
+      "step": 1327
+    },
+    {
+      "epoch": 0.04829617776484708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982454428655332e-05,
+      "loss": 1.1058,
+      "step": 1328
+    },
+    {
+      "epoch": 0.048332545368585667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982384616114036e-05,
+      "loss": 1.2204,
+      "step": 1329
+    },
+    {
+      "epoch": 0.048368912972324256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982314665081314e-05,
+      "loss": 1.2168,
+      "step": 1330
+    },
+    {
+      "epoch": 0.048405280576062845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998224457555814e-05,
+      "loss": 1.2405,
+      "step": 1331
+    },
+    {
+      "epoch": 0.048441648179801435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9982174347545484e-05,
+      "loss": 1.2598,
+      "step": 1332
+    },
+    {
+      "epoch": 0.048478015783540024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998210398104432e-05,
+      "loss": 1.1952,
+      "step": 1333
+    },
+    {
+      "epoch": 0.04851438338727861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998203347605563e-05,
+      "loss": 1.3289,
+      "step": 1334
+    },
+    {
+      "epoch": 0.0485507509910172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998196283258039e-05,
+      "loss": 1.2755,
+      "step": 1335
+    },
+    {
+      "epoch": 0.04858711859475579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981892050619574e-05,
+      "loss": 1.3099,
+      "step": 1336
+    },
+    {
+      "epoch": 0.04862348619849438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981821130174168e-05,
+      "loss": 1.3464,
+      "step": 1337
+    },
+    {
+      "epoch": 0.04865985380223297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998175007124516e-05,
+      "loss": 1.3101,
+      "step": 1338
+    },
+    {
+      "epoch": 0.04869622140597156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998167887383353e-05,
+      "loss": 1.3857,
+      "step": 1339
+    },
+    {
+      "epoch": 0.04873258900971015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981607537940265e-05,
+      "loss": 1.2725,
+      "step": 1340
+    },
+    {
+      "epoch": 0.04876895661344874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998153606356636e-05,
+      "loss": 1.1102,
+      "step": 1341
+    },
+    {
+      "epoch": 0.04880532421718733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981464450712804e-05,
+      "loss": 1.2263,
+      "step": 1342
+    },
+    {
+      "epoch": 0.04884169182092592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981392699380593e-05,
+      "loss": 1.2594,
+      "step": 1343
+    },
+    {
+      "epoch": 0.048878059424664506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981320809570716e-05,
+      "loss": 1.1209,
+      "step": 1344
+    },
+    {
+      "epoch": 0.048914427028403096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981248781284175e-05,
+      "loss": 1.2621,
+      "step": 1345
+    },
+    {
+      "epoch": 0.048950794632141685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998117661452197e-05,
+      "loss": 1.2904,
+      "step": 1346
+    },
+    {
+      "epoch": 0.048987162235880274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99811043092851e-05,
+      "loss": 1.1974,
+      "step": 1347
+    },
+    {
+      "epoch": 0.04902352983961887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9981031865574572e-05,
+      "loss": 1.1878,
+      "step": 1348
+    },
+    {
+      "epoch": 0.04905989744335746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980959283391385e-05,
+      "loss": 1.3819,
+      "step": 1349
+    },
+    {
+      "epoch": 0.04909626504709605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998088656273655e-05,
+      "loss": 1.1629,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04913263265083464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998081370361107e-05,
+      "loss": 1.2478,
+      "step": 1351
+    },
+    {
+      "epoch": 0.04916900025457323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998074070601597e-05,
+      "loss": 1.1334,
+      "step": 1352
+    },
+    {
+      "epoch": 0.04920536785831182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998066756995225e-05,
+      "loss": 1.2276,
+      "step": 1353
+    },
+    {
+      "epoch": 0.04924173546205041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980594295420928e-05,
+      "loss": 1.3326,
+      "step": 1354
+    },
+    {
+      "epoch": 0.049278103065788996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980520882423018e-05,
+      "loss": 1.2912,
+      "step": 1355
+    },
+    {
+      "epoch": 0.049314470669527585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980447330959545e-05,
+      "loss": 1.1998,
+      "step": 1356
+    },
+    {
+      "epoch": 0.049350838273266175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980373641031525e-05,
+      "loss": 1.1942,
+      "step": 1357
+    },
+    {
+      "epoch": 0.049387205877004764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980299812639977e-05,
+      "loss": 1.2567,
+      "step": 1358
+    },
+    {
+      "epoch": 0.04942357348074335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980225845785936e-05,
+      "loss": 1.2064,
+      "step": 1359
+    },
+    {
+      "epoch": 0.04945994108448194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980151740470414e-05,
+      "loss": 1.3242,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04949630868822053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998007749669445e-05,
+      "loss": 1.2435,
+      "step": 1361
+    },
+    {
+      "epoch": 0.04953267629195912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.998000311445907e-05,
+      "loss": 1.2558,
+      "step": 1362
+    },
+    {
+      "epoch": 0.04956904389569771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997992859376531e-05,
+      "loss": 1.0865,
+      "step": 1363
+    },
+    {
+      "epoch": 0.0496054114994363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99798539346142e-05,
+      "loss": 1.2741,
+      "step": 1364
+    },
+    {
+      "epoch": 0.04964177910317489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997977913700677e-05,
+      "loss": 1.2242,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04967814670691348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997970420094407e-05,
+      "loss": 1.3038,
+      "step": 1366
+    },
+    {
+      "epoch": 0.04971451431065207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979629126427132e-05,
+      "loss": 1.1995,
+      "step": 1367
+    },
+    {
+      "epoch": 0.049750881914390664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979553913456996e-05,
+      "loss": 1.2124,
+      "step": 1368
+    },
+    {
+      "epoch": 0.049787249518129253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979478562034713e-05,
+      "loss": 1.2199,
+      "step": 1369
+    },
+    {
+      "epoch": 0.04982361712186784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979403072161324e-05,
+      "loss": 1.2758,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04985998472560643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979327443837872e-05,
+      "loss": 1.2089,
+      "step": 1371
+    },
+    {
+      "epoch": 0.04989635232934502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979251677065415e-05,
+      "loss": 1.2321,
+      "step": 1372
+    },
+    {
+      "epoch": 0.04993271993308361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979175771844995e-05,
+      "loss": 1.3752,
+      "step": 1373
+    },
+    {
+      "epoch": 0.0499690875368222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979099728177674e-05,
+      "loss": 1.1165,
+      "step": 1374
+    },
+    {
+      "epoch": 0.05000545514056079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9979023546064505e-05,
+      "loss": 1.1783,
+      "step": 1375
+    },
+    {
+      "epoch": 0.05004182274429938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997894722550654e-05,
+      "loss": 1.2298,
+      "step": 1376
+    },
+    {
+      "epoch": 0.05007819034803797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978870766504838e-05,
+      "loss": 1.3096,
+      "step": 1377
+    },
+    {
+      "epoch": 0.05011455795177656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978794169060464e-05,
+      "loss": 1.3909,
+      "step": 1378
+    },
+    {
+      "epoch": 0.05015092555551515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978717433174478e-05,
+      "loss": 1.2527,
+      "step": 1379
+    },
+    {
+      "epoch": 0.050187293159253736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978640558847946e-05,
+      "loss": 1.1975,
+      "step": 1380
+    },
+    {
+      "epoch": 0.050223660762992325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978563546081932e-05,
+      "loss": 1.3054,
+      "step": 1381
+    },
+    {
+      "epoch": 0.050260028366730915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997848639487751e-05,
+      "loss": 1.291,
+      "step": 1382
+    },
+    {
+      "epoch": 0.050296395970469504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997840910523575e-05,
+      "loss": 1.2397,
+      "step": 1383
+    },
+    {
+      "epoch": 0.05033276357420809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978331677157714e-05,
+      "loss": 1.2255,
+      "step": 1384
+    },
+    {
+      "epoch": 0.05036913117794668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978254110644487e-05,
+      "loss": 1.3408,
+      "step": 1385
+    },
+    {
+      "epoch": 0.05040549878168527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997817640569714e-05,
+      "loss": 1.2708,
+      "step": 1386
+    },
+    {
+      "epoch": 0.05044186638542386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978098562316753e-05,
+      "loss": 1.2703,
+      "step": 1387
+    },
+    {
+      "epoch": 0.05047823398916245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9978020580504407e-05,
+      "loss": 1.2585,
+      "step": 1388
+    },
+    {
+      "epoch": 0.05051460159290105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977942460261182e-05,
+      "loss": 1.2408,
+      "step": 1389
+    },
+    {
+      "epoch": 0.050550969196639636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977864201588162e-05,
+      "loss": 1.2839,
+      "step": 1390
+    },
+    {
+      "epoch": 0.050587336800378226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977785804486434e-05,
+      "loss": 1.3529,
+      "step": 1391
+    },
+    {
+      "epoch": 0.050623704404116815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977707268957084e-05,
+      "loss": 1.1492,
+      "step": 1392
+    },
+    {
+      "epoch": 0.050660072007855404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99776285950012e-05,
+      "loss": 1.19,
+      "step": 1393
+    },
+    {
+      "epoch": 0.050696439611593994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977549782619877e-05,
+      "loss": 1.2477,
+      "step": 1394
+    },
+    {
+      "epoch": 0.05073280721533258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977470831814206e-05,
+      "loss": 1.2114,
+      "step": 1395
+    },
+    {
+      "epoch": 0.05076917481907117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977391742585282e-05,
+      "loss": 1.1466,
+      "step": 1396
+    },
+    {
+      "epoch": 0.05080554242280976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977312514934207e-05,
+      "loss": 1.2078,
+      "step": 1397
+    },
+    {
+      "epoch": 0.05084191002654835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977233148862076e-05,
+      "loss": 1.1738,
+      "step": 1398
+    },
+    {
+      "epoch": 0.05087827763028694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997715364436999e-05,
+      "loss": 1.302,
+      "step": 1399
+    },
+    {
+      "epoch": 0.05091464523402553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9977074001459056e-05,
+      "loss": 1.1474,
+      "step": 1400
+    },
+    {
+      "epoch": 0.05095101283776412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976994220130374e-05,
+      "loss": 1.2794,
+      "step": 1401
+    },
+    {
+      "epoch": 0.05098738044150271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997691430038505e-05,
+      "loss": 1.2014,
+      "step": 1402
+    },
+    {
+      "epoch": 0.0510237480452413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99768342422242e-05,
+      "loss": 1.236,
+      "step": 1403
+    },
+    {
+      "epoch": 0.05106011564897989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997675404564893e-05,
+      "loss": 1.2649,
+      "step": 1404
+    },
+    {
+      "epoch": 0.051096483252718476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997667371066035e-05,
+      "loss": 1.1614,
+      "step": 1405
+    },
+    {
+      "epoch": 0.051132850856457066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976593237259575e-05,
+      "loss": 1.1319,
+      "step": 1406
+    },
+    {
+      "epoch": 0.051169218460195655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976512625447726e-05,
+      "loss": 1.2368,
+      "step": 1407
+    },
+    {
+      "epoch": 0.051205586063934244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976431875225923e-05,
+      "loss": 1.3448,
+      "step": 1408
+    },
+    {
+      "epoch": 0.05124195366767284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976350986595277e-05,
+      "loss": 1.2514,
+      "step": 1409
+    },
+    {
+      "epoch": 0.05127832127141143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976269959556916e-05,
+      "loss": 1.1827,
+      "step": 1410
+    },
+    {
+      "epoch": 0.05131468887515002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997618879411197e-05,
+      "loss": 1.1578,
+      "step": 1411
+    },
+    {
+      "epoch": 0.05135105647888861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9976107490261554e-05,
+      "loss": 1.2585,
+      "step": 1412
+    },
+    {
+      "epoch": 0.0513874240826272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99760260480068e-05,
+      "loss": 1.2745,
+      "step": 1413
+    },
+    {
+      "epoch": 0.05142379168636579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975944467348843e-05,
+      "loss": 1.2856,
+      "step": 1414
+    },
+    {
+      "epoch": 0.051460159290104376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975862748288806e-05,
+      "loss": 1.3126,
+      "step": 1415
+    },
+    {
+      "epoch": 0.051496526893842966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997578089082783e-05,
+      "loss": 1.1874,
+      "step": 1416
+    },
+    {
+      "epoch": 0.051532894497581555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975698894967044e-05,
+      "loss": 1.3329,
+      "step": 1417
+    },
+    {
+      "epoch": 0.051569262101320144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975616760707594e-05,
+      "loss": 1.308,
+      "step": 1418
+    },
+    {
+      "epoch": 0.051605629705058734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997553448805061e-05,
+      "loss": 1.1556,
+      "step": 1419
+    },
+    {
+      "epoch": 0.05164199730879732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975452076997244e-05,
+      "loss": 1.1954,
+      "step": 1420
+    },
+    {
+      "epoch": 0.05167836491253591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975369527548634e-05,
+      "loss": 1.1165,
+      "step": 1421
+    },
+    {
+      "epoch": 0.0517147325162745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997528683970592e-05,
+      "loss": 1.2388,
+      "step": 1422
+    },
+    {
+      "epoch": 0.05175110012001309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997520401347026e-05,
+      "loss": 1.2914,
+      "step": 1423
+    },
+    {
+      "epoch": 0.05178746772375168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997512104884279e-05,
+      "loss": 1.408,
+      "step": 1424
+    },
+    {
+      "epoch": 0.05182383532749027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9975037945824672e-05,
+      "loss": 1.3477,
+      "step": 1425
+    },
+    {
+      "epoch": 0.05186020293122886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974954704417056e-05,
+      "loss": 1.1705,
+      "step": 1426
+    },
+    {
+      "epoch": 0.05189657053496745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974871324621096e-05,
+      "loss": 1.2338,
+      "step": 1427
+    },
+    {
+      "epoch": 0.05193293813870604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997478780643795e-05,
+      "loss": 1.3159,
+      "step": 1428
+    },
+    {
+      "epoch": 0.05196930574244463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974704149868773e-05,
+      "loss": 1.2003,
+      "step": 1429
+    },
+    {
+      "epoch": 0.05200567334618322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997462035491473e-05,
+      "loss": 1.2571,
+      "step": 1430
+    },
+    {
+      "epoch": 0.05204204094992181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997453642157698e-05,
+      "loss": 1.1482,
+      "step": 1431
+    },
+    {
+      "epoch": 0.0520784085536604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997445234985669e-05,
+      "loss": 1.2623,
+      "step": 1432
+    },
+    {
+      "epoch": 0.05211477615739899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974368139755028e-05,
+      "loss": 1.2548,
+      "step": 1433
+    },
+    {
+      "epoch": 0.05215114376113758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974283791273158e-05,
+      "loss": 1.17,
+      "step": 1434
+    },
+    {
+      "epoch": 0.05218751136487617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997419930441225e-05,
+      "loss": 1.2244,
+      "step": 1435
+    },
+    {
+      "epoch": 0.05222387896861476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974114679173478e-05,
+      "loss": 1.1524,
+      "step": 1436
+    },
+    {
+      "epoch": 0.05226024657235335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9974029915558016e-05,
+      "loss": 1.1354,
+      "step": 1437
+    },
+    {
+      "epoch": 0.05229661417609194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973945013567044e-05,
+      "loss": 1.1594,
+      "step": 1438
+    },
+    {
+      "epoch": 0.05233298177983053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973859973201733e-05,
+      "loss": 1.258,
+      "step": 1439
+    },
+    {
+      "epoch": 0.052369349383569117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997377479446327e-05,
+      "loss": 1.3349,
+      "step": 1440
+    },
+    {
+      "epoch": 0.052405716987307706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997368947735283e-05,
+      "loss": 1.4238,
+      "step": 1441
+    },
+    {
+      "epoch": 0.052442084591046295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99736040218716e-05,
+      "loss": 1.2056,
+      "step": 1442
+    },
+    {
+      "epoch": 0.052478452194784884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973518428020765e-05,
+      "loss": 1.1894,
+      "step": 1443
+    },
+    {
+      "epoch": 0.052514819798523474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973432695801508e-05,
+      "loss": 1.3918,
+      "step": 1444
+    },
+    {
+      "epoch": 0.05255118740226206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973346825215025e-05,
+      "loss": 1.2071,
+      "step": 1445
+    },
+    {
+      "epoch": 0.05258755500600065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973260816262508e-05,
+      "loss": 1.1146,
+      "step": 1446
+    },
+    {
+      "epoch": 0.05262392260973924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973174668945147e-05,
+      "loss": 1.2022,
+      "step": 1447
+    },
+    {
+      "epoch": 0.05266029021347783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9973088383264137e-05,
+      "loss": 1.1081,
+      "step": 1448
+    },
+    {
+      "epoch": 0.05269665781721642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997300195922067e-05,
+      "loss": 1.1986,
+      "step": 1449
+    },
+    {
+      "epoch": 0.05273302542095502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972915396815956e-05,
+      "loss": 1.275,
+      "step": 1450
+    },
+    {
+      "epoch": 0.052769393024693606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972828696051194e-05,
+      "loss": 1.1177,
+      "step": 1451
+    },
+    {
+      "epoch": 0.052805760628432195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972741856927578e-05,
+      "loss": 1.2761,
+      "step": 1452
+    },
+    {
+      "epoch": 0.052842128232170785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997265487944632e-05,
+      "loss": 1.3831,
+      "step": 1453
+    },
+    {
+      "epoch": 0.052878495835909374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972567763608627e-05,
+      "loss": 1.3391,
+      "step": 1454
+    },
+    {
+      "epoch": 0.05291486343964796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972480509415702e-05,
+      "loss": 1.2962,
+      "step": 1455
+    },
+    {
+      "epoch": 0.05295123104338655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997239311686876e-05,
+      "loss": 1.1202,
+      "step": 1456
+    },
+    {
+      "epoch": 0.05298759864712514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972305585969013e-05,
+      "loss": 1.1987,
+      "step": 1457
+    },
+    {
+      "epoch": 0.05302396625086373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9972217916717675e-05,
+      "loss": 1.1935,
+      "step": 1458
+    },
+    {
+      "epoch": 0.05306033385460232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997213010911596e-05,
+      "loss": 1.3631,
+      "step": 1459
+    },
+    {
+      "epoch": 0.05309670145834091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997204216316509e-05,
+      "loss": 1.3391,
+      "step": 1460
+    },
+    {
+      "epoch": 0.0531330690620795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971954078866284e-05,
+      "loss": 1.186,
+      "step": 1461
+    },
+    {
+      "epoch": 0.05316943666581809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971865856220763e-05,
+      "loss": 1.2822,
+      "step": 1462
+    },
+    {
+      "epoch": 0.05320580426955668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997177749522975e-05,
+      "loss": 1.1778,
+      "step": 1463
+    },
+    {
+      "epoch": 0.05324217187329527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971688995894476e-05,
+      "loss": 1.3138,
+      "step": 1464
+    },
+    {
+      "epoch": 0.05327853947703386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997160035821616e-05,
+      "loss": 1.2436,
+      "step": 1465
+    },
+    {
+      "epoch": 0.053314907080772446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971511582196045e-05,
+      "loss": 1.2402,
+      "step": 1466
+    },
+    {
+      "epoch": 0.053351274684511035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971422667835348e-05,
+      "loss": 1.2655,
+      "step": 1467
+    },
+    {
+      "epoch": 0.053387642288249625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997133361513531e-05,
+      "loss": 1.1903,
+      "step": 1468
+    },
+    {
+      "epoch": 0.053424009891988214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971244424097165e-05,
+      "loss": 1.2121,
+      "step": 1469
+    },
+    {
+      "epoch": 0.0534603774957268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9971155094722153e-05,
+      "loss": 1.2672,
+      "step": 1470
+    },
+    {
+      "epoch": 0.0534967450994654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997106562701151e-05,
+      "loss": 1.2569,
+      "step": 1471
+    },
+    {
+      "epoch": 0.05353311270320399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970976020966478e-05,
+      "loss": 1.1548,
+      "step": 1472
+    },
+    {
+      "epoch": 0.05356948030694258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970886276588303e-05,
+      "loss": 1.1471,
+      "step": 1473
+    },
+    {
+      "epoch": 0.05360584791068117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970796393878226e-05,
+      "loss": 1.1516,
+      "step": 1474
+    },
+    {
+      "epoch": 0.05364221551441976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970706372837494e-05,
+      "loss": 1.3049,
+      "step": 1475
+    },
+    {
+      "epoch": 0.053678583118158346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970616213467358e-05,
+      "loss": 1.2637,
+      "step": 1476
+    },
+    {
+      "epoch": 0.053714950721896935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997052591576907e-05,
+      "loss": 1.2639,
+      "step": 1477
+    },
+    {
+      "epoch": 0.053751318325635525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970435479743883e-05,
+      "loss": 1.2813,
+      "step": 1478
+    },
+    {
+      "epoch": 0.053787685929374114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997034490539305e-05,
+      "loss": 1.3447,
+      "step": 1479
+    },
+    {
+      "epoch": 0.0538240535331127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970254192717826e-05,
+      "loss": 1.1747,
+      "step": 1480
+    },
+    {
+      "epoch": 0.05386042113685129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997016334171947e-05,
+      "loss": 1.2168,
+      "step": 1481
+    },
+    {
+      "epoch": 0.05389678874058988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9970072352399246e-05,
+      "loss": 1.1957,
+      "step": 1482
+    },
+    {
+      "epoch": 0.05393315634432847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996998122475841e-05,
+      "loss": 1.2615,
+      "step": 1483
+    },
+    {
+      "epoch": 0.05396952394806706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969889958798233e-05,
+      "loss": 1.2667,
+      "step": 1484
+    },
+    {
+      "epoch": 0.05400589155180565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969798554519976e-05,
+      "loss": 1.1639,
+      "step": 1485
+    },
+    {
+      "epoch": 0.05404225915554424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996970701192491e-05,
+      "loss": 1.1703,
+      "step": 1486
+    },
+    {
+      "epoch": 0.05407862675928283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969615331014306e-05,
+      "loss": 1.2354,
+      "step": 1487
+    },
+    {
+      "epoch": 0.05411499436302142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996952351178943e-05,
+      "loss": 1.249,
+      "step": 1488
+    },
+    {
+      "epoch": 0.05415136196676001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969431554251568e-05,
+      "loss": 1.2357,
+      "step": 1489
+    },
+    {
+      "epoch": 0.0541877295704986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996933945840198e-05,
+      "loss": 1.2059,
+      "step": 1490
+    },
+    {
+      "epoch": 0.054224097174237186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969247224241958e-05,
+      "loss": 1.1614,
+      "step": 1491
+    },
+    {
+      "epoch": 0.05426046477797578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996915485177277e-05,
+      "loss": 1.2716,
+      "step": 1492
+    },
+    {
+      "epoch": 0.05429683238171437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996906234099571e-05,
+      "loss": 1.2369,
+      "step": 1493
+    },
+    {
+      "epoch": 0.05433319998545296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996896969191205e-05,
+      "loss": 1.2751,
+      "step": 1494
+    },
+    {
+      "epoch": 0.05436956758919155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968876904523075e-05,
+      "loss": 1.1096,
+      "step": 1495
+    },
+    {
+      "epoch": 0.05440593519293014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968783978830082e-05,
+      "loss": 1.2961,
+      "step": 1496
+    },
+    {
+      "epoch": 0.05444230279666873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968690914834353e-05,
+      "loss": 1.1612,
+      "step": 1497
+    },
+    {
+      "epoch": 0.05447867040040732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968597712537186e-05,
+      "loss": 1.243,
+      "step": 1498
+    },
+    {
+      "epoch": 0.05451503800414591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968504371939866e-05,
+      "loss": 1.2439,
+      "step": 1499
+    },
+    {
+      "epoch": 0.0545514056078845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996841089304369e-05,
+      "loss": 1.3229,
+      "step": 1500
+    },
+    {
+      "epoch": 0.054587773211623086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968317275849957e-05,
+      "loss": 1.2446,
+      "step": 1501
+    },
+    {
+      "epoch": 0.054624140815361676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968223520359966e-05,
+      "loss": 1.2927,
+      "step": 1502
+    },
+    {
+      "epoch": 0.054660508419100265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968129626575017e-05,
+      "loss": 1.257,
+      "step": 1503
+    },
+    {
+      "epoch": 0.054696876022838854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9968035594496408e-05,
+      "loss": 1.3259,
+      "step": 1504
+    },
+    {
+      "epoch": 0.054733243626577444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967941424125454e-05,
+      "loss": 1.1233,
+      "step": 1505
+    },
+    {
+      "epoch": 0.05476961123031603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967847115463453e-05,
+      "loss": 1.1305,
+      "step": 1506
+    },
+    {
+      "epoch": 0.05480597883405462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967752668511715e-05,
+      "loss": 1.1664,
+      "step": 1507
+    },
+    {
+      "epoch": 0.05484234643779321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967658083271552e-05,
+      "loss": 1.2766,
+      "step": 1508
+    },
+    {
+      "epoch": 0.0548787140415318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967563359744275e-05,
+      "loss": 1.184,
+      "step": 1509
+    },
+    {
+      "epoch": 0.05491508164527039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99674684979312e-05,
+      "loss": 1.2268,
+      "step": 1510
+    },
+    {
+      "epoch": 0.05495144924900898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967373497833637e-05,
+      "loss": 1.3588,
+      "step": 1511
+    },
+    {
+      "epoch": 0.054987816852747576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967278359452913e-05,
+      "loss": 1.2211,
+      "step": 1512
+    },
+    {
+      "epoch": 0.055024184456486165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967183082790343e-05,
+      "loss": 1.2405,
+      "step": 1513
+    },
+    {
+      "epoch": 0.055060552060224754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9967087667847246e-05,
+      "loss": 1.2312,
+      "step": 1514
+    },
+    {
+      "epoch": 0.055096919663963344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966992114624952e-05,
+      "loss": 1.3375,
+      "step": 1515
+    },
+    {
+      "epoch": 0.05513328726770193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966896423124785e-05,
+      "loss": 1.1836,
+      "step": 1516
+    },
+    {
+      "epoch": 0.05516965487144052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996680059334807e-05,
+      "loss": 1.228,
+      "step": 1517
+    },
+    {
+      "epoch": 0.05520602247517911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966704625296135e-05,
+      "loss": 1.1602,
+      "step": 1518
+    },
+    {
+      "epoch": 0.0552423900789177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966608518970315e-05,
+      "loss": 1.2924,
+      "step": 1519
+    },
+    {
+      "epoch": 0.05527875768265629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966512274371944e-05,
+      "loss": 1.2706,
+      "step": 1520
+    },
+    {
+      "epoch": 0.05531512528639488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966415891502357e-05,
+      "loss": 1.3522,
+      "step": 1521
+    },
+    {
+      "epoch": 0.05535149289013347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996631937036289e-05,
+      "loss": 1.2893,
+      "step": 1522
+    },
+    {
+      "epoch": 0.05538786049387206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996622271095488e-05,
+      "loss": 1.1679,
+      "step": 1523
+    },
+    {
+      "epoch": 0.05542422809761065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966125913279672e-05,
+      "loss": 1.3136,
+      "step": 1524
+    },
+    {
+      "epoch": 0.05546059570134924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9966028977338607e-05,
+      "loss": 1.277,
+      "step": 1525
+    },
+    {
+      "epoch": 0.055496963305087826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996593190313303e-05,
+      "loss": 1.208,
+      "step": 1526
+    },
+    {
+      "epoch": 0.055533330908826416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965834690664286e-05,
+      "loss": 1.1572,
+      "step": 1527
+    },
+    {
+      "epoch": 0.055569698512565005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965737339933728e-05,
+      "loss": 1.0994,
+      "step": 1528
+    },
+    {
+      "epoch": 0.055606066116303594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99656398509427e-05,
+      "loss": 1.1554,
+      "step": 1529
+    },
+    {
+      "epoch": 0.055642433720042184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996554222369256e-05,
+      "loss": 1.1966,
+      "step": 1530
+    },
+    {
+      "epoch": 0.05567880132378077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965444458184662e-05,
+      "loss": 1.1982,
+      "step": 1531
+    },
+    {
+      "epoch": 0.05571516892751936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996534655442036e-05,
+      "loss": 1.2741,
+      "step": 1532
+    },
+    {
+      "epoch": 0.05575153653125796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965248512401016e-05,
+      "loss": 1.3082,
+      "step": 1533
+    },
+    {
+      "epoch": 0.05578790413499655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965150332127985e-05,
+      "loss": 1.2068,
+      "step": 1534
+    },
+    {
+      "epoch": 0.05582427173873514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9965052013602633e-05,
+      "loss": 1.2148,
+      "step": 1535
+    },
+    {
+      "epoch": 0.05586063934247373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964953556826325e-05,
+      "loss": 1.3882,
+      "step": 1536
+    },
+    {
+      "epoch": 0.055897006946212316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996485496180042e-05,
+      "loss": 1.3874,
+      "step": 1537
+    },
+    {
+      "epoch": 0.055933374549950905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964756228526293e-05,
+      "loss": 1.1946,
+      "step": 1538
+    },
+    {
+      "epoch": 0.055969742153689495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996465735700531e-05,
+      "loss": 1.3228,
+      "step": 1539
+    },
+    {
+      "epoch": 0.056006109757428084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964558347238846e-05,
+      "loss": 1.2917,
+      "step": 1540
+    },
+    {
+      "epoch": 0.05604247736116667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964459199228274e-05,
+      "loss": 1.204,
+      "step": 1541
+    },
+    {
+      "epoch": 0.05607884496490526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964359912974966e-05,
+      "loss": 1.1685,
+      "step": 1542
+    },
+    {
+      "epoch": 0.05611521256864385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99642604884803e-05,
+      "loss": 1.2363,
+      "step": 1543
+    },
+    {
+      "epoch": 0.05615158017238244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964160925745664e-05,
+      "loss": 1.2592,
+      "step": 1544
+    },
+    {
+      "epoch": 0.05618794777612103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9964061224772428e-05,
+      "loss": 1.1877,
+      "step": 1545
+    },
+    {
+      "epoch": 0.05622431537985962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963961385561978e-05,
+      "loss": 1.3568,
+      "step": 1546
+    },
+    {
+      "epoch": 0.05626068298359821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963861408115704e-05,
+      "loss": 1.1543,
+      "step": 1547
+    },
+    {
+      "epoch": 0.0562970505873368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963761292434988e-05,
+      "loss": 1.1658,
+      "step": 1548
+    },
+    {
+      "epoch": 0.05633341819107539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963661038521222e-05,
+      "loss": 1.2437,
+      "step": 1549
+    },
+    {
+      "epoch": 0.05636978579481398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963560646375796e-05,
+      "loss": 1.3668,
+      "step": 1550
+    },
+    {
+      "epoch": 0.056406153398552566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99634601160001e-05,
+      "loss": 1.2649,
+      "step": 1551
+    },
+    {
+      "epoch": 0.056442521002291156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963359447395532e-05,
+      "loss": 1.293,
+      "step": 1552
+    },
+    {
+      "epoch": 0.05647888860602975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996325864056349e-05,
+      "loss": 1.243,
+      "step": 1553
+    },
+    {
+      "epoch": 0.05651525620976834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9963157695505368e-05,
+      "loss": 1.2419,
+      "step": 1554
+    },
+    {
+      "epoch": 0.05655162381350693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996305661222257e-05,
+      "loss": 1.2337,
+      "step": 1555
+    },
+    {
+      "epoch": 0.05658799141724552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962955390716497e-05,
+      "loss": 1.1678,
+      "step": 1556
+    },
+    {
+      "epoch": 0.05662435902098411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962854030988554e-05,
+      "loss": 1.2594,
+      "step": 1557
+    },
+    {
+      "epoch": 0.0566607266247227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962752533040144e-05,
+      "loss": 1.2107,
+      "step": 1558
+    },
+    {
+      "epoch": 0.05669709422846129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962650896872682e-05,
+      "loss": 1.2977,
+      "step": 1559
+    },
+    {
+      "epoch": 0.05673346183219988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962549122487572e-05,
+      "loss": 1.1635,
+      "step": 1560
+    },
+    {
+      "epoch": 0.05676982943593847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962447209886225e-05,
+      "loss": 1.2775,
+      "step": 1561
+    },
+    {
+      "epoch": 0.056806197039677056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962345159070063e-05,
+      "loss": 1.1356,
+      "step": 1562
+    },
+    {
+      "epoch": 0.056842564643415645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9962242970040497e-05,
+      "loss": 1.2906,
+      "step": 1563
+    },
+    {
+      "epoch": 0.056878932247154235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996214064279894e-05,
+      "loss": 1.2164,
+      "step": 1564
+    },
+    {
+      "epoch": 0.056915299850892824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996203817734682e-05,
+      "loss": 1.3262,
+      "step": 1565
+    },
+    {
+      "epoch": 0.05695166745463141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996193557368555e-05,
+      "loss": 1.2816,
+      "step": 1566
+    },
+    {
+      "epoch": 0.05698803505837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996183283181656e-05,
+      "loss": 1.1892,
+      "step": 1567
+    },
+    {
+      "epoch": 0.05702440266210859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961729951741275e-05,
+      "loss": 1.2908,
+      "step": 1568
+    },
+    {
+      "epoch": 0.05706077026584718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996162693346112e-05,
+      "loss": 1.1345,
+      "step": 1569
+    },
+    {
+      "epoch": 0.05709713786958577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961523776977524e-05,
+      "loss": 1.1456,
+      "step": 1570
+    },
+    {
+      "epoch": 0.05713350547332436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961420482291924e-05,
+      "loss": 1.2292,
+      "step": 1571
+    },
+    {
+      "epoch": 0.05716987307706295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961317049405742e-05,
+      "loss": 1.0717,
+      "step": 1572
+    },
+    {
+      "epoch": 0.05720624068080154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996121347832042e-05,
+      "loss": 1.2141,
+      "step": 1573
+    },
+    {
+      "epoch": 0.057242608284540135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961109769037397e-05,
+      "loss": 1.1577,
+      "step": 1574
+    },
+    {
+      "epoch": 0.057278975888278724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9961005921558106e-05,
+      "loss": 1.1951,
+      "step": 1575
+    },
+    {
+      "epoch": 0.057315343492017314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9960901935883993e-05,
+      "loss": 1.2202,
+      "step": 1576
+    },
+    {
+      "epoch": 0.0573517110957559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9960797812016496e-05,
+      "loss": 1.2611,
+      "step": 1577
+    },
+    {
+      "epoch": 0.05738807869949449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996069354995706e-05,
+      "loss": 1.2351,
+      "step": 1578
+    },
+    {
+      "epoch": 0.05742444630323308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996058914970714e-05,
+      "loss": 1.1263,
+      "step": 1579
+    },
+    {
+      "epoch": 0.05746081390697167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9960484611268173e-05,
+      "loss": 1.2867,
+      "step": 1580
+    },
+    {
+      "epoch": 0.05749718151071026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9960379934641615e-05,
+      "loss": 1.124,
+      "step": 1581
+    },
+    {
+      "epoch": 0.05753354911444885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996027511982892e-05,
+      "loss": 1.1742,
+      "step": 1582
+    },
+    {
+      "epoch": 0.05756991671818744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9960170166831535e-05,
+      "loss": 1.3235,
+      "step": 1583
+    },
+    {
+      "epoch": 0.05760628432192603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996006507565092e-05,
+      "loss": 1.3975,
+      "step": 1584
+    },
+    {
+      "epoch": 0.05764265192566462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995995984628854e-05,
+      "loss": 1.1606,
+      "step": 1585
+    },
+    {
+      "epoch": 0.05767901952940321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959854478745847e-05,
+      "loss": 1.2224,
+      "step": 1586
+    },
+    {
+      "epoch": 0.057715387133141796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959748973024303e-05,
+      "loss": 1.2009,
+      "step": 1587
+    },
+    {
+      "epoch": 0.057751754736880385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995964332912537e-05,
+      "loss": 1.2846,
+      "step": 1588
+    },
+    {
+      "epoch": 0.057788122340618975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995953754705052e-05,
+      "loss": 1.2731,
+      "step": 1589
+    },
+    {
+      "epoch": 0.057824489944357564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959431626801215e-05,
+      "loss": 1.2775,
+      "step": 1590
+    },
+    {
+      "epoch": 0.05786085754809615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959325568378927e-05,
+      "loss": 1.1996,
+      "step": 1591
+    },
+    {
+      "epoch": 0.05789722515183474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9959219371785128e-05,
+      "loss": 1.1893,
+      "step": 1592
+    },
+    {
+      "epoch": 0.05793359275557333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995911303702129e-05,
+      "loss": 1.1566,
+      "step": 1593
+    },
+    {
+      "epoch": 0.05796996035931193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995900656408889e-05,
+      "loss": 1.2762,
+      "step": 1594
+    },
+    {
+      "epoch": 0.05800632796305052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9958899952989403e-05,
+      "loss": 1.2292,
+      "step": 1595
+    },
+    {
+      "epoch": 0.05804269556678911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995879320372431e-05,
+      "loss": 1.2499,
+      "step": 1596
+    },
+    {
+      "epoch": 0.058079063170527696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995868631629509e-05,
+      "loss": 1.1405,
+      "step": 1597
+    },
+    {
+      "epoch": 0.058115430774266286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9958579290703227e-05,
+      "loss": 1.3411,
+      "step": 1598
+    },
+    {
+      "epoch": 0.058151798378004875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9958472126950207e-05,
+      "loss": 1.289,
+      "step": 1599
+    },
+    {
+      "epoch": 0.058188165981743464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9958364825037514e-05,
+      "loss": 1.231,
+      "step": 1600
+    },
+    {
+      "epoch": 0.058224533585482054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995825738496664e-05,
+      "loss": 1.2896,
+      "step": 1601
+    },
+    {
+      "epoch": 0.05826090118922064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995814980673907e-05,
+      "loss": 1.2544,
+      "step": 1602
+    },
+    {
+      "epoch": 0.05829726879295923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9958042090356304e-05,
+      "loss": 1.2705,
+      "step": 1603
+    },
+    {
+      "epoch": 0.05833363639669782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957934235819828e-05,
+      "loss": 1.2173,
+      "step": 1604
+    },
+    {
+      "epoch": 0.05837000400043641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995782624313115e-05,
+      "loss": 1.1382,
+      "step": 1605
+    },
+    {
+      "epoch": 0.058406371604175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957718112291754e-05,
+      "loss": 1.1953,
+      "step": 1606
+    },
+    {
+      "epoch": 0.05844273920791359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995760984330315e-05,
+      "loss": 1.2022,
+      "step": 1607
+    },
+    {
+      "epoch": 0.05847910681165218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995750143616684e-05,
+      "loss": 1.1355,
+      "step": 1608
+    },
+    {
+      "epoch": 0.05851547441539077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957392890884322e-05,
+      "loss": 1.3801,
+      "step": 1609
+    },
+    {
+      "epoch": 0.05855184201912936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957284207457108e-05,
+      "loss": 1.2456,
+      "step": 1610
+    },
+    {
+      "epoch": 0.05858820962286795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99571753858867e-05,
+      "loss": 1.0919,
+      "step": 1611
+    },
+    {
+      "epoch": 0.058624577226606536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957066426174615e-05,
+      "loss": 1.2264,
+      "step": 1612
+    },
+    {
+      "epoch": 0.058660944830345126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956957328322355e-05,
+      "loss": 1.2081,
+      "step": 1613
+    },
+    {
+      "epoch": 0.058697312434083715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956848092331446e-05,
+      "loss": 1.2808,
+      "step": 1614
+    },
+    {
+      "epoch": 0.05873368003782231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995673871820339e-05,
+      "loss": 1.0837,
+      "step": 1615
+    },
+    {
+      "epoch": 0.0587700476415609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956629205939714e-05,
+      "loss": 1.1866,
+      "step": 1616
+    },
+    {
+      "epoch": 0.05880641524529949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956519555541934e-05,
+      "loss": 1.1388,
+      "step": 1617
+    },
+    {
+      "epoch": 0.05884278284903808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956409767011572e-05,
+      "loss": 1.3046,
+      "step": 1618
+    },
+    {
+      "epoch": 0.05887915045277667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995629984035015e-05,
+      "loss": 1.2063,
+      "step": 1619
+    },
+    {
+      "epoch": 0.05891551805651526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956189775559196e-05,
+      "loss": 1.257,
+      "step": 1620
+    },
+    {
+      "epoch": 0.05895188566025385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9956079572640235e-05,
+      "loss": 1.302,
+      "step": 1621
+    },
+    {
+      "epoch": 0.058988253263992436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995596923159479e-05,
+      "loss": 1.2645,
+      "step": 1622
+    },
+    {
+      "epoch": 0.059024620867731026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955858752424403e-05,
+      "loss": 1.2886,
+      "step": 1623
+    },
+    {
+      "epoch": 0.059060988471469615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955748135130602e-05,
+      "loss": 1.3137,
+      "step": 1624
+    },
+    {
+      "epoch": 0.059097356075208204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995563737971492e-05,
+      "loss": 1.1919,
+      "step": 1625
+    },
+    {
+      "epoch": 0.059133723678946794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955526486178894e-05,
+      "loss": 1.3694,
+      "step": 1626
+    },
+    {
+      "epoch": 0.05917009128268538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995541545452406e-05,
+      "loss": 1.1501,
+      "step": 1627
+    },
+    {
+      "epoch": 0.05920645888642397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955304284751965e-05,
+      "loss": 1.2216,
+      "step": 1628
+    },
+    {
+      "epoch": 0.05924282649016256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9955192976864145e-05,
+      "loss": 1.2001,
+      "step": 1629
+    },
+    {
+      "epoch": 0.05927919409390115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995508153086215e-05,
+      "loss": 1.2725,
+      "step": 1630
+    },
+    {
+      "epoch": 0.05931556169763974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954969946747522e-05,
+      "loss": 1.3035,
+      "step": 1631
+    },
+    {
+      "epoch": 0.05935192930137833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995485822452181e-05,
+      "loss": 1.2465,
+      "step": 1632
+    },
+    {
+      "epoch": 0.05938829690511692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995474636418657e-05,
+      "loss": 1.209,
+      "step": 1633
+    },
+    {
+      "epoch": 0.05942466450885551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954634365743338e-05,
+      "loss": 1.2449,
+      "step": 1634
+    },
+    {
+      "epoch": 0.0594610321125941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954522229193685e-05,
+      "loss": 1.2387,
+      "step": 1635
+    },
+    {
+      "epoch": 0.059497399716332694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954409954539156e-05,
+      "loss": 1.1932,
+      "step": 1636
+    },
+    {
+      "epoch": 0.05953376732007128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954297541781312e-05,
+      "loss": 1.3036,
+      "step": 1637
+    },
+    {
+      "epoch": 0.05957013492380987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954184990921716e-05,
+      "loss": 1.2764,
+      "step": 1638
+    },
+    {
+      "epoch": 0.05960650252754846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9954072301961926e-05,
+      "loss": 1.1992,
+      "step": 1639
+    },
+    {
+      "epoch": 0.05964287013128705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953959474903506e-05,
+      "loss": 1.1885,
+      "step": 1640
+    },
+    {
+      "epoch": 0.05967923773502564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953846509748017e-05,
+      "loss": 1.2467,
+      "step": 1641
+    },
+    {
+      "epoch": 0.05971560533876423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953733406497036e-05,
+      "loss": 1.1633,
+      "step": 1642
+    },
+    {
+      "epoch": 0.05975197294250282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953620165152122e-05,
+      "loss": 1.2745,
+      "step": 1643
+    },
+    {
+      "epoch": 0.05978834054624141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953506785714854e-05,
+      "loss": 1.1975,
+      "step": 1644
+    },
+    {
+      "epoch": 0.05982470814998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953393268186798e-05,
+      "loss": 1.1947,
+      "step": 1645
+    },
+    {
+      "epoch": 0.05986107575371859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953279612569535e-05,
+      "loss": 1.0568,
+      "step": 1646
+    },
+    {
+      "epoch": 0.05989744335745718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9953165818864638e-05,
+      "loss": 1.2308,
+      "step": 1647
+    },
+    {
+      "epoch": 0.059933810961195766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995305188707369e-05,
+      "loss": 1.1298,
+      "step": 1648
+    },
+    {
+      "epoch": 0.059970178564934355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9952937817198266e-05,
+      "loss": 1.1873,
+      "step": 1649
+    },
+    {
+      "epoch": 0.060006546168672945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995282360923995e-05,
+      "loss": 1.2639,
+      "step": 1650
+    },
+    {
+      "epoch": 0.060042913772411534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995270926320033e-05,
+      "loss": 1.1906,
+      "step": 1651
+    },
+    {
+      "epoch": 0.06007928137615012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995259477908099e-05,
+      "loss": 1.2048,
+      "step": 1652
+    },
+    {
+      "epoch": 0.06011564897988871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9952480156883513e-05,
+      "loss": 1.2898,
+      "step": 1653
+    },
+    {
+      "epoch": 0.0601520165836273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99523653966095e-05,
+      "loss": 1.3198,
+      "step": 1654
+    },
+    {
+      "epoch": 0.06018838418736589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9952250498260536e-05,
+      "loss": 1.3782,
+      "step": 1655
+    },
+    {
+      "epoch": 0.06022475179110449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995213546183822e-05,
+      "loss": 1.2792,
+      "step": 1656
+    },
+    {
+      "epoch": 0.06026111939484308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995202028734414e-05,
+      "loss": 1.1663,
+      "step": 1657
+    },
+    {
+      "epoch": 0.060297486998581666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99519049747799e-05,
+      "loss": 1.2351,
+      "step": 1658
+    },
+    {
+      "epoch": 0.060333854602320255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99517895241471e-05,
+      "loss": 1.2668,
+      "step": 1659
+    },
+    {
+      "epoch": 0.060370222206058845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951673935447337e-05,
+      "loss": 1.2009,
+      "step": 1660
+    },
+    {
+      "epoch": 0.060406589809797434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995155820868222e-05,
+      "loss": 1.1685,
+      "step": 1661
+    },
+    {
+      "epoch": 0.06044295741353602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951442343853352e-05,
+      "loss": 1.3528,
+      "step": 1662
+    },
+    {
+      "epoch": 0.06047932501727461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951326340962343e-05,
+      "loss": 1.3229,
+      "step": 1663
+    },
+    {
+      "epoch": 0.0605156926210132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9951210200010798e-05,
+      "loss": 1.1343,
+      "step": 1664
+    },
+    {
+      "epoch": 0.06055206022475179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995109392100033e-05,
+      "loss": 1.227,
+      "step": 1665
+    },
+    {
+      "epoch": 0.06058842782849038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995097750393255e-05,
+      "loss": 1.0443,
+      "step": 1666
+    },
+    {
+      "epoch": 0.06062479543222897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995086094880908e-05,
+      "loss": 1.2145,
+      "step": 1667
+    },
+    {
+      "epoch": 0.06066116303596756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995074425563153e-05,
+      "loss": 1.262,
+      "step": 1668
+    },
+    {
+      "epoch": 0.06069753063970615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9950627424401524e-05,
+      "loss": 1.2223,
+      "step": 1669
+    },
+    {
+      "epoch": 0.06073389824344474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995051045512068e-05,
+      "loss": 1.1607,
+      "step": 1670
+    },
+    {
+      "epoch": 0.06077026584718333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9950393347790622e-05,
+      "loss": 1.285,
+      "step": 1671
+    },
+    {
+      "epoch": 0.06080663345092192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995027610241297e-05,
+      "loss": 1.3008,
+      "step": 1672
+    },
+    {
+      "epoch": 0.060843001054660506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9950158718989358e-05,
+      "loss": 1.2589,
+      "step": 1673
+    },
+    {
+      "epoch": 0.060879368658399095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995004119752141e-05,
+      "loss": 1.3017,
+      "step": 1674
+    },
+    {
+      "epoch": 0.060915736262137685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994992353801076e-05,
+      "loss": 1.2335,
+      "step": 1675
+    },
+    {
+      "epoch": 0.060952103865876274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994980574045904e-05,
+      "loss": 1.1439,
+      "step": 1676
+    },
+    {
+      "epoch": 0.06098847146961487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949687804867875e-05,
+      "loss": 1.2484,
+      "step": 1677
+    },
+    {
+      "epoch": 0.06102483907335346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949569731238913e-05,
+      "loss": 1.074,
+      "step": 1678
+    },
+    {
+      "epoch": 0.06106120667709205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994945151957379e-05,
+      "loss": 1.2318,
+      "step": 1679
+    },
+    {
+      "epoch": 0.06109757428083064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949333169874137e-05,
+      "loss": 1.2595,
+      "step": 1680
+    },
+    {
+      "epoch": 0.06113394188456923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9949214682141606e-05,
+      "loss": 1.1564,
+      "step": 1681
+    },
+    {
+      "epoch": 0.06117030948830782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994909605637784e-05,
+      "loss": 1.1798,
+      "step": 1682
+    },
+    {
+      "epoch": 0.061206677092046406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948977292584477e-05,
+      "loss": 1.1965,
+      "step": 1683
+    },
+    {
+      "epoch": 0.061243044695784996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948858390763173e-05,
+      "loss": 1.242,
+      "step": 1684
+    },
+    {
+      "epoch": 0.061279412299523585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948739350915573e-05,
+      "loss": 1.2454,
+      "step": 1685
+    },
+    {
+      "epoch": 0.061315779903262174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994862017304333e-05,
+      "loss": 1.1744,
+      "step": 1686
+    },
+    {
+      "epoch": 0.061352147507000764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99485008571481e-05,
+      "loss": 1.2372,
+      "step": 1687
+    },
+    {
+      "epoch": 0.06138851511073935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948381403231534e-05,
+      "loss": 1.2838,
+      "step": 1688
+    },
+    {
+      "epoch": 0.06142488271447794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994826181129529e-05,
+      "loss": 1.2302,
+      "step": 1689
+    },
+    {
+      "epoch": 0.06146125031821653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994814208134103e-05,
+      "loss": 1.3248,
+      "step": 1690
+    },
+    {
+      "epoch": 0.06149761792195512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948022213370408e-05,
+      "loss": 1.2241,
+      "step": 1691
+    },
+    {
+      "epoch": 0.06153398552569371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99479022073851e-05,
+      "loss": 1.2047,
+      "step": 1692
+    },
+    {
+      "epoch": 0.0615703531294323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9947782063386758e-05,
+      "loss": 1.3295,
+      "step": 1693
+    },
+    {
+      "epoch": 0.06160672073317089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9947661781377054e-05,
+      "loss": 1.3486,
+      "step": 1694
+    },
+    {
+      "epoch": 0.06164308833690948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994754136135766e-05,
+      "loss": 1.1924,
+      "step": 1695
+    },
+    {
+      "epoch": 0.06167945594064807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994742080333024e-05,
+      "loss": 1.1867,
+      "step": 1696
+    },
+    {
+      "epoch": 0.061715823544386664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994730010729647e-05,
+      "loss": 1.2458,
+      "step": 1697
+    },
+    {
+      "epoch": 0.06175219114812525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9947179273258026e-05,
+      "loss": 1.389,
+      "step": 1698
+    },
+    {
+      "epoch": 0.06178855875186384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994705830121658e-05,
+      "loss": 1.2712,
+      "step": 1699
+    },
+    {
+      "epoch": 0.06182492635560243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9946937191173817e-05,
+      "loss": 1.1501,
+      "step": 1700
+    },
+    {
+      "epoch": 0.06186129395934102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994681594313141e-05,
+      "loss": 1.2519,
+      "step": 1701
+    },
+    {
+      "epoch": 0.06189766156307961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9946694557091045e-05,
+      "loss": 1.3047,
+      "step": 1702
+    },
+    {
+      "epoch": 0.0619340291668182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994657303305441e-05,
+      "loss": 1.2142,
+      "step": 1703
+    },
+    {
+      "epoch": 0.06197039677055679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994645137102318e-05,
+      "loss": 1.2344,
+      "step": 1704
+    },
+    {
+      "epoch": 0.06200676437429538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9946329570999054e-05,
+      "loss": 1.19,
+      "step": 1705
+    },
+    {
+      "epoch": 0.06204313197803397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9946207632983715e-05,
+      "loss": 1.2136,
+      "step": 1706
+    },
+    {
+      "epoch": 0.06207949958177256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994608555697886e-05,
+      "loss": 1.2239,
+      "step": 1707
+    },
+    {
+      "epoch": 0.062115867185511146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945963342986175e-05,
+      "loss": 1.2242,
+      "step": 1708
+    },
+    {
+      "epoch": 0.062152234789249736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945840991007363e-05,
+      "loss": 1.2839,
+      "step": 1709
+    },
+    {
+      "epoch": 0.062188602392988325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994571850104412e-05,
+      "loss": 1.295,
+      "step": 1710
+    },
+    {
+      "epoch": 0.062224969996726914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994559587309814e-05,
+      "loss": 1.2571,
+      "step": 1711
+    },
+    {
+      "epoch": 0.062261337600465504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994547310717113e-05,
+      "loss": 1.2515,
+      "step": 1712
+    },
+    {
+      "epoch": 0.06229770520420409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994535020326479e-05,
+      "loss": 1.1362,
+      "step": 1713
+    },
+    {
+      "epoch": 0.06233407280794268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9945227161380828e-05,
+      "loss": 1.0798,
+      "step": 1714
+    },
+    {
+      "epoch": 0.06237044041168127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994510398152095e-05,
+      "loss": 1.2215,
+      "step": 1715
+    },
+    {
+      "epoch": 0.06240680801541986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944980663686864e-05,
+      "loss": 1.1868,
+      "step": 1716
+    },
+    {
+      "epoch": 0.06244317561915845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994485720788028e-05,
+      "loss": 1.2971,
+      "step": 1717
+    },
+    {
+      "epoch": 0.06247954322289705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944733614102914e-05,
+      "loss": 1.3182,
+      "step": 1718
+    },
+    {
+      "epoch": 0.06251591082663563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944609882356478e-05,
+      "loss": 1.3335,
+      "step": 1719
+    },
+    {
+      "epoch": 0.06255227843037423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994448601264269e-05,
+      "loss": 1.2116,
+      "step": 1720
+    },
+    {
+      "epoch": 0.06258864603411281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994436200496327e-05,
+      "loss": 1.2076,
+      "step": 1721
+    },
+    {
+      "epoch": 0.0626250136378514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944237859319932e-05,
+      "loss": 1.2312,
+      "step": 1722
+    },
+    {
+      "epoch": 0.06266138124158999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9944113575714406e-05,
+      "loss": 1.1976,
+      "step": 1723
+    },
+    {
+      "epoch": 0.06269774884532858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994398915414841e-05,
+      "loss": 1.3217,
+      "step": 1724
+    },
+    {
+      "epoch": 0.06273411644906716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943864594623677e-05,
+      "loss": 1.1882,
+      "step": 1725
+    },
+    {
+      "epoch": 0.06277048405280576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994373989714193e-05,
+      "loss": 1.1841,
+      "step": 1726
+    },
+    {
+      "epoch": 0.06280685165654434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943615061704898e-05,
+      "loss": 1.3237,
+      "step": 1727
+    },
+    {
+      "epoch": 0.06284321926028294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943490088314317e-05,
+      "loss": 1.3177,
+      "step": 1728
+    },
+    {
+      "epoch": 0.06287958686402154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994336497697192e-05,
+      "loss": 1.2647,
+      "step": 1729
+    },
+    {
+      "epoch": 0.06291595446776012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943239727679438e-05,
+      "loss": 1.2461,
+      "step": 1730
+    },
+    {
+      "epoch": 0.06295232207149871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9943114340438615e-05,
+      "loss": 1.2814,
+      "step": 1731
+    },
+    {
+      "epoch": 0.0629886896752373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994298881525119e-05,
+      "loss": 1.1076,
+      "step": 1732
+    },
+    {
+      "epoch": 0.0630250572789759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942863152118898e-05,
+      "loss": 1.2639,
+      "step": 1733
+    },
+    {
+      "epoch": 0.06306142488271448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942737351043487e-05,
+      "loss": 1.2898,
+      "step": 1734
+    },
+    {
+      "epoch": 0.06309779248645307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942611412026707e-05,
+      "loss": 1.3094,
+      "step": 1735
+    },
+    {
+      "epoch": 0.06313416009019165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942485335070296e-05,
+      "loss": 1.1627,
+      "step": 1736
+    },
+    {
+      "epoch": 0.06317052769393025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994235912017601e-05,
+      "loss": 1.3791,
+      "step": 1737
+    },
+    {
+      "epoch": 0.06320689529766883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9942232767345596e-05,
+      "loss": 1.1188,
+      "step": 1738
+    },
+    {
+      "epoch": 0.06324326290140743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994210627658081e-05,
+      "loss": 1.1931,
+      "step": 1739
+    },
+    {
+      "epoch": 0.06327963050514601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99419796478834e-05,
+      "loss": 1.1193,
+      "step": 1740
+    },
+    {
+      "epoch": 0.06331599810888461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9941852881255137e-05,
+      "loss": 1.3253,
+      "step": 1741
+    },
+    {
+      "epoch": 0.06335236571262319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9941725976697768e-05,
+      "loss": 1.2723,
+      "step": 1742
+    },
+    {
+      "epoch": 0.06338873331636179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9941598934213055e-05,
+      "loss": 1.2343,
+      "step": 1743
+    },
+    {
+      "epoch": 0.06342510092010037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994147175380276e-05,
+      "loss": 1.1061,
+      "step": 1744
+    },
+    {
+      "epoch": 0.06346146852383897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9941344435468656e-05,
+      "loss": 1.1878,
+      "step": 1745
+    },
+    {
+      "epoch": 0.06349783612757755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9941216979212498e-05,
+      "loss": 1.1923,
+      "step": 1746
+    },
+    {
+      "epoch": 0.06353420373131614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994108938503606e-05,
+      "loss": 1.2567,
+      "step": 1747
+    },
+    {
+      "epoch": 0.06357057133505473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940961652941108e-05,
+      "loss": 1.1992,
+      "step": 1748
+    },
+    {
+      "epoch": 0.06360693893879332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940833782929424e-05,
+      "loss": 1.4183,
+      "step": 1749
+    },
+    {
+      "epoch": 0.06364330654253192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.994070577500277e-05,
+      "loss": 1.287,
+      "step": 1750
+    },
+    {
+      "epoch": 0.0636796741462705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940577629162928e-05,
+      "loss": 1.2825,
+      "step": 1751
+    },
+    {
+      "epoch": 0.0637160417500091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940449345411678e-05,
+      "loss": 1.2925,
+      "step": 1752
+    },
+    {
+      "epoch": 0.06375240935374768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940320923750796e-05,
+      "loss": 1.325,
+      "step": 1753
+    },
+    {
+      "epoch": 0.06378877695748628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940192364182064e-05,
+      "loss": 1.1599,
+      "step": 1754
+    },
+    {
+      "epoch": 0.06382514456122486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9940063666707264e-05,
+      "loss": 1.2202,
+      "step": 1755
+    },
+    {
+      "epoch": 0.06386151216496345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993993483132819e-05,
+      "loss": 1.1714,
+      "step": 1756
+    },
+    {
+      "epoch": 0.06389787976870204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993980585804662e-05,
+      "loss": 1.1873,
+      "step": 1757
+    },
+    {
+      "epoch": 0.06393424737244063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939676746864345e-05,
+      "loss": 1.2031,
+      "step": 1758
+    },
+    {
+      "epoch": 0.06397061497617922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939547497783157e-05,
+      "loss": 1.2084,
+      "step": 1759
+    },
+    {
+      "epoch": 0.06400698257991781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939418110804852e-05,
+      "loss": 1.2005,
+      "step": 1760
+    },
+    {
+      "epoch": 0.0640433501836564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939288585931224e-05,
+      "loss": 1.2151,
+      "step": 1761
+    },
+    {
+      "epoch": 0.06407971778739499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939158923164068e-05,
+      "loss": 1.2961,
+      "step": 1762
+    },
+    {
+      "epoch": 0.06411608539113357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9939029122505182e-05,
+      "loss": 1.2579,
+      "step": 1763
+    },
+    {
+      "epoch": 0.06415245299487217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993889918395637e-05,
+      "loss": 1.2639,
+      "step": 1764
+    },
+    {
+      "epoch": 0.06418882059861075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993876910751943e-05,
+      "loss": 1.2004,
+      "step": 1765
+    },
+    {
+      "epoch": 0.06422518820234935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938638893196175e-05,
+      "loss": 1.2281,
+      "step": 1766
+    },
+    {
+      "epoch": 0.06426155580608793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938508540988404e-05,
+      "loss": 1.1599,
+      "step": 1767
+    },
+    {
+      "epoch": 0.06429792340982653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938378050897927e-05,
+      "loss": 1.255,
+      "step": 1768
+    },
+    {
+      "epoch": 0.06433429101356512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938247422926556e-05,
+      "loss": 1.2714,
+      "step": 1769
+    },
+    {
+      "epoch": 0.0643706586173037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9938116657076104e-05,
+      "loss": 1.351,
+      "step": 1770
+    },
+    {
+      "epoch": 0.0644070262210423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937985753348383e-05,
+      "loss": 1.2526,
+      "step": 1771
+    },
+    {
+      "epoch": 0.06444339382478088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993785471174521e-05,
+      "loss": 1.2003,
+      "step": 1772
+    },
+    {
+      "epoch": 0.06447976142851948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937723532268403e-05,
+      "loss": 1.3251,
+      "step": 1773
+    },
+    {
+      "epoch": 0.06451612903225806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937592214919782e-05,
+      "loss": 1.2751,
+      "step": 1774
+    },
+    {
+      "epoch": 0.06455249663599666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937460759701164e-05,
+      "loss": 1.2277,
+      "step": 1775
+    },
+    {
+      "epoch": 0.06458886423973524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993732916661438e-05,
+      "loss": 1.2219,
+      "step": 1776
+    },
+    {
+      "epoch": 0.06462523184347384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937197435661257e-05,
+      "loss": 1.3012,
+      "step": 1777
+    },
+    {
+      "epoch": 0.06466159944721242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9937065566843613e-05,
+      "loss": 1.0673,
+      "step": 1778
+    },
+    {
+      "epoch": 0.06469796705095102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936933560163287e-05,
+      "loss": 1.2227,
+      "step": 1779
+    },
+    {
+      "epoch": 0.0647343346546896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936801415622104e-05,
+      "loss": 1.3054,
+      "step": 1780
+    },
+    {
+      "epoch": 0.0647707022584282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99366691332219e-05,
+      "loss": 1.22,
+      "step": 1781
+    },
+    {
+      "epoch": 0.06480706986216678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936536712964515e-05,
+      "loss": 1.399,
+      "step": 1782
+    },
+    {
+      "epoch": 0.06484343746590537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936404154851777e-05,
+      "loss": 1.2375,
+      "step": 1783
+    },
+    {
+      "epoch": 0.06487980506964396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993627145888553e-05,
+      "loss": 1.1613,
+      "step": 1784
+    },
+    {
+      "epoch": 0.06491617267338255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9936138625067612e-05,
+      "loss": 1.2284,
+      "step": 1785
+    },
+    {
+      "epoch": 0.06495254027712113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993600565339987e-05,
+      "loss": 1.1697,
+      "step": 1786
+    },
+    {
+      "epoch": 0.06498890788085973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9935872543884148e-05,
+      "loss": 1.3156,
+      "step": 1787
+    },
+    {
+      "epoch": 0.06502527548459831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993573929652229e-05,
+      "loss": 1.1292,
+      "step": 1788
+    },
+    {
+      "epoch": 0.06506164308833691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993560591131615e-05,
+      "loss": 1.3876,
+      "step": 1789
+    },
+    {
+      "epoch": 0.0650980106920755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993547238826757e-05,
+      "loss": 1.2069,
+      "step": 1790
+    },
+    {
+      "epoch": 0.06513437829581409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993533872737841e-05,
+      "loss": 1.231,
+      "step": 1791
+    },
+    {
+      "epoch": 0.06517074589955268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9935204928650517e-05,
+      "loss": 1.181,
+      "step": 1792
+    },
+    {
+      "epoch": 0.06520711350329127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9935070992085757e-05,
+      "loss": 1.252,
+      "step": 1793
+    },
+    {
+      "epoch": 0.06524348110702986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993493691768598e-05,
+      "loss": 1.2939,
+      "step": 1794
+    },
+    {
+      "epoch": 0.06527984871076845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993480270545305e-05,
+      "loss": 1.2898,
+      "step": 1795
+    },
+    {
+      "epoch": 0.06531621631450704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9934668355388828e-05,
+      "loss": 1.2749,
+      "step": 1796
+    },
+    {
+      "epoch": 0.06535258391824562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993453386749518e-05,
+      "loss": 1.249,
+      "step": 1797
+    },
+    {
+      "epoch": 0.06538895152198422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9934399241773968e-05,
+      "loss": 1.3004,
+      "step": 1798
+    },
+    {
+      "epoch": 0.0654253191257228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993426447822706e-05,
+      "loss": 1.1116,
+      "step": 1799
+    },
+    {
+      "epoch": 0.0654616867294614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993412957685633e-05,
+      "loss": 1.2596,
+      "step": 1800
+    },
+    {
+      "epoch": 0.06549805433319998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993399453766365e-05,
+      "loss": 1.2653,
+      "step": 1801
+    },
+    {
+      "epoch": 0.06553442193693858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993385936065088e-05,
+      "loss": 1.3171,
+      "step": 1802
+    },
+    {
+      "epoch": 0.06557078954067716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933724045819916e-05,
+      "loss": 1.2373,
+      "step": 1803
+    },
+    {
+      "epoch": 0.06560715714441576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993358859317262e-05,
+      "loss": 1.1629,
+      "step": 1804
+    },
+    {
+      "epoch": 0.06564352474815434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933453002710878e-05,
+      "loss": 1.1903,
+      "step": 1805
+    },
+    {
+      "epoch": 0.06567989235189294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993331727443657e-05,
+      "loss": 1.1709,
+      "step": 1806
+    },
+    {
+      "epoch": 0.06571625995563152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933181408351574e-05,
+      "loss": 1.2693,
+      "step": 1807
+    },
+    {
+      "epoch": 0.06575262755937011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933045404457782e-05,
+      "loss": 1.246,
+      "step": 1808
+    },
+    {
+      "epoch": 0.0657889951631087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993290926275708e-05,
+      "loss": 1.1996,
+      "step": 1809
+    },
+    {
+      "epoch": 0.06582536276684729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9932772983251353e-05,
+      "loss": 1.184,
+      "step": 1810
+    },
+    {
+      "epoch": 0.06586173037058589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993263656594249e-05,
+      "loss": 1.2749,
+      "step": 1811
+    },
+    {
+      "epoch": 0.06589809797432447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9932500010832392e-05,
+      "loss": 1.311,
+      "step": 1812
+    },
+    {
+      "epoch": 0.06593446557806307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9932363317922945e-05,
+      "loss": 1.1844,
+      "step": 1813
+    },
+    {
+      "epoch": 0.06597083318180165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9932226487216055e-05,
+      "loss": 1.1396,
+      "step": 1814
+    },
+    {
+      "epoch": 0.06600720078554025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993208951871361e-05,
+      "loss": 1.2139,
+      "step": 1815
+    },
+    {
+      "epoch": 0.06604356838927883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931952412417517e-05,
+      "loss": 1.1156,
+      "step": 1816
+    },
+    {
+      "epoch": 0.06607993599301742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931815168329674e-05,
+      "loss": 1.1904,
+      "step": 1817
+    },
+    {
+      "epoch": 0.06611630359675601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931677786451987e-05,
+      "loss": 1.1513,
+      "step": 1818
+    },
+    {
+      "epoch": 0.0661526712004946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931540266786366e-05,
+      "loss": 1.2491,
+      "step": 1819
+    },
+    {
+      "epoch": 0.06618903880423319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993140260933471e-05,
+      "loss": 1.2323,
+      "step": 1820
+    },
+    {
+      "epoch": 0.06622540640797178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931264814098936e-05,
+      "loss": 1.3236,
+      "step": 1821
+    },
+    {
+      "epoch": 0.06626177401171036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9931126881080952e-05,
+      "loss": 1.2526,
+      "step": 1822
+    },
+    {
+      "epoch": 0.06629814161544896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993098881028267e-05,
+      "loss": 1.1384,
+      "step": 1823
+    },
+    {
+      "epoch": 0.06633450921918754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930850601706015e-05,
+      "loss": 1.3436,
+      "step": 1824
+    },
+    {
+      "epoch": 0.06637087682292614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930712255352896e-05,
+      "loss": 1.0617,
+      "step": 1825
+    },
+    {
+      "epoch": 0.06640724442666472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.993057377122523e-05,
+      "loss": 1.1383,
+      "step": 1826
+    },
+    {
+      "epoch": 0.06644361203040332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930435149324947e-05,
+      "loss": 1.1978,
+      "step": 1827
+    },
+    {
+      "epoch": 0.0664799796341419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930296389653967e-05,
+      "loss": 1.2666,
+      "step": 1828
+    },
+    {
+      "epoch": 0.0665163472378805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930157492214207e-05,
+      "loss": 1.1742,
+      "step": 1829
+    },
+    {
+      "epoch": 0.06655271484161908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9930018457007607e-05,
+      "loss": 1.1824,
+      "step": 1830
+    },
+    {
+      "epoch": 0.06658908244535768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992987928403609e-05,
+      "loss": 1.1673,
+      "step": 1831
+    },
+    {
+      "epoch": 0.06662545004909627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929739973301582e-05,
+      "loss": 1.1454,
+      "step": 1832
+    },
+    {
+      "epoch": 0.06666181765283485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992960052480602e-05,
+      "loss": 1.1581,
+      "step": 1833
+    },
+    {
+      "epoch": 0.06669818525657345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929460938551347e-05,
+      "loss": 1.1556,
+      "step": 1834
+    },
+    {
+      "epoch": 0.06673455286031203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9929321214539486e-05,
+      "loss": 1.2023,
+      "step": 1835
+    },
+    {
+      "epoch": 0.06677092046405063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992918135277238e-05,
+      "loss": 1.2284,
+      "step": 1836
+    },
+    {
+      "epoch": 0.06680728806778921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992904135325197e-05,
+      "loss": 1.0509,
+      "step": 1837
+    },
+    {
+      "epoch": 0.06684365567152781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928901215980202e-05,
+      "loss": 1.1373,
+      "step": 1838
+    },
+    {
+      "epoch": 0.06688002327526639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928760940959016e-05,
+      "loss": 1.2747,
+      "step": 1839
+    },
+    {
+      "epoch": 0.06691639087900499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928620528190358e-05,
+      "loss": 1.3114,
+      "step": 1840
+    },
+    {
+      "epoch": 0.06695275848274357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992847997767618e-05,
+      "loss": 1.2065,
+      "step": 1841
+    },
+    {
+      "epoch": 0.06698912608648216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928339289418422e-05,
+      "loss": 1.2114,
+      "step": 1842
+    },
+    {
+      "epoch": 0.06702549369022075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928198463419052e-05,
+      "loss": 1.2915,
+      "step": 1843
+    },
+    {
+      "epoch": 0.06706186129395934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9928057499680008e-05,
+      "loss": 1.1555,
+      "step": 1844
+    },
+    {
+      "epoch": 0.06709822889769793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9927916398203253e-05,
+      "loss": 1.2139,
+      "step": 1845
+    },
+    {
+      "epoch": 0.06713459650143652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9927775158990745e-05,
+      "loss": 1.1563,
+      "step": 1846
+    },
+    {
+      "epoch": 0.0671709641051751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992763378204444e-05,
+      "loss": 1.2068,
+      "step": 1847
+    },
+    {
+      "epoch": 0.0672073317089137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9927492267366306e-05,
+      "loss": 1.2435,
+      "step": 1848
+    },
+    {
+      "epoch": 0.06724369931265228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99273506149583e-05,
+      "loss": 1.3901,
+      "step": 1849
+    },
+    {
+      "epoch": 0.06728006691639088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9927208824822388e-05,
+      "loss": 1.2629,
+      "step": 1850
+    },
+    {
+      "epoch": 0.06731643452012948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992706689696054e-05,
+      "loss": 1.1955,
+      "step": 1851
+    },
+    {
+      "epoch": 0.06735280212386806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9926924831374723e-05,
+      "loss": 1.3475,
+      "step": 1852
+    },
+    {
+      "epoch": 0.06738916972760665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992678262806691e-05,
+      "loss": 1.2072,
+      "step": 1853
+    },
+    {
+      "epoch": 0.06742553733134524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992664028703907e-05,
+      "loss": 1.2756,
+      "step": 1854
+    },
+    {
+      "epoch": 0.06746190493508383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9926497808293182e-05,
+      "loss": 1.2885,
+      "step": 1855
+    },
+    {
+      "epoch": 0.06749827253882242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992635519183122e-05,
+      "loss": 1.3569,
+      "step": 1856
+    },
+    {
+      "epoch": 0.06753464014256101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9926212437655163e-05,
+      "loss": 1.1739,
+      "step": 1857
+    },
+    {
+      "epoch": 0.0675710077462996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992606954576699e-05,
+      "loss": 1.2004,
+      "step": 1858
+    },
+    {
+      "epoch": 0.06760737535003819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992592651616869e-05,
+      "loss": 1.2419,
+      "step": 1859
+    },
+    {
+      "epoch": 0.06764374295377677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992578334886224e-05,
+      "loss": 1.1438,
+      "step": 1860
+    },
+    {
+      "epoch": 0.06768011055751537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992564004384963e-05,
+      "loss": 1.293,
+      "step": 1861
+    },
+    {
+      "epoch": 0.06771647816125395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9925496601132845e-05,
+      "loss": 1.2458,
+      "step": 1862
+    },
+    {
+      "epoch": 0.06775284576499255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9925353020713878e-05,
+      "loss": 1.3142,
+      "step": 1863
+    },
+    {
+      "epoch": 0.06778921336873113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992520930259472e-05,
+      "loss": 1.2255,
+      "step": 1864
+    },
+    {
+      "epoch": 0.06782558097246973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9925065446777365e-05,
+      "loss": 1.0754,
+      "step": 1865
+    },
+    {
+      "epoch": 0.06786194857620831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9924921453263807e-05,
+      "loss": 1.226,
+      "step": 1866
+    },
+    {
+      "epoch": 0.0678983161799469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9924777322056048e-05,
+      "loss": 1.2142,
+      "step": 1867
+    },
+    {
+      "epoch": 0.06793468378368549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9924633053156084e-05,
+      "loss": 1.2001,
+      "step": 1868
+    },
+    {
+      "epoch": 0.06797105138742408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992448864656592e-05,
+      "loss": 1.3129,
+      "step": 1869
+    },
+    {
+      "epoch": 0.06800741899116267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9924344102287555e-05,
+      "loss": 1.1682,
+      "step": 1870
+    },
+    {
+      "epoch": 0.06804378659490126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9924199420322995e-05,
+      "loss": 1.2722,
+      "step": 1871
+    },
+    {
+      "epoch": 0.06808015419863986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992405460067425e-05,
+      "loss": 1.3156,
+      "step": 1872
+    },
+    {
+      "epoch": 0.06811652180237844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923909643343328e-05,
+      "loss": 1.2192,
+      "step": 1873
+    },
+    {
+      "epoch": 0.06815288940611704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992376454833224e-05,
+      "loss": 1.2696,
+      "step": 1874
+    },
+    {
+      "epoch": 0.06818925700985562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923619315642998e-05,
+      "loss": 1.1939,
+      "step": 1875
+    },
+    {
+      "epoch": 0.06822562461359422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923473945277618e-05,
+      "loss": 1.1343,
+      "step": 1876
+    },
+    {
+      "epoch": 0.0682619922173328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923328437238115e-05,
+      "loss": 1.18,
+      "step": 1877
+    },
+    {
+      "epoch": 0.0682983598210714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9923182791526516e-05,
+      "loss": 1.1817,
+      "step": 1878
+    },
+    {
+      "epoch": 0.06833472742480998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992303700814483e-05,
+      "loss": 1.1929,
+      "step": 1879
+    },
+    {
+      "epoch": 0.06837109502854857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9922891087095083e-05,
+      "loss": 1.2526,
+      "step": 1880
+    },
+    {
+      "epoch": 0.06840746263228716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99227450283793e-05,
+      "loss": 1.1078,
+      "step": 1881
+    },
+    {
+      "epoch": 0.06844383023602575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992259883199951e-05,
+      "loss": 1.1687,
+      "step": 1882
+    },
+    {
+      "epoch": 0.06848019783976433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992245249795774e-05,
+      "loss": 1.2145,
+      "step": 1883
+    },
+    {
+      "epoch": 0.06851656544350293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9922306026256023e-05,
+      "loss": 1.1905,
+      "step": 1884
+    },
+    {
+      "epoch": 0.06855293304724151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9922159416896384e-05,
+      "loss": 1.2336,
+      "step": 1885
+    },
+    {
+      "epoch": 0.06858930065098011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992201266988086e-05,
+      "loss": 1.2902,
+      "step": 1886
+    },
+    {
+      "epoch": 0.06862566825471869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921865785211493e-05,
+      "loss": 1.3865,
+      "step": 1887
+    },
+    {
+      "epoch": 0.06866203585845729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992171876289031e-05,
+      "loss": 1.2401,
+      "step": 1888
+    },
+    {
+      "epoch": 0.06869840346219587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921571602919356e-05,
+      "loss": 1.1814,
+      "step": 1889
+    },
+    {
+      "epoch": 0.06873477106593447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921424305300677e-05,
+      "loss": 1.2476,
+      "step": 1890
+    },
+    {
+      "epoch": 0.06877113866967305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9921276870036312e-05,
+      "loss": 1.1111,
+      "step": 1891
+    },
+    {
+      "epoch": 0.06880750627341164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99211292971283e-05,
+      "loss": 1.2194,
+      "step": 1892
+    },
+    {
+      "epoch": 0.06884387387715024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9920981586578704e-05,
+      "loss": 1.3134,
+      "step": 1893
+    },
+    {
+      "epoch": 0.06888024148088882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992083373838956e-05,
+      "loss": 1.2624,
+      "step": 1894
+    },
+    {
+      "epoch": 0.06891660908462742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9920685752562925e-05,
+      "loss": 1.0251,
+      "step": 1895
+    },
+    {
+      "epoch": 0.068952976688366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9920537629100848e-05,
+      "loss": 1.1723,
+      "step": 1896
+    },
+    {
+      "epoch": 0.0689893442921046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992038936800539e-05,
+      "loss": 1.3649,
+      "step": 1897
+    },
+    {
+      "epoch": 0.06902571189584318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9920240969278605e-05,
+      "loss": 1.218,
+      "step": 1898
+    },
+    {
+      "epoch": 0.06906207949958178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.992009243292255e-05,
+      "loss": 1.2426,
+      "step": 1899
+    },
+    {
+      "epoch": 0.06909844710332036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9919943758939288e-05,
+      "loss": 1.1733,
+      "step": 1900
+    },
+    {
+      "epoch": 0.06913481470705896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9919794947330883e-05,
+      "loss": 1.3403,
+      "step": 1901
+    },
+    {
+      "epoch": 0.06917118231079754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9919645998099394e-05,
+      "loss": 1.1436,
+      "step": 1902
+    },
+    {
+      "epoch": 0.06920754991453613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991949691124689e-05,
+      "loss": 1.2699,
+      "step": 1903
+    },
+    {
+      "epoch": 0.06924391751827472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9919347686775445e-05,
+      "loss": 1.2642,
+      "step": 1904
+    },
+    {
+      "epoch": 0.06928028512201331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991919832468712e-05,
+      "loss": 1.2963,
+      "step": 1905
+    },
+    {
+      "epoch": 0.0693166527257519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9919048824983993e-05,
+      "loss": 1.2603,
+      "step": 1906
+    },
+    {
+      "epoch": 0.06935302032949049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991889918766814e-05,
+      "loss": 1.2443,
+      "step": 1907
+    },
+    {
+      "epoch": 0.06938938793322907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991874941274163e-05,
+      "loss": 1.1371,
+      "step": 1908
+    },
+    {
+      "epoch": 0.06942575553696767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9918599500206545e-05,
+      "loss": 1.217,
+      "step": 1909
+    },
+    {
+      "epoch": 0.06946212314070625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9918449450064966e-05,
+      "loss": 1.142,
+      "step": 1910
+    },
+    {
+      "epoch": 0.06949849074444485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9918299262318974e-05,
+      "loss": 1.2208,
+      "step": 1911
+    },
+    {
+      "epoch": 0.06953485834818343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991814893697065e-05,
+      "loss": 1.1089,
+      "step": 1912
+    },
+    {
+      "epoch": 0.06957122595192203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917998474022083e-05,
+      "loss": 1.0822,
+      "step": 1913
+    },
+    {
+      "epoch": 0.06960759355566062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917847873475357e-05,
+      "loss": 1.2971,
+      "step": 1914
+    },
+    {
+      "epoch": 0.0696439611593992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917697135332567e-05,
+      "loss": 1.384,
+      "step": 1915
+    },
+    {
+      "epoch": 0.0696803287631378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917546259595798e-05,
+      "loss": 1.2544,
+      "step": 1916
+    },
+    {
+      "epoch": 0.06971669636687639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917395246267145e-05,
+      "loss": 1.1928,
+      "step": 1917
+    },
+    {
+      "epoch": 0.06975306397061498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917244095348707e-05,
+      "loss": 1.3102,
+      "step": 1918
+    },
+    {
+      "epoch": 0.06978943157435356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9917092806842577e-05,
+      "loss": 1.2943,
+      "step": 1919
+    },
+    {
+      "epoch": 0.06982579917809216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9916941380750852e-05,
+      "loss": 1.2545,
+      "step": 1920
+    },
+    {
+      "epoch": 0.06986216678183074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9916789817075637e-05,
+      "loss": 1.1992,
+      "step": 1921
+    },
+    {
+      "epoch": 0.06989853438556934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9916638115819032e-05,
+      "loss": 1.2801,
+      "step": 1922
+    },
+    {
+      "epoch": 0.06993490198930792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9916486276983148e-05,
+      "loss": 1.2552,
+      "step": 1923
+    },
+    {
+      "epoch": 0.06997126959304652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9916334300570083e-05,
+      "loss": 1.3411,
+      "step": 1924
+    },
+    {
+      "epoch": 0.0700076371967851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9916182186581946e-05,
+      "loss": 1.2547,
+      "step": 1925
+    },
+    {
+      "epoch": 0.0700440048005237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9916029935020856e-05,
+      "loss": 1.2865,
+      "step": 1926
+    },
+    {
+      "epoch": 0.07008037240426228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9915877545888917e-05,
+      "loss": 1.1493,
+      "step": 1927
+    },
+    {
+      "epoch": 0.07011674000800087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9915725019188245e-05,
+      "loss": 1.2647,
+      "step": 1928
+    },
+    {
+      "epoch": 0.07015310761173946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991557235492096e-05,
+      "loss": 1.2236,
+      "step": 1929
+    },
+    {
+      "epoch": 0.07018947521547805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9915419553089178e-05,
+      "loss": 1.1588,
+      "step": 1930
+    },
+    {
+      "epoch": 0.07022584281921664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9915266613695014e-05,
+      "loss": 1.1949,
+      "step": 1931
+    },
+    {
+      "epoch": 0.07026221042295523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9915113536740595e-05,
+      "loss": 1.1403,
+      "step": 1932
+    },
+    {
+      "epoch": 0.07029857802669381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914960322228042e-05,
+      "loss": 1.3029,
+      "step": 1933
+    },
+    {
+      "epoch": 0.07033494563043241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914806970159484e-05,
+      "loss": 1.2168,
+      "step": 1934
+    },
+    {
+      "epoch": 0.070371313234171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914653480537047e-05,
+      "loss": 1.1143,
+      "step": 1935
+    },
+    {
+      "epoch": 0.07040768083790959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914499853362853e-05,
+      "loss": 1.241,
+      "step": 1936
+    },
+    {
+      "epoch": 0.07044404844164819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914346088639046e-05,
+      "loss": 1.1781,
+      "step": 1937
+    },
+    {
+      "epoch": 0.07048041604538677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914192186367755e-05,
+      "loss": 1.1447,
+      "step": 1938
+    },
+    {
+      "epoch": 0.07051678364912536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9914038146551113e-05,
+      "loss": 1.4045,
+      "step": 1939
+    },
+    {
+      "epoch": 0.07055315125286395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991388396919126e-05,
+      "loss": 1.163,
+      "step": 1940
+    },
+    {
+      "epoch": 0.07058951885660254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913729654290327e-05,
+      "loss": 1.1871,
+      "step": 1941
+    },
+    {
+      "epoch": 0.07062588646034113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913575201850465e-05,
+      "loss": 1.266,
+      "step": 1942
+    },
+    {
+      "epoch": 0.07066225406407972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913420611873812e-05,
+      "loss": 1.0867,
+      "step": 1943
+    },
+    {
+      "epoch": 0.0706986216678183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913265884362515e-05,
+      "loss": 1.288,
+      "step": 1944
+    },
+    {
+      "epoch": 0.0707349892715569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913111019318715e-05,
+      "loss": 1.2964,
+      "step": 1945
+    },
+    {
+      "epoch": 0.07077135687529548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912956016744568e-05,
+      "loss": 1.2441,
+      "step": 1946
+    },
+    {
+      "epoch": 0.07080772447903408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912800876642217e-05,
+      "loss": 1.2212,
+      "step": 1947
+    },
+    {
+      "epoch": 0.07084409208277266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991264559901382e-05,
+      "loss": 1.0316,
+      "step": 1948
+    },
+    {
+      "epoch": 0.07088045968651126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991249018386153e-05,
+      "loss": 1.1478,
+      "step": 1949
+    },
+    {
+      "epoch": 0.07091682729024984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912334631187504e-05,
+      "loss": 1.1988,
+      "step": 1950
+    },
+    {
+      "epoch": 0.07095319489398844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912178940993897e-05,
+      "loss": 1.1549,
+      "step": 1951
+    },
+    {
+      "epoch": 0.07098956249772702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9912023113282872e-05,
+      "loss": 1.1672,
+      "step": 1952
+    },
+    {
+      "epoch": 0.07102593010146561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9911867148056587e-05,
+      "loss": 1.3302,
+      "step": 1953
+    },
+    {
+      "epoch": 0.07106229770520421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9911711045317213e-05,
+      "loss": 1.2766,
+      "step": 1954
+    },
+    {
+      "epoch": 0.0710986653089428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9911554805066905e-05,
+      "loss": 1.26,
+      "step": 1955
+    },
+    {
+      "epoch": 0.07113503291268139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9911398427307842e-05,
+      "loss": 1.2302,
+      "step": 1956
+    },
+    {
+      "epoch": 0.07117140051641997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9911241912042186e-05,
+      "loss": 1.24,
+      "step": 1957
+    },
+    {
+      "epoch": 0.07120776812015857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9911085259272112e-05,
+      "loss": 1.1752,
+      "step": 1958
+    },
+    {
+      "epoch": 0.07124413572389715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991092846899979e-05,
+      "loss": 1.353,
+      "step": 1959
+    },
+    {
+      "epoch": 0.07128050332763575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99107715412274e-05,
+      "loss": 1.2616,
+      "step": 1960
+    },
+    {
+      "epoch": 0.07131687093137433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9910614475957114e-05,
+      "loss": 1.2952,
+      "step": 1961
+    },
+    {
+      "epoch": 0.07135323853511293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9910457273191115e-05,
+      "loss": 1.2481,
+      "step": 1962
+    },
+    {
+      "epoch": 0.07138960613885151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.991029993293158e-05,
+      "loss": 1.2062,
+      "step": 1963
+    },
+    {
+      "epoch": 0.0714259737425901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9910142455180696e-05,
+      "loss": 1.1627,
+      "step": 1964
+    },
+    {
+      "epoch": 0.07146234134632869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990998483994065e-05,
+      "loss": 1.1982,
+      "step": 1965
+    },
+    {
+      "epoch": 0.07149870895006728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9909827087213617e-05,
+      "loss": 1.2689,
+      "step": 1966
+    },
+    {
+      "epoch": 0.07153507655380587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99096691970018e-05,
+      "loss": 1.2979,
+      "step": 1967
+    },
+    {
+      "epoch": 0.07157144415754446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9909511169307378e-05,
+      "loss": 1.3124,
+      "step": 1968
+    },
+    {
+      "epoch": 0.07160781176128304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990935300413255e-05,
+      "loss": 1.0845,
+      "step": 1969
+    },
+    {
+      "epoch": 0.07164417936502164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990919470147951e-05,
+      "loss": 1.312,
+      "step": 1970
+    },
+    {
+      "epoch": 0.07168054696876022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990903626135045e-05,
+      "loss": 1.321,
+      "step": 1971
+    },
+    {
+      "epoch": 0.07171691457249882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9908877683747572e-05,
+      "loss": 1.2634,
+      "step": 1972
+    },
+    {
+      "epoch": 0.0717532821762374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9908718968673076e-05,
+      "loss": 1.1345,
+      "step": 1973
+    },
+    {
+      "epoch": 0.071789649779976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9908560116129164e-05,
+      "loss": 1.1894,
+      "step": 1974
+    },
+    {
+      "epoch": 0.0718260173837146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990840112611804e-05,
+      "loss": 1.212,
+      "step": 1975
+    },
+    {
+      "epoch": 0.07186238498745318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9908241998641903e-05,
+      "loss": 1.2773,
+      "step": 1976
+    },
+    {
+      "epoch": 0.07189875259119177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9908082733702972e-05,
+      "loss": 1.3506,
+      "step": 1977
+    },
+    {
+      "epoch": 0.07193512019493035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990792333130345e-05,
+      "loss": 1.2277,
+      "step": 1978
+    },
+    {
+      "epoch": 0.07197148779866895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907763791445544e-05,
+      "loss": 1.2024,
+      "step": 1979
+    },
+    {
+      "epoch": 0.07200785540240753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907604114131476e-05,
+      "loss": 1.0872,
+      "step": 1980
+    },
+    {
+      "epoch": 0.07204422300614613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907444299363455e-05,
+      "loss": 1.1728,
+      "step": 1981
+    },
+    {
+      "epoch": 0.07208059060988471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990728434714371e-05,
+      "loss": 1.0056,
+      "step": 1982
+    },
+    {
+      "epoch": 0.07211695821362331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9907124257474445e-05,
+      "loss": 1.3021,
+      "step": 1983
+    },
+    {
+      "epoch": 0.07215332581736189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906964030357884e-05,
+      "loss": 1.1486,
+      "step": 1984
+    },
+    {
+      "epoch": 0.07218969342110049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906803665796257e-05,
+      "loss": 1.3409,
+      "step": 1985
+    },
+    {
+      "epoch": 0.07222606102483907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906643163791785e-05,
+      "loss": 1.1567,
+      "step": 1986
+    },
+    {
+      "epoch": 0.07226242862857767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906482524346694e-05,
+      "loss": 1.1763,
+      "step": 1987
+    },
+    {
+      "epoch": 0.07229879623231625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906321747463218e-05,
+      "loss": 1.351,
+      "step": 1988
+    },
+    {
+      "epoch": 0.07233516383605484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990616083314358e-05,
+      "loss": 1.2803,
+      "step": 1989
+    },
+    {
+      "epoch": 0.07237153143979343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990599978139001e-05,
+      "loss": 1.1858,
+      "step": 1990
+    },
+    {
+      "epoch": 0.07240789904353202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9905838592204754e-05,
+      "loss": 1.2064,
+      "step": 1991
+    },
+    {
+      "epoch": 0.0724442666472706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9905677265590042e-05,
+      "loss": 1.1575,
+      "step": 1992
+    },
+    {
+      "epoch": 0.0724806342510092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990551580154811e-05,
+      "loss": 1.2495,
+      "step": 1993
+    },
+    {
+      "epoch": 0.07251700185474778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9905354200081198e-05,
+      "loss": 1.1447,
+      "step": 1994
+    },
+    {
+      "epoch": 0.07255336945848638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990519246119156e-05,
+      "loss": 1.2406,
+      "step": 1995
+    },
+    {
+      "epoch": 0.07258973706222498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990503058488142e-05,
+      "loss": 1.2408,
+      "step": 1996
+    },
+    {
+      "epoch": 0.07262610466596356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9904868571153038e-05,
+      "loss": 1.2636,
+      "step": 1997
+    },
+    {
+      "epoch": 0.07266247226970216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990470642000866e-05,
+      "loss": 1.2159,
+      "step": 1998
+    },
+    {
+      "epoch": 0.07269883987344074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9904544131450527e-05,
+      "loss": 1.2455,
+      "step": 1999
+    },
+    {
+      "epoch": 0.07273520747717933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99043817054809e-05,
+      "loss": 1.2334,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07277157508091792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9904219142102032e-05,
+      "loss": 1.2419,
+      "step": 2001
+    },
+    {
+      "epoch": 0.07280794268465651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9904056441316174e-05,
+      "loss": 1.2618,
+      "step": 2002
+    },
+    {
+      "epoch": 0.0728443102883951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9903893603125587e-05,
+      "loss": 1.2053,
+      "step": 2003
+    },
+    {
+      "epoch": 0.07288067789213369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9903730627532524e-05,
+      "loss": 1.2075,
+      "step": 2004
+    },
+    {
+      "epoch": 0.07291704549587227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9903567514539254e-05,
+      "loss": 1.2372,
+      "step": 2005
+    },
+    {
+      "epoch": 0.07295341309961087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990340426414803e-05,
+      "loss": 1.1037,
+      "step": 2006
+    },
+    {
+      "epoch": 0.07298978070334945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9903240876361126e-05,
+      "loss": 1.2258,
+      "step": 2007
+    },
+    {
+      "epoch": 0.07302614830708805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9903077351180808e-05,
+      "loss": 1.163,
+      "step": 2008
+    },
+    {
+      "epoch": 0.07306251591082663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990291368860934e-05,
+      "loss": 1.1773,
+      "step": 2009
+    },
+    {
+      "epoch": 0.07309888351456523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9902749888648996e-05,
+      "loss": 1.2251,
+      "step": 2010
+    },
+    {
+      "epoch": 0.07313525111830381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9902585951302046e-05,
+      "loss": 1.2511,
+      "step": 2011
+    },
+    {
+      "epoch": 0.0731716187220424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990242187657077e-05,
+      "loss": 1.2549,
+      "step": 2012
+    },
+    {
+      "epoch": 0.07320798632578099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9902257664457435e-05,
+      "loss": 1.1798,
+      "step": 2013
+    },
+    {
+      "epoch": 0.07324435392951958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9902093314964327e-05,
+      "loss": 1.1976,
+      "step": 2014
+    },
+    {
+      "epoch": 0.07328072153325817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9901928828093724e-05,
+      "loss": 1.1253,
+      "step": 2015
+    },
+    {
+      "epoch": 0.07331708913699676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9901764203847906e-05,
+      "loss": 1.3341,
+      "step": 2016
+    },
+    {
+      "epoch": 0.07335345674073536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990159944222916e-05,
+      "loss": 1.1984,
+      "step": 2017
+    },
+    {
+      "epoch": 0.07338982434447394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990143454323977e-05,
+      "loss": 1.2157,
+      "step": 2018
+    },
+    {
+      "epoch": 0.07342619194821254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9901269506882023e-05,
+      "loss": 1.2957,
+      "step": 2019
+    },
+    {
+      "epoch": 0.07346255955195112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990110433315821e-05,
+      "loss": 1.3347,
+      "step": 2020
+    },
+    {
+      "epoch": 0.07349892715568972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9900939022070624e-05,
+      "loss": 1.1553,
+      "step": 2021
+    },
+    {
+      "epoch": 0.0735352947594283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990077357362155e-05,
+      "loss": 1.3663,
+      "step": 2022
+    },
+    {
+      "epoch": 0.0735716623631669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99006079878133e-05,
+      "loss": 1.2517,
+      "step": 2023
+    },
+    {
+      "epoch": 0.07360802996690548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9900442264648155e-05,
+      "loss": 1.1162,
+      "step": 2024
+    },
+    {
+      "epoch": 0.07364439757064407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.990027640412842e-05,
+      "loss": 1.2318,
+      "step": 2025
+    },
+    {
+      "epoch": 0.07368076517438266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.99001104062564e-05,
+      "loss": 1.1713,
+      "step": 2026
+    },
+    {
+      "epoch": 0.07371713277812125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989994427103439e-05,
+      "loss": 1.2451,
+      "step": 2027
+    },
+    {
+      "epoch": 0.07375350038185984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98997779984647e-05,
+      "loss": 1.2318,
+      "step": 2028
+    },
+    {
+      "epoch": 0.07378986798559843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989961158854964e-05,
+      "loss": 1.2116,
+      "step": 2029
+    },
+    {
+      "epoch": 0.07382623558933701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9899445041291516e-05,
+      "loss": 1.1672,
+      "step": 2030
+    },
+    {
+      "epoch": 0.07386260319307561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9899278356692636e-05,
+      "loss": 1.1702,
+      "step": 2031
+    },
+    {
+      "epoch": 0.07389897079681419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9899111534755312e-05,
+      "loss": 1.2151,
+      "step": 2032
+    },
+    {
+      "epoch": 0.07393533840055279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9898944575481862e-05,
+      "loss": 1.2173,
+      "step": 2033
+    },
+    {
+      "epoch": 0.07397170600429137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98987774788746e-05,
+      "loss": 1.1871,
+      "step": 2034
+    },
+    {
+      "epoch": 0.07400807360802997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9898610244935847e-05,
+      "loss": 1.2559,
+      "step": 2035
+    },
+    {
+      "epoch": 0.07404444121176855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9898442873667916e-05,
+      "loss": 1.3875,
+      "step": 2036
+    },
+    {
+      "epoch": 0.07408080881550715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989827536507314e-05,
+      "loss": 1.2352,
+      "step": 2037
+    },
+    {
+      "epoch": 0.07411717641924574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9898107719153833e-05,
+      "loss": 1.2762,
+      "step": 2038
+    },
+    {
+      "epoch": 0.07415354402298432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9897939935912326e-05,
+      "loss": 1.2663,
+      "step": 2039
+    },
+    {
+      "epoch": 0.07418991162672292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9897772015350945e-05,
+      "loss": 1.1572,
+      "step": 2040
+    },
+    {
+      "epoch": 0.0742262792304615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9897603957472022e-05,
+      "loss": 1.3319,
+      "step": 2041
+    },
+    {
+      "epoch": 0.0742626468342001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9897435762277888e-05,
+      "loss": 1.2775,
+      "step": 2042
+    },
+    {
+      "epoch": 0.07429901443793868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9897267429770873e-05,
+      "loss": 1.2314,
+      "step": 2043
+    },
+    {
+      "epoch": 0.07433538204167728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9897098959953316e-05,
+      "loss": 1.0743,
+      "step": 2044
+    },
+    {
+      "epoch": 0.07437174964541586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9896930352827554e-05,
+      "loss": 1.1403,
+      "step": 2045
+    },
+    {
+      "epoch": 0.07440811724915446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989676160839592e-05,
+      "loss": 1.2299,
+      "step": 2046
+    },
+    {
+      "epoch": 0.07444448485289304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989659272666077e-05,
+      "loss": 1.2584,
+      "step": 2047
+    },
+    {
+      "epoch": 0.07448085245663164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989642370762443e-05,
+      "loss": 1.2156,
+      "step": 2048
+    },
+    {
+      "epoch": 0.07451722006037022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9896254551289255e-05,
+      "loss": 1.1418,
+      "step": 2049
+    },
+    {
+      "epoch": 0.07455358766410881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9896085257657587e-05,
+      "loss": 1.1367,
+      "step": 2050
+    },
+    {
+      "epoch": 0.0745899552678474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9895915826731778e-05,
+      "loss": 1.2978,
+      "step": 2051
+    },
+    {
+      "epoch": 0.07462632287158599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9895746258514177e-05,
+      "loss": 1.2524,
+      "step": 2052
+    },
+    {
+      "epoch": 0.07466269047532458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9895576553007138e-05,
+      "loss": 1.2762,
+      "step": 2053
+    },
+    {
+      "epoch": 0.07469905807906317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9895406710213012e-05,
+      "loss": 1.2215,
+      "step": 2054
+    },
+    {
+      "epoch": 0.07473542568280175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989523673013416e-05,
+      "loss": 1.2388,
+      "step": 2055
+    },
+    {
+      "epoch": 0.07477179328654035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9895066612772937e-05,
+      "loss": 1.2316,
+      "step": 2056
+    },
+    {
+      "epoch": 0.07480816089027895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9894896358131707e-05,
+      "loss": 1.1315,
+      "step": 2057
+    },
+    {
+      "epoch": 0.07484452849401753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9894725966212824e-05,
+      "loss": 1.2849,
+      "step": 2058
+    },
+    {
+      "epoch": 0.07488089609775612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9894555437018658e-05,
+      "loss": 1.1672,
+      "step": 2059
+    },
+    {
+      "epoch": 0.07491726370149471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9894384770551574e-05,
+      "loss": 1.1538,
+      "step": 2060
+    },
+    {
+      "epoch": 0.0749536313052333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9894213966813942e-05,
+      "loss": 1.1815,
+      "step": 2061
+    },
+    {
+      "epoch": 0.07498999890897189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989404302580813e-05,
+      "loss": 1.1407,
+      "step": 2062
+    },
+    {
+      "epoch": 0.07502636651271048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9893871947536507e-05,
+      "loss": 1.2199,
+      "step": 2063
+    },
+    {
+      "epoch": 0.07506273411644906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9893700732001447e-05,
+      "loss": 1.1882,
+      "step": 2064
+    },
+    {
+      "epoch": 0.07509910172018766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989352937920533e-05,
+      "loss": 1.1973,
+      "step": 2065
+    },
+    {
+      "epoch": 0.07513546932392624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9893357889150523e-05,
+      "loss": 1.1132,
+      "step": 2066
+    },
+    {
+      "epoch": 0.07517183692766484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9893186261839415e-05,
+      "loss": 1.2905,
+      "step": 2067
+    },
+    {
+      "epoch": 0.07520820453140342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9893014497274387e-05,
+      "loss": 1.1759,
+      "step": 2068
+    },
+    {
+      "epoch": 0.07524457213514202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989284259545782e-05,
+      "loss": 1.1646,
+      "step": 2069
+    },
+    {
+      "epoch": 0.0752809397388806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98926705563921e-05,
+      "loss": 1.2124,
+      "step": 2070
+    },
+    {
+      "epoch": 0.0753173073426192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9892498380079605e-05,
+      "loss": 1.3049,
+      "step": 2071
+    },
+    {
+      "epoch": 0.07535367494635778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9892326066522732e-05,
+      "loss": 1.2003,
+      "step": 2072
+    },
+    {
+      "epoch": 0.07539004255009638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9892153615723873e-05,
+      "loss": 1.1426,
+      "step": 2073
+    },
+    {
+      "epoch": 0.07542641015383496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9891981027685418e-05,
+      "loss": 1.0889,
+      "step": 2074
+    },
+    {
+      "epoch": 0.07546277775757355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989180830240976e-05,
+      "loss": 1.2254,
+      "step": 2075
+    },
+    {
+      "epoch": 0.07549914536131214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9891635439899294e-05,
+      "loss": 1.1417,
+      "step": 2076
+    },
+    {
+      "epoch": 0.07553551296505073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9891462440156423e-05,
+      "loss": 1.142,
+      "step": 2077
+    },
+    {
+      "epoch": 0.07557188056878933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9891289303183545e-05,
+      "loss": 1.2709,
+      "step": 2078
+    },
+    {
+      "epoch": 0.07560824817252791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.989111602898306e-05,
+      "loss": 1.2341,
+      "step": 2079
+    },
+    {
+      "epoch": 0.07564461577626651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890942617557378e-05,
+      "loss": 1.2706,
+      "step": 2080
+    },
+    {
+      "epoch": 0.07568098338000509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890769068908896e-05,
+      "loss": 1.1116,
+      "step": 2081
+    },
+    {
+      "epoch": 0.07571735098374369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890595383040027e-05,
+      "loss": 1.1389,
+      "step": 2082
+    },
+    {
+      "epoch": 0.07575371858748227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890421559953182e-05,
+      "loss": 1.2526,
+      "step": 2083
+    },
+    {
+      "epoch": 0.07579008619122087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890247599650768e-05,
+      "loss": 1.2633,
+      "step": 2084
+    },
+    {
+      "epoch": 0.07582645379495945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9890073502135206e-05,
+      "loss": 1.2759,
+      "step": 2085
+    },
+    {
+      "epoch": 0.07586282139869804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9889899267408903e-05,
+      "loss": 1.1805,
+      "step": 2086
+    },
+    {
+      "epoch": 0.07589918900243663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988972489547428e-05,
+      "loss": 1.2673,
+      "step": 2087
+    },
+    {
+      "epoch": 0.07593555660617522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9889550386333753e-05,
+      "loss": 1.2474,
+      "step": 2088
+    },
+    {
+      "epoch": 0.0759719242099138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988937573998975e-05,
+      "loss": 1.1636,
+      "step": 2089
+    },
+    {
+      "epoch": 0.0760082918136524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9889200956444687e-05,
+      "loss": 1.2395,
+      "step": 2090
+    },
+    {
+      "epoch": 0.07604465941739098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9889026035700992e-05,
+      "loss": 1.2723,
+      "step": 2091
+    },
+    {
+      "epoch": 0.07608102702112958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9888850977761094e-05,
+      "loss": 1.1371,
+      "step": 2092
+    },
+    {
+      "epoch": 0.07611739462486816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9888675782627415e-05,
+      "loss": 1.2794,
+      "step": 2093
+    },
+    {
+      "epoch": 0.07615376222860676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9888500450302394e-05,
+      "loss": 1.084,
+      "step": 2094
+    },
+    {
+      "epoch": 0.07619012983234534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9888324980788455e-05,
+      "loss": 1.2137,
+      "step": 2095
+    },
+    {
+      "epoch": 0.07622649743608394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988814937408804e-05,
+      "loss": 1.2448,
+      "step": 2096
+    },
+    {
+      "epoch": 0.07626286503982252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887973630203578e-05,
+      "loss": 1.2117,
+      "step": 2097
+    },
+    {
+      "epoch": 0.07629923264356112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887797749137516e-05,
+      "loss": 1.1838,
+      "step": 2098
+    },
+    {
+      "epoch": 0.07633560024729971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887621730892285e-05,
+      "loss": 1.193,
+      "step": 2099
+    },
+    {
+      "epoch": 0.0763719678510383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887445575470334e-05,
+      "loss": 1.3798,
+      "step": 2100
+    },
+    {
+      "epoch": 0.07640833545477689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887269282874104e-05,
+      "loss": 1.2454,
+      "step": 2101
+    },
+    {
+      "epoch": 0.07644470305851547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9887092853106037e-05,
+      "loss": 1.2145,
+      "step": 2102
+    },
+    {
+      "epoch": 0.07648107066225407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988691628616859e-05,
+      "loss": 1.2305,
+      "step": 2103
+    },
+    {
+      "epoch": 0.07651743826599265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9886739582064206e-05,
+      "loss": 1.2865,
+      "step": 2104
+    },
+    {
+      "epoch": 0.07655380586973125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9886562740795337e-05,
+      "loss": 1.2548,
+      "step": 2105
+    },
+    {
+      "epoch": 0.07659017347346983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9886385762364434e-05,
+      "loss": 1.1869,
+      "step": 2106
+    },
+    {
+      "epoch": 0.07662654107720843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988620864677396e-05,
+      "loss": 1.2501,
+      "step": 2107
+    },
+    {
+      "epoch": 0.07666290868094701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9886031394026365e-05,
+      "loss": 1.3306,
+      "step": 2108
+    },
+    {
+      "epoch": 0.0766992762846856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988585400412411e-05,
+      "loss": 1.2908,
+      "step": 2109
+    },
+    {
+      "epoch": 0.07673564388842419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9885676477069664e-05,
+      "loss": 1.2036,
+      "step": 2110
+    },
+    {
+      "epoch": 0.07677201149216278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9885498812865474e-05,
+      "loss": 1.1912,
+      "step": 2111
+    },
+    {
+      "epoch": 0.07680837909590137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988532101151402e-05,
+      "loss": 1.143,
+      "step": 2112
+    },
+    {
+      "epoch": 0.07684474669963996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9885143073017755e-05,
+      "loss": 1.1116,
+      "step": 2113
+    },
+    {
+      "epoch": 0.07688111430337854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9884964997379163e-05,
+      "loss": 1.2535,
+      "step": 2114
+    },
+    {
+      "epoch": 0.07691748190711714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98847867846007e-05,
+      "loss": 1.1883,
+      "step": 2115
+    },
+    {
+      "epoch": 0.07695384951085572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988460843468485e-05,
+      "loss": 1.2176,
+      "step": 2116
+    },
+    {
+      "epoch": 0.07699021711459432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9884429947634083e-05,
+      "loss": 1.0721,
+      "step": 2117
+    },
+    {
+      "epoch": 0.0770265847183329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988425132345087e-05,
+      "loss": 1.2585,
+      "step": 2118
+    },
+    {
+      "epoch": 0.0770629523220715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9884072562137696e-05,
+      "loss": 1.2017,
+      "step": 2119
+    },
+    {
+      "epoch": 0.0770993199258101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9883893663697042e-05,
+      "loss": 1.1383,
+      "step": 2120
+    },
+    {
+      "epoch": 0.07713568752954868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9883714628131382e-05,
+      "loss": 1.2941,
+      "step": 2121
+    },
+    {
+      "epoch": 0.07717205513328727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9883535455443205e-05,
+      "loss": 1.1683,
+      "step": 2122
+    },
+    {
+      "epoch": 0.07720842273702586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9883356145635e-05,
+      "loss": 1.1225,
+      "step": 2123
+    },
+    {
+      "epoch": 0.07724479034076445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988317669870925e-05,
+      "loss": 1.4058,
+      "step": 2124
+    },
+    {
+      "epoch": 0.07728115794450303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988299711466844e-05,
+      "loss": 1.1798,
+      "step": 2125
+    },
+    {
+      "epoch": 0.07731752554824163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9882817393515073e-05,
+      "loss": 1.1354,
+      "step": 2126
+    },
+    {
+      "epoch": 0.07735389315198021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9882637535251635e-05,
+      "loss": 1.3242,
+      "step": 2127
+    },
+    {
+      "epoch": 0.07739026075571881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9882457539880627e-05,
+      "loss": 1.2233,
+      "step": 2128
+    },
+    {
+      "epoch": 0.07742662835945739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9882277407404536e-05,
+      "loss": 1.191,
+      "step": 2129
+    },
+    {
+      "epoch": 0.07746299596319599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988209713782587e-05,
+      "loss": 1.2054,
+      "step": 2130
+    },
+    {
+      "epoch": 0.07749936356693457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9881916731147125e-05,
+      "loss": 1.2063,
+      "step": 2131
+    },
+    {
+      "epoch": 0.07753573117067317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988173618737081e-05,
+      "loss": 1.1978,
+      "step": 2132
+    },
+    {
+      "epoch": 0.07757209877441175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988155550649942e-05,
+      "loss": 1.2426,
+      "step": 2133
+    },
+    {
+      "epoch": 0.07760846637815035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9881374688535473e-05,
+      "loss": 1.4213,
+      "step": 2134
+    },
+    {
+      "epoch": 0.07764483398188893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9881193733481468e-05,
+      "loss": 1.1237,
+      "step": 2135
+    },
+    {
+      "epoch": 0.07768120158562752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9881012641339925e-05,
+      "loss": 1.1057,
+      "step": 2136
+    },
+    {
+      "epoch": 0.0777175691893661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9880831412113346e-05,
+      "loss": 1.1339,
+      "step": 2137
+    },
+    {
+      "epoch": 0.0777539367931047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9880650045804253e-05,
+      "loss": 1.1667,
+      "step": 2138
+    },
+    {
+      "epoch": 0.0777903043968433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9880468542415156e-05,
+      "loss": 1.2261,
+      "step": 2139
+    },
+    {
+      "epoch": 0.07782667200058188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988028690194858e-05,
+      "loss": 1.1071,
+      "step": 2140
+    },
+    {
+      "epoch": 0.07786303960432048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.988010512440704e-05,
+      "loss": 1.3553,
+      "step": 2141
+    },
+    {
+      "epoch": 0.07789940720805906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9879923209793058e-05,
+      "loss": 1.1638,
+      "step": 2142
+    },
+    {
+      "epoch": 0.07793577481179766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987974115810916e-05,
+      "loss": 1.1349,
+      "step": 2143
+    },
+    {
+      "epoch": 0.07797214241553624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987955896935787e-05,
+      "loss": 1.0633,
+      "step": 2144
+    },
+    {
+      "epoch": 0.07800851001927483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987937664354172e-05,
+      "loss": 1.241,
+      "step": 2145
+    },
+    {
+      "epoch": 0.07804487762301342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9879194180663232e-05,
+      "loss": 1.2015,
+      "step": 2146
+    },
+    {
+      "epoch": 0.07808124522675201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9879011580724946e-05,
+      "loss": 1.2545,
+      "step": 2147
+    },
+    {
+      "epoch": 0.0781176128304906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9878828843729387e-05,
+      "loss": 1.2073,
+      "step": 2148
+    },
+    {
+      "epoch": 0.07815398043422919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9878645969679096e-05,
+      "loss": 1.3253,
+      "step": 2149
+    },
+    {
+      "epoch": 0.07819034803796777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9878462958576608e-05,
+      "loss": 1.2373,
+      "step": 2150
+    },
+    {
+      "epoch": 0.07822671564170637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9878279810424463e-05,
+      "loss": 1.0756,
+      "step": 2151
+    },
+    {
+      "epoch": 0.07826308324544495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9878096525225196e-05,
+      "loss": 1.2751,
+      "step": 2152
+    },
+    {
+      "epoch": 0.07829945084918355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987791310298136e-05,
+      "loss": 1.2781,
+      "step": 2153
+    },
+    {
+      "epoch": 0.07833581845292213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9877729543695494e-05,
+      "loss": 1.1935,
+      "step": 2154
+    },
+    {
+      "epoch": 0.07837218605666073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9877545847370145e-05,
+      "loss": 1.2609,
+      "step": 2155
+    },
+    {
+      "epoch": 0.07840855366039931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987736201400786e-05,
+      "loss": 1.1266,
+      "step": 2156
+    },
+    {
+      "epoch": 0.0784449212641379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9877178043611195e-05,
+      "loss": 1.2969,
+      "step": 2157
+    },
+    {
+      "epoch": 0.07848128886787649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9876993936182694e-05,
+      "loss": 1.1443,
+      "step": 2158
+    },
+    {
+      "epoch": 0.07851765647161509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987680969172492e-05,
+      "loss": 1.1818,
+      "step": 2159
+    },
+    {
+      "epoch": 0.07855402407535368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987662531024042e-05,
+      "loss": 1.1043,
+      "step": 2160
+    },
+    {
+      "epoch": 0.07859039167909226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9876440791731763e-05,
+      "loss": 1.2328,
+      "step": 2161
+    },
+    {
+      "epoch": 0.07862675928283086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9876256136201496e-05,
+      "loss": 1.2711,
+      "step": 2162
+    },
+    {
+      "epoch": 0.07866312688656944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9876071343652192e-05,
+      "loss": 1.2451,
+      "step": 2163
+    },
+    {
+      "epoch": 0.07869949449030804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987588641408641e-05,
+      "loss": 1.1714,
+      "step": 2164
+    },
+    {
+      "epoch": 0.07873586209404662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9875701347506715e-05,
+      "loss": 1.2161,
+      "step": 2165
+    },
+    {
+      "epoch": 0.07877222969778522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987551614391568e-05,
+      "loss": 1.3887,
+      "step": 2166
+    },
+    {
+      "epoch": 0.0788085973015238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9875330803315864e-05,
+      "loss": 1.2242,
+      "step": 2167
+    },
+    {
+      "epoch": 0.0788449649052624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9875145325709845e-05,
+      "loss": 1.3228,
+      "step": 2168
+    },
+    {
+      "epoch": 0.07888133250900098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98749597111002e-05,
+      "loss": 1.2247,
+      "step": 2169
+    },
+    {
+      "epoch": 0.07891770011273957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9874773959489494e-05,
+      "loss": 1.1828,
+      "step": 2170
+    },
+    {
+      "epoch": 0.07895406771647816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9874588070880315e-05,
+      "loss": 1.3114,
+      "step": 2171
+    },
+    {
+      "epoch": 0.07899043532021675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9874402045275235e-05,
+      "loss": 1.1878,
+      "step": 2172
+    },
+    {
+      "epoch": 0.07902680292395534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9874215882676834e-05,
+      "loss": 1.1381,
+      "step": 2173
+    },
+    {
+      "epoch": 0.07906317052769393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98740295830877e-05,
+      "loss": 1.2535,
+      "step": 2174
+    },
+    {
+      "epoch": 0.07909953813143251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9873843146510415e-05,
+      "loss": 1.2568,
+      "step": 2175
+    },
+    {
+      "epoch": 0.07913590573517111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9873656572947565e-05,
+      "loss": 1.1935,
+      "step": 2176
+    },
+    {
+      "epoch": 0.0791722733389097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9873469862401737e-05,
+      "loss": 1.1917,
+      "step": 2177
+    },
+    {
+      "epoch": 0.07920864094264829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9873283014875527e-05,
+      "loss": 1.291,
+      "step": 2178
+    },
+    {
+      "epoch": 0.07924500854638687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9873096030371522e-05,
+      "loss": 1.1552,
+      "step": 2179
+    },
+    {
+      "epoch": 0.07928137615012547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9872908908892315e-05,
+      "loss": 1.1402,
+      "step": 2180
+    },
+    {
+      "epoch": 0.07931774375386406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9872721650440508e-05,
+      "loss": 1.1372,
+      "step": 2181
+    },
+    {
+      "epoch": 0.07935411135760265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9872534255018692e-05,
+      "loss": 1.2986,
+      "step": 2182
+    },
+    {
+      "epoch": 0.07939047896134124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9872346722629478e-05,
+      "loss": 1.1341,
+      "step": 2183
+    },
+    {
+      "epoch": 0.07942684656507983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9872159053275452e-05,
+      "loss": 1.1632,
+      "step": 2184
+    },
+    {
+      "epoch": 0.07946321416881842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9871971246959228e-05,
+      "loss": 1.1882,
+      "step": 2185
+    },
+    {
+      "epoch": 0.079499581772557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9871783303683414e-05,
+      "loss": 1.2072,
+      "step": 2186
+    },
+    {
+      "epoch": 0.0795359493762956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987159522345061e-05,
+      "loss": 1.1492,
+      "step": 2187
+    },
+    {
+      "epoch": 0.07957231698003418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9871407006263426e-05,
+      "loss": 1.2415,
+      "step": 2188
+    },
+    {
+      "epoch": 0.07960868458377278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987121865212448e-05,
+      "loss": 1.2676,
+      "step": 2189
+    },
+    {
+      "epoch": 0.07964505218751136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9871030161036375e-05,
+      "loss": 1.2025,
+      "step": 2190
+    },
+    {
+      "epoch": 0.07968141979124996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9870841533001735e-05,
+      "loss": 1.2047,
+      "step": 2191
+    },
+    {
+      "epoch": 0.07971778739498854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9870652768023172e-05,
+      "loss": 1.2713,
+      "step": 2192
+    },
+    {
+      "epoch": 0.07975415499872714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987046386610331e-05,
+      "loss": 1.186,
+      "step": 2193
+    },
+    {
+      "epoch": 0.07979052260246572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987027482724476e-05,
+      "loss": 1.3167,
+      "step": 2194
+    },
+    {
+      "epoch": 0.07982689020620432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9870085651450157e-05,
+      "loss": 1.2265,
+      "step": 2195
+    },
+    {
+      "epoch": 0.0798632578099429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9869896338722116e-05,
+      "loss": 1.1296,
+      "step": 2196
+    },
+    {
+      "epoch": 0.0798996254136815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9869706889063268e-05,
+      "loss": 1.1067,
+      "step": 2197
+    },
+    {
+      "epoch": 0.07993599301742008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986951730247624e-05,
+      "loss": 1.2764,
+      "step": 2198
+    },
+    {
+      "epoch": 0.07997236062115867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9869327578963666e-05,
+      "loss": 1.1713,
+      "step": 2199
+    },
+    {
+      "epoch": 0.08000872822489725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986913771852817e-05,
+      "loss": 1.2525,
+      "step": 2200
+    },
+    {
+      "epoch": 0.08004509582863585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986894772117239e-05,
+      "loss": 1.3203,
+      "step": 2201
+    },
+    {
+      "epoch": 0.08008146343237445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9868757586898966e-05,
+      "loss": 1.2903,
+      "step": 2202
+    },
+    {
+      "epoch": 0.08011783103611303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9868567315710533e-05,
+      "loss": 1.3405,
+      "step": 2203
+    },
+    {
+      "epoch": 0.08015419863985163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9868376907609727e-05,
+      "loss": 1.0645,
+      "step": 2204
+    },
+    {
+      "epoch": 0.08019056624359021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986818636259919e-05,
+      "loss": 1.2385,
+      "step": 2205
+    },
+    {
+      "epoch": 0.0802269338473288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9867995680681575e-05,
+      "loss": 1.1964,
+      "step": 2206
+    },
+    {
+      "epoch": 0.08026330145106739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9867804861859518e-05,
+      "loss": 1.2109,
+      "step": 2207
+    },
+    {
+      "epoch": 0.08029966905480598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986761390613567e-05,
+      "loss": 1.1181,
+      "step": 2208
+    },
+    {
+      "epoch": 0.08033603665854457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9867422813512678e-05,
+      "loss": 1.3109,
+      "step": 2209
+    },
+    {
+      "epoch": 0.08037240426228316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9867231583993194e-05,
+      "loss": 1.1982,
+      "step": 2210
+    },
+    {
+      "epoch": 0.08040877186602174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986704021757987e-05,
+      "loss": 1.1752,
+      "step": 2211
+    },
+    {
+      "epoch": 0.08044513946976034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9866848714275368e-05,
+      "loss": 1.2205,
+      "step": 2212
+    },
+    {
+      "epoch": 0.08048150707349892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9866657074082335e-05,
+      "loss": 1.2433,
+      "step": 2213
+    },
+    {
+      "epoch": 0.08051787467723752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9866465297003437e-05,
+      "loss": 1.2913,
+      "step": 2214
+    },
+    {
+      "epoch": 0.0805542422809761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986627338304133e-05,
+      "loss": 1.2582,
+      "step": 2215
+    },
+    {
+      "epoch": 0.0805906098847147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9866081332198676e-05,
+      "loss": 1.2357,
+      "step": 2216
+    },
+    {
+      "epoch": 0.08062697748845328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9865889144478144e-05,
+      "loss": 1.1959,
+      "step": 2217
+    },
+    {
+      "epoch": 0.08066334509219188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98656968198824e-05,
+      "loss": 1.2059,
+      "step": 2218
+    },
+    {
+      "epoch": 0.08069971269593046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9865504358414107e-05,
+      "loss": 1.3645,
+      "step": 2219
+    },
+    {
+      "epoch": 0.08073608029966906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986531176007594e-05,
+      "loss": 1.2554,
+      "step": 2220
+    },
+    {
+      "epoch": 0.08077244790340764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986511902487057e-05,
+      "loss": 1.2691,
+      "step": 2221
+    },
+    {
+      "epoch": 0.08080881550714623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986492615280067e-05,
+      "loss": 1.2069,
+      "step": 2222
+    },
+    {
+      "epoch": 0.08084518311088483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9864733143868912e-05,
+      "loss": 1.2192,
+      "step": 2223
+    },
+    {
+      "epoch": 0.08088155071462341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9864539998077983e-05,
+      "loss": 1.1929,
+      "step": 2224
+    },
+    {
+      "epoch": 0.08091791831836201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9864346715430554e-05,
+      "loss": 1.1764,
+      "step": 2225
+    },
+    {
+      "epoch": 0.08095428592210059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9864153295929314e-05,
+      "loss": 1.1343,
+      "step": 2226
+    },
+    {
+      "epoch": 0.08099065352583919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9863959739576937e-05,
+      "loss": 1.2005,
+      "step": 2227
+    },
+    {
+      "epoch": 0.08102702112957777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986376604637612e-05,
+      "loss": 1.0314,
+      "step": 2228
+    },
+    {
+      "epoch": 0.08106338873331637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986357221632954e-05,
+      "loss": 1.196,
+      "step": 2229
+    },
+    {
+      "epoch": 0.08109975633705495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986337824943989e-05,
+      "loss": 1.1574,
+      "step": 2230
+    },
+    {
+      "epoch": 0.08113612394079354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9863184145709863e-05,
+      "loss": 1.0367,
+      "step": 2231
+    },
+    {
+      "epoch": 0.08117249154453213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9862989905142148e-05,
+      "loss": 1.201,
+      "step": 2232
+    },
+    {
+      "epoch": 0.08120885914827072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986279552773944e-05,
+      "loss": 1.2412,
+      "step": 2233
+    },
+    {
+      "epoch": 0.0812452267520093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9862601013504444e-05,
+      "loss": 1.1767,
+      "step": 2234
+    },
+    {
+      "epoch": 0.0812815943557479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9862406362439846e-05,
+      "loss": 1.2716,
+      "step": 2235
+    },
+    {
+      "epoch": 0.08131796195948648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9862211574548355e-05,
+      "loss": 1.2414,
+      "step": 2236
+    },
+    {
+      "epoch": 0.08135432956322508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986201664983267e-05,
+      "loss": 1.1249,
+      "step": 2237
+    },
+    {
+      "epoch": 0.08139069716696366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98618215882955e-05,
+      "loss": 1.2746,
+      "step": 2238
+    },
+    {
+      "epoch": 0.08142706477070226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9861626389939546e-05,
+      "loss": 1.2318,
+      "step": 2239
+    },
+    {
+      "epoch": 0.08146343237444084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9861431054767518e-05,
+      "loss": 1.1707,
+      "step": 2240
+    },
+    {
+      "epoch": 0.08149979997817944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986123558278212e-05,
+      "loss": 1.3071,
+      "step": 2241
+    },
+    {
+      "epoch": 0.08153616758191803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9861039973986074e-05,
+      "loss": 1.2699,
+      "step": 2242
+    },
+    {
+      "epoch": 0.08157253518565662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986084422838209e-05,
+      "loss": 1.2758,
+      "step": 2243
+    },
+    {
+      "epoch": 0.08160890278939521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9860648345972883e-05,
+      "loss": 1.2307,
+      "step": 2244
+    },
+    {
+      "epoch": 0.0816452703931338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986045232676117e-05,
+      "loss": 1.0742,
+      "step": 2245
+    },
+    {
+      "epoch": 0.08168163799687239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986025617074967e-05,
+      "loss": 1.2129,
+      "step": 2246
+    },
+    {
+      "epoch": 0.08171800560061097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9860059877941106e-05,
+      "loss": 1.1476,
+      "step": 2247
+    },
+    {
+      "epoch": 0.08175437320434957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9859863448338203e-05,
+      "loss": 1.2659,
+      "step": 2248
+    },
+    {
+      "epoch": 0.08179074080808815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9859666881943684e-05,
+      "loss": 1.1822,
+      "step": 2249
+    },
+    {
+      "epoch": 0.08182710841182675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9859470178760272e-05,
+      "loss": 1.1263,
+      "step": 2250
+    },
+    {
+      "epoch": 0.08186347601556533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9859273338790703e-05,
+      "loss": 1.2734,
+      "step": 2251
+    },
+    {
+      "epoch": 0.08189984361930393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9859076362037703e-05,
+      "loss": 1.1603,
+      "step": 2252
+    },
+    {
+      "epoch": 0.08193621122304251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985887924850401e-05,
+      "loss": 1.0869,
+      "step": 2253
+    },
+    {
+      "epoch": 0.0819725788267811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985868199819235e-05,
+      "loss": 1.1328,
+      "step": 2254
+    },
+    {
+      "epoch": 0.08200894643051969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985848461110547e-05,
+      "loss": 1.1968,
+      "step": 2255
+    },
+    {
+      "epoch": 0.08204531403425828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98582870872461e-05,
+      "loss": 1.3086,
+      "step": 2256
+    },
+    {
+      "epoch": 0.08208168163799687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9858089426616986e-05,
+      "loss": 1.1696,
+      "step": 2257
+    },
+    {
+      "epoch": 0.08211804924173546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9857891629220868e-05,
+      "loss": 1.1221,
+      "step": 2258
+    },
+    {
+      "epoch": 0.08215441684547405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985769369506049e-05,
+      "loss": 1.2255,
+      "step": 2259
+    },
+    {
+      "epoch": 0.08219078444921264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98574956241386e-05,
+      "loss": 1.087,
+      "step": 2260
+    },
+    {
+      "epoch": 0.08222715205295122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985729741645794e-05,
+      "loss": 1.2635,
+      "step": 2261
+    },
+    {
+      "epoch": 0.08226351965668982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9857099072021264e-05,
+      "loss": 1.2999,
+      "step": 2262
+    },
+    {
+      "epoch": 0.08229988726042842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985690059083133e-05,
+      "loss": 1.281,
+      "step": 2263
+    },
+    {
+      "epoch": 0.082336254864167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9856701972890878e-05,
+      "loss": 1.201,
+      "step": 2264
+    },
+    {
+      "epoch": 0.0823726224679056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9856503218202674e-05,
+      "loss": 1.1315,
+      "step": 2265
+    },
+    {
+      "epoch": 0.08240899007164418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9856304326769474e-05,
+      "loss": 1.169,
+      "step": 2266
+    },
+    {
+      "epoch": 0.08244535767538277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9856105298594033e-05,
+      "loss": 1.2465,
+      "step": 2267
+    },
+    {
+      "epoch": 0.08248172527912136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9855906133679116e-05,
+      "loss": 1.237,
+      "step": 2268
+    },
+    {
+      "epoch": 0.08251809288285995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9855706832027484e-05,
+      "loss": 1.343,
+      "step": 2269
+    },
+    {
+      "epoch": 0.08255446048659854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9855507393641905e-05,
+      "loss": 1.0877,
+      "step": 2270
+    },
+    {
+      "epoch": 0.08259082809033713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9855307818525144e-05,
+      "loss": 1.1793,
+      "step": 2271
+    },
+    {
+      "epoch": 0.08262719569407571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985510810667997e-05,
+      "loss": 1.3511,
+      "step": 2272
+    },
+    {
+      "epoch": 0.08266356329781431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9854908258109154e-05,
+      "loss": 1.2313,
+      "step": 2273
+    },
+    {
+      "epoch": 0.08269993090155289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9854708272815465e-05,
+      "loss": 1.1837,
+      "step": 2274
+    },
+    {
+      "epoch": 0.08273629850529149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9854508150801683e-05,
+      "loss": 1.1074,
+      "step": 2275
+    },
+    {
+      "epoch": 0.08277266610903007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985430789207058e-05,
+      "loss": 1.1381,
+      "step": 2276
+    },
+    {
+      "epoch": 0.08280903371276867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985410749662494e-05,
+      "loss": 1.2575,
+      "step": 2277
+    },
+    {
+      "epoch": 0.08284540131650725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985390696446754e-05,
+      "loss": 1.0723,
+      "step": 2278
+    },
+    {
+      "epoch": 0.08288176892024585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985370629560116e-05,
+      "loss": 1.2066,
+      "step": 2279
+    },
+    {
+      "epoch": 0.08291813652398443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9853505490028583e-05,
+      "loss": 1.2181,
+      "step": 2280
+    },
+    {
+      "epoch": 0.08295450412772302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98533045477526e-05,
+      "loss": 1.2453,
+      "step": 2281
+    },
+    {
+      "epoch": 0.08299087173146161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9853103468775998e-05,
+      "loss": 1.1468,
+      "step": 2282
+    },
+    {
+      "epoch": 0.0830272393352002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9852902253101562e-05,
+      "loss": 1.1936,
+      "step": 2283
+    },
+    {
+      "epoch": 0.0830636069389388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985270090073209e-05,
+      "loss": 1.3242,
+      "step": 2284
+    },
+    {
+      "epoch": 0.08309997454267738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985249941167037e-05,
+      "loss": 1.2158,
+      "step": 2285
+    },
+    {
+      "epoch": 0.08313634214641598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98522977859192e-05,
+      "loss": 1.0679,
+      "step": 2286
+    },
+    {
+      "epoch": 0.08317270975015456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9852096023481374e-05,
+      "loss": 1.2612,
+      "step": 2287
+    },
+    {
+      "epoch": 0.08320907735389316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98518941243597e-05,
+      "loss": 1.2529,
+      "step": 2288
+    },
+    {
+      "epoch": 0.08324544495763174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985169208855697e-05,
+      "loss": 1.0957,
+      "step": 2289
+    },
+    {
+      "epoch": 0.08328181256137034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9851489916075986e-05,
+      "loss": 1.2096,
+      "step": 2290
+    },
+    {
+      "epoch": 0.08331818016510892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9851287606919563e-05,
+      "loss": 1.1912,
+      "step": 2291
+    },
+    {
+      "epoch": 0.08335454776884751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9851085161090497e-05,
+      "loss": 1.1757,
+      "step": 2292
+    },
+    {
+      "epoch": 0.0833909153725861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9850882578591605e-05,
+      "loss": 1.1497,
+      "step": 2293
+    },
+    {
+      "epoch": 0.0834272829763247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9850679859425692e-05,
+      "loss": 1.2866,
+      "step": 2294
+    },
+    {
+      "epoch": 0.08346365058006328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9850477003595573e-05,
+      "loss": 1.1069,
+      "step": 2295
+    },
+    {
+      "epoch": 0.08350001818380187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985027401110406e-05,
+      "loss": 1.1622,
+      "step": 2296
+    },
+    {
+      "epoch": 0.08353638578754045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9850070881953974e-05,
+      "loss": 1.3587,
+      "step": 2297
+    },
+    {
+      "epoch": 0.08357275339127905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9849867616148125e-05,
+      "loss": 1.1716,
+      "step": 2298
+    },
+    {
+      "epoch": 0.08360912099501763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9849664213689342e-05,
+      "loss": 1.245,
+      "step": 2299
+    },
+    {
+      "epoch": 0.08364548859875623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9849460674580442e-05,
+      "loss": 1.1501,
+      "step": 2300
+    },
+    {
+      "epoch": 0.08368185620249481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984925699882425e-05,
+      "loss": 1.1807,
+      "step": 2301
+    },
+    {
+      "epoch": 0.08371822380623341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984905318642359e-05,
+      "loss": 1.1577,
+      "step": 2302
+    },
+    {
+      "epoch": 0.08375459140997199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9848849237381295e-05,
+      "loss": 1.2355,
+      "step": 2303
+    },
+    {
+      "epoch": 0.08379095901371059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9848645151700184e-05,
+      "loss": 1.186,
+      "step": 2304
+    },
+    {
+      "epoch": 0.08382732661744918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98484409293831e-05,
+      "loss": 1.1699,
+      "step": 2305
+    },
+    {
+      "epoch": 0.08386369422118777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9848236570432867e-05,
+      "loss": 1.1961,
+      "step": 2306
+    },
+    {
+      "epoch": 0.08390006182492636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9848032074852327e-05,
+      "loss": 1.2549,
+      "step": 2307
+    },
+    {
+      "epoch": 0.08393642942866494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9847827442644318e-05,
+      "loss": 1.2671,
+      "step": 2308
+    },
+    {
+      "epoch": 0.08397279703240354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984762267381167e-05,
+      "loss": 1.2979,
+      "step": 2309
+    },
+    {
+      "epoch": 0.08400916463614212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9847417768357233e-05,
+      "loss": 1.1815,
+      "step": 2310
+    },
+    {
+      "epoch": 0.08404553223988072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984721272628384e-05,
+      "loss": 1.239,
+      "step": 2311
+    },
+    {
+      "epoch": 0.0840818998436193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9847007547594346e-05,
+      "loss": 1.1329,
+      "step": 2312
+    },
+    {
+      "epoch": 0.0841182674473579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9846802232291593e-05,
+      "loss": 1.2564,
+      "step": 2313
+    },
+    {
+      "epoch": 0.08415463505109648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984659678037843e-05,
+      "loss": 1.2098,
+      "step": 2314
+    },
+    {
+      "epoch": 0.08419100265483508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9846391191857708e-05,
+      "loss": 1.2042,
+      "step": 2315
+    },
+    {
+      "epoch": 0.08422737025857366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984618546673228e-05,
+      "loss": 1.2593,
+      "step": 2316
+    },
+    {
+      "epoch": 0.08426373786231225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9845979605004997e-05,
+      "loss": 1.2179,
+      "step": 2317
+    },
+    {
+      "epoch": 0.08430010546605084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9845773606678714e-05,
+      "loss": 1.2517,
+      "step": 2318
+    },
+    {
+      "epoch": 0.08433647306978943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9845567471756296e-05,
+      "loss": 1.2181,
+      "step": 2319
+    },
+    {
+      "epoch": 0.08437284067352802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9845361200240593e-05,
+      "loss": 1.2751,
+      "step": 2320
+    },
+    {
+      "epoch": 0.08440920827726661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9845154792134477e-05,
+      "loss": 1.1013,
+      "step": 2321
+    },
+    {
+      "epoch": 0.0844455758810052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9844948247440802e-05,
+      "loss": 1.1876,
+      "step": 2322
+    },
+    {
+      "epoch": 0.08448194348474379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984474156616244e-05,
+      "loss": 1.4127,
+      "step": 2323
+    },
+    {
+      "epoch": 0.08451831108848239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984453474830226e-05,
+      "loss": 1.2646,
+      "step": 2324
+    },
+    {
+      "epoch": 0.08455467869222097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9844327793863127e-05,
+      "loss": 1.1419,
+      "step": 2325
+    },
+    {
+      "epoch": 0.08459104629595957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984412070284791e-05,
+      "loss": 1.2643,
+      "step": 2326
+    },
+    {
+      "epoch": 0.08462741389969815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9843913475259487e-05,
+      "loss": 1.1784,
+      "step": 2327
+    },
+    {
+      "epoch": 0.08466378150343674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9843706111100732e-05,
+      "loss": 1.2852,
+      "step": 2328
+    },
+    {
+      "epoch": 0.08470014910717533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984349861037452e-05,
+      "loss": 1.3076,
+      "step": 2329
+    },
+    {
+      "epoch": 0.08473651671091392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9843290973083728e-05,
+      "loss": 1.3468,
+      "step": 2330
+    },
+    {
+      "epoch": 0.0847728843146525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9843083199231243e-05,
+      "loss": 1.3096,
+      "step": 2331
+    },
+    {
+      "epoch": 0.0848092519183911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842875288819943e-05,
+      "loss": 1.3163,
+      "step": 2332
+    },
+    {
+      "epoch": 0.08484561952212968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842667241852716e-05,
+      "loss": 1.2938,
+      "step": 2333
+    },
+    {
+      "epoch": 0.08488198712586828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842459058332445e-05,
+      "loss": 1.268,
+      "step": 2334
+    },
+    {
+      "epoch": 0.08491835472960686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842250738262016e-05,
+      "loss": 1.1862,
+      "step": 2335
+    },
+    {
+      "epoch": 0.08495472233334546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842042281644326e-05,
+      "loss": 1.2849,
+      "step": 2336
+    },
+    {
+      "epoch": 0.08499108993708404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984183368848226e-05,
+      "loss": 1.1432,
+      "step": 2337
+    },
+    {
+      "epoch": 0.08502745754082264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9841624958778716e-05,
+      "loss": 1.1675,
+      "step": 2338
+    },
+    {
+      "epoch": 0.08506382514456122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9841416092536594e-05,
+      "loss": 1.2537,
+      "step": 2339
+    },
+    {
+      "epoch": 0.08510019274829982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9841207089758785e-05,
+      "loss": 1.1381,
+      "step": 2340
+    },
+    {
+      "epoch": 0.0851365603520384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9840997950448186e-05,
+      "loss": 1.1472,
+      "step": 2341
+    },
+    {
+      "epoch": 0.085172927955777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9840788674607707e-05,
+      "loss": 1.209,
+      "step": 2342
+    },
+    {
+      "epoch": 0.08520929555951558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9840579262240245e-05,
+      "loss": 1.2808,
+      "step": 2343
+    },
+    {
+      "epoch": 0.08524566316325417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.984036971334871e-05,
+      "loss": 1.0955,
+      "step": 2344
+    },
+    {
+      "epoch": 0.08528203076699277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9840160027936005e-05,
+      "loss": 1.0669,
+      "step": 2345
+    },
+    {
+      "epoch": 0.08531839837073135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9839950206005043e-05,
+      "loss": 1.1252,
+      "step": 2346
+    },
+    {
+      "epoch": 0.08535476597446995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9839740247558733e-05,
+      "loss": 1.1873,
+      "step": 2347
+    },
+    {
+      "epoch": 0.08539113357820853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9839530152599983e-05,
+      "loss": 1.1621,
+      "step": 2348
+    },
+    {
+      "epoch": 0.08542750118194713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983931992113172e-05,
+      "loss": 1.3287,
+      "step": 2349
+    },
+    {
+      "epoch": 0.08546386878568571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9839109553156847e-05,
+      "loss": 1.1201,
+      "step": 2350
+    },
+    {
+      "epoch": 0.0855002363894243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9838899048678294e-05,
+      "loss": 1.1168,
+      "step": 2351
+    },
+    {
+      "epoch": 0.08553660399316289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983868840769897e-05,
+      "loss": 1.1694,
+      "step": 2352
+    },
+    {
+      "epoch": 0.08557297159690148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9838477630221807e-05,
+      "loss": 1.1322,
+      "step": 2353
+    },
+    {
+      "epoch": 0.08560933920064007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983826671624973e-05,
+      "loss": 1.2139,
+      "step": 2354
+    },
+    {
+      "epoch": 0.08564570680437866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9838055665785653e-05,
+      "loss": 1.1486,
+      "step": 2355
+    },
+    {
+      "epoch": 0.08568207440811725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9837844478832518e-05,
+      "loss": 1.1192,
+      "step": 2356
+    },
+    {
+      "epoch": 0.08571844201185584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9837633155393247e-05,
+      "loss": 1.2004,
+      "step": 2357
+    },
+    {
+      "epoch": 0.08575480961559442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9837421695470772e-05,
+      "loss": 1.2285,
+      "step": 2358
+    },
+    {
+      "epoch": 0.08579117721933302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9837210099068033e-05,
+      "loss": 1.2551,
+      "step": 2359
+    },
+    {
+      "epoch": 0.0858275448230716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9836998366187957e-05,
+      "loss": 1.3588,
+      "step": 2360
+    },
+    {
+      "epoch": 0.0858639124268102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9836786496833485e-05,
+      "loss": 1.2744,
+      "step": 2361
+    },
+    {
+      "epoch": 0.08590028003054878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983657449100756e-05,
+      "loss": 1.3507,
+      "step": 2362
+    },
+    {
+      "epoch": 0.08593664763428738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983636234871312e-05,
+      "loss": 1.2188,
+      "step": 2363
+    },
+    {
+      "epoch": 0.08597301523802596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9836150069953105e-05,
+      "loss": 1.124,
+      "step": 2364
+    },
+    {
+      "epoch": 0.08600938284176456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983593765473047e-05,
+      "loss": 1.1884,
+      "step": 2365
+    },
+    {
+      "epoch": 0.08604575044550315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9835725103048147e-05,
+      "loss": 1.2561,
+      "step": 2366
+    },
+    {
+      "epoch": 0.08608211804924173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98355124149091e-05,
+      "loss": 1.2738,
+      "step": 2367
+    },
+    {
+      "epoch": 0.08611848565298033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983529959031627e-05,
+      "loss": 1.1884,
+      "step": 2368
+    },
+    {
+      "epoch": 0.08615485325671891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9835086629272612e-05,
+      "loss": 1.1621,
+      "step": 2369
+    },
+    {
+      "epoch": 0.08619122086045751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983487353178108e-05,
+      "loss": 1.2253,
+      "step": 2370
+    },
+    {
+      "epoch": 0.08622758846419609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9834660297844634e-05,
+      "loss": 1.2056,
+      "step": 2371
+    },
+    {
+      "epoch": 0.08626395606793469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983444692746623e-05,
+      "loss": 1.2484,
+      "step": 2372
+    },
+    {
+      "epoch": 0.08630032367167327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983423342064883e-05,
+      "loss": 1.2067,
+      "step": 2373
+    },
+    {
+      "epoch": 0.08633669127541187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983401977739539e-05,
+      "loss": 1.2135,
+      "step": 2374
+    },
+    {
+      "epoch": 0.08637305887915045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983380599770888e-05,
+      "loss": 1.125,
+      "step": 2375
+    },
+    {
+      "epoch": 0.08640942648288905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9833592081592265e-05,
+      "loss": 1.318,
+      "step": 2376
+    },
+    {
+      "epoch": 0.08644579408662763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9833378029048512e-05,
+      "loss": 1.112,
+      "step": 2377
+    },
+    {
+      "epoch": 0.08648216169036622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9833163840080593e-05,
+      "loss": 1.186,
+      "step": 2378
+    },
+    {
+      "epoch": 0.0865185292941048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983294951469147e-05,
+      "loss": 1.1125,
+      "step": 2379
+    },
+    {
+      "epoch": 0.0865548968978434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9832735052884133e-05,
+      "loss": 1.2324,
+      "step": 2380
+    },
+    {
+      "epoch": 0.08659126450158199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9832520454661538e-05,
+      "loss": 1.1371,
+      "step": 2381
+    },
+    {
+      "epoch": 0.08662763210532058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983230572002668e-05,
+      "loss": 1.3195,
+      "step": 2382
+    },
+    {
+      "epoch": 0.08666399970905916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9832090848982524e-05,
+      "loss": 1.2474,
+      "step": 2383
+    },
+    {
+      "epoch": 0.08670036731279776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9831875841532062e-05,
+      "loss": 1.2286,
+      "step": 2384
+    },
+    {
+      "epoch": 0.08673673491653634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983166069767827e-05,
+      "loss": 1.1282,
+      "step": 2385
+    },
+    {
+      "epoch": 0.08677310252027494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9831445417424136e-05,
+      "loss": 1.3088,
+      "step": 2386
+    },
+    {
+      "epoch": 0.08680947012401354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9831230000772646e-05,
+      "loss": 1.0713,
+      "step": 2387
+    },
+    {
+      "epoch": 0.08684583772775212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9831014447726788e-05,
+      "loss": 1.2117,
+      "step": 2388
+    },
+    {
+      "epoch": 0.08688220533149071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9830798758289554e-05,
+      "loss": 1.2216,
+      "step": 2389
+    },
+    {
+      "epoch": 0.0869185729352293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9830582932463935e-05,
+      "loss": 1.2728,
+      "step": 2390
+    },
+    {
+      "epoch": 0.08695494053896789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9830366970252927e-05,
+      "loss": 1.131,
+      "step": 2391
+    },
+    {
+      "epoch": 0.08699130814270647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983015087165952e-05,
+      "loss": 1.11,
+      "step": 2392
+    },
+    {
+      "epoch": 0.08702767574644507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9829934636686722e-05,
+      "loss": 1.1064,
+      "step": 2393
+    },
+    {
+      "epoch": 0.08706404335018365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9829718265337523e-05,
+      "loss": 1.1492,
+      "step": 2394
+    },
+    {
+      "epoch": 0.08710041095392225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982950175761493e-05,
+      "loss": 1.2456,
+      "step": 2395
+    },
+    {
+      "epoch": 0.08713677855766083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9829285113521953e-05,
+      "loss": 1.1329,
+      "step": 2396
+    },
+    {
+      "epoch": 0.08717314616139943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9829068333061587e-05,
+      "loss": 1.2295,
+      "step": 2397
+    },
+    {
+      "epoch": 0.08720951376513801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982885141623684e-05,
+      "loss": 1.2814,
+      "step": 2398
+    },
+    {
+      "epoch": 0.08724588136887661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982863436305073e-05,
+      "loss": 1.2339,
+      "step": 2399
+    },
+    {
+      "epoch": 0.08728224897261519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982841717350626e-05,
+      "loss": 1.1454,
+      "step": 2400
+    },
+    {
+      "epoch": 0.08731861657635379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9828199847606448e-05,
+      "loss": 1.1353,
+      "step": 2401
+    },
+    {
+      "epoch": 0.08735498418009237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9827982385354307e-05,
+      "loss": 1.2119,
+      "step": 2402
+    },
+    {
+      "epoch": 0.08739135178383096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9827764786752853e-05,
+      "loss": 1.2514,
+      "step": 2403
+    },
+    {
+      "epoch": 0.08742771938756955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982754705180511e-05,
+      "loss": 1.1382,
+      "step": 2404
+    },
+    {
+      "epoch": 0.08746408699130814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9827329180514093e-05,
+      "loss": 1.224,
+      "step": 2405
+    },
+    {
+      "epoch": 0.08750045459504673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9827111172882828e-05,
+      "loss": 1.2709,
+      "step": 2406
+    },
+    {
+      "epoch": 0.08753682219878532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9826893028914336e-05,
+      "loss": 1.1434,
+      "step": 2407
+    },
+    {
+      "epoch": 0.08757318980252392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9826674748611646e-05,
+      "loss": 1.1892,
+      "step": 2408
+    },
+    {
+      "epoch": 0.0876095574062625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9826456331977786e-05,
+      "loss": 1.191,
+      "step": 2409
+    },
+    {
+      "epoch": 0.0876459250100011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982623777901579e-05,
+      "loss": 1.208,
+      "step": 2410
+    },
+    {
+      "epoch": 0.08768229261373968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9826019089728682e-05,
+      "loss": 1.2203,
+      "step": 2411
+    },
+    {
+      "epoch": 0.08771866021747828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9825800264119502e-05,
+      "loss": 1.1682,
+      "step": 2412
+    },
+    {
+      "epoch": 0.08775502782121686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9825581302191286e-05,
+      "loss": 1.0966,
+      "step": 2413
+    },
+    {
+      "epoch": 0.08779139542495545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9825362203947066e-05,
+      "loss": 1.1564,
+      "step": 2414
+    },
+    {
+      "epoch": 0.08782776302869404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982514296938989e-05,
+      "loss": 1.2183,
+      "step": 2415
+    },
+    {
+      "epoch": 0.08786413063243263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9824923598522795e-05,
+      "loss": 1.0729,
+      "step": 2416
+    },
+    {
+      "epoch": 0.08790049823617122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982470409134882e-05,
+      "loss": 1.2005,
+      "step": 2417
+    },
+    {
+      "epoch": 0.08793686583990981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982448444787102e-05,
+      "loss": 1.1696,
+      "step": 2418
+    },
+    {
+      "epoch": 0.0879732334436484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9824264668092437e-05,
+      "loss": 1.2094,
+      "step": 2419
+    },
+    {
+      "epoch": 0.08800960104738699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9824044752016117e-05,
+      "loss": 1.1572,
+      "step": 2420
+    },
+    {
+      "epoch": 0.08804596865112557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982382469964512e-05,
+      "loss": 1.2932,
+      "step": 2421
+    },
+    {
+      "epoch": 0.08808233625486417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982360451098249e-05,
+      "loss": 1.1089,
+      "step": 2422
+    },
+    {
+      "epoch": 0.08811870385860275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982338418603129e-05,
+      "loss": 1.2082,
+      "step": 2423
+    },
+    {
+      "epoch": 0.08815507146234135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9823163724794562e-05,
+      "loss": 1.2537,
+      "step": 2424
+    },
+    {
+      "epoch": 0.08819143906607993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982294312727538e-05,
+      "loss": 1.1609,
+      "step": 2425
+    },
+    {
+      "epoch": 0.08822780666981853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9822722393476802e-05,
+      "loss": 1.1966,
+      "step": 2426
+    },
+    {
+      "epoch": 0.08826417427355712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9822501523401883e-05,
+      "loss": 1.2194,
+      "step": 2427
+    },
+    {
+      "epoch": 0.0883005418772957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9822280517053695e-05,
+      "loss": 1.0457,
+      "step": 2428
+    },
+    {
+      "epoch": 0.0883369094810343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98220593744353e-05,
+      "loss": 1.1728,
+      "step": 2429
+    },
+    {
+      "epoch": 0.08837327708477288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9821838095549766e-05,
+      "loss": 1.1648,
+      "step": 2430
+    },
+    {
+      "epoch": 0.08840964468851148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9821616680400164e-05,
+      "loss": 1.213,
+      "step": 2431
+    },
+    {
+      "epoch": 0.08844601229225006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9821395128989565e-05,
+      "loss": 1.198,
+      "step": 2432
+    },
+    {
+      "epoch": 0.08848237989598866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9821173441321044e-05,
+      "loss": 1.1141,
+      "step": 2433
+    },
+    {
+      "epoch": 0.08851874749972724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982095161739768e-05,
+      "loss": 1.1861,
+      "step": 2434
+    },
+    {
+      "epoch": 0.08855511510346584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9820729657222538e-05,
+      "loss": 1.1935,
+      "step": 2435
+    },
+    {
+      "epoch": 0.08859148270720442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982050756079871e-05,
+      "loss": 1.1679,
+      "step": 2436
+    },
+    {
+      "epoch": 0.08862785031094302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9820285328129274e-05,
+      "loss": 1.174,
+      "step": 2437
+    },
+    {
+      "epoch": 0.0886642179146816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982006295921731e-05,
+      "loss": 1.1539,
+      "step": 2438
+    },
+    {
+      "epoch": 0.0887005855184202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981984045406591e-05,
+      "loss": 1.1948,
+      "step": 2439
+    },
+    {
+      "epoch": 0.08873695312215878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981961781267815e-05,
+      "loss": 1.2255,
+      "step": 2440
+    },
+    {
+      "epoch": 0.08877332072589737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981939503505713e-05,
+      "loss": 1.3202,
+      "step": 2441
+    },
+    {
+      "epoch": 0.08880968832963596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9819172121205934e-05,
+      "loss": 1.3073,
+      "step": 2442
+    },
+    {
+      "epoch": 0.08884605593337455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9818949071127655e-05,
+      "loss": 1.208,
+      "step": 2443
+    },
+    {
+      "epoch": 0.08888242353711313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9818725884825388e-05,
+      "loss": 1.089,
+      "step": 2444
+    },
+    {
+      "epoch": 0.08891879114085173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9818502562302235e-05,
+      "loss": 1.2103,
+      "step": 2445
+    },
+    {
+      "epoch": 0.08895515874459031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9818279103561286e-05,
+      "loss": 1.1138,
+      "step": 2446
+    },
+    {
+      "epoch": 0.08899152634832891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9818055508605646e-05,
+      "loss": 1.2548,
+      "step": 2447
+    },
+    {
+      "epoch": 0.0890278939520675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9817831777438416e-05,
+      "loss": 1.3096,
+      "step": 2448
+    },
+    {
+      "epoch": 0.08906426155580609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98176079100627e-05,
+      "loss": 1.2459,
+      "step": 2449
+    },
+    {
+      "epoch": 0.08910062915954468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9817383906481604e-05,
+      "loss": 1.3644,
+      "step": 2450
+    },
+    {
+      "epoch": 0.08913699676328327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9817159766698234e-05,
+      "loss": 1.2847,
+      "step": 2451
+    },
+    {
+      "epoch": 0.08917336436702186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98169354907157e-05,
+      "loss": 1.1539,
+      "step": 2452
+    },
+    {
+      "epoch": 0.08920973197076044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9816711078537115e-05,
+      "loss": 1.0891,
+      "step": 2453
+    },
+    {
+      "epoch": 0.08924609957449904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9816486530165593e-05,
+      "loss": 1.2709,
+      "step": 2454
+    },
+    {
+      "epoch": 0.08928246717823762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981626184560425e-05,
+      "loss": 1.1569,
+      "step": 2455
+    },
+    {
+      "epoch": 0.08931883478197622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9816037024856198e-05,
+      "loss": 1.0986,
+      "step": 2456
+    },
+    {
+      "epoch": 0.0893552023857148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9815812067924562e-05,
+      "loss": 1.1265,
+      "step": 2457
+    },
+    {
+      "epoch": 0.0893915699894534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9815586974812454e-05,
+      "loss": 1.203,
+      "step": 2458
+    },
+    {
+      "epoch": 0.08942793759319198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981536174552301e-05,
+      "loss": 1.2093,
+      "step": 2459
+    },
+    {
+      "epoch": 0.08946430519693058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9815136380059344e-05,
+      "loss": 1.1534,
+      "step": 2460
+    },
+    {
+      "epoch": 0.08950067280066916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981491087842459e-05,
+      "loss": 1.0863,
+      "step": 2461
+    },
+    {
+      "epoch": 0.08953704040440776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9814685240621875e-05,
+      "loss": 1.1616,
+      "step": 2462
+    },
+    {
+      "epoch": 0.08957340800814634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9814459466654323e-05,
+      "loss": 1.1807,
+      "step": 2463
+    },
+    {
+      "epoch": 0.08960977561188493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9814233556525072e-05,
+      "loss": 1.3066,
+      "step": 2464
+    },
+    {
+      "epoch": 0.08964614321562352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9814007510237255e-05,
+      "loss": 1.2433,
+      "step": 2465
+    },
+    {
+      "epoch": 0.08968251081936211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9813781327794007e-05,
+      "loss": 1.2631,
+      "step": 2466
+    },
+    {
+      "epoch": 0.0897188784231007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981355500919847e-05,
+      "loss": 1.1902,
+      "step": 2467
+    },
+    {
+      "epoch": 0.08975524602683929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981332855445378e-05,
+      "loss": 1.2766,
+      "step": 2468
+    },
+    {
+      "epoch": 0.08979161363057789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9813101963563076e-05,
+      "loss": 1.1609,
+      "step": 2469
+    },
+    {
+      "epoch": 0.08982798123431647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9812875236529508e-05,
+      "loss": 0.9531,
+      "step": 2470
+    },
+    {
+      "epoch": 0.08986434883805507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9812648373356218e-05,
+      "loss": 1.2916,
+      "step": 2471
+    },
+    {
+      "epoch": 0.08990071644179365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9812421374046356e-05,
+      "loss": 1.091,
+      "step": 2472
+    },
+    {
+      "epoch": 0.08993708404553225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981219423860307e-05,
+      "loss": 1.2191,
+      "step": 2473
+    },
+    {
+      "epoch": 0.08997345164927083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981196696702951e-05,
+      "loss": 1.1745,
+      "step": 2474
+    },
+    {
+      "epoch": 0.09000981925300942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9811739559328824e-05,
+      "loss": 1.2719,
+      "step": 2475
+    },
+    {
+      "epoch": 0.090046186856748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9811512015504177e-05,
+      "loss": 1.141,
+      "step": 2476
+    },
+    {
+      "epoch": 0.0900825544604866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9811284335558723e-05,
+      "loss": 1.099,
+      "step": 2477
+    },
+    {
+      "epoch": 0.09011892206422518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9811056519495617e-05,
+      "loss": 1.1631,
+      "step": 2478
+    },
+    {
+      "epoch": 0.09015528966796378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9810828567318023e-05,
+      "loss": 1.157,
+      "step": 2479
+    },
+    {
+      "epoch": 0.09019165727170236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9810600479029102e-05,
+      "loss": 1.261,
+      "step": 2480
+    },
+    {
+      "epoch": 0.09022802487544096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981037225463202e-05,
+      "loss": 1.1899,
+      "step": 2481
+    },
+    {
+      "epoch": 0.09026439247917954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981014389412994e-05,
+      "loss": 1.2029,
+      "step": 2482
+    },
+    {
+      "epoch": 0.09030076008291814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9809915397526033e-05,
+      "loss": 1.1624,
+      "step": 2483
+    },
+    {
+      "epoch": 0.09033712768665672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980968676482347e-05,
+      "loss": 1.3054,
+      "step": 2484
+    },
+    {
+      "epoch": 0.09037349529039532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980945799602542e-05,
+      "loss": 1.0893,
+      "step": 2485
+    },
+    {
+      "epoch": 0.0904098628941339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9809229091135056e-05,
+      "loss": 1.1313,
+      "step": 2486
+    },
+    {
+      "epoch": 0.0904462304978725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9809000050155558e-05,
+      "loss": 1.2336,
+      "step": 2487
+    },
+    {
+      "epoch": 0.09048259810161108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98087708730901e-05,
+      "loss": 1.1236,
+      "step": 2488
+    },
+    {
+      "epoch": 0.09051896570534967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9808541559941864e-05,
+      "loss": 1.1124,
+      "step": 2489
+    },
+    {
+      "epoch": 0.09055533330908827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980831211071403e-05,
+      "loss": 1.2222,
+      "step": 2490
+    },
+    {
+      "epoch": 0.09059170091282685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9808082525409782e-05,
+      "loss": 1.2115,
+      "step": 2491
+    },
+    {
+      "epoch": 0.09062806851656545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 1.2967,
+      "step": 2492
+    },
+    {
+      "epoch": 0.09066443612030403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9807622946584787e-05,
+      "loss": 1.2034,
+      "step": 2493
+    },
+    {
+      "epoch": 0.09070080372404263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9807392953070414e-05,
+      "loss": 1.1572,
+      "step": 2494
+    },
+    {
+      "epoch": 0.09073717132778121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9807162823492382e-05,
+      "loss": 1.231,
+      "step": 2495
+    },
+    {
+      "epoch": 0.0907735389315198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9806932557853876e-05,
+      "loss": 1.2675,
+      "step": 2496
+    },
+    {
+      "epoch": 0.09080990653525839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9806702156158096e-05,
+      "loss": 1.1195,
+      "step": 2497
+    },
+    {
+      "epoch": 0.09084627413899699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980647161840824e-05,
+      "loss": 1.2362,
+      "step": 2498
+    },
+    {
+      "epoch": 0.09088264174273557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98062409446075e-05,
+      "loss": 1.2053,
+      "step": 2499
+    },
+    {
+      "epoch": 0.09091900934647416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9806010134759083e-05,
+      "loss": 1.025,
+      "step": 2500
+    },
+    {
+      "epoch": 0.09095537695021275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9805779188866187e-05,
+      "loss": 1.075,
+      "step": 2501
+    },
+    {
+      "epoch": 0.09099174455395134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9805548106932018e-05,
+      "loss": 1.1757,
+      "step": 2502
+    },
+    {
+      "epoch": 0.09102811215768992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980531688895978e-05,
+      "loss": 1.192,
+      "step": 2503
+    },
+    {
+      "epoch": 0.09106447976142852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9805085534952684e-05,
+      "loss": 1.238,
+      "step": 2504
+    },
+    {
+      "epoch": 0.0911008473651671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9804854044913934e-05,
+      "loss": 1.1698,
+      "step": 2505
+    },
+    {
+      "epoch": 0.0911372149689057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9804622418846747e-05,
+      "loss": 1.1034,
+      "step": 2506
+    },
+    {
+      "epoch": 0.09117358257264428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9804390656754337e-05,
+      "loss": 1.1556,
+      "step": 2507
+    },
+    {
+      "epoch": 0.09120995017638288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9804158758639917e-05,
+      "loss": 1.1487,
+      "step": 2508
+    },
+    {
+      "epoch": 0.09124631778012146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9803926724506707e-05,
+      "loss": 1.1066,
+      "step": 2509
+    },
+    {
+      "epoch": 0.09128268538386006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980369455435792e-05,
+      "loss": 1.1938,
+      "step": 2510
+    },
+    {
+      "epoch": 0.09131905298759865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980346224819678e-05,
+      "loss": 1.1577,
+      "step": 2511
+    },
+    {
+      "epoch": 0.09135542059133724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980322980602651e-05,
+      "loss": 1.1252,
+      "step": 2512
+    },
+    {
+      "epoch": 0.09139178819507583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9802997227850337e-05,
+      "loss": 1.1775,
+      "step": 2513
+    },
+    {
+      "epoch": 0.09142815579881441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9802764513671487e-05,
+      "loss": 1.1675,
+      "step": 2514
+    },
+    {
+      "epoch": 0.09146452340255301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9802531663493187e-05,
+      "loss": 1.2296,
+      "step": 2515
+    },
+    {
+      "epoch": 0.0915008910062916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980229867731867e-05,
+      "loss": 1.2455,
+      "step": 2516
+    },
+    {
+      "epoch": 0.09153725861003019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9802065555151163e-05,
+      "loss": 1.2273,
+      "step": 2517
+    },
+    {
+      "epoch": 0.09157362621376877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9801832296993907e-05,
+      "loss": 1.2101,
+      "step": 2518
+    },
+    {
+      "epoch": 0.09160999381750737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9801598902850133e-05,
+      "loss": 1.046,
+      "step": 2519
+    },
+    {
+      "epoch": 0.09164636142124595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980136537272308e-05,
+      "loss": 1.2667,
+      "step": 2520
+    },
+    {
+      "epoch": 0.09168272902498455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980113170661599e-05,
+      "loss": 1.1249,
+      "step": 2521
+    },
+    {
+      "epoch": 0.09171909662872313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9800897904532102e-05,
+      "loss": 1.2581,
+      "step": 2522
+    },
+    {
+      "epoch": 0.09175546423246173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9800663966474666e-05,
+      "loss": 1.1878,
+      "step": 2523
+    },
+    {
+      "epoch": 0.09179183183620031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9800429892446917e-05,
+      "loss": 1.2199,
+      "step": 2524
+    },
+    {
+      "epoch": 0.0918281994399389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980019568245211e-05,
+      "loss": 1.2073,
+      "step": 2525
+    },
+    {
+      "epoch": 0.09186456704367749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9799961336493495e-05,
+      "loss": 1.1841,
+      "step": 2526
+    },
+    {
+      "epoch": 0.09190093464741608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9799726854574316e-05,
+      "loss": 1.0828,
+      "step": 2527
+    },
+    {
+      "epoch": 0.09193730225115467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979949223669784e-05,
+      "loss": 1.2095,
+      "step": 2528
+    },
+    {
+      "epoch": 0.09197366985489326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97992574828673e-05,
+      "loss": 1.2237,
+      "step": 2529
+    },
+    {
+      "epoch": 0.09201003745863186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9799022593085976e-05,
+      "loss": 1.2343,
+      "step": 2530
+    },
+    {
+      "epoch": 0.09204640506237044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979878756735711e-05,
+      "loss": 1.1401,
+      "step": 2531
+    },
+    {
+      "epoch": 0.09208277266610904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9798552405683976e-05,
+      "loss": 1.1659,
+      "step": 2532
+    },
+    {
+      "epoch": 0.09211914026984762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9798317108069823e-05,
+      "loss": 1.1322,
+      "step": 2533
+    },
+    {
+      "epoch": 0.09215550787358621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9798081674517927e-05,
+      "loss": 1.2328,
+      "step": 2534
+    },
+    {
+      "epoch": 0.0921918754773248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9797846105031545e-05,
+      "loss": 1.1462,
+      "step": 2535
+    },
+    {
+      "epoch": 0.0922282430810634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9797610399613955e-05,
+      "loss": 1.1855,
+      "step": 2536
+    },
+    {
+      "epoch": 0.09226461068480198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9797374558268418e-05,
+      "loss": 1.1662,
+      "step": 2537
+    },
+    {
+      "epoch": 0.09230097828854057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979713858099821e-05,
+      "loss": 1.1683,
+      "step": 2538
+    },
+    {
+      "epoch": 0.09233734589227915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9796902467806606e-05,
+      "loss": 1.113,
+      "step": 2539
+    },
+    {
+      "epoch": 0.09237371349601775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9796666218696877e-05,
+      "loss": 1.0807,
+      "step": 2540
+    },
+    {
+      "epoch": 0.09241008109975633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9796429833672305e-05,
+      "loss": 1.125,
+      "step": 2541
+    },
+    {
+      "epoch": 0.09244644870349493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979619331273617e-05,
+      "loss": 1.2245,
+      "step": 2542
+    },
+    {
+      "epoch": 0.09248281630723351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979595665589175e-05,
+      "loss": 1.2133,
+      "step": 2543
+    },
+    {
+      "epoch": 0.09251918391097211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9795719863142334e-05,
+      "loss": 1.11,
+      "step": 2544
+    },
+    {
+      "epoch": 0.09255555151471069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9795482934491198e-05,
+      "loss": 1.2604,
+      "step": 2545
+    },
+    {
+      "epoch": 0.09259191911844929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9795245869941637e-05,
+      "loss": 1.1414,
+      "step": 2546
+    },
+    {
+      "epoch": 0.09262828672218787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9795008669496937e-05,
+      "loss": 1.2383,
+      "step": 2547
+    },
+    {
+      "epoch": 0.09266465432592647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979477133316039e-05,
+      "loss": 1.2389,
+      "step": 2548
+    },
+    {
+      "epoch": 0.09270102192966505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9794533860935287e-05,
+      "loss": 1.1726,
+      "step": 2549
+    },
+    {
+      "epoch": 0.09273738953340364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9794296252824922e-05,
+      "loss": 1.1448,
+      "step": 2550
+    },
+    {
+      "epoch": 0.09277375713714224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9794058508832596e-05,
+      "loss": 1.1578,
+      "step": 2551
+    },
+    {
+      "epoch": 0.09281012474088082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97938206289616e-05,
+      "loss": 1.1212,
+      "step": 2552
+    },
+    {
+      "epoch": 0.09284649234461942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9793582613215242e-05,
+      "loss": 1.1217,
+      "step": 2553
+    },
+    {
+      "epoch": 0.092882859948358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9793344461596814e-05,
+      "loss": 1.1382,
+      "step": 2554
+    },
+    {
+      "epoch": 0.0929192275520966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9793106174109633e-05,
+      "loss": 1.2115,
+      "step": 2555
+    },
+    {
+      "epoch": 0.09295559515583518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9792867750757e-05,
+      "loss": 1.2178,
+      "step": 2556
+    },
+    {
+      "epoch": 0.09299196275957378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9792629191542217e-05,
+      "loss": 1.1317,
+      "step": 2557
+    },
+    {
+      "epoch": 0.09302833036331236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97923904964686e-05,
+      "loss": 1.2274,
+      "step": 2558
+    },
+    {
+      "epoch": 0.09306469796705095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9792151665539454e-05,
+      "loss": 1.2017,
+      "step": 2559
+    },
+    {
+      "epoch": 0.09310106557078954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97919126987581e-05,
+      "loss": 1.2048,
+      "step": 2560
+    },
+    {
+      "epoch": 0.09313743317452813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9791673596127848e-05,
+      "loss": 1.254,
+      "step": 2561
+    },
+    {
+      "epoch": 0.09317380077826672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9791434357652023e-05,
+      "loss": 1.1351,
+      "step": 2562
+    },
+    {
+      "epoch": 0.09321016838200531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9791194983333933e-05,
+      "loss": 1.3312,
+      "step": 2563
+    },
+    {
+      "epoch": 0.0932465359857439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979095547317691e-05,
+      "loss": 1.2749,
+      "step": 2564
+    },
+    {
+      "epoch": 0.09328290358948249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.979071582718427e-05,
+      "loss": 1.0803,
+      "step": 2565
+    },
+    {
+      "epoch": 0.09331927119322107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9790476045359333e-05,
+      "loss": 1.1643,
+      "step": 2566
+    },
+    {
+      "epoch": 0.09335563879695967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9790236127705437e-05,
+      "loss": 1.1893,
+      "step": 2567
+    },
+    {
+      "epoch": 0.09339200640069825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9789996074225905e-05,
+      "loss": 1.2207,
+      "step": 2568
+    },
+    {
+      "epoch": 0.09342837400443685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9789755884924067e-05,
+      "loss": 1.1264,
+      "step": 2569
+    },
+    {
+      "epoch": 0.09346474160817543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9789515559803253e-05,
+      "loss": 1.3124,
+      "step": 2570
+    },
+    {
+      "epoch": 0.09350110921191403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9789275098866806e-05,
+      "loss": 1.1909,
+      "step": 2571
+    },
+    {
+      "epoch": 0.09353747681565262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978903450211805e-05,
+      "loss": 1.1915,
+      "step": 2572
+    },
+    {
+      "epoch": 0.0935738444193912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9788793769560333e-05,
+      "loss": 1.0906,
+      "step": 2573
+    },
+    {
+      "epoch": 0.0936102120231298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9788552901196992e-05,
+      "loss": 1.2756,
+      "step": 2574
+    },
+    {
+      "epoch": 0.09364657962686838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9788311897031366e-05,
+      "loss": 1.2332,
+      "step": 2575
+    },
+    {
+      "epoch": 0.09368294723060698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97880707570668e-05,
+      "loss": 1.1427,
+      "step": 2576
+    },
+    {
+      "epoch": 0.09371931483434556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9787829481306638e-05,
+      "loss": 1.1775,
+      "step": 2577
+    },
+    {
+      "epoch": 0.09375568243808416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978758806975423e-05,
+      "loss": 1.0649,
+      "step": 2578
+    },
+    {
+      "epoch": 0.09379205004182274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9787346522412925e-05,
+      "loss": 1.161,
+      "step": 2579
+    },
+    {
+      "epoch": 0.09382841764556134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9787104839286075e-05,
+      "loss": 1.1552,
+      "step": 2580
+    },
+    {
+      "epoch": 0.09386478524929992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978686302037703e-05,
+      "loss": 1.1353,
+      "step": 2581
+    },
+    {
+      "epoch": 0.09390115285303852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9786621065689145e-05,
+      "loss": 1.1814,
+      "step": 2582
+    },
+    {
+      "epoch": 0.0939375204567771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978637897522578e-05,
+      "loss": 1.1483,
+      "step": 2583
+    },
+    {
+      "epoch": 0.0939738880605157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9786136748990292e-05,
+      "loss": 1.1854,
+      "step": 2584
+    },
+    {
+      "epoch": 0.09401025566425428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9785894386986043e-05,
+      "loss": 1.2058,
+      "step": 2585
+    },
+    {
+      "epoch": 0.09404662326799287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978565188921639e-05,
+      "loss": 1.3086,
+      "step": 2586
+    },
+    {
+      "epoch": 0.09408299087173146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97854092556847e-05,
+      "loss": 1.1781,
+      "step": 2587
+    },
+    {
+      "epoch": 0.09411935847547005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9785166486394343e-05,
+      "loss": 1.2436,
+      "step": 2588
+    },
+    {
+      "epoch": 0.09415572607920863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9784923581348684e-05,
+      "loss": 1.1969,
+      "step": 2589
+    },
+    {
+      "epoch": 0.09419209368294723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9784680540551093e-05,
+      "loss": 1.3082,
+      "step": 2590
+    },
+    {
+      "epoch": 0.09422846128668581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9784437364004946e-05,
+      "loss": 1.1456,
+      "step": 2591
+    },
+    {
+      "epoch": 0.09426482889042441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9784194051713607e-05,
+      "loss": 1.2881,
+      "step": 2592
+    },
+    {
+      "epoch": 0.094301196494163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9783950603680458e-05,
+      "loss": 1.2557,
+      "step": 2593
+    },
+    {
+      "epoch": 0.09433756409790159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9783707019908877e-05,
+      "loss": 1.2191,
+      "step": 2594
+    },
+    {
+      "epoch": 0.09437393170164018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9783463300402244e-05,
+      "loss": 1.3375,
+      "step": 2595
+    },
+    {
+      "epoch": 0.09441029930537877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9783219445163938e-05,
+      "loss": 1.2232,
+      "step": 2596
+    },
+    {
+      "epoch": 0.09444666690911736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9782975454197344e-05,
+      "loss": 1.1257,
+      "step": 2597
+    },
+    {
+      "epoch": 0.09448303451285595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9782731327505842e-05,
+      "loss": 1.0649,
+      "step": 2598
+    },
+    {
+      "epoch": 0.09451940211659454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9782487065092825e-05,
+      "loss": 1.1322,
+      "step": 2599
+    },
+    {
+      "epoch": 0.09455576972033312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978224266696168e-05,
+      "loss": 1.2131,
+      "step": 2600
+    },
+    {
+      "epoch": 0.09459213732407172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9781998133115797e-05,
+      "loss": 1.1769,
+      "step": 2601
+    },
+    {
+      "epoch": 0.0946285049278103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978175346355857e-05,
+      "loss": 1.1065,
+      "step": 2602
+    },
+    {
+      "epoch": 0.0946648725315489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978150865829339e-05,
+      "loss": 1.2714,
+      "step": 2603
+    },
+    {
+      "epoch": 0.09470124013528748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9781263717323657e-05,
+      "loss": 1.1956,
+      "step": 2604
+    },
+    {
+      "epoch": 0.09473760773902608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9781018640652765e-05,
+      "loss": 1.1652,
+      "step": 2605
+    },
+    {
+      "epoch": 0.09477397534276466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978077342828412e-05,
+      "loss": 1.3763,
+      "step": 2606
+    },
+    {
+      "epoch": 0.09481034294650326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9780528080221116e-05,
+      "loss": 1.2201,
+      "step": 2607
+    },
+    {
+      "epoch": 0.09484671055024184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978028259646717e-05,
+      "loss": 1.2145,
+      "step": 2608
+    },
+    {
+      "epoch": 0.09488307815398044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9780036977025673e-05,
+      "loss": 1.2062,
+      "step": 2609
+    },
+    {
+      "epoch": 0.09491944575771902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977979122190004e-05,
+      "loss": 1.1404,
+      "step": 2610
+    },
+    {
+      "epoch": 0.09495581336145761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977954533109368e-05,
+      "loss": 1.2302,
+      "step": 2611
+    },
+    {
+      "epoch": 0.09499218096519621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977929930461e-05,
+      "loss": 1.2545,
+      "step": 2612
+    },
+    {
+      "epoch": 0.09502854856893479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9779053142452422e-05,
+      "loss": 1.1441,
+      "step": 2613
+    },
+    {
+      "epoch": 0.09506491617267339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9778806844624356e-05,
+      "loss": 1.1791,
+      "step": 2614
+    },
+    {
+      "epoch": 0.09510128377641197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977856041112922e-05,
+      "loss": 1.0464,
+      "step": 2615
+    },
+    {
+      "epoch": 0.09513765138015057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977831384197043e-05,
+      "loss": 1.1216,
+      "step": 2616
+    },
+    {
+      "epoch": 0.09517401898388915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977806713715141e-05,
+      "loss": 1.1348,
+      "step": 2617
+    },
+    {
+      "epoch": 0.09521038658762775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9777820296675584e-05,
+      "loss": 1.2072,
+      "step": 2618
+    },
+    {
+      "epoch": 0.09524675419136633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977757332054637e-05,
+      "loss": 1.2513,
+      "step": 2619
+    },
+    {
+      "epoch": 0.09528312179510492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97773262087672e-05,
+      "loss": 1.2599,
+      "step": 2620
+    },
+    {
+      "epoch": 0.09531948939884351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97770789613415e-05,
+      "loss": 1.131,
+      "step": 2621
+    },
+    {
+      "epoch": 0.0953558570025821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9776831578272705e-05,
+      "loss": 1.2936,
+      "step": 2622
+    },
+    {
+      "epoch": 0.09539222460632069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977658405956424e-05,
+      "loss": 1.0669,
+      "step": 2623
+    },
+    {
+      "epoch": 0.09542859221005928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9776336405219546e-05,
+      "loss": 1.2096,
+      "step": 2624
+    },
+    {
+      "epoch": 0.09546495981379786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977608861524205e-05,
+      "loss": 1.2443,
+      "step": 2625
+    },
+    {
+      "epoch": 0.09550132741753646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9775840689635202e-05,
+      "loss": 1.1431,
+      "step": 2626
+    },
+    {
+      "epoch": 0.09553769502127504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977559262840243e-05,
+      "loss": 1.0341,
+      "step": 2627
+    },
+    {
+      "epoch": 0.09557406262501364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977534443154718e-05,
+      "loss": 1.1779,
+      "step": 2628
+    },
+    {
+      "epoch": 0.09561043022875222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9775096099072896e-05,
+      "loss": 1.2106,
+      "step": 2629
+    },
+    {
+      "epoch": 0.09564679783249082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977484763098302e-05,
+      "loss": 1.1607,
+      "step": 2630
+    },
+    {
+      "epoch": 0.0956831654362294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9774599027281006e-05,
+      "loss": 1.2084,
+      "step": 2631
+    },
+    {
+      "epoch": 0.095719533039968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9774350287970295e-05,
+      "loss": 1.2404,
+      "step": 2632
+    },
+    {
+      "epoch": 0.09575590064370659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9774101413054347e-05,
+      "loss": 1.2808,
+      "step": 2633
+    },
+    {
+      "epoch": 0.09579226824744518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9773852402536605e-05,
+      "loss": 1.1971,
+      "step": 2634
+    },
+    {
+      "epoch": 0.09582863585118377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977360325642053e-05,
+      "loss": 1.2037,
+      "step": 2635
+    },
+    {
+      "epoch": 0.09586500345492235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977335397470958e-05,
+      "loss": 1.2009,
+      "step": 2636
+    },
+    {
+      "epoch": 0.09590137105866095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9773104557407202e-05,
+      "loss": 1.213,
+      "step": 2637
+    },
+    {
+      "epoch": 0.09593773866239953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977285500451687e-05,
+      "loss": 1.252,
+      "step": 2638
+    },
+    {
+      "epoch": 0.09597410626613813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9772605316042038e-05,
+      "loss": 1.2713,
+      "step": 2639
+    },
+    {
+      "epoch": 0.09601047386987671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9772355491986176e-05,
+      "loss": 1.2696,
+      "step": 2640
+    },
+    {
+      "epoch": 0.09604684147361531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9772105532352747e-05,
+      "loss": 1.2297,
+      "step": 2641
+    },
+    {
+      "epoch": 0.09608320907735389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9771855437145217e-05,
+      "loss": 1.2475,
+      "step": 2642
+    },
+    {
+      "epoch": 0.09611957668109249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9771605206367054e-05,
+      "loss": 1.2035,
+      "step": 2643
+    },
+    {
+      "epoch": 0.09615594428483107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9771354840021736e-05,
+      "loss": 1.0924,
+      "step": 2644
+    },
+    {
+      "epoch": 0.09619231188856966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9771104338112736e-05,
+      "loss": 1.1567,
+      "step": 2645
+    },
+    {
+      "epoch": 0.09622867949230825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977085370064352e-05,
+      "loss": 1.0983,
+      "step": 2646
+    },
+    {
+      "epoch": 0.09626504709604684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977060292761758e-05,
+      "loss": 1.1119,
+      "step": 2647
+    },
+    {
+      "epoch": 0.09630141469978543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9770352019038382e-05,
+      "loss": 1.3447,
+      "step": 2648
+    },
+    {
+      "epoch": 0.09633778230352402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9770100974909412e-05,
+      "loss": 1.1199,
+      "step": 2649
+    },
+    {
+      "epoch": 0.0963741499072626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9769849795234154e-05,
+      "loss": 1.1516,
+      "step": 2650
+    },
+    {
+      "epoch": 0.0964105175110012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976959848001609e-05,
+      "loss": 1.196,
+      "step": 2651
+    },
+    {
+      "epoch": 0.09644688511473978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976934702925871e-05,
+      "loss": 1.2168,
+      "step": 2652
+    },
+    {
+      "epoch": 0.09648325271847838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9769095442965502e-05,
+      "loss": 1.1359,
+      "step": 2653
+    },
+    {
+      "epoch": 0.09651962032221698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9768843721139955e-05,
+      "loss": 1.2068,
+      "step": 2654
+    },
+    {
+      "epoch": 0.09655598792595556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976859186378556e-05,
+      "loss": 1.2479,
+      "step": 2655
+    },
+    {
+      "epoch": 0.09659235552969415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9768339870905814e-05,
+      "loss": 1.1815,
+      "step": 2656
+    },
+    {
+      "epoch": 0.09662872313343274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976808774250421e-05,
+      "loss": 1.22,
+      "step": 2657
+    },
+    {
+      "epoch": 0.09666509073717133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9767835478584247e-05,
+      "loss": 1.1964,
+      "step": 2658
+    },
+    {
+      "epoch": 0.09670145834090992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976758307914943e-05,
+      "loss": 1.1391,
+      "step": 2659
+    },
+    {
+      "epoch": 0.09673782594464851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976733054420325e-05,
+      "loss": 1.1759,
+      "step": 2660
+    },
+    {
+      "epoch": 0.0967741935483871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976707787374922e-05,
+      "loss": 1.2191,
+      "step": 2661
+    },
+    {
+      "epoch": 0.09681056115212569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9766825067790846e-05,
+      "loss": 1.1272,
+      "step": 2662
+    },
+    {
+      "epoch": 0.09684692875586427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976657212633163e-05,
+      "loss": 1.3129,
+      "step": 2663
+    },
+    {
+      "epoch": 0.09688329635960287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9766319049375075e-05,
+      "loss": 1.2183,
+      "step": 2664
+    },
+    {
+      "epoch": 0.09691966396334145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9766065836924705e-05,
+      "loss": 1.164,
+      "step": 2665
+    },
+    {
+      "epoch": 0.09695603156708005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976581248898403e-05,
+      "loss": 1.1611,
+      "step": 2666
+    },
+    {
+      "epoch": 0.09699239917081863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9765559005556557e-05,
+      "loss": 1.249,
+      "step": 2667
+    },
+    {
+      "epoch": 0.09702876677455723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9765305386645813e-05,
+      "loss": 1.3183,
+      "step": 2668
+    },
+    {
+      "epoch": 0.09706513437829581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9765051632255313e-05,
+      "loss": 1.1536,
+      "step": 2669
+    },
+    {
+      "epoch": 0.0971015019820344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9764797742388572e-05,
+      "loss": 1.1436,
+      "step": 2670
+    },
+    {
+      "epoch": 0.09713786958577299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976454371704912e-05,
+      "loss": 1.1424,
+      "step": 2671
+    },
+    {
+      "epoch": 0.09717423718951158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9764289556240478e-05,
+      "loss": 1.0334,
+      "step": 2672
+    },
+    {
+      "epoch": 0.09721060479325017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976403525996617e-05,
+      "loss": 1.1718,
+      "step": 2673
+    },
+    {
+      "epoch": 0.09724697239698876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9763780828229725e-05,
+      "loss": 1.1444,
+      "step": 2674
+    },
+    {
+      "epoch": 0.09728334000072736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976352626103468e-05,
+      "loss": 1.2784,
+      "step": 2675
+    },
+    {
+      "epoch": 0.09731970760446594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9763271558384554e-05,
+      "loss": 1.2135,
+      "step": 2676
+    },
+    {
+      "epoch": 0.09735607520820454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976301672028289e-05,
+      "loss": 1.1749,
+      "step": 2677
+    },
+    {
+      "epoch": 0.09739244281194312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9762761746733225e-05,
+      "loss": 1.2643,
+      "step": 2678
+    },
+    {
+      "epoch": 0.09742881041568172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9762506637739087e-05,
+      "loss": 1.2944,
+      "step": 2679
+    },
+    {
+      "epoch": 0.0974651780194203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9762251393304023e-05,
+      "loss": 1.1978,
+      "step": 2680
+    },
+    {
+      "epoch": 0.0975015456231589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976199601343157e-05,
+      "loss": 1.0885,
+      "step": 2681
+    },
+    {
+      "epoch": 0.09753791322689748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9761740498125274e-05,
+      "loss": 1.2186,
+      "step": 2682
+    },
+    {
+      "epoch": 0.09757428083063607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9761484847388677e-05,
+      "loss": 1.2776,
+      "step": 2683
+    },
+    {
+      "epoch": 0.09761064843437466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.976122906122533e-05,
+      "loss": 1.2054,
+      "step": 2684
+    },
+    {
+      "epoch": 0.09764701603811325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9760973139638777e-05,
+      "loss": 1.3117,
+      "step": 2685
+    },
+    {
+      "epoch": 0.09768338364185183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9760717082632572e-05,
+      "loss": 1.1773,
+      "step": 2686
+    },
+    {
+      "epoch": 0.09771975124559043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9760460890210264e-05,
+      "loss": 1.2353,
+      "step": 2687
+    },
+    {
+      "epoch": 0.09775611884932901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9760204562375415e-05,
+      "loss": 1.2086,
+      "step": 2688
+    },
+    {
+      "epoch": 0.09779248645306761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975994809913157e-05,
+      "loss": 1.0431,
+      "step": 2689
+    },
+    {
+      "epoch": 0.09782885405680619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9759691500482294e-05,
+      "loss": 1.1949,
+      "step": 2690
+    },
+    {
+      "epoch": 0.09786522166054479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9759434766431148e-05,
+      "loss": 1.1865,
+      "step": 2691
+    },
+    {
+      "epoch": 0.09790158926428337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9759177896981687e-05,
+      "loss": 1.17,
+      "step": 2692
+    },
+    {
+      "epoch": 0.09793795686802197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9758920892137482e-05,
+      "loss": 1.2496,
+      "step": 2693
+    },
+    {
+      "epoch": 0.09797432447176055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9758663751902097e-05,
+      "loss": 1.0997,
+      "step": 2694
+    },
+    {
+      "epoch": 0.09801069207549915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9758406476279095e-05,
+      "loss": 1.2118,
+      "step": 2695
+    },
+    {
+      "epoch": 0.09804705967923774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975814906527205e-05,
+      "loss": 1.1918,
+      "step": 2696
+    },
+    {
+      "epoch": 0.09808342728297632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9757891518884535e-05,
+      "loss": 1.2274,
+      "step": 2697
+    },
+    {
+      "epoch": 0.09811979488671492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9757633837120113e-05,
+      "loss": 1.2734,
+      "step": 2698
+    },
+    {
+      "epoch": 0.0981561624904535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9757376019982374e-05,
+      "loss": 1.3082,
+      "step": 2699
+    },
+    {
+      "epoch": 0.0981925300941921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9757118067474882e-05,
+      "loss": 1.2795,
+      "step": 2700
+    },
+    {
+      "epoch": 0.09822889769793068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975685997960122e-05,
+      "loss": 1.2865,
+      "step": 2701
+    },
+    {
+      "epoch": 0.09826526530166928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9756601756364964e-05,
+      "loss": 1.1985,
+      "step": 2702
+    },
+    {
+      "epoch": 0.09830163290540786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975634339776971e-05,
+      "loss": 1.1539,
+      "step": 2703
+    },
+    {
+      "epoch": 0.09833800050914646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975608490381903e-05,
+      "loss": 1.2331,
+      "step": 2704
+    },
+    {
+      "epoch": 0.09837436811288504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9755826274516513e-05,
+      "loss": 1.0641,
+      "step": 2705
+    },
+    {
+      "epoch": 0.09841073571662363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975556750986575e-05,
+      "loss": 1.1911,
+      "step": 2706
+    },
+    {
+      "epoch": 0.09844710332036222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9755308609870328e-05,
+      "loss": 1.2026,
+      "step": 2707
+    },
+    {
+      "epoch": 0.09848347092410081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9755049574533843e-05,
+      "loss": 1.2741,
+      "step": 2708
+    },
+    {
+      "epoch": 0.0985198385278394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9754790403859883e-05,
+      "loss": 1.2115,
+      "step": 2709
+    },
+    {
+      "epoch": 0.09855620613157799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975453109785205e-05,
+      "loss": 1.1855,
+      "step": 2710
+    },
+    {
+      "epoch": 0.09859257373531657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9754271656513935e-05,
+      "loss": 1.2215,
+      "step": 2711
+    },
+    {
+      "epoch": 0.09862894133905517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9754012079849142e-05,
+      "loss": 1.1619,
+      "step": 2712
+    },
+    {
+      "epoch": 0.09866530894279375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9753752367861265e-05,
+      "loss": 1.1031,
+      "step": 2713
+    },
+    {
+      "epoch": 0.09870167654653235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9753492520553918e-05,
+      "loss": 1.1688,
+      "step": 2714
+    },
+    {
+      "epoch": 0.09873804415027095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97532325379307e-05,
+      "loss": 1.1375,
+      "step": 2715
+    },
+    {
+      "epoch": 0.09877441175400953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9752972419995218e-05,
+      "loss": 1.1153,
+      "step": 2716
+    },
+    {
+      "epoch": 0.09881077935774812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975271216675108e-05,
+      "loss": 1.2685,
+      "step": 2717
+    },
+    {
+      "epoch": 0.0988471469614867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97524517782019e-05,
+      "loss": 1.17,
+      "step": 2718
+    },
+    {
+      "epoch": 0.0988835145652253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9752191254351288e-05,
+      "loss": 1.2285,
+      "step": 2719
+    },
+    {
+      "epoch": 0.09891988216896389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9751930595202857e-05,
+      "loss": 1.213,
+      "step": 2720
+    },
+    {
+      "epoch": 0.09895624977270248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9751669800760226e-05,
+      "loss": 1.0554,
+      "step": 2721
+    },
+    {
+      "epoch": 0.09899261737644106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9751408871027013e-05,
+      "loss": 1.2611,
+      "step": 2722
+    },
+    {
+      "epoch": 0.09902898498017966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9751147806006835e-05,
+      "loss": 1.1284,
+      "step": 2723
+    },
+    {
+      "epoch": 0.09906535258391824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975088660570332e-05,
+      "loss": 1.0932,
+      "step": 2724
+    },
+    {
+      "epoch": 0.09910172018765684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9750625270120086e-05,
+      "loss": 1.2908,
+      "step": 2725
+    },
+    {
+      "epoch": 0.09913808779139542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9750363799260762e-05,
+      "loss": 1.1391,
+      "step": 2726
+    },
+    {
+      "epoch": 0.09917445539513402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9750102193128975e-05,
+      "loss": 1.1509,
+      "step": 2727
+    },
+    {
+      "epoch": 0.0992108229988726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974984045172835e-05,
+      "loss": 1.3075,
+      "step": 2728
+    },
+    {
+      "epoch": 0.0992471906026112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9749578575062524e-05,
+      "loss": 1.3519,
+      "step": 2729
+    },
+    {
+      "epoch": 0.09928355820634978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9749316563135128e-05,
+      "loss": 1.3315,
+      "step": 2730
+    },
+    {
+      "epoch": 0.09931992581008837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9749054415949797e-05,
+      "loss": 1.2101,
+      "step": 2731
+    },
+    {
+      "epoch": 0.09935629341382696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974879213351017e-05,
+      "loss": 1.1049,
+      "step": 2732
+    },
+    {
+      "epoch": 0.09939266101756555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974852971581988e-05,
+      "loss": 1.0042,
+      "step": 2733
+    },
+    {
+      "epoch": 0.09942902862130414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9748267162882574e-05,
+      "loss": 1.2108,
+      "step": 2734
+    },
+    {
+      "epoch": 0.09946539622504273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9748004474701894e-05,
+      "loss": 1.1118,
+      "step": 2735
+    },
+    {
+      "epoch": 0.09950176382878133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974774165128148e-05,
+      "loss": 1.2019,
+      "step": 2736
+    },
+    {
+      "epoch": 0.09953813143251991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974747869262498e-05,
+      "loss": 1.2314,
+      "step": 2737
+    },
+    {
+      "epoch": 0.09957449903625851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9747215598736048e-05,
+      "loss": 1.1942,
+      "step": 2738
+    },
+    {
+      "epoch": 0.09961086663999709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974695236961832e-05,
+      "loss": 1.3363,
+      "step": 2739
+    },
+    {
+      "epoch": 0.09964723424373569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9746689005275465e-05,
+      "loss": 1.0884,
+      "step": 2740
+    },
+    {
+      "epoch": 0.09968360184747427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9746425505711127e-05,
+      "loss": 1.1708,
+      "step": 2741
+    },
+    {
+      "epoch": 0.09971996945121286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974616187092896e-05,
+      "loss": 1.1563,
+      "step": 2742
+    },
+    {
+      "epoch": 0.09975633705495145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974589810093263e-05,
+      "loss": 1.1447,
+      "step": 2743
+    },
+    {
+      "epoch": 0.09979270465869004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974563419572579e-05,
+      "loss": 1.0939,
+      "step": 2744
+    },
+    {
+      "epoch": 0.09982907226242863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97453701553121e-05,
+      "loss": 1.055,
+      "step": 2745
+    },
+    {
+      "epoch": 0.09986543986616722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974510597969523e-05,
+      "loss": 1.1172,
+      "step": 2746
+    },
+    {
+      "epoch": 0.0999018074699058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974484166887884e-05,
+      "loss": 1.0721,
+      "step": 2747
+    },
+    {
+      "epoch": 0.0999381750736444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9744577222866594e-05,
+      "loss": 1.0465,
+      "step": 2748
+    },
+    {
+      "epoch": 0.09997454267738298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9744312641662165e-05,
+      "loss": 1.2007,
+      "step": 2749
+    },
+    {
+      "epoch": 0.10001091028112158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9744047925269228e-05,
+      "loss": 1.2871,
+      "step": 2750
+    },
+    {
+      "epoch": 0.10004727788486016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9743783073691452e-05,
+      "loss": 1.2236,
+      "step": 2751
+    },
+    {
+      "epoch": 0.10008364548859876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9743518086932507e-05,
+      "loss": 1.2012,
+      "step": 2752
+    },
+    {
+      "epoch": 0.10012001309233734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974325296499607e-05,
+      "loss": 1.1831,
+      "step": 2753
+    },
+    {
+      "epoch": 0.10015638069607594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974298770788583e-05,
+      "loss": 1.1085,
+      "step": 2754
+    },
+    {
+      "epoch": 0.10019274829981452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9742722315605452e-05,
+      "loss": 1.1522,
+      "step": 2755
+    },
+    {
+      "epoch": 0.10022911590355311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9742456788158627e-05,
+      "loss": 1.2151,
+      "step": 2756
+    },
+    {
+      "epoch": 0.10026548350729171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974219112554904e-05,
+      "loss": 1.2296,
+      "step": 2757
+    },
+    {
+      "epoch": 0.1003018511110303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9741925327780368e-05,
+      "loss": 1.163,
+      "step": 2758
+    },
+    {
+      "epoch": 0.10033821871476889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9741659394856306e-05,
+      "loss": 1.1656,
+      "step": 2759
+    },
+    {
+      "epoch": 0.10037458631850747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9741393326780542e-05,
+      "loss": 1.3182,
+      "step": 2760
+    },
+    {
+      "epoch": 0.10041095392224607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974112712355677e-05,
+      "loss": 1.0651,
+      "step": 2761
+    },
+    {
+      "epoch": 0.10044732152598465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9740860785188676e-05,
+      "loss": 1.2354,
+      "step": 2762
+    },
+    {
+      "epoch": 0.10048368912972325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974059431167996e-05,
+      "loss": 1.1743,
+      "step": 2763
+    },
+    {
+      "epoch": 0.10052005673346183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974032770303432e-05,
+      "loss": 1.0708,
+      "step": 2764
+    },
+    {
+      "epoch": 0.10055642433720043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.974006095925545e-05,
+      "loss": 1.2113,
+      "step": 2765
+    },
+    {
+      "epoch": 0.10059279194093901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9739794080347056e-05,
+      "loss": 1.1536,
+      "step": 2766
+    },
+    {
+      "epoch": 0.1006291595446776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9739527066312837e-05,
+      "loss": 1.1513,
+      "step": 2767
+    },
+    {
+      "epoch": 0.10066552714841619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9739259917156498e-05,
+      "loss": 1.1961,
+      "step": 2768
+    },
+    {
+      "epoch": 0.10070189475215478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9738992632881744e-05,
+      "loss": 1.1996,
+      "step": 2769
+    },
+    {
+      "epoch": 0.10073826235589337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9738725213492286e-05,
+      "loss": 1.2452,
+      "step": 2770
+    },
+    {
+      "epoch": 0.10077462995963196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9738457658991836e-05,
+      "loss": 1.0573,
+      "step": 2771
+    },
+    {
+      "epoch": 0.10081099756337054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9738189969384102e-05,
+      "loss": 1.1822,
+      "step": 2772
+    },
+    {
+      "epoch": 0.10084736516710914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9737922144672797e-05,
+      "loss": 1.0988,
+      "step": 2773
+    },
+    {
+      "epoch": 0.10088373277084772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9737654184861637e-05,
+      "loss": 1.1525,
+      "step": 2774
+    },
+    {
+      "epoch": 0.10092010037458632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9737386089954347e-05,
+      "loss": 1.241,
+      "step": 2775
+    },
+    {
+      "epoch": 0.1009564679783249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973711785995464e-05,
+      "loss": 1.2178,
+      "step": 2776
+    },
+    {
+      "epoch": 0.1009928355820635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9736849494866234e-05,
+      "loss": 1.1824,
+      "step": 2777
+    },
+    {
+      "epoch": 0.1010292031858021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9736580994692852e-05,
+      "loss": 1.2026,
+      "step": 2778
+    },
+    {
+      "epoch": 0.10106557078954068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973631235943823e-05,
+      "loss": 1.2235,
+      "step": 2779
+    },
+    {
+      "epoch": 0.10110193839327927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973604358910609e-05,
+      "loss": 1.1979,
+      "step": 2780
+    },
+    {
+      "epoch": 0.10113830599701785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9735774683700154e-05,
+      "loss": 1.0887,
+      "step": 2781
+    },
+    {
+      "epoch": 0.10117467360075645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9735505643224157e-05,
+      "loss": 1.1194,
+      "step": 2782
+    },
+    {
+      "epoch": 0.10121104120449503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9735236467681834e-05,
+      "loss": 1.1128,
+      "step": 2783
+    },
+    {
+      "epoch": 0.10124740880823363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973496715707692e-05,
+      "loss": 1.1986,
+      "step": 2784
+    },
+    {
+      "epoch": 0.10128377641197221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973469771141314e-05,
+      "loss": 1.2321,
+      "step": 2785
+    },
+    {
+      "epoch": 0.10132014401571081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973442813069425e-05,
+      "loss": 1.2924,
+      "step": 2786
+    },
+    {
+      "epoch": 0.10135651161944939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9734158414923978e-05,
+      "loss": 1.1843,
+      "step": 2787
+    },
+    {
+      "epoch": 0.10139287922318799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973388856410607e-05,
+      "loss": 1.1205,
+      "step": 2788
+    },
+    {
+      "epoch": 0.10142924682692657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973361857824427e-05,
+      "loss": 1.1786,
+      "step": 2789
+    },
+    {
+      "epoch": 0.10146561443066517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973334845734232e-05,
+      "loss": 1.1897,
+      "step": 2790
+    },
+    {
+      "epoch": 0.10150198203440375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973307820140397e-05,
+      "loss": 1.1882,
+      "step": 2791
+    },
+    {
+      "epoch": 0.10153834963814234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9732807810432966e-05,
+      "loss": 1.2993,
+      "step": 2792
+    },
+    {
+      "epoch": 0.10157471724188093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973253728443307e-05,
+      "loss": 1.0601,
+      "step": 2793
+    },
+    {
+      "epoch": 0.10161108484561952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9732266623408024e-05,
+      "loss": 1.1443,
+      "step": 2794
+    },
+    {
+      "epoch": 0.1016474524493581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9731995827361586e-05,
+      "loss": 1.2322,
+      "step": 2795
+    },
+    {
+      "epoch": 0.1016838200530967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9731724896297514e-05,
+      "loss": 1.2586,
+      "step": 2796
+    },
+    {
+      "epoch": 0.10172018765683528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9731453830219567e-05,
+      "loss": 1.0621,
+      "step": 2797
+    },
+    {
+      "epoch": 0.10175655526057388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9731182629131505e-05,
+      "loss": 1.193,
+      "step": 2798
+    },
+    {
+      "epoch": 0.10179292286431248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9730911293037094e-05,
+      "loss": 1.2956,
+      "step": 2799
+    },
+    {
+      "epoch": 0.10182929046805106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973063982194009e-05,
+      "loss": 1.2891,
+      "step": 2800
+    },
+    {
+      "epoch": 0.10186565807178966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973036821584427e-05,
+      "loss": 1.1932,
+      "step": 2801
+    },
+    {
+      "epoch": 0.10190202567552824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.973009647475339e-05,
+      "loss": 1.1945,
+      "step": 2802
+    },
+    {
+      "epoch": 0.10193839327926683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9729824598671233e-05,
+      "loss": 1.1797,
+      "step": 2803
+    },
+    {
+      "epoch": 0.10197476088300542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972955258760156e-05,
+      "loss": 1.192,
+      "step": 2804
+    },
+    {
+      "epoch": 0.10201112848674401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972928044154815e-05,
+      "loss": 1.1313,
+      "step": 2805
+    },
+    {
+      "epoch": 0.1020474960904826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9729008160514782e-05,
+      "loss": 1.1823,
+      "step": 2806
+    },
+    {
+      "epoch": 0.10208386369422119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9728735744505225e-05,
+      "loss": 1.1371,
+      "step": 2807
+    },
+    {
+      "epoch": 0.10212023129795977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9728463193523265e-05,
+      "loss": 1.1406,
+      "step": 2808
+    },
+    {
+      "epoch": 0.10215659890169837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9728190507572677e-05,
+      "loss": 1.0478,
+      "step": 2809
+    },
+    {
+      "epoch": 0.10219296650543695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9727917686657255e-05,
+      "loss": 1.1387,
+      "step": 2810
+    },
+    {
+      "epoch": 0.10222933410917555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972764473078077e-05,
+      "loss": 1.2641,
+      "step": 2811
+    },
+    {
+      "epoch": 0.10226570171291413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9727371639947023e-05,
+      "loss": 1.2313,
+      "step": 2812
+    },
+    {
+      "epoch": 0.10230206931665273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972709841415979e-05,
+      "loss": 1.3342,
+      "step": 2813
+    },
+    {
+      "epoch": 0.10233843692039131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972682505342287e-05,
+      "loss": 1.2888,
+      "step": 2814
+    },
+    {
+      "epoch": 0.1023748045241299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9726551557740053e-05,
+      "loss": 1.2408,
+      "step": 2815
+    },
+    {
+      "epoch": 0.10241117212786849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9726277927115137e-05,
+      "loss": 1.2131,
+      "step": 2816
+    },
+    {
+      "epoch": 0.10244753973160708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972600416155191e-05,
+      "loss": 1.1658,
+      "step": 2817
+    },
+    {
+      "epoch": 0.10248390733534568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9725730261054174e-05,
+      "loss": 1.1454,
+      "step": 2818
+    },
+    {
+      "epoch": 0.10252027493908426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972545622562573e-05,
+      "loss": 1.1978,
+      "step": 2819
+    },
+    {
+      "epoch": 0.10255664254282286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9725182055270383e-05,
+      "loss": 1.2023,
+      "step": 2820
+    },
+    {
+      "epoch": 0.10259301014656144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972490774999193e-05,
+      "loss": 1.3031,
+      "step": 2821
+    },
+    {
+      "epoch": 0.10262937775030004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972463330979418e-05,
+      "loss": 1.1621,
+      "step": 2822
+    },
+    {
+      "epoch": 0.10266574535403862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972435873468094e-05,
+      "loss": 1.1215,
+      "step": 2823
+    },
+    {
+      "epoch": 0.10270211295777722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972408402465602e-05,
+      "loss": 1.1708,
+      "step": 2824
+    },
+    {
+      "epoch": 0.1027384805615158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972380917972323e-05,
+      "loss": 1.2036,
+      "step": 2825
+    },
+    {
+      "epoch": 0.1027748481652544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9723534199886387e-05,
+      "loss": 1.1159,
+      "step": 2826
+    },
+    {
+      "epoch": 0.10281121576899298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97232590851493e-05,
+      "loss": 1.2036,
+      "step": 2827
+    },
+    {
+      "epoch": 0.10284758337273157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972298383551579e-05,
+      "loss": 1.1259,
+      "step": 2828
+    },
+    {
+      "epoch": 0.10288395097647016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9722708450989678e-05,
+      "loss": 1.2831,
+      "step": 2829
+    },
+    {
+      "epoch": 0.10292031858020875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9722432931574774e-05,
+      "loss": 1.1969,
+      "step": 2830
+    },
+    {
+      "epoch": 0.10295668618394734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9722157277274912e-05,
+      "loss": 1.267,
+      "step": 2831
+    },
+    {
+      "epoch": 0.10299305378768593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972188148809391e-05,
+      "loss": 1.1815,
+      "step": 2832
+    },
+    {
+      "epoch": 0.10302942139142451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9721605564035595e-05,
+      "loss": 1.0732,
+      "step": 2833
+    },
+    {
+      "epoch": 0.10306578899516311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9721329505103798e-05,
+      "loss": 1.1938,
+      "step": 2834
+    },
+    {
+      "epoch": 0.10310215659890169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9721053311302346e-05,
+      "loss": 1.1038,
+      "step": 2835
+    },
+    {
+      "epoch": 0.10313852420264029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9720776982635072e-05,
+      "loss": 1.1947,
+      "step": 2836
+    },
+    {
+      "epoch": 0.10317489180637887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9720500519105807e-05,
+      "loss": 1.1241,
+      "step": 2837
+    },
+    {
+      "epoch": 0.10321125941011747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972022392071839e-05,
+      "loss": 1.2564,
+      "step": 2838
+    },
+    {
+      "epoch": 0.10324762701385606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971994718747666e-05,
+      "loss": 1.2187,
+      "step": 2839
+    },
+    {
+      "epoch": 0.10328399461759465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971967031938445e-05,
+      "loss": 1.2762,
+      "step": 2840
+    },
+    {
+      "epoch": 0.10332036222133324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9719393316445607e-05,
+      "loss": 1.3185,
+      "step": 2841
+    },
+    {
+      "epoch": 0.10335672982507182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9719116178663975e-05,
+      "loss": 1.2312,
+      "step": 2842
+    },
+    {
+      "epoch": 0.10339309742881042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971883890604339e-05,
+      "loss": 1.2448,
+      "step": 2843
+    },
+    {
+      "epoch": 0.103429465032549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9718561498587707e-05,
+      "loss": 1.1671,
+      "step": 2844
+    },
+    {
+      "epoch": 0.1034658326362876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971828395630077e-05,
+      "loss": 1.1868,
+      "step": 2845
+    },
+    {
+      "epoch": 0.10350220024002618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9718006279186433e-05,
+      "loss": 1.2698,
+      "step": 2846
+    },
+    {
+      "epoch": 0.10353856784376478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9717728467248547e-05,
+      "loss": 1.1861,
+      "step": 2847
+    },
+    {
+      "epoch": 0.10357493544750336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9717450520490967e-05,
+      "loss": 1.2233,
+      "step": 2848
+    },
+    {
+      "epoch": 0.10361130305124196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9717172438917547e-05,
+      "loss": 1.1367,
+      "step": 2849
+    },
+    {
+      "epoch": 0.10364767065498054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9716894222532145e-05,
+      "loss": 1.0406,
+      "step": 2850
+    },
+    {
+      "epoch": 0.10368403825871914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9716615871338623e-05,
+      "loss": 1.178,
+      "step": 2851
+    },
+    {
+      "epoch": 0.10372040586245772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9716337385340845e-05,
+      "loss": 1.2311,
+      "step": 2852
+    },
+    {
+      "epoch": 0.10375677346619631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971605876454267e-05,
+      "loss": 1.1475,
+      "step": 2853
+    },
+    {
+      "epoch": 0.1037931410699349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971578000894796e-05,
+      "loss": 1.1853,
+      "step": 2854
+    },
+    {
+      "epoch": 0.10382950867367349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9715501118560595e-05,
+      "loss": 1.0782,
+      "step": 2855
+    },
+    {
+      "epoch": 0.10386587627741208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971522209338443e-05,
+      "loss": 1.3549,
+      "step": 2856
+    },
+    {
+      "epoch": 0.10390224388115067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9714942933423343e-05,
+      "loss": 1.1566,
+      "step": 2857
+    },
+    {
+      "epoch": 0.10393861148488925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971466363868121e-05,
+      "loss": 1.2599,
+      "step": 2858
+    },
+    {
+      "epoch": 0.10397497908862785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9714384209161903e-05,
+      "loss": 1.0763,
+      "step": 2859
+    },
+    {
+      "epoch": 0.10401134669236645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9714104644869296e-05,
+      "loss": 1.1256,
+      "step": 2860
+    },
+    {
+      "epoch": 0.10404771429610503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971382494580727e-05,
+      "loss": 1.1671,
+      "step": 2861
+    },
+    {
+      "epoch": 0.10408408189984363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9713545111979704e-05,
+      "loss": 1.217,
+      "step": 2862
+    },
+    {
+      "epoch": 0.10412044950358221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9713265143390485e-05,
+      "loss": 1.1465,
+      "step": 2863
+    },
+    {
+      "epoch": 0.1041568171073208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9712985040043487e-05,
+      "loss": 1.2579,
+      "step": 2864
+    },
+    {
+      "epoch": 0.10419318471105939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9712704801942605e-05,
+      "loss": 1.2159,
+      "step": 2865
+    },
+    {
+      "epoch": 0.10422955231479798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971242442909173e-05,
+      "loss": 1.1005,
+      "step": 2866
+    },
+    {
+      "epoch": 0.10426591991853656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9712143921494738e-05,
+      "loss": 1.1332,
+      "step": 2867
+    },
+    {
+      "epoch": 0.10430228752227516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9711863279155533e-05,
+      "loss": 1.2653,
+      "step": 2868
+    },
+    {
+      "epoch": 0.10433865512601374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9711582502078004e-05,
+      "loss": 1.1196,
+      "step": 2869
+    },
+    {
+      "epoch": 0.10437502272975234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9711301590266044e-05,
+      "loss": 1.1275,
+      "step": 2870
+    },
+    {
+      "epoch": 0.10441139033349092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9711020543723557e-05,
+      "loss": 1.1364,
+      "step": 2871
+    },
+    {
+      "epoch": 0.10444775793722952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9710739362454435e-05,
+      "loss": 1.2611,
+      "step": 2872
+    },
+    {
+      "epoch": 0.1044841255409681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9710458046462585e-05,
+      "loss": 1.2091,
+      "step": 2873
+    },
+    {
+      "epoch": 0.1045204931447067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9710176595751903e-05,
+      "loss": 1.2484,
+      "step": 2874
+    },
+    {
+      "epoch": 0.10455686074844528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97098950103263e-05,
+      "loss": 1.255,
+      "step": 2875
+    },
+    {
+      "epoch": 0.10459322835218388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9709613290189684e-05,
+      "loss": 1.2358,
+      "step": 2876
+    },
+    {
+      "epoch": 0.10462959595592246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9709331435345952e-05,
+      "loss": 1.1853,
+      "step": 2877
+    },
+    {
+      "epoch": 0.10466596355966105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970904944579903e-05,
+      "loss": 1.0973,
+      "step": 2878
+    },
+    {
+      "epoch": 0.10470233116339964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9708767321552816e-05,
+      "loss": 1.1965,
+      "step": 2879
+    },
+    {
+      "epoch": 0.10473869876713823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970848506261123e-05,
+      "loss": 1.1793,
+      "step": 2880
+    },
+    {
+      "epoch": 0.10477506637087683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9708202668978193e-05,
+      "loss": 1.2667,
+      "step": 2881
+    },
+    {
+      "epoch": 0.10481143397461541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9707920140657616e-05,
+      "loss": 1.2338,
+      "step": 2882
+    },
+    {
+      "epoch": 0.10484780157835401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970763747765342e-05,
+      "loss": 1.0826,
+      "step": 2883
+    },
+    {
+      "epoch": 0.10488416918209259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970735467996953e-05,
+      "loss": 1.1154,
+      "step": 2884
+    },
+    {
+      "epoch": 0.10492053678583119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9707071747609864e-05,
+      "loss": 1.306,
+      "step": 2885
+    },
+    {
+      "epoch": 0.10495690438956977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9706788680578352e-05,
+      "loss": 1.1797,
+      "step": 2886
+    },
+    {
+      "epoch": 0.10499327199330837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9706505478878915e-05,
+      "loss": 1.2127,
+      "step": 2887
+    },
+    {
+      "epoch": 0.10502963959704695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9706222142515494e-05,
+      "loss": 1.1363,
+      "step": 2888
+    },
+    {
+      "epoch": 0.10506600720078554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9705938671492007e-05,
+      "loss": 1.2747,
+      "step": 2889
+    },
+    {
+      "epoch": 0.10510237480452413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970565506581239e-05,
+      "loss": 1.1428,
+      "step": 2890
+    },
+    {
+      "epoch": 0.10513874240826272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9705371325480582e-05,
+      "loss": 1.307,
+      "step": 2891
+    },
+    {
+      "epoch": 0.1051751100120013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9705087450500517e-05,
+      "loss": 1.1458,
+      "step": 2892
+    },
+    {
+      "epoch": 0.1052114776157399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9704803440876136e-05,
+      "loss": 1.1722,
+      "step": 2893
+    },
+    {
+      "epoch": 0.10524784521947848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9704519296611374e-05,
+      "loss": 1.1432,
+      "step": 2894
+    },
+    {
+      "epoch": 0.10528421282321708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9704235017710174e-05,
+      "loss": 1.1979,
+      "step": 2895
+    },
+    {
+      "epoch": 0.10532058042695566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970395060417648e-05,
+      "loss": 1.0851,
+      "step": 2896
+    },
+    {
+      "epoch": 0.10535694803069426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9703666056014244e-05,
+      "loss": 1.2267,
+      "step": 2897
+    },
+    {
+      "epoch": 0.10539331563443284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970338137322741e-05,
+      "loss": 1.1413,
+      "step": 2898
+    },
+    {
+      "epoch": 0.10542968323817144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9703096555819923e-05,
+      "loss": 1.2092,
+      "step": 2899
+    },
+    {
+      "epoch": 0.10546605084191003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970281160379574e-05,
+      "loss": 1.0388,
+      "step": 2900
+    },
+    {
+      "epoch": 0.10550241844564862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9702526517158813e-05,
+      "loss": 1.2274,
+      "step": 2901
+    },
+    {
+      "epoch": 0.10553878604938721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9702241295913093e-05,
+      "loss": 1.1541,
+      "step": 2902
+    },
+    {
+      "epoch": 0.1055751536531258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9701955940062544e-05,
+      "loss": 1.0094,
+      "step": 2903
+    },
+    {
+      "epoch": 0.10561152125686439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9701670449611122e-05,
+      "loss": 1.1579,
+      "step": 2904
+    },
+    {
+      "epoch": 0.10564788886060297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9701384824562786e-05,
+      "loss": 1.2501,
+      "step": 2905
+    },
+    {
+      "epoch": 0.10568425646434157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97010990649215e-05,
+      "loss": 1.1207,
+      "step": 2906
+    },
+    {
+      "epoch": 0.10572062406808015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9700813170691232e-05,
+      "loss": 1.1706,
+      "step": 2907
+    },
+    {
+      "epoch": 0.10575699167181875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.970052714187594e-05,
+      "loss": 1.1904,
+      "step": 2908
+    },
+    {
+      "epoch": 0.10579335927555733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97002409784796e-05,
+      "loss": 1.1806,
+      "step": 2909
+    },
+    {
+      "epoch": 0.10582972687929593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969995468050618e-05,
+      "loss": 0.9736,
+      "step": 2910
+    },
+    {
+      "epoch": 0.10586609448303451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969966824795965e-05,
+      "loss": 1.186,
+      "step": 2911
+    },
+    {
+      "epoch": 0.1059024620867731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9699381680843983e-05,
+      "loss": 1.2211,
+      "step": 2912
+    },
+    {
+      "epoch": 0.10593882969051169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9699094979163158e-05,
+      "loss": 1.1825,
+      "step": 2913
+    },
+    {
+      "epoch": 0.10597519729425028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969880814292115e-05,
+      "loss": 1.1284,
+      "step": 2914
+    },
+    {
+      "epoch": 0.10601156489798887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9698521172121942e-05,
+      "loss": 1.0893,
+      "step": 2915
+    },
+    {
+      "epoch": 0.10604793250172746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9698234066769514e-05,
+      "loss": 1.2401,
+      "step": 2916
+    },
+    {
+      "epoch": 0.10608430010546605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969794682686785e-05,
+      "loss": 1.1321,
+      "step": 2917
+    },
+    {
+      "epoch": 0.10612066770920464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9697659452420928e-05,
+      "loss": 1.2074,
+      "step": 2918
+    },
+    {
+      "epoch": 0.10615703531294322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9697371943432743e-05,
+      "loss": 1.065,
+      "step": 2919
+    },
+    {
+      "epoch": 0.10619340291668182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969708429990728e-05,
+      "loss": 1.0574,
+      "step": 2920
+    },
+    {
+      "epoch": 0.10622977052042042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9696796521848532e-05,
+      "loss": 1.2262,
+      "step": 2921
+    },
+    {
+      "epoch": 0.106266138124159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969650860926049e-05,
+      "loss": 1.2394,
+      "step": 2922
+    },
+    {
+      "epoch": 0.1063025057278976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9696220562147152e-05,
+      "loss": 1.3188,
+      "step": 2923
+    },
+    {
+      "epoch": 0.10633887333163618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9695932380512502e-05,
+      "loss": 1.1067,
+      "step": 2924
+    },
+    {
+      "epoch": 0.10637524093537477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9695644064360554e-05,
+      "loss": 1.2115,
+      "step": 2925
+    },
+    {
+      "epoch": 0.10641160853911336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96953556136953e-05,
+      "loss": 1.2382,
+      "step": 2926
+    },
+    {
+      "epoch": 0.10644797614285195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969506702852074e-05,
+      "loss": 1.1614,
+      "step": 2927
+    },
+    {
+      "epoch": 0.10648434374659053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9694778308840884e-05,
+      "loss": 1.1628,
+      "step": 2928
+    },
+    {
+      "epoch": 0.10652071135032913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969448945465973e-05,
+      "loss": 1.0963,
+      "step": 2929
+    },
+    {
+      "epoch": 0.10655707895406771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9694200465981293e-05,
+      "loss": 1.2178,
+      "step": 2930
+    },
+    {
+      "epoch": 0.10659344655780631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969391134280958e-05,
+      "loss": 1.1986,
+      "step": 2931
+    },
+    {
+      "epoch": 0.10662981416154489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9693622085148594e-05,
+      "loss": 1.2532,
+      "step": 2932
+    },
+    {
+      "epoch": 0.10666618176528349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969333269300236e-05,
+      "loss": 1.1127,
+      "step": 2933
+    },
+    {
+      "epoch": 0.10670254936902207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969304316637489e-05,
+      "loss": 1.1811,
+      "step": 2934
+    },
+    {
+      "epoch": 0.10673891697276067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9692753505270197e-05,
+      "loss": 1.2137,
+      "step": 2935
+    },
+    {
+      "epoch": 0.10677528457649925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96924637096923e-05,
+      "loss": 1.2374,
+      "step": 2936
+    },
+    {
+      "epoch": 0.10681165218023785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9692173779645222e-05,
+      "loss": 1.153,
+      "step": 2937
+    },
+    {
+      "epoch": 0.10684801978397643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9691883715132984e-05,
+      "loss": 1.2881,
+      "step": 2938
+    },
+    {
+      "epoch": 0.10688438738771502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969159351615961e-05,
+      "loss": 1.1212,
+      "step": 2939
+    },
+    {
+      "epoch": 0.1069207549914536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9691303182729127e-05,
+      "loss": 1.1758,
+      "step": 2940
+    },
+    {
+      "epoch": 0.1069571225951922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969101271484556e-05,
+      "loss": 1.0481,
+      "step": 2941
+    },
+    {
+      "epoch": 0.1069934901989308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9690722112512946e-05,
+      "loss": 1.2606,
+      "step": 2942
+    },
+    {
+      "epoch": 0.10702985780266938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969043137573531e-05,
+      "loss": 1.1736,
+      "step": 2943
+    },
+    {
+      "epoch": 0.10706622540640798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9690140504516685e-05,
+      "loss": 1.1469,
+      "step": 2944
+    },
+    {
+      "epoch": 0.10710259301014656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968984949886111e-05,
+      "loss": 1.2215,
+      "step": 2945
+    },
+    {
+      "epoch": 0.10713896061388516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9689558358772623e-05,
+      "loss": 1.3114,
+      "step": 2946
+    },
+    {
+      "epoch": 0.10717532821762374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9689267084255263e-05,
+      "loss": 1.2952,
+      "step": 2947
+    },
+    {
+      "epoch": 0.10721169582136234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9688975675313066e-05,
+      "loss": 1.2222,
+      "step": 2948
+    },
+    {
+      "epoch": 0.10724806342510092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968868413195008e-05,
+      "loss": 1.2336,
+      "step": 2949
+    },
+    {
+      "epoch": 0.10728443102883951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968839245417035e-05,
+      "loss": 1.1546,
+      "step": 2950
+    },
+    {
+      "epoch": 0.1073207986325781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9688100641977918e-05,
+      "loss": 1.2513,
+      "step": 2951
+    },
+    {
+      "epoch": 0.10735716623631669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9687808695376834e-05,
+      "loss": 1.1295,
+      "step": 2952
+    },
+    {
+      "epoch": 0.10739353384005527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968751661437115e-05,
+      "loss": 1.318,
+      "step": 2953
+    },
+    {
+      "epoch": 0.10742990144379387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968722439896492e-05,
+      "loss": 1.1162,
+      "step": 2954
+    },
+    {
+      "epoch": 0.10746626904753245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9686932049162196e-05,
+      "loss": 1.2083,
+      "step": 2955
+    },
+    {
+      "epoch": 0.10750263665127105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968663956496703e-05,
+      "loss": 1.1933,
+      "step": 2956
+    },
+    {
+      "epoch": 0.10753900425500963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968634694638349e-05,
+      "loss": 1.128,
+      "step": 2957
+    },
+    {
+      "epoch": 0.10757537185874823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9686054193415625e-05,
+      "loss": 1.0948,
+      "step": 2958
+    },
+    {
+      "epoch": 0.10761173946248681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9685761306067502e-05,
+      "loss": 1.2562,
+      "step": 2959
+    },
+    {
+      "epoch": 0.1076481070662254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9685468284343183e-05,
+      "loss": 1.1333,
+      "step": 2960
+    },
+    {
+      "epoch": 0.10768447466996399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9685175128246732e-05,
+      "loss": 1.2228,
+      "step": 2961
+    },
+    {
+      "epoch": 0.10772084227370259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968488183778222e-05,
+      "loss": 1.2524,
+      "step": 2962
+    },
+    {
+      "epoch": 0.10775720987744118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968458841295371e-05,
+      "loss": 1.2493,
+      "step": 2963
+    },
+    {
+      "epoch": 0.10779357748117976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9684294853765278e-05,
+      "loss": 1.1711,
+      "step": 2964
+    },
+    {
+      "epoch": 0.10782994508491836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9684001160220994e-05,
+      "loss": 1.19,
+      "step": 2965
+    },
+    {
+      "epoch": 0.10786631268865694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968370733232494e-05,
+      "loss": 1.1364,
+      "step": 2966
+    },
+    {
+      "epoch": 0.10790268029239554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968341337008118e-05,
+      "loss": 1.2651,
+      "step": 2967
+    },
+    {
+      "epoch": 0.10793904789613412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96831192734938e-05,
+      "loss": 1.0731,
+      "step": 2968
+    },
+    {
+      "epoch": 0.10797541549987272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9682825042566874e-05,
+      "loss": 1.2142,
+      "step": 2969
+    },
+    {
+      "epoch": 0.1080117831036113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9682530677304498e-05,
+      "loss": 1.2539,
+      "step": 2970
+    },
+    {
+      "epoch": 0.1080481507073499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968223617771074e-05,
+      "loss": 1.2151,
+      "step": 2971
+    },
+    {
+      "epoch": 0.10808451831108848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9681941543789695e-05,
+      "loss": 1.193,
+      "step": 2972
+    },
+    {
+      "epoch": 0.10812088591482708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9681646775545444e-05,
+      "loss": 1.1646,
+      "step": 2973
+    },
+    {
+      "epoch": 0.10815725351856566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9681351872982083e-05,
+      "loss": 1.2263,
+      "step": 2974
+    },
+    {
+      "epoch": 0.10819362112230425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96810568361037e-05,
+      "loss": 1.1812,
+      "step": 2975
+    },
+    {
+      "epoch": 0.10822998872604284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9680761664914395e-05,
+      "loss": 1.1173,
+      "step": 2976
+    },
+    {
+      "epoch": 0.10826635632978143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968046635941825e-05,
+      "loss": 1.3007,
+      "step": 2977
+    },
+    {
+      "epoch": 0.10830272393352001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968017091961937e-05,
+      "loss": 1.239,
+      "step": 2978
+    },
+    {
+      "epoch": 0.10833909153725861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967987534552186e-05,
+      "loss": 1.1682,
+      "step": 2979
+    },
+    {
+      "epoch": 0.1083754591409972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9679579637129804e-05,
+      "loss": 1.0706,
+      "step": 2980
+    },
+    {
+      "epoch": 0.10841182674473579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967928379444732e-05,
+      "loss": 1.1953,
+      "step": 2981
+    },
+    {
+      "epoch": 0.10844819434847437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9678987817478504e-05,
+      "loss": 1.115,
+      "step": 2982
+    },
+    {
+      "epoch": 0.10848456195221297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9678691706227463e-05,
+      "loss": 1.2601,
+      "step": 2983
+    },
+    {
+      "epoch": 0.10852092955595156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967839546069831e-05,
+      "loss": 1.1725,
+      "step": 2984
+    },
+    {
+      "epoch": 0.10855729715969015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967809908089515e-05,
+      "loss": 1.2796,
+      "step": 2985
+    },
+    {
+      "epoch": 0.10859366476342874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9677802566822097e-05,
+      "loss": 1.2384,
+      "step": 2986
+    },
+    {
+      "epoch": 0.10863003236716733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9677505918483266e-05,
+      "loss": 1.116,
+      "step": 2987
+    },
+    {
+      "epoch": 0.10866639997090592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967720913588277e-05,
+      "loss": 1.1421,
+      "step": 2988
+    },
+    {
+      "epoch": 0.1087027675746445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967691221902473e-05,
+      "loss": 1.1946,
+      "step": 2989
+    },
+    {
+      "epoch": 0.1087391351783831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967661516791326e-05,
+      "loss": 1.1539,
+      "step": 2990
+    },
+    {
+      "epoch": 0.10877550278212168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9676317982552483e-05,
+      "loss": 1.1167,
+      "step": 2991
+    },
+    {
+      "epoch": 0.10881187038586028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967602066294652e-05,
+      "loss": 1.2758,
+      "step": 2992
+    },
+    {
+      "epoch": 0.10884823798959886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9675723209099504e-05,
+      "loss": 1.0532,
+      "step": 2993
+    },
+    {
+      "epoch": 0.10888460559333746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9675425621015554e-05,
+      "loss": 1.1814,
+      "step": 2994
+    },
+    {
+      "epoch": 0.10892097319707604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9675127898698804e-05,
+      "loss": 1.1608,
+      "step": 2995
+    },
+    {
+      "epoch": 0.10895734080081464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967483004215338e-05,
+      "loss": 1.1814,
+      "step": 2996
+    },
+    {
+      "epoch": 0.10899370840455322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9674532051383414e-05,
+      "loss": 1.2123,
+      "step": 2997
+    },
+    {
+      "epoch": 0.10903007600829182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9674233926393045e-05,
+      "loss": 1.1447,
+      "step": 2998
+    },
+    {
+      "epoch": 0.1090664436120304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9673935667186406e-05,
+      "loss": 1.3033,
+      "step": 2999
+    },
+    {
+      "epoch": 0.109102811215769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967363727376763e-05,
+      "loss": 1.2116,
+      "step": 3000
+    },
+    {
+      "epoch": 0.10913917881950758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9673338746140868e-05,
+      "loss": 1.1724,
+      "step": 3001
+    },
+    {
+      "epoch": 0.10917554642324617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9673040084310253e-05,
+      "loss": 1.2076,
+      "step": 3002
+    },
+    {
+      "epoch": 0.10921191402698477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9672741288279933e-05,
+      "loss": 1.2705,
+      "step": 3003
+    },
+    {
+      "epoch": 0.10924828163072335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967244235805405e-05,
+      "loss": 1.2551,
+      "step": 3004
+    },
+    {
+      "epoch": 0.10928464923446195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9672143293636753e-05,
+      "loss": 1.3766,
+      "step": 3005
+    },
+    {
+      "epoch": 0.10932101683820053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967184409503219e-05,
+      "loss": 1.1533,
+      "step": 3006
+    },
+    {
+      "epoch": 0.10935738444193913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9671544762244516e-05,
+      "loss": 1.1151,
+      "step": 3007
+    },
+    {
+      "epoch": 0.10939375204567771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9671245295277876e-05,
+      "loss": 1.1847,
+      "step": 3008
+    },
+    {
+      "epoch": 0.1094301196494163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967094569413643e-05,
+      "loss": 1.1669,
+      "step": 3009
+    },
+    {
+      "epoch": 0.10946648725315489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9670645958824336e-05,
+      "loss": 1.1987,
+      "step": 3010
+    },
+    {
+      "epoch": 0.10950285485689348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.967034608934575e-05,
+      "loss": 1.1258,
+      "step": 3011
+    },
+    {
+      "epoch": 0.10953922246063207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9670046085704832e-05,
+      "loss": 1.1552,
+      "step": 3012
+    },
+    {
+      "epoch": 0.10957559006437066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9669745947905744e-05,
+      "loss": 1.2318,
+      "step": 3013
+    },
+    {
+      "epoch": 0.10961195766810924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966944567595265e-05,
+      "loss": 0.994,
+      "step": 3014
+    },
+    {
+      "epoch": 0.10964832527184784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966914526984972e-05,
+      "loss": 1.1691,
+      "step": 3015
+    },
+    {
+      "epoch": 0.10968469287558642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966884472960111e-05,
+      "loss": 1.1895,
+      "step": 3016
+    },
+    {
+      "epoch": 0.10972106047932502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9668544055211004e-05,
+      "loss": 1.2801,
+      "step": 3017
+    },
+    {
+      "epoch": 0.1097574280830636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9668243246683567e-05,
+      "loss": 1.084,
+      "step": 3018
+    },
+    {
+      "epoch": 0.1097937956868022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966794230402297e-05,
+      "loss": 1.1138,
+      "step": 3019
+    },
+    {
+      "epoch": 0.10983016329054078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9667641227233393e-05,
+      "loss": 1.09,
+      "step": 3020
+    },
+    {
+      "epoch": 0.10986653089427938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966734001631901e-05,
+      "loss": 1.2925,
+      "step": 3021
+    },
+    {
+      "epoch": 0.10990289849801796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9667038671284e-05,
+      "loss": 1.2748,
+      "step": 3022
+    },
+    {
+      "epoch": 0.10993926610175656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9666737192132545e-05,
+      "loss": 1.2055,
+      "step": 3023
+    },
+    {
+      "epoch": 0.10997563370549515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9666435578868826e-05,
+      "loss": 1.1521,
+      "step": 3024
+    },
+    {
+      "epoch": 0.11001200130923373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9666133831497026e-05,
+      "loss": 1.1693,
+      "step": 3025
+    },
+    {
+      "epoch": 0.11004836891297233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9665831950021337e-05,
+      "loss": 1.261,
+      "step": 3026
+    },
+    {
+      "epoch": 0.11008473651671091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966552993444594e-05,
+      "loss": 1.2241,
+      "step": 3027
+    },
+    {
+      "epoch": 0.11012110412044951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9665227784775033e-05,
+      "loss": 1.1777,
+      "step": 3028
+    },
+    {
+      "epoch": 0.11015747172418809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9664925501012803e-05,
+      "loss": 1.0398,
+      "step": 3029
+    },
+    {
+      "epoch": 0.11019383932792669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966462308316344e-05,
+      "loss": 1.1823,
+      "step": 3030
+    },
+    {
+      "epoch": 0.11023020693166527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966432053123115e-05,
+      "loss": 1.1454,
+      "step": 3031
+    },
+    {
+      "epoch": 0.11026657453540387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966401784522012e-05,
+      "loss": 1.1337,
+      "step": 3032
+    },
+    {
+      "epoch": 0.11030294213914245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9663715025134555e-05,
+      "loss": 1.2197,
+      "step": 3033
+    },
+    {
+      "epoch": 0.11033930974288104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9663412070978656e-05,
+      "loss": 1.2458,
+      "step": 3034
+    },
+    {
+      "epoch": 0.11037567734661963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9663108982756624e-05,
+      "loss": 1.1499,
+      "step": 3035
+    },
+    {
+      "epoch": 0.11041204495035822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9662805760472666e-05,
+      "loss": 1.1609,
+      "step": 3036
+    },
+    {
+      "epoch": 0.1104484125540968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966250240413099e-05,
+      "loss": 1.2204,
+      "step": 3037
+    },
+    {
+      "epoch": 0.1104847801578354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9662198913735797e-05,
+      "loss": 1.174,
+      "step": 3038
+    },
+    {
+      "epoch": 0.11052114776157398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9661895289291305e-05,
+      "loss": 1.1547,
+      "step": 3039
+    },
+    {
+      "epoch": 0.11055751536531258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9661591530801728e-05,
+      "loss": 1.1558,
+      "step": 3040
+    },
+    {
+      "epoch": 0.11059388296905116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9661287638271275e-05,
+      "loss": 1.1245,
+      "step": 3041
+    },
+    {
+      "epoch": 0.11063025057278976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9660983611704162e-05,
+      "loss": 1.2106,
+      "step": 3042
+    },
+    {
+      "epoch": 0.11066661817652834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9660679451104607e-05,
+      "loss": 1.0816,
+      "step": 3043
+    },
+    {
+      "epoch": 0.11070298578026694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9660375156476835e-05,
+      "loss": 1.1415,
+      "step": 3044
+    },
+    {
+      "epoch": 0.11073935338400553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966007072782506e-05,
+      "loss": 1.1809,
+      "step": 3045
+    },
+    {
+      "epoch": 0.11077572098774412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965976616515351e-05,
+      "loss": 1.0323,
+      "step": 3046
+    },
+    {
+      "epoch": 0.11081208859148271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9659461468466415e-05,
+      "loss": 1.1065,
+      "step": 3047
+    },
+    {
+      "epoch": 0.1108484561952213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9659156637767992e-05,
+      "loss": 1.1414,
+      "step": 3048
+    },
+    {
+      "epoch": 0.11088482379895989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9658851673062475e-05,
+      "loss": 1.1275,
+      "step": 3049
+    },
+    {
+      "epoch": 0.11092119140269847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96585465743541e-05,
+      "loss": 1.2146,
+      "step": 3050
+    },
+    {
+      "epoch": 0.11095755900643707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965824134164709e-05,
+      "loss": 1.2805,
+      "step": 3051
+    },
+    {
+      "epoch": 0.11099392661017565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965793597494569e-05,
+      "loss": 1.2646,
+      "step": 3052
+    },
+    {
+      "epoch": 0.11103029421391425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9657630474254125e-05,
+      "loss": 1.1406,
+      "step": 3053
+    },
+    {
+      "epoch": 0.11106666181765283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9657324839576643e-05,
+      "loss": 1.1376,
+      "step": 3054
+    },
+    {
+      "epoch": 0.11110302942139143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9657019070917478e-05,
+      "loss": 1.1342,
+      "step": 3055
+    },
+    {
+      "epoch": 0.11113939702513001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965671316828088e-05,
+      "loss": 1.142,
+      "step": 3056
+    },
+    {
+      "epoch": 0.1111757646288686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965640713167108e-05,
+      "loss": 1.1491,
+      "step": 3057
+    },
+    {
+      "epoch": 0.11121213223260719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965610096109234e-05,
+      "loss": 1.0789,
+      "step": 3058
+    },
+    {
+      "epoch": 0.11124849983634578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9655794656548894e-05,
+      "loss": 1.1097,
+      "step": 3059
+    },
+    {
+      "epoch": 0.11128486744008437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9655488218045e-05,
+      "loss": 1.0938,
+      "step": 3060
+    },
+    {
+      "epoch": 0.11132123504382296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9655181645584903e-05,
+      "loss": 1.1035,
+      "step": 3061
+    },
+    {
+      "epoch": 0.11135760264756155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9654874939172862e-05,
+      "loss": 1.2692,
+      "step": 3062
+    },
+    {
+      "epoch": 0.11139397025130014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9654568098813128e-05,
+      "loss": 1.2215,
+      "step": 3063
+    },
+    {
+      "epoch": 0.11143033785503872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9654261124509966e-05,
+      "loss": 1.2214,
+      "step": 3064
+    },
+    {
+      "epoch": 0.11146670545877732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9653954016267623e-05,
+      "loss": 1.1077,
+      "step": 3065
+    },
+    {
+      "epoch": 0.11150307306251592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9653646774090366e-05,
+      "loss": 1.1078,
+      "step": 3066
+    },
+    {
+      "epoch": 0.1115394406662545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9653339397982452e-05,
+      "loss": 1.176,
+      "step": 3067
+    },
+    {
+      "epoch": 0.1115758082699931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9653031887948157e-05,
+      "loss": 1.1933,
+      "step": 3068
+    },
+    {
+      "epoch": 0.11161217587373168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9652724243991738e-05,
+      "loss": 1.1235,
+      "step": 3069
+    },
+    {
+      "epoch": 0.11164854347747027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9652416466117466e-05,
+      "loss": 1.1577,
+      "step": 3070
+    },
+    {
+      "epoch": 0.11168491108120886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965210855432961e-05,
+      "loss": 1.1526,
+      "step": 3071
+    },
+    {
+      "epoch": 0.11172127868494745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9651800508632443e-05,
+      "loss": 1.0648,
+      "step": 3072
+    },
+    {
+      "epoch": 0.11175764628868604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9651492329030238e-05,
+      "loss": 1.0347,
+      "step": 3073
+    },
+    {
+      "epoch": 0.11179401389242463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965118401552727e-05,
+      "loss": 1.2786,
+      "step": 3074
+    },
+    {
+      "epoch": 0.11183038149616321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9650875568127817e-05,
+      "loss": 1.1216,
+      "step": 3075
+    },
+    {
+      "epoch": 0.11186674909990181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965056698683616e-05,
+      "loss": 1.0548,
+      "step": 3076
+    },
+    {
+      "epoch": 0.11190311670364039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9650258271656577e-05,
+      "loss": 1.2445,
+      "step": 3077
+    },
+    {
+      "epoch": 0.11193948430737899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964994942259335e-05,
+      "loss": 1.2216,
+      "step": 3078
+    },
+    {
+      "epoch": 0.11197585191111757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964964043965077e-05,
+      "loss": 1.1558,
+      "step": 3079
+    },
+    {
+      "epoch": 0.11201221951485617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964933132283312e-05,
+      "loss": 1.2637,
+      "step": 3080
+    },
+    {
+      "epoch": 0.11204858711859475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9649022072144684e-05,
+      "loss": 1.1247,
+      "step": 3081
+    },
+    {
+      "epoch": 0.11208495472233335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9648712687589762e-05,
+      "loss": 1.1987,
+      "step": 3082
+    },
+    {
+      "epoch": 0.11212132232607193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9648403169172637e-05,
+      "loss": 1.1657,
+      "step": 3083
+    },
+    {
+      "epoch": 0.11215768992981053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964809351689761e-05,
+      "loss": 1.1321,
+      "step": 3084
+    },
+    {
+      "epoch": 0.11219405753354912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9647783730768973e-05,
+      "loss": 1.1992,
+      "step": 3085
+    },
+    {
+      "epoch": 0.1122304251372877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9647473810791027e-05,
+      "loss": 1.2371,
+      "step": 3086
+    },
+    {
+      "epoch": 0.1122667927410263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964716375696807e-05,
+      "loss": 1.2397,
+      "step": 3087
+    },
+    {
+      "epoch": 0.11230316034476488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96468535693044e-05,
+      "loss": 1.1661,
+      "step": 3088
+    },
+    {
+      "epoch": 0.11233952794850348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9646543247804327e-05,
+      "loss": 1.0933,
+      "step": 3089
+    },
+    {
+      "epoch": 0.11237589555224206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9646232792472153e-05,
+      "loss": 1.2813,
+      "step": 3090
+    },
+    {
+      "epoch": 0.11241226315598066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9645922203312185e-05,
+      "loss": 0.9412,
+      "step": 3091
+    },
+    {
+      "epoch": 0.11244863075971924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964561148032873e-05,
+      "loss": 1.1535,
+      "step": 3092
+    },
+    {
+      "epoch": 0.11248499836345784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9645300623526102e-05,
+      "loss": 1.0605,
+      "step": 3093
+    },
+    {
+      "epoch": 0.11252136596719642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9644989632908614e-05,
+      "loss": 1.1952,
+      "step": 3094
+    },
+    {
+      "epoch": 0.11255773357093501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9644678508480578e-05,
+      "loss": 1.1029,
+      "step": 3095
+    },
+    {
+      "epoch": 0.1125941011746736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9644367250246314e-05,
+      "loss": 1.1119,
+      "step": 3096
+    },
+    {
+      "epoch": 0.1126304687784122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9644055858210137e-05,
+      "loss": 1.1619,
+      "step": 3097
+    },
+    {
+      "epoch": 0.11266683638215078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9643744332376365e-05,
+      "loss": 1.1992,
+      "step": 3098
+    },
+    {
+      "epoch": 0.11270320398588937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9643432672749325e-05,
+      "loss": 1.129,
+      "step": 3099
+    },
+    {
+      "epoch": 0.11273957158962795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9643120879333342e-05,
+      "loss": 1.1919,
+      "step": 3100
+    },
+    {
+      "epoch": 0.11277593919336655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9642808952132734e-05,
+      "loss": 1.1298,
+      "step": 3101
+    },
+    {
+      "epoch": 0.11281230679710513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9642496891151833e-05,
+      "loss": 1.2269,
+      "step": 3102
+    },
+    {
+      "epoch": 0.11284867440084373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9642184696394974e-05,
+      "loss": 1.1423,
+      "step": 3103
+    },
+    {
+      "epoch": 0.11288504200458231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964187236786648e-05,
+      "loss": 1.2106,
+      "step": 3104
+    },
+    {
+      "epoch": 0.11292140960832091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9641559905570682e-05,
+      "loss": 1.2384,
+      "step": 3105
+    },
+    {
+      "epoch": 0.1129577772120595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9641247309511927e-05,
+      "loss": 1.098,
+      "step": 3106
+    },
+    {
+      "epoch": 0.11299414481579809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9640934579694543e-05,
+      "loss": 1.118,
+      "step": 3107
+    },
+    {
+      "epoch": 0.11303051241953668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964062171612287e-05,
+      "loss": 1.1119,
+      "step": 3108
+    },
+    {
+      "epoch": 0.11306688002327527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.964030871880125e-05,
+      "loss": 1.2308,
+      "step": 3109
+    },
+    {
+      "epoch": 0.11310324762701386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963999558773402e-05,
+      "loss": 1.2268,
+      "step": 3110
+    },
+    {
+      "epoch": 0.11313961523075244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963968232292553e-05,
+      "loss": 1.2311,
+      "step": 3111
+    },
+    {
+      "epoch": 0.11317598283449104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9639368924380126e-05,
+      "loss": 1.2435,
+      "step": 3112
+    },
+    {
+      "epoch": 0.11321235043822962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9639055392102156e-05,
+      "loss": 1.1377,
+      "step": 3113
+    },
+    {
+      "epoch": 0.11324871804196822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9638741726095966e-05,
+      "loss": 1.2635,
+      "step": 3114
+    },
+    {
+      "epoch": 0.1132850856457068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9638427926365913e-05,
+      "loss": 1.0573,
+      "step": 3115
+    },
+    {
+      "epoch": 0.1133214532494454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9638113992916344e-05,
+      "loss": 1.1908,
+      "step": 3116
+    },
+    {
+      "epoch": 0.11335782085318398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963779992575162e-05,
+      "loss": 1.0349,
+      "step": 3117
+    },
+    {
+      "epoch": 0.11339418845692258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9637485724876097e-05,
+      "loss": 1.1077,
+      "step": 3118
+    },
+    {
+      "epoch": 0.11343055606066116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9637171390294133e-05,
+      "loss": 1.1081,
+      "step": 3119
+    },
+    {
+      "epoch": 0.11346692366439975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963685692201009e-05,
+      "loss": 1.2054,
+      "step": 3120
+    },
+    {
+      "epoch": 0.11350329126813834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963654232002833e-05,
+      "loss": 1.1972,
+      "step": 3121
+    },
+    {
+      "epoch": 0.11353965887187693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9636227584353215e-05,
+      "loss": 1.1695,
+      "step": 3122
+    },
+    {
+      "epoch": 0.11357602647561552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9635912714989116e-05,
+      "loss": 1.2372,
+      "step": 3123
+    },
+    {
+      "epoch": 0.11361239407935411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96355977119404e-05,
+      "loss": 1.101,
+      "step": 3124
+    },
+    {
+      "epoch": 0.1136487616830927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9635282575211434e-05,
+      "loss": 1.1511,
+      "step": 3125
+    },
+    {
+      "epoch": 0.11368512928683129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96349673048066e-05,
+      "loss": 1.3268,
+      "step": 3126
+    },
+    {
+      "epoch": 0.11372149689056989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9634651900730258e-05,
+      "loss": 1.2449,
+      "step": 3127
+    },
+    {
+      "epoch": 0.11375786449430847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9634336362986792e-05,
+      "loss": 1.059,
+      "step": 3128
+    },
+    {
+      "epoch": 0.11379423209804707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963402069158058e-05,
+      "loss": 1.1848,
+      "step": 3129
+    },
+    {
+      "epoch": 0.11383059970178565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9633704886516e-05,
+      "loss": 1.2197,
+      "step": 3130
+    },
+    {
+      "epoch": 0.11386696730552424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963338894779743e-05,
+      "loss": 1.0615,
+      "step": 3131
+    },
+    {
+      "epoch": 0.11390333490926283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963307287542926e-05,
+      "loss": 1.1779,
+      "step": 3132
+    },
+    {
+      "epoch": 0.11393970251300142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963275666941587e-05,
+      "loss": 1.1697,
+      "step": 3133
+    },
+    {
+      "epoch": 0.11397607011674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9632440329761647e-05,
+      "loss": 1.1724,
+      "step": 3134
+    },
+    {
+      "epoch": 0.1140124377204786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9632123856470982e-05,
+      "loss": 1.1671,
+      "step": 3135
+    },
+    {
+      "epoch": 0.11404880532421718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9631807249548266e-05,
+      "loss": 1.2489,
+      "step": 3136
+    },
+    {
+      "epoch": 0.11408517292795578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963149050899789e-05,
+      "loss": 1.2508,
+      "step": 3137
+    },
+    {
+      "epoch": 0.11412154053169436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9631173634824247e-05,
+      "loss": 1.1289,
+      "step": 3138
+    },
+    {
+      "epoch": 0.11415790813543296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9630856627031736e-05,
+      "loss": 1.2129,
+      "step": 3139
+    },
+    {
+      "epoch": 0.11419427573917154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9630539485624752e-05,
+      "loss": 1.3215,
+      "step": 3140
+    },
+    {
+      "epoch": 0.11423064334291014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96302222106077e-05,
+      "loss": 1.2937,
+      "step": 3141
+    },
+    {
+      "epoch": 0.11426701094664872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9629904801984975e-05,
+      "loss": 1.0959,
+      "step": 3142
+    },
+    {
+      "epoch": 0.11430337855038732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9629587259760983e-05,
+      "loss": 1.2625,
+      "step": 3143
+    },
+    {
+      "epoch": 0.1143397461541259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9629269583940137e-05,
+      "loss": 1.2329,
+      "step": 3144
+    },
+    {
+      "epoch": 0.1143761137578645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962895177452683e-05,
+      "loss": 1.0323,
+      "step": 3145
+    },
+    {
+      "epoch": 0.11441248136160308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9628633831525482e-05,
+      "loss": 1.1959,
+      "step": 3146
+    },
+    {
+      "epoch": 0.11444884896534167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96283157549405e-05,
+      "loss": 1.2207,
+      "step": 3147
+    },
+    {
+      "epoch": 0.11448521656908027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96279975447763e-05,
+      "loss": 1.0838,
+      "step": 3148
+    },
+    {
+      "epoch": 0.11452158417281885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9627679201037294e-05,
+      "loss": 1.2145,
+      "step": 3149
+    },
+    {
+      "epoch": 0.11455795177655745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9627360723727896e-05,
+      "loss": 1.2155,
+      "step": 3150
+    },
+    {
+      "epoch": 0.11459431938029603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962704211285253e-05,
+      "loss": 1.178,
+      "step": 3151
+    },
+    {
+      "epoch": 0.11463068698403463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962672336841561e-05,
+      "loss": 1.1443,
+      "step": 3152
+    },
+    {
+      "epoch": 0.11466705458777321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9626404490421563e-05,
+      "loss": 1.1644,
+      "step": 3153
+    },
+    {
+      "epoch": 0.1147034221915118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9626085478874817e-05,
+      "loss": 1.0511,
+      "step": 3154
+    },
+    {
+      "epoch": 0.11473978979525039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962576633377979e-05,
+      "loss": 1.2216,
+      "step": 3155
+    },
+    {
+      "epoch": 0.11477615739898898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962544705514091e-05,
+      "loss": 1.2102,
+      "step": 3156
+    },
+    {
+      "epoch": 0.11481252500272757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962512764296261e-05,
+      "loss": 1.1501,
+      "step": 3157
+    },
+    {
+      "epoch": 0.11484889260646616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962480809724932e-05,
+      "loss": 1.2301,
+      "step": 3158
+    },
+    {
+      "epoch": 0.11488526021020475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9624488418005473e-05,
+      "loss": 1.2309,
+      "step": 3159
+    },
+    {
+      "epoch": 0.11492162781394334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9624168605235504e-05,
+      "loss": 1.2393,
+      "step": 3160
+    },
+    {
+      "epoch": 0.11495799541768192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9623848658943855e-05,
+      "loss": 1.2062,
+      "step": 3161
+    },
+    {
+      "epoch": 0.11499436302142052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9623528579134955e-05,
+      "loss": 1.1956,
+      "step": 3162
+    },
+    {
+      "epoch": 0.1150307306251591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9623208365813253e-05,
+      "loss": 1.2425,
+      "step": 3163
+    },
+    {
+      "epoch": 0.1150670982288977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9622888018983186e-05,
+      "loss": 1.1684,
+      "step": 3164
+    },
+    {
+      "epoch": 0.11510346583263628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9622567538649204e-05,
+      "loss": 1.2957,
+      "step": 3165
+    },
+    {
+      "epoch": 0.11513983343637488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9622246924815748e-05,
+      "loss": 1.2176,
+      "step": 3166
+    },
+    {
+      "epoch": 0.11517620104011346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962192617748727e-05,
+      "loss": 1.1218,
+      "step": 3167
+    },
+    {
+      "epoch": 0.11521256864385206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9621605296668217e-05,
+      "loss": 1.1798,
+      "step": 3168
+    },
+    {
+      "epoch": 0.11524893624759065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9621284282363042e-05,
+      "loss": 1.0874,
+      "step": 3169
+    },
+    {
+      "epoch": 0.11528530385132923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9620963134576198e-05,
+      "loss": 1.2407,
+      "step": 3170
+    },
+    {
+      "epoch": 0.11532167145506783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962064185331214e-05,
+      "loss": 1.1541,
+      "step": 3171
+    },
+    {
+      "epoch": 0.11535803905880641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9620320438575327e-05,
+      "loss": 1.2358,
+      "step": 3172
+    },
+    {
+      "epoch": 0.11539440666254501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9619998890370215e-05,
+      "loss": 1.1866,
+      "step": 3173
+    },
+    {
+      "epoch": 0.11543077426628359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961967720870127e-05,
+      "loss": 1.2144,
+      "step": 3174
+    },
+    {
+      "epoch": 0.11546714187002219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9619355393572952e-05,
+      "loss": 1.1239,
+      "step": 3175
+    },
+    {
+      "epoch": 0.11550350947376077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9619033444989724e-05,
+      "loss": 1.166,
+      "step": 3176
+    },
+    {
+      "epoch": 0.11553987707749937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9618711362956056e-05,
+      "loss": 1.1684,
+      "step": 3177
+    },
+    {
+      "epoch": 0.11557624468123795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9618389147476417e-05,
+      "loss": 1.2197,
+      "step": 3178
+    },
+    {
+      "epoch": 0.11561261228497655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9618066798555274e-05,
+      "loss": 1.1735,
+      "step": 3179
+    },
+    {
+      "epoch": 0.11564897988871513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96177443161971e-05,
+      "loss": 1.2228,
+      "step": 3180
+    },
+    {
+      "epoch": 0.11568534749245372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961742170040637e-05,
+      "loss": 1.1248,
+      "step": 3181
+    },
+    {
+      "epoch": 0.1157217150961923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9617098951187558e-05,
+      "loss": 1.1959,
+      "step": 3182
+    },
+    {
+      "epoch": 0.1157580826999309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9616776068545144e-05,
+      "loss": 1.3199,
+      "step": 3183
+    },
+    {
+      "epoch": 0.11579445030366949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9616453052483607e-05,
+      "loss": 1.2162,
+      "step": 3184
+    },
+    {
+      "epoch": 0.11583081790740808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9616129903007425e-05,
+      "loss": 1.2695,
+      "step": 3185
+    },
+    {
+      "epoch": 0.11586718551114666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961580662012109e-05,
+      "loss": 0.9817,
+      "step": 3186
+    },
+    {
+      "epoch": 0.11590355311488526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9615483203829076e-05,
+      "loss": 1.368,
+      "step": 3187
+    },
+    {
+      "epoch": 0.11593992071862386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9615159654135877e-05,
+      "loss": 1.1658,
+      "step": 3188
+    },
+    {
+      "epoch": 0.11597628832236244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9614835971045977e-05,
+      "loss": 1.1574,
+      "step": 3189
+    },
+    {
+      "epoch": 0.11601265592610104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9614512154563875e-05,
+      "loss": 1.1942,
+      "step": 3190
+    },
+    {
+      "epoch": 0.11604902352983962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9614188204694053e-05,
+      "loss": 1.1647,
+      "step": 3191
+    },
+    {
+      "epoch": 0.11608539113357821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9613864121441015e-05,
+      "loss": 1.2432,
+      "step": 3192
+    },
+    {
+      "epoch": 0.1161217587373168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961353990480925e-05,
+      "loss": 1.2571,
+      "step": 3193
+    },
+    {
+      "epoch": 0.11615812634105539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9613215554803258e-05,
+      "loss": 1.123,
+      "step": 3194
+    },
+    {
+      "epoch": 0.11619449394479398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961289107142754e-05,
+      "loss": 1.2636,
+      "step": 3195
+    },
+    {
+      "epoch": 0.11623086154853257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9612566454686598e-05,
+      "loss": 1.2494,
+      "step": 3196
+    },
+    {
+      "epoch": 0.11626722915227115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9612241704584934e-05,
+      "loss": 1.1024,
+      "step": 3197
+    },
+    {
+      "epoch": 0.11630359675600975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9611916821127053e-05,
+      "loss": 1.1973,
+      "step": 3198
+    },
+    {
+      "epoch": 0.11633996435974833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9611591804317464e-05,
+      "loss": 1.3188,
+      "step": 3199
+    },
+    {
+      "epoch": 0.11637633196348693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961126665416068e-05,
+      "loss": 1.1046,
+      "step": 3200
+    },
+    {
+      "epoch": 0.11641269956722551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9610941370661205e-05,
+      "loss": 1.0397,
+      "step": 3201
+    },
+    {
+      "epoch": 0.11644906717096411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9610615953823554e-05,
+      "loss": 1.1736,
+      "step": 3202
+    },
+    {
+      "epoch": 0.11648543477470269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.961029040365224e-05,
+      "loss": 1.1647,
+      "step": 3203
+    },
+    {
+      "epoch": 0.11652180237844129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9609964720151788e-05,
+      "loss": 1.2732,
+      "step": 3204
+    },
+    {
+      "epoch": 0.11655816998217987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9609638903326703e-05,
+      "loss": 1.2256,
+      "step": 3205
+    },
+    {
+      "epoch": 0.11659453758591846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960931295318152e-05,
+      "loss": 1.0237,
+      "step": 3206
+    },
+    {
+      "epoch": 0.11663090518965705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9608986869720748e-05,
+      "loss": 1.25,
+      "step": 3207
+    },
+    {
+      "epoch": 0.11666727279339564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960866065294892e-05,
+      "loss": 1.0952,
+      "step": 3208
+    },
+    {
+      "epoch": 0.11670364039713424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960833430287056e-05,
+      "loss": 1.0688,
+      "step": 3209
+    },
+    {
+      "epoch": 0.11674000800087282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960800781949019e-05,
+      "loss": 1.0399,
+      "step": 3210
+    },
+    {
+      "epoch": 0.11677637560461142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9607681202812342e-05,
+      "loss": 1.0707,
+      "step": 3211
+    },
+    {
+      "epoch": 0.11681274320835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9607354452841553e-05,
+      "loss": 1.1582,
+      "step": 3212
+    },
+    {
+      "epoch": 0.1168491108120886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960702756958235e-05,
+      "loss": 1.125,
+      "step": 3213
+    },
+    {
+      "epoch": 0.11688547841582718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960670055303927e-05,
+      "loss": 1.2525,
+      "step": 3214
+    },
+    {
+      "epoch": 0.11692184601956578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960637340321685e-05,
+      "loss": 1.2294,
+      "step": 3215
+    },
+    {
+      "epoch": 0.11695821362330436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960604612011963e-05,
+      "loss": 1.2319,
+      "step": 3216
+    },
+    {
+      "epoch": 0.11699458122704295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9605718703752147e-05,
+      "loss": 1.1683,
+      "step": 3217
+    },
+    {
+      "epoch": 0.11703094883078154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9605391154118948e-05,
+      "loss": 1.1557,
+      "step": 3218
+    },
+    {
+      "epoch": 0.11706731643452013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9605063471224573e-05,
+      "loss": 1.1324,
+      "step": 3219
+    },
+    {
+      "epoch": 0.11710368403825872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960473565507357e-05,
+      "loss": 1.079,
+      "step": 3220
+    },
+    {
+      "epoch": 0.11714005164199731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960440770567049e-05,
+      "loss": 1.0887,
+      "step": 3221
+    },
+    {
+      "epoch": 0.1171764192457359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960407962301988e-05,
+      "loss": 1.1247,
+      "step": 3222
+    },
+    {
+      "epoch": 0.11721278684947449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9603751407126287e-05,
+      "loss": 1.1794,
+      "step": 3223
+    },
+    {
+      "epoch": 0.11724915445321307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9603423057994275e-05,
+      "loss": 1.1241,
+      "step": 3224
+    },
+    {
+      "epoch": 0.11728552205695167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960309457562839e-05,
+      "loss": 1.0431,
+      "step": 3225
+    },
+    {
+      "epoch": 0.11732188966069025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9602765960033193e-05,
+      "loss": 1.1941,
+      "step": 3226
+    },
+    {
+      "epoch": 0.11735825726442885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9602437211213244e-05,
+      "loss": 1.2426,
+      "step": 3227
+    },
+    {
+      "epoch": 0.11739462486816743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96021083291731e-05,
+      "loss": 1.2371,
+      "step": 3228
+    },
+    {
+      "epoch": 0.11743099247190603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960177931391733e-05,
+      "loss": 1.2602,
+      "step": 3229
+    },
+    {
+      "epoch": 0.11746736007564462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9601450165450493e-05,
+      "loss": 1.092,
+      "step": 3230
+    },
+    {
+      "epoch": 0.1175037276793832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9601120883777158e-05,
+      "loss": 1.0589,
+      "step": 3231
+    },
+    {
+      "epoch": 0.1175400952831218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9600791468901894e-05,
+      "loss": 1.2366,
+      "step": 3232
+    },
+    {
+      "epoch": 0.11757646288686038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.960046192082927e-05,
+      "loss": 1.2314,
+      "step": 3233
+    },
+    {
+      "epoch": 0.11761283049059898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9600132239563855e-05,
+      "loss": 1.2174,
+      "step": 3234
+    },
+    {
+      "epoch": 0.11764919809433756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959980242511023e-05,
+      "loss": 1.2846,
+      "step": 3235
+    },
+    {
+      "epoch": 0.11768556569807616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9599472477472965e-05,
+      "loss": 1.0462,
+      "step": 3236
+    },
+    {
+      "epoch": 0.11772193330181474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9599142396656637e-05,
+      "loss": 1.2196,
+      "step": 3237
+    },
+    {
+      "epoch": 0.11775830090555334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9598812182665835e-05,
+      "loss": 1.1763,
+      "step": 3238
+    },
+    {
+      "epoch": 0.11779466850929192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9598481835505127e-05,
+      "loss": 1.2485,
+      "step": 3239
+    },
+    {
+      "epoch": 0.11783103611303052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9598151355179103e-05,
+      "loss": 1.1642,
+      "step": 3240
+    },
+    {
+      "epoch": 0.1178674037167691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9597820741692347e-05,
+      "loss": 1.1558,
+      "step": 3241
+    },
+    {
+      "epoch": 0.1179037713205077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9597489995049444e-05,
+      "loss": 1.0242,
+      "step": 3242
+    },
+    {
+      "epoch": 0.11794013892424628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9597159115254988e-05,
+      "loss": 1.2065,
+      "step": 3243
+    },
+    {
+      "epoch": 0.11797650652798487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9596828102313562e-05,
+      "loss": 1.1235,
+      "step": 3244
+    },
+    {
+      "epoch": 0.11801287413172346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9596496956229765e-05,
+      "loss": 1.0726,
+      "step": 3245
+    },
+    {
+      "epoch": 0.11804924173546205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9596165677008188e-05,
+      "loss": 1.2665,
+      "step": 3246
+    },
+    {
+      "epoch": 0.11808560933920063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9595834264653428e-05,
+      "loss": 1.2539,
+      "step": 3247
+    },
+    {
+      "epoch": 0.11812197694293923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959550271917008e-05,
+      "loss": 1.1322,
+      "step": 3248
+    },
+    {
+      "epoch": 0.11815834454667781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9595171040562747e-05,
+      "loss": 1.1096,
+      "step": 3249
+    },
+    {
+      "epoch": 0.11819471215041641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9594839228836028e-05,
+      "loss": 1.1741,
+      "step": 3250
+    },
+    {
+      "epoch": 0.118231079754155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9594507283994534e-05,
+      "loss": 1.0882,
+      "step": 3251
+    },
+    {
+      "epoch": 0.11826744735789359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959417520604286e-05,
+      "loss": 1.2457,
+      "step": 3252
+    },
+    {
+      "epoch": 0.11830381496163218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9593842994985615e-05,
+      "loss": 1.2173,
+      "step": 3253
+    },
+    {
+      "epoch": 0.11834018256537077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9593510650827413e-05,
+      "loss": 1.1205,
+      "step": 3254
+    },
+    {
+      "epoch": 0.11837655016910936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959317817357286e-05,
+      "loss": 1.201,
+      "step": 3255
+    },
+    {
+      "epoch": 0.11841291777284794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9592845563226572e-05,
+      "loss": 1.3056,
+      "step": 3256
+    },
+    {
+      "epoch": 0.11844928537658654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9592512819793166e-05,
+      "loss": 1.2209,
+      "step": 3257
+    },
+    {
+      "epoch": 0.11848565298032512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959217994327725e-05,
+      "loss": 1.222,
+      "step": 3258
+    },
+    {
+      "epoch": 0.11852202058406372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9591846933683446e-05,
+      "loss": 1.1216,
+      "step": 3259
+    },
+    {
+      "epoch": 0.1185583881878023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959151379101638e-05,
+      "loss": 1.1142,
+      "step": 3260
+    },
+    {
+      "epoch": 0.1185947557915409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9591180515280667e-05,
+      "loss": 1.2452,
+      "step": 3261
+    },
+    {
+      "epoch": 0.11863112339527948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.959084710648093e-05,
+      "loss": 1.1888,
+      "step": 3262
+    },
+    {
+      "epoch": 0.11866749099901808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95905135646218e-05,
+      "loss": 1.1889,
+      "step": 3263
+    },
+    {
+      "epoch": 0.11870385860275666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9590179889707897e-05,
+      "loss": 1.1573,
+      "step": 3264
+    },
+    {
+      "epoch": 0.11874022620649526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958984608174386e-05,
+      "loss": 1.2754,
+      "step": 3265
+    },
+    {
+      "epoch": 0.11877659381023384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958951214073431e-05,
+      "loss": 1.1146,
+      "step": 3266
+    },
+    {
+      "epoch": 0.11881296141397243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958917806668389e-05,
+      "loss": 1.1553,
+      "step": 3267
+    },
+    {
+      "epoch": 0.11884932901771102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9588843859597223e-05,
+      "loss": 1.1832,
+      "step": 3268
+    },
+    {
+      "epoch": 0.11888569662144961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958850951947896e-05,
+      "loss": 1.1707,
+      "step": 3269
+    },
+    {
+      "epoch": 0.1189220642251882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9588175046333727e-05,
+      "loss": 1.1084,
+      "step": 3270
+    },
+    {
+      "epoch": 0.11895843182892679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958784044016617e-05,
+      "loss": 1.1939,
+      "step": 3271
+    },
+    {
+      "epoch": 0.11899479943266539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958750570098093e-05,
+      "loss": 1.1797,
+      "step": 3272
+    },
+    {
+      "epoch": 0.11903116703640397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9587170828782652e-05,
+      "loss": 1.0771,
+      "step": 3273
+    },
+    {
+      "epoch": 0.11906753464014257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9586835823575982e-05,
+      "loss": 1.1858,
+      "step": 3274
+    },
+    {
+      "epoch": 0.11910390224388115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9586500685365562e-05,
+      "loss": 1.1776,
+      "step": 3275
+    },
+    {
+      "epoch": 0.11914026984761975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958616541415605e-05,
+      "loss": 1.1015,
+      "step": 3276
+    },
+    {
+      "epoch": 0.11917663745135833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9585830009952095e-05,
+      "loss": 1.1879,
+      "step": 3277
+    },
+    {
+      "epoch": 0.11921300505509692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958549447275835e-05,
+      "loss": 1.153,
+      "step": 3278
+    },
+    {
+      "epoch": 0.1192493726588355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9585158802579464e-05,
+      "loss": 1.2527,
+      "step": 3279
+    },
+    {
+      "epoch": 0.1192857402625741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95848229994201e-05,
+      "loss": 1.0642,
+      "step": 3280
+    },
+    {
+      "epoch": 0.11932210786631268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958448706328492e-05,
+      "loss": 1.1328,
+      "step": 3281
+    },
+    {
+      "epoch": 0.11935847547005128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958415099417858e-05,
+      "loss": 1.1857,
+      "step": 3282
+    },
+    {
+      "epoch": 0.11939484307378986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9583814792105743e-05,
+      "loss": 1.0234,
+      "step": 3283
+    },
+    {
+      "epoch": 0.11943121067752846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958347845707107e-05,
+      "loss": 1.1277,
+      "step": 3284
+    },
+    {
+      "epoch": 0.11946757828126704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9583141989079232e-05,
+      "loss": 1.121,
+      "step": 3285
+    },
+    {
+      "epoch": 0.11950394588500564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9582805388134896e-05,
+      "loss": 1.0948,
+      "step": 3286
+    },
+    {
+      "epoch": 0.11954031348874422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9582468654242736e-05,
+      "loss": 1.1545,
+      "step": 3287
+    },
+    {
+      "epoch": 0.11957668109248282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9582131787407413e-05,
+      "loss": 1.151,
+      "step": 3288
+    },
+    {
+      "epoch": 0.1196130486962214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958179478763361e-05,
+      "loss": 1.2031,
+      "step": 3289
+    },
+    {
+      "epoch": 0.11964941629996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9581457654925998e-05,
+      "loss": 1.1519,
+      "step": 3290
+    },
+    {
+      "epoch": 0.11968578390369859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9581120389289254e-05,
+      "loss": 1.14,
+      "step": 3291
+    },
+    {
+      "epoch": 0.11972215150743717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9580782990728063e-05,
+      "loss": 1.1359,
+      "step": 3292
+    },
+    {
+      "epoch": 0.11975851911117577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9580445459247098e-05,
+      "loss": 1.2934,
+      "step": 3293
+    },
+    {
+      "epoch": 0.11979488671491435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9580107794851044e-05,
+      "loss": 1.1189,
+      "step": 3294
+    },
+    {
+      "epoch": 0.11983125431865295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957976999754459e-05,
+      "loss": 1.2042,
+      "step": 3295
+    },
+    {
+      "epoch": 0.11986762192239153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9579432067332416e-05,
+      "loss": 1.1221,
+      "step": 3296
+    },
+    {
+      "epoch": 0.11990398952613013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9579094004219214e-05,
+      "loss": 1.252,
+      "step": 3297
+    },
+    {
+      "epoch": 0.11994035712986871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9578755808209673e-05,
+      "loss": 1.1255,
+      "step": 3298
+    },
+    {
+      "epoch": 0.1199767247336073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9578417479308486e-05,
+      "loss": 1.224,
+      "step": 3299
+    },
+    {
+      "epoch": 0.12001309233734589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957807901752035e-05,
+      "loss": 1.1387,
+      "step": 3300
+    },
+    {
+      "epoch": 0.12004945994108449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9577740422849954e-05,
+      "loss": 1.1705,
+      "step": 3301
+    },
+    {
+      "epoch": 0.12008582754482307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9577401695302e-05,
+      "loss": 1.2081,
+      "step": 3302
+    },
+    {
+      "epoch": 0.12012219514856166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9577062834881184e-05,
+      "loss": 1.113,
+      "step": 3303
+    },
+    {
+      "epoch": 0.12015856275230025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9576723841592212e-05,
+      "loss": 1.1656,
+      "step": 3304
+    },
+    {
+      "epoch": 0.12019493035603884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957638471543978e-05,
+      "loss": 1.0849,
+      "step": 3305
+    },
+    {
+      "epoch": 0.12023129795977743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9576045456428603e-05,
+      "loss": 1.1758,
+      "step": 3306
+    },
+    {
+      "epoch": 0.12026766556351602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9575706064563378e-05,
+      "loss": 1.1193,
+      "step": 3307
+    },
+    {
+      "epoch": 0.1203040331672546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9575366539848818e-05,
+      "loss": 1.1664,
+      "step": 3308
+    },
+    {
+      "epoch": 0.1203404007709932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9575026882289632e-05,
+      "loss": 1.1946,
+      "step": 3309
+    },
+    {
+      "epoch": 0.12037676837473178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9574687091890537e-05,
+      "loss": 1.0509,
+      "step": 3310
+    },
+    {
+      "epoch": 0.12041313597847038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957434716865624e-05,
+      "loss": 1.2482,
+      "step": 3311
+    },
+    {
+      "epoch": 0.12044950358220897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9574007112591463e-05,
+      "loss": 1.1837,
+      "step": 3312
+    },
+    {
+      "epoch": 0.12048587118594756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9573666923700917e-05,
+      "loss": 1.2355,
+      "step": 3313
+    },
+    {
+      "epoch": 0.12052223878968615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957332660198933e-05,
+      "loss": 1.1289,
+      "step": 3314
+    },
+    {
+      "epoch": 0.12055860639342474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9572986147461415e-05,
+      "loss": 1.2495,
+      "step": 3315
+    },
+    {
+      "epoch": 0.12059497399716333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95726455601219e-05,
+      "loss": 1.1978,
+      "step": 3316
+    },
+    {
+      "epoch": 0.12063134160090191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9572304839975513e-05,
+      "loss": 1.1205,
+      "step": 3317
+    },
+    {
+      "epoch": 0.12066770920464051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9571963987026975e-05,
+      "loss": 1.1406,
+      "step": 3318
+    },
+    {
+      "epoch": 0.1207040768083791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957162300128102e-05,
+      "loss": 1.1878,
+      "step": 3319
+    },
+    {
+      "epoch": 0.12074044441211769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9571281882742374e-05,
+      "loss": 1.1379,
+      "step": 3320
+    },
+    {
+      "epoch": 0.12077681201585627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9570940631415773e-05,
+      "loss": 1.2042,
+      "step": 3321
+    },
+    {
+      "epoch": 0.12081317961959487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957059924730595e-05,
+      "loss": 1.217,
+      "step": 3322
+    },
+    {
+      "epoch": 0.12084954722333345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957025773041764e-05,
+      "loss": 1.1404,
+      "step": 3323
+    },
+    {
+      "epoch": 0.12088591482707205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9569916080755583e-05,
+      "loss": 1.1198,
+      "step": 3324
+    },
+    {
+      "epoch": 0.12092228243081063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9569574298324518e-05,
+      "loss": 1.149,
+      "step": 3325
+    },
+    {
+      "epoch": 0.12095865003454923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9569232383129187e-05,
+      "loss": 1.216,
+      "step": 3326
+    },
+    {
+      "epoch": 0.12099501763828781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9568890335174336e-05,
+      "loss": 1.1516,
+      "step": 3327
+    },
+    {
+      "epoch": 0.1210313852420264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9568548154464707e-05,
+      "loss": 1.1125,
+      "step": 3328
+    },
+    {
+      "epoch": 0.12106775284576499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9568205841005045e-05,
+      "loss": 1.1003,
+      "step": 3329
+    },
+    {
+      "epoch": 0.12110412044950358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9567863394800107e-05,
+      "loss": 1.2508,
+      "step": 3330
+    },
+    {
+      "epoch": 0.12114048805324217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9567520815854637e-05,
+      "loss": 1.374,
+      "step": 3331
+    },
+    {
+      "epoch": 0.12117685565698076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9567178104173388e-05,
+      "loss": 1.1232,
+      "step": 3332
+    },
+    {
+      "epoch": 0.12121322326071936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9566835259761125e-05,
+      "loss": 1.1119,
+      "step": 3333
+    },
+    {
+      "epoch": 0.12124959086445794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956649228262259e-05,
+      "loss": 1.2113,
+      "step": 3334
+    },
+    {
+      "epoch": 0.12128595846819654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9566149172762547e-05,
+      "loss": 1.0797,
+      "step": 3335
+    },
+    {
+      "epoch": 0.12132232607193512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956580593018576e-05,
+      "loss": 1.2089,
+      "step": 3336
+    },
+    {
+      "epoch": 0.12135869367567372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9565462554896986e-05,
+      "loss": 1.2114,
+      "step": 3337
+    },
+    {
+      "epoch": 0.1213950612794123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956511904690099e-05,
+      "loss": 1.1896,
+      "step": 3338
+    },
+    {
+      "epoch": 0.1214314288831509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9564775406202542e-05,
+      "loss": 1.2565,
+      "step": 3339
+    },
+    {
+      "epoch": 0.12146779648688948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95644316328064e-05,
+      "loss": 1.1101,
+      "step": 3340
+    },
+    {
+      "epoch": 0.12150416409062807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9564087726717344e-05,
+      "loss": 1.2223,
+      "step": 3341
+    },
+    {
+      "epoch": 0.12154053169436665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9563743687940137e-05,
+      "loss": 1.1995,
+      "step": 3342
+    },
+    {
+      "epoch": 0.12157689929810525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956339951647956e-05,
+      "loss": 1.2227,
+      "step": 3343
+    },
+    {
+      "epoch": 0.12161326690184383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9563055212340378e-05,
+      "loss": 1.1783,
+      "step": 3344
+    },
+    {
+      "epoch": 0.12164963450558243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9562710775527377e-05,
+      "loss": 1.1593,
+      "step": 3345
+    },
+    {
+      "epoch": 0.12168600210932101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9562366206045333e-05,
+      "loss": 1.2547,
+      "step": 3346
+    },
+    {
+      "epoch": 0.12172236971305961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9562021503899024e-05,
+      "loss": 1.1896,
+      "step": 3347
+    },
+    {
+      "epoch": 0.12175873731679819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9561676669093226e-05,
+      "loss": 1.2417,
+      "step": 3348
+    },
+    {
+      "epoch": 0.12179510492053679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.956133170163274e-05,
+      "loss": 1.1806,
+      "step": 3349
+    },
+    {
+      "epoch": 0.12183147252427537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9560986601522338e-05,
+      "loss": 1.3092,
+      "step": 3350
+    },
+    {
+      "epoch": 0.12186784012801397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9560641368766814e-05,
+      "loss": 1.2151,
+      "step": 3351
+    },
+    {
+      "epoch": 0.12190420773175255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9560296003370953e-05,
+      "loss": 1.2443,
+      "step": 3352
+    },
+    {
+      "epoch": 0.12194057533549114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9559950505339554e-05,
+      "loss": 1.1107,
+      "step": 3353
+    },
+    {
+      "epoch": 0.12197694293922974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95596048746774e-05,
+      "loss": 1.1261,
+      "step": 3354
+    },
+    {
+      "epoch": 0.12201331054296832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9559259111389293e-05,
+      "loss": 1.1676,
+      "step": 3355
+    },
+    {
+      "epoch": 0.12204967814670692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9558913215480034e-05,
+      "loss": 1.0627,
+      "step": 3356
+    },
+    {
+      "epoch": 0.1220860457504455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955856718695441e-05,
+      "loss": 1.1857,
+      "step": 3357
+    },
+    {
+      "epoch": 0.1221224133541841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955822102581723e-05,
+      "loss": 1.1,
+      "step": 3358
+    },
+    {
+      "epoch": 0.12215878095792268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95578747320733e-05,
+      "loss": 1.0459,
+      "step": 3359
+    },
+    {
+      "epoch": 0.12219514856166128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955752830572741e-05,
+      "loss": 1.236,
+      "step": 3360
+    },
+    {
+      "epoch": 0.12223151616539986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9557181746784383e-05,
+      "loss": 1.0682,
+      "step": 3361
+    },
+    {
+      "epoch": 0.12226788376913846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9556835055249013e-05,
+      "loss": 1.1652,
+      "step": 3362
+    },
+    {
+      "epoch": 0.12230425137287704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9556488231126118e-05,
+      "loss": 1.256,
+      "step": 3363
+    },
+    {
+      "epoch": 0.12234061897661563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9556141274420507e-05,
+      "loss": 1.1479,
+      "step": 3364
+    },
+    {
+      "epoch": 0.12237698658035422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9555794185136994e-05,
+      "loss": 1.2296,
+      "step": 3365
+    },
+    {
+      "epoch": 0.12241335418409281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95554469632804e-05,
+      "loss": 1.2994,
+      "step": 3366
+    },
+    {
+      "epoch": 0.1224497217878314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955509960885553e-05,
+      "loss": 1.1528,
+      "step": 3367
+    },
+    {
+      "epoch": 0.12248608939156999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9554752121867214e-05,
+      "loss": 1.2253,
+      "step": 3368
+    },
+    {
+      "epoch": 0.12252245699530857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9554404502320266e-05,
+      "loss": 1.0782,
+      "step": 3369
+    },
+    {
+      "epoch": 0.12255882459904717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9554056750219513e-05,
+      "loss": 1.0641,
+      "step": 3370
+    },
+    {
+      "epoch": 0.12259519220278575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955370886556978e-05,
+      "loss": 1.237,
+      "step": 3371
+    },
+    {
+      "epoch": 0.12263155980652435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9553360848375886e-05,
+      "loss": 1.1312,
+      "step": 3372
+    },
+    {
+      "epoch": 0.12266792741026294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9553012698642672e-05,
+      "loss": 1.2791,
+      "step": 3373
+    },
+    {
+      "epoch": 0.12270429501400153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9552664416374954e-05,
+      "loss": 1.2853,
+      "step": 3374
+    },
+    {
+      "epoch": 0.12274066261774012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9552316001577575e-05,
+      "loss": 1.3703,
+      "step": 3375
+    },
+    {
+      "epoch": 0.1227770302214787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9551967454255362e-05,
+      "loss": 1.256,
+      "step": 3376
+    },
+    {
+      "epoch": 0.1228133978252173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9551618774413154e-05,
+      "loss": 1.1343,
+      "step": 3377
+    },
+    {
+      "epoch": 0.12284976542895588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9551269962055785e-05,
+      "loss": 1.3007,
+      "step": 3378
+    },
+    {
+      "epoch": 0.12288613303269448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95509210171881e-05,
+      "loss": 1.2003,
+      "step": 3379
+    },
+    {
+      "epoch": 0.12292250063643306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9550571939814937e-05,
+      "loss": 1.1676,
+      "step": 3380
+    },
+    {
+      "epoch": 0.12295886824017166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9550222729941137e-05,
+      "loss": 0.9901,
+      "step": 3381
+    },
+    {
+      "epoch": 0.12299523584391024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9549873387571546e-05,
+      "loss": 1.0848,
+      "step": 3382
+    },
+    {
+      "epoch": 0.12303160344764884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9549523912711014e-05,
+      "loss": 1.1402,
+      "step": 3383
+    },
+    {
+      "epoch": 0.12306797105138742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9549174305364384e-05,
+      "loss": 1.0448,
+      "step": 3384
+    },
+    {
+      "epoch": 0.12310433865512602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954882456553651e-05,
+      "loss": 1.1194,
+      "step": 3385
+    },
+    {
+      "epoch": 0.1231407062588646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9548474693232245e-05,
+      "loss": 1.1878,
+      "step": 3386
+    },
+    {
+      "epoch": 0.1231770738626032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954812468845644e-05,
+      "loss": 1.1239,
+      "step": 3387
+    },
+    {
+      "epoch": 0.12321344146634178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954777455121395e-05,
+      "loss": 1.2279,
+      "step": 3388
+    },
+    {
+      "epoch": 0.12324980907008037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9547424281509636e-05,
+      "loss": 1.205,
+      "step": 3389
+    },
+    {
+      "epoch": 0.12328617667381896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9547073879348356e-05,
+      "loss": 1.1628,
+      "step": 3390
+    },
+    {
+      "epoch": 0.12332254427755755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9546723344734973e-05,
+      "loss": 1.0802,
+      "step": 3391
+    },
+    {
+      "epoch": 0.12335891188129613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9546372677674347e-05,
+      "loss": 1.2965,
+      "step": 3392
+    },
+    {
+      "epoch": 0.12339527948503473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9546021878171345e-05,
+      "loss": 1.2324,
+      "step": 3393
+    },
+    {
+      "epoch": 0.12343164708877333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9545670946230832e-05,
+      "loss": 1.1732,
+      "step": 3394
+    },
+    {
+      "epoch": 0.12346801469251191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954531988185768e-05,
+      "loss": 1.2103,
+      "step": 3395
+    },
+    {
+      "epoch": 0.1235043822962505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9544968685056757e-05,
+      "loss": 1.1385,
+      "step": 3396
+    },
+    {
+      "epoch": 0.12354074989998909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9544617355832936e-05,
+      "loss": 1.1214,
+      "step": 3397
+    },
+    {
+      "epoch": 0.12357711750372768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9544265894191093e-05,
+      "loss": 1.1717,
+      "step": 3398
+    },
+    {
+      "epoch": 0.12361348510746627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95439143001361e-05,
+      "loss": 1.072,
+      "step": 3399
+    },
+    {
+      "epoch": 0.12364985271120486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954356257367284e-05,
+      "loss": 1.1799,
+      "step": 3400
+    },
+    {
+      "epoch": 0.12368622031494345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9543210714806188e-05,
+      "loss": 1.2925,
+      "step": 3401
+    },
+    {
+      "epoch": 0.12372258791868204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9542858723541026e-05,
+      "loss": 1.2508,
+      "step": 3402
+    },
+    {
+      "epoch": 0.12375895552242062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954250659988224e-05,
+      "loss": 1.223,
+      "step": 3403
+    },
+    {
+      "epoch": 0.12379532312615922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954215434383471e-05,
+      "loss": 1.1858,
+      "step": 3404
+    },
+    {
+      "epoch": 0.1238316907298978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9541801955403334e-05,
+      "loss": 1.137,
+      "step": 3405
+    },
+    {
+      "epoch": 0.1238680583336364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954144943459299e-05,
+      "loss": 1.3369,
+      "step": 3406
+    },
+    {
+      "epoch": 0.12390442593737498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9541096781408572e-05,
+      "loss": 1.2001,
+      "step": 3407
+    },
+    {
+      "epoch": 0.12394079354111358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9540743995854978e-05,
+      "loss": 1.1013,
+      "step": 3408
+    },
+    {
+      "epoch": 0.12397716114485216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9540391077937093e-05,
+      "loss": 1.1163,
+      "step": 3409
+    },
+    {
+      "epoch": 0.12401352874859076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.954003802765982e-05,
+      "loss": 1.1807,
+      "step": 3410
+    },
+    {
+      "epoch": 0.12404989635232934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9539684845028052e-05,
+      "loss": 1.1516,
+      "step": 3411
+    },
+    {
+      "epoch": 0.12408626395606794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9539331530046695e-05,
+      "loss": 1.1723,
+      "step": 3412
+    },
+    {
+      "epoch": 0.12412263155980652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9538978082720645e-05,
+      "loss": 1.1573,
+      "step": 3413
+    },
+    {
+      "epoch": 0.12415899916354511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953862450305481e-05,
+      "loss": 1.269,
+      "step": 3414
+    },
+    {
+      "epoch": 0.12419536676728371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9538270791054095e-05,
+      "loss": 1.2459,
+      "step": 3415
+    },
+    {
+      "epoch": 0.12423173437102229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9537916946723404e-05,
+      "loss": 1.1703,
+      "step": 3416
+    },
+    {
+      "epoch": 0.12426810197476089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9537562970067644e-05,
+      "loss": 1.1768,
+      "step": 3417
+    },
+    {
+      "epoch": 0.12430446957849947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9537208861091738e-05,
+      "loss": 1.0651,
+      "step": 3418
+    },
+    {
+      "epoch": 0.12434083718223807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9536854619800584e-05,
+      "loss": 1.1538,
+      "step": 3419
+    },
+    {
+      "epoch": 0.12437720478597665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9536500246199106e-05,
+      "loss": 1.0908,
+      "step": 3420
+    },
+    {
+      "epoch": 0.12441357238971525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9536145740292217e-05,
+      "loss": 1.217,
+      "step": 3421
+    },
+    {
+      "epoch": 0.12444993999345383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9535791102084837e-05,
+      "loss": 1.096,
+      "step": 3422
+    },
+    {
+      "epoch": 0.12448630759719242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953543633158188e-05,
+      "loss": 1.1107,
+      "step": 3423
+    },
+    {
+      "epoch": 0.12452267520093101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9535081428788278e-05,
+      "loss": 1.2207,
+      "step": 3424
+    },
+    {
+      "epoch": 0.1245590428046696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953472639370895e-05,
+      "loss": 1.2965,
+      "step": 3425
+    },
+    {
+      "epoch": 0.12459541040840819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953437122634882e-05,
+      "loss": 1.0972,
+      "step": 3426
+    },
+    {
+      "epoch": 0.12463177801214678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9534015926712816e-05,
+      "loss": 1.1739,
+      "step": 3427
+    },
+    {
+      "epoch": 0.12466814561588536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953366049480587e-05,
+      "loss": 1.1392,
+      "step": 3428
+    },
+    {
+      "epoch": 0.12470451321962396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953330493063291e-05,
+      "loss": 1.0837,
+      "step": 3429
+    },
+    {
+      "epoch": 0.12474088082336254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9532949234198865e-05,
+      "loss": 1.1082,
+      "step": 3430
+    },
+    {
+      "epoch": 0.12477724842710114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9532593405508683e-05,
+      "loss": 1.2084,
+      "step": 3431
+    },
+    {
+      "epoch": 0.12481361603083972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953223744456729e-05,
+      "loss": 1.1801,
+      "step": 3432
+    },
+    {
+      "epoch": 0.12484998363457832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9531881351379624e-05,
+      "loss": 1.3573,
+      "step": 3433
+    },
+    {
+      "epoch": 0.1248863512383169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953152512595063e-05,
+      "loss": 1.2099,
+      "step": 3434
+    },
+    {
+      "epoch": 0.1249227188420555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953116876828525e-05,
+      "loss": 1.1752,
+      "step": 3435
+    },
+    {
+      "epoch": 0.1249590864457941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9530812278388422e-05,
+      "loss": 1.1666,
+      "step": 3436
+    },
+    {
+      "epoch": 0.12499545404953268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95304556562651e-05,
+      "loss": 1.082,
+      "step": 3437
+    },
+    {
+      "epoch": 0.12503182165327126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9530098901920227e-05,
+      "loss": 1.1118,
+      "step": 3438
+    },
+    {
+      "epoch": 0.12506818925700985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9529742015358752e-05,
+      "loss": 1.1016,
+      "step": 3439
+    },
+    {
+      "epoch": 0.12510455686074845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9529384996585626e-05,
+      "loss": 1.1511,
+      "step": 3440
+    },
+    {
+      "epoch": 0.12514092446448705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952902784560581e-05,
+      "loss": 1.1584,
+      "step": 3441
+    },
+    {
+      "epoch": 0.12517729206822562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952867056242425e-05,
+      "loss": 1.0483,
+      "step": 3442
+    },
+    {
+      "epoch": 0.1252136596719642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9528313147045902e-05,
+      "loss": 1.1907,
+      "step": 3443
+    },
+    {
+      "epoch": 0.1252500272757028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952795559947573e-05,
+      "loss": 1.2348,
+      "step": 3444
+    },
+    {
+      "epoch": 0.1252863948794414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9527597919718692e-05,
+      "loss": 1.189,
+      "step": 3445
+    },
+    {
+      "epoch": 0.12532276248317997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952724010777975e-05,
+      "loss": 1.1711,
+      "step": 3446
+    },
+    {
+      "epoch": 0.12535913008691857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952688216366387e-05,
+      "loss": 1.1352,
+      "step": 3447
+    },
+    {
+      "epoch": 0.12539549769065717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9526524087376018e-05,
+      "loss": 1.3198,
+      "step": 3448
+    },
+    {
+      "epoch": 0.12543186529439576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952616587892116e-05,
+      "loss": 1.156,
+      "step": 3449
+    },
+    {
+      "epoch": 0.12546823289813433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9525807538304266e-05,
+      "loss": 1.1587,
+      "step": 3450
+    },
+    {
+      "epoch": 0.12550460050187293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9525449065530306e-05,
+      "loss": 1.3038,
+      "step": 3451
+    },
+    {
+      "epoch": 0.12554096810561152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9525090460604255e-05,
+      "loss": 1.18,
+      "step": 3452
+    },
+    {
+      "epoch": 0.12557733570935012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9524731723531093e-05,
+      "loss": 1.2083,
+      "step": 3453
+    },
+    {
+      "epoch": 0.1256137033130887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9524372854315786e-05,
+      "loss": 1.1616,
+      "step": 3454
+    },
+    {
+      "epoch": 0.12565007091682728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9524013852963323e-05,
+      "loss": 1.1632,
+      "step": 3455
+    },
+    {
+      "epoch": 0.12568643852056588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952365471947868e-05,
+      "loss": 1.28,
+      "step": 3456
+    },
+    {
+      "epoch": 0.12572280612430448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9523295453866836e-05,
+      "loss": 1.2332,
+      "step": 3457
+    },
+    {
+      "epoch": 0.12575917372804307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9522936056132783e-05,
+      "loss": 1.2346,
+      "step": 3458
+    },
+    {
+      "epoch": 0.12579554133178164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952257652628151e-05,
+      "loss": 1.2978,
+      "step": 3459
+    },
+    {
+      "epoch": 0.12583190893552024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9522216864317987e-05,
+      "loss": 1.0049,
+      "step": 3460
+    },
+    {
+      "epoch": 0.12586827653925883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952185707024722e-05,
+      "loss": 1.0896,
+      "step": 3461
+    },
+    {
+      "epoch": 0.12590464414299743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9521497144074197e-05,
+      "loss": 1.1189,
+      "step": 3462
+    },
+    {
+      "epoch": 0.125941011746736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.952113708580391e-05,
+      "loss": 1.2471,
+      "step": 3463
+    },
+    {
+      "epoch": 0.1259773793504746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9520776895441353e-05,
+      "loss": 1.1788,
+      "step": 3464
+    },
+    {
+      "epoch": 0.1260137469542132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9520416572991525e-05,
+      "loss": 1.0742,
+      "step": 3465
+    },
+    {
+      "epoch": 0.1260501145579518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9520056118459427e-05,
+      "loss": 1.0516,
+      "step": 3466
+    },
+    {
+      "epoch": 0.12608648216169036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9519695531850055e-05,
+      "loss": 1.1692,
+      "step": 3467
+    },
+    {
+      "epoch": 0.12612284976542895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9519334813168415e-05,
+      "loss": 1.2113,
+      "step": 3468
+    },
+    {
+      "epoch": 0.12615921736916755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9518973962419513e-05,
+      "loss": 1.2332,
+      "step": 3469
+    },
+    {
+      "epoch": 0.12619558497290614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9518612979608348e-05,
+      "loss": 1.2552,
+      "step": 3470
+    },
+    {
+      "epoch": 0.1262319525766447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9518251864739936e-05,
+      "loss": 1.0704,
+      "step": 3471
+    },
+    {
+      "epoch": 0.1262683201803833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9517890617819287e-05,
+      "loss": 1.1833,
+      "step": 3472
+    },
+    {
+      "epoch": 0.1263046877841219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9517529238851406e-05,
+      "loss": 1.1899,
+      "step": 3473
+    },
+    {
+      "epoch": 0.1263410553878605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951716772784131e-05,
+      "loss": 1.0588,
+      "step": 3474
+    },
+    {
+      "epoch": 0.12637742299159907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9516806084794018e-05,
+      "loss": 1.2217,
+      "step": 3475
+    },
+    {
+      "epoch": 0.12641379059533767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9516444309714543e-05,
+      "loss": 1.0063,
+      "step": 3476
+    },
+    {
+      "epoch": 0.12645015819907626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9516082402607904e-05,
+      "loss": 1.0258,
+      "step": 3477
+    },
+    {
+      "epoch": 0.12648652580281486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9515720363479122e-05,
+      "loss": 1.1998,
+      "step": 3478
+    },
+    {
+      "epoch": 0.12652289340655345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951535819233322e-05,
+      "loss": 1.1324,
+      "step": 3479
+    },
+    {
+      "epoch": 0.12655926101029202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951499588917523e-05,
+      "loss": 1.1493,
+      "step": 3480
+    },
+    {
+      "epoch": 0.12659562861403062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951463345401017e-05,
+      "loss": 1.1001,
+      "step": 3481
+    },
+    {
+      "epoch": 0.12663199621776922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9514270886843064e-05,
+      "loss": 1.3032,
+      "step": 3482
+    },
+    {
+      "epoch": 0.1266683638215078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9513908187678955e-05,
+      "loss": 1.1373,
+      "step": 3483
+    },
+    {
+      "epoch": 0.12670473142524638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9513545356522864e-05,
+      "loss": 1.1307,
+      "step": 3484
+    },
+    {
+      "epoch": 0.12674109902898498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9513182393379836e-05,
+      "loss": 1.0669,
+      "step": 3485
+    },
+    {
+      "epoch": 0.12677746663272357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9512819298254896e-05,
+      "loss": 1.069,
+      "step": 3486
+    },
+    {
+      "epoch": 0.12681383423646217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951245607115308e-05,
+      "loss": 1.0962,
+      "step": 3487
+    },
+    {
+      "epoch": 0.12685020184020074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9512092712079437e-05,
+      "loss": 1.1039,
+      "step": 3488
+    },
+    {
+      "epoch": 0.12688656944393933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9511729221039005e-05,
+      "loss": 1.2154,
+      "step": 3489
+    },
+    {
+      "epoch": 0.12692293704767793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9511365598036823e-05,
+      "loss": 1.2012,
+      "step": 3490
+    },
+    {
+      "epoch": 0.12695930465141653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951100184307794e-05,
+      "loss": 1.1176,
+      "step": 3491
+    },
+    {
+      "epoch": 0.1269956722551551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95106379561674e-05,
+      "loss": 1.2827,
+      "step": 3492
+    },
+    {
+      "epoch": 0.1270320398588937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9510273937310252e-05,
+      "loss": 1.1723,
+      "step": 3493
+    },
+    {
+      "epoch": 0.1270684074626323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9509909786511544e-05,
+      "loss": 1.121,
+      "step": 3494
+    },
+    {
+      "epoch": 0.12710477506637088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9509545503776335e-05,
+      "loss": 1.0879,
+      "step": 3495
+    },
+    {
+      "epoch": 0.12714114267010945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9509181089109673e-05,
+      "loss": 1.1685,
+      "step": 3496
+    },
+    {
+      "epoch": 0.12717751027384805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9508816542516613e-05,
+      "loss": 1.2543,
+      "step": 3497
+    },
+    {
+      "epoch": 0.12721387787758665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950845186400222e-05,
+      "loss": 0.9652,
+      "step": 3498
+    },
+    {
+      "epoch": 0.12725024548132524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950808705357154e-05,
+      "loss": 1.2795,
+      "step": 3499
+    },
+    {
+      "epoch": 0.12728661308506384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950772211122965e-05,
+      "loss": 1.1588,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1273229806888024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95073570369816e-05,
+      "loss": 1.2684,
+      "step": 3501
+    },
+    {
+      "epoch": 0.127359348292541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9506991830832463e-05,
+      "loss": 1.2327,
+      "step": 3502
+    },
+    {
+      "epoch": 0.1273957158962796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9506626492787304e-05,
+      "loss": 1.0891,
+      "step": 3503
+    },
+    {
+      "epoch": 0.1274320835000182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9506261022851187e-05,
+      "loss": 1.1772,
+      "step": 3504
+    },
+    {
+      "epoch": 0.12746845110375676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9505895421029187e-05,
+      "loss": 1.1436,
+      "step": 3505
+    },
+    {
+      "epoch": 0.12750481870749536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950552968732638e-05,
+      "loss": 1.1384,
+      "step": 3506
+    },
+    {
+      "epoch": 0.12754118631123396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9505163821747828e-05,
+      "loss": 1.1417,
+      "step": 3507
+    },
+    {
+      "epoch": 0.12757755391497255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950479782429862e-05,
+      "loss": 1.1599,
+      "step": 3508
+    },
+    {
+      "epoch": 0.12761392151871112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9504431694983822e-05,
+      "loss": 1.0452,
+      "step": 3509
+    },
+    {
+      "epoch": 0.12765028912244972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9504065433808523e-05,
+      "loss": 1.1904,
+      "step": 3510
+    },
+    {
+      "epoch": 0.1276866567261883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95036990407778e-05,
+      "loss": 1.2012,
+      "step": 3511
+    },
+    {
+      "epoch": 0.1277230243299269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9503332515896737e-05,
+      "loss": 1.2525,
+      "step": 3512
+    },
+    {
+      "epoch": 0.12775939193366548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9502965859170418e-05,
+      "loss": 1.2167,
+      "step": 3513
+    },
+    {
+      "epoch": 0.12779575953740407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9502599070603932e-05,
+      "loss": 1.1268,
+      "step": 3514
+    },
+    {
+      "epoch": 0.12783212714114267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9502232150202362e-05,
+      "loss": 1.2855,
+      "step": 3515
+    },
+    {
+      "epoch": 0.12786849474488127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9501865097970807e-05,
+      "loss": 1.2493,
+      "step": 3516
+    },
+    {
+      "epoch": 0.12790486234861986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9501497913914355e-05,
+      "loss": 1.1777,
+      "step": 3517
+    },
+    {
+      "epoch": 0.12794122995235843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.95011305980381e-05,
+      "loss": 0.9696,
+      "step": 3518
+    },
+    {
+      "epoch": 0.12797759755609703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950076315034714e-05,
+      "loss": 1.2348,
+      "step": 3519
+    },
+    {
+      "epoch": 0.12801396515983562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9500395570846566e-05,
+      "loss": 1.1148,
+      "step": 3520
+    },
+    {
+      "epoch": 0.12805033276357422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9500027859541486e-05,
+      "loss": 1.1256,
+      "step": 3521
+    },
+    {
+      "epoch": 0.1280867003673128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9499660016437e-05,
+      "loss": 1.1913,
+      "step": 3522
+    },
+    {
+      "epoch": 0.12812306797105139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9499292041538207e-05,
+      "loss": 0.9951,
+      "step": 3523
+    },
+    {
+      "epoch": 0.12815943557478998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9498923934850213e-05,
+      "loss": 1.2651,
+      "step": 3524
+    },
+    {
+      "epoch": 0.12819580317852858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949855569637813e-05,
+      "loss": 1.1499,
+      "step": 3525
+    },
+    {
+      "epoch": 0.12823217078226715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949818732612706e-05,
+      "loss": 1.0844,
+      "step": 3526
+    },
+    {
+      "epoch": 0.12826853838600574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9497818824102123e-05,
+      "loss": 1.1741,
+      "step": 3527
+    },
+    {
+      "epoch": 0.12830490598974434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949745019030842e-05,
+      "loss": 1.1802,
+      "step": 3528
+    },
+    {
+      "epoch": 0.12834127359348294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9497081424751073e-05,
+      "loss": 1.2787,
+      "step": 3529
+    },
+    {
+      "epoch": 0.1283776411972215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9496712527435194e-05,
+      "loss": 1.0519,
+      "step": 3530
+    },
+    {
+      "epoch": 0.1284140088009601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9496343498365905e-05,
+      "loss": 1.1521,
+      "step": 3531
+    },
+    {
+      "epoch": 0.1284503764046987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9495974337548325e-05,
+      "loss": 1.1996,
+      "step": 3532
+    },
+    {
+      "epoch": 0.1284867440084373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949560504498757e-05,
+      "loss": 1.2043,
+      "step": 3533
+    },
+    {
+      "epoch": 0.12852311161217586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949523562068877e-05,
+      "loss": 1.1732,
+      "step": 3534
+    },
+    {
+      "epoch": 0.12855947921591446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9494866064657046e-05,
+      "loss": 1.0938,
+      "step": 3535
+    },
+    {
+      "epoch": 0.12859584681965305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949449637689753e-05,
+      "loss": 1.0968,
+      "step": 3536
+    },
+    {
+      "epoch": 0.12863221442339165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949412655741534e-05,
+      "loss": 1.0474,
+      "step": 3537
+    },
+    {
+      "epoch": 0.12866858202713025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9493756606215625e-05,
+      "loss": 1.2127,
+      "step": 3538
+    },
+    {
+      "epoch": 0.12870494963086881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9493386523303498e-05,
+      "loss": 1.1408,
+      "step": 3539
+    },
+    {
+      "epoch": 0.1287413172346074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9493016308684107e-05,
+      "loss": 1.1374,
+      "step": 3540
+    },
+    {
+      "epoch": 0.128777684838346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9492645962362582e-05,
+      "loss": 1.174,
+      "step": 3541
+    },
+    {
+      "epoch": 0.1288140524420846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9492275484344066e-05,
+      "loss": 1.1777,
+      "step": 3542
+    },
+    {
+      "epoch": 0.12885042004582317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9491904874633692e-05,
+      "loss": 1.0569,
+      "step": 3543
+    },
+    {
+      "epoch": 0.12888678764956177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9491534133236606e-05,
+      "loss": 1.1743,
+      "step": 3544
+    },
+    {
+      "epoch": 0.12892315525330036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949116326015795e-05,
+      "loss": 1.1068,
+      "step": 3545
+    },
+    {
+      "epoch": 0.12895952285703896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9490792255402868e-05,
+      "loss": 1.0904,
+      "step": 3546
+    },
+    {
+      "epoch": 0.12899589046077753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9490421118976514e-05,
+      "loss": 1.1916,
+      "step": 3547
+    },
+    {
+      "epoch": 0.12903225806451613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949004985088403e-05,
+      "loss": 0.9884,
+      "step": 3548
+    },
+    {
+      "epoch": 0.12906862566825472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9489678451130568e-05,
+      "loss": 1.0841,
+      "step": 3549
+    },
+    {
+      "epoch": 0.12910499327199332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948930691972128e-05,
+      "loss": 1.0764,
+      "step": 3550
+    },
+    {
+      "epoch": 0.1291413608757319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9488935256661325e-05,
+      "loss": 1.1273,
+      "step": 3551
+    },
+    {
+      "epoch": 0.12917772847947048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9488563461955856e-05,
+      "loss": 1.2384,
+      "step": 3552
+    },
+    {
+      "epoch": 0.12921409608320908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9488191535610032e-05,
+      "loss": 1.1629,
+      "step": 3553
+    },
+    {
+      "epoch": 0.12925046368694768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948781947762901e-05,
+      "loss": 1.233,
+      "step": 3554
+    },
+    {
+      "epoch": 0.12928683129068624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9487447288017956e-05,
+      "loss": 1.0856,
+      "step": 3555
+    },
+    {
+      "epoch": 0.12932319889442484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948707496678203e-05,
+      "loss": 1.2526,
+      "step": 3556
+    },
+    {
+      "epoch": 0.12935956649816344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9486702513926402e-05,
+      "loss": 1.1734,
+      "step": 3557
+    },
+    {
+      "epoch": 0.12939593410190203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9486329929456235e-05,
+      "loss": 0.9968,
+      "step": 3558
+    },
+    {
+      "epoch": 0.12943230170564063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9485957213376702e-05,
+      "loss": 1.2005,
+      "step": 3559
+    },
+    {
+      "epoch": 0.1294686693093792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9485584365692968e-05,
+      "loss": 1.2244,
+      "step": 3560
+    },
+    {
+      "epoch": 0.1295050369131178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948521138641021e-05,
+      "loss": 1.1907,
+      "step": 3561
+    },
+    {
+      "epoch": 0.1295414045168564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94848382755336e-05,
+      "loss": 1.1377,
+      "step": 3562
+    },
+    {
+      "epoch": 0.129577772120595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948446503306832e-05,
+      "loss": 1.1949,
+      "step": 3563
+    },
+    {
+      "epoch": 0.12961413972433355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9484091659019545e-05,
+      "loss": 1.1622,
+      "step": 3564
+    },
+    {
+      "epoch": 0.12965050732807215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948371815339245e-05,
+      "loss": 1.2259,
+      "step": 3565
+    },
+    {
+      "epoch": 0.12968687493181075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9483344516192223e-05,
+      "loss": 1.0264,
+      "step": 3566
+    },
+    {
+      "epoch": 0.12972324253554934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9482970747424045e-05,
+      "loss": 1.2052,
+      "step": 3567
+    },
+    {
+      "epoch": 0.1297596101392879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9482596847093105e-05,
+      "loss": 1.1853,
+      "step": 3568
+    },
+    {
+      "epoch": 0.1297959777430265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9482222815204585e-05,
+      "loss": 1.1277,
+      "step": 3569
+    },
+    {
+      "epoch": 0.1298323453467651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948184865176368e-05,
+      "loss": 1.2823,
+      "step": 3570
+    },
+    {
+      "epoch": 0.1298687129505037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948147435677558e-05,
+      "loss": 1.2704,
+      "step": 3571
+    },
+    {
+      "epoch": 0.12990508055424227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9481099930245467e-05,
+      "loss": 1.1482,
+      "step": 3572
+    },
+    {
+      "epoch": 0.12994144815798087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9480725372178552e-05,
+      "loss": 1.1001,
+      "step": 3573
+    },
+    {
+      "epoch": 0.12997781576171946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.948035068258002e-05,
+      "loss": 1.2006,
+      "step": 3574
+    },
+    {
+      "epoch": 0.13001418336545806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9479975861455077e-05,
+      "loss": 1.1405,
+      "step": 3575
+    },
+    {
+      "epoch": 0.13005055096919663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9479600908808916e-05,
+      "loss": 1.0395,
+      "step": 3576
+    },
+    {
+      "epoch": 0.13008691857293522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947922582464674e-05,
+      "loss": 1.0909,
+      "step": 3577
+    },
+    {
+      "epoch": 0.13012328617667382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9478850608973762e-05,
+      "loss": 1.2145,
+      "step": 3578
+    },
+    {
+      "epoch": 0.13015965378041242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9478475261795175e-05,
+      "loss": 1.1242,
+      "step": 3579
+    },
+    {
+      "epoch": 0.130196021384151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9478099783116193e-05,
+      "loss": 1.092,
+      "step": 3580
+    },
+    {
+      "epoch": 0.13023238898788958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9477724172942024e-05,
+      "loss": 1.1246,
+      "step": 3581
+    },
+    {
+      "epoch": 0.13026875659162818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9477348431277883e-05,
+      "loss": 1.1832,
+      "step": 3582
+    },
+    {
+      "epoch": 0.13030512419536677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9476972558128976e-05,
+      "loss": 1.1119,
+      "step": 3583
+    },
+    {
+      "epoch": 0.13034149179910537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9476596553500518e-05,
+      "loss": 1.1071,
+      "step": 3584
+    },
+    {
+      "epoch": 0.13037785940284394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947622041739773e-05,
+      "loss": 1.1051,
+      "step": 3585
+    },
+    {
+      "epoch": 0.13041422700658253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947584414982583e-05,
+      "loss": 1.2023,
+      "step": 3586
+    },
+    {
+      "epoch": 0.13045059461032113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9475467750790038e-05,
+      "loss": 1.226,
+      "step": 3587
+    },
+    {
+      "epoch": 0.13048696221405973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9475091220295575e-05,
+      "loss": 1.1044,
+      "step": 3588
+    },
+    {
+      "epoch": 0.1305233298177983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9474714558347663e-05,
+      "loss": 1.2103,
+      "step": 3589
+    },
+    {
+      "epoch": 0.1305596974215369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947433776495153e-05,
+      "loss": 1.2513,
+      "step": 3590
+    },
+    {
+      "epoch": 0.1305960650252755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94739608401124e-05,
+      "loss": 1.1146,
+      "step": 3591
+    },
+    {
+      "epoch": 0.13063243262901408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9473583783835507e-05,
+      "loss": 1.1694,
+      "step": 3592
+    },
+    {
+      "epoch": 0.13066880023275265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9473206596126083e-05,
+      "loss": 1.2655,
+      "step": 3593
+    },
+    {
+      "epoch": 0.13070516783649125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9472829276989356e-05,
+      "loss": 1.254,
+      "step": 3594
+    },
+    {
+      "epoch": 0.13074153544022984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9472451826430564e-05,
+      "loss": 1.1172,
+      "step": 3595
+    },
+    {
+      "epoch": 0.13077790304396844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947207424445494e-05,
+      "loss": 1.2227,
+      "step": 3596
+    },
+    {
+      "epoch": 0.130814270647707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9471696531067728e-05,
+      "loss": 1.1458,
+      "step": 3597
+    },
+    {
+      "epoch": 0.1308506382514456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9471318686274165e-05,
+      "loss": 1.1389,
+      "step": 3598
+    },
+    {
+      "epoch": 0.1308870058551842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947094071007949e-05,
+      "loss": 1.0787,
+      "step": 3599
+    },
+    {
+      "epoch": 0.1309233734589228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9470562602488954e-05,
+      "loss": 1.1031,
+      "step": 3600
+    },
+    {
+      "epoch": 0.1309597410626614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9470184363507796e-05,
+      "loss": 1.2056,
+      "step": 3601
+    },
+    {
+      "epoch": 0.13099610866639996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946980599314127e-05,
+      "loss": 1.2081,
+      "step": 3602
+    },
+    {
+      "epoch": 0.13103247627013856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946942749139462e-05,
+      "loss": 1.1958,
+      "step": 3603
+    },
+    {
+      "epoch": 0.13106884387387716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9469048858273096e-05,
+      "loss": 1.1862,
+      "step": 3604
+    },
+    {
+      "epoch": 0.13110521147761575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9468670093781953e-05,
+      "loss": 1.0761,
+      "step": 3605
+    },
+    {
+      "epoch": 0.13114157908135432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9468291197926453e-05,
+      "loss": 1.0031,
+      "step": 3606
+    },
+    {
+      "epoch": 0.13117794668509292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9467912170711843e-05,
+      "loss": 1.1842,
+      "step": 3607
+    },
+    {
+      "epoch": 0.1312143142888315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9467533012143386e-05,
+      "loss": 1.16,
+      "step": 3608
+    },
+    {
+      "epoch": 0.1312506818925701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946715372222634e-05,
+      "loss": 1.2094,
+      "step": 3609
+    },
+    {
+      "epoch": 0.13128704949630868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9466774300965968e-05,
+      "loss": 1.2208,
+      "step": 3610
+    },
+    {
+      "epoch": 0.13132341710004727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9466394748367535e-05,
+      "loss": 1.0848,
+      "step": 3611
+    },
+    {
+      "epoch": 0.13135978470378587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9466015064436304e-05,
+      "loss": 1.1439,
+      "step": 3612
+    },
+    {
+      "epoch": 0.13139615230752447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9465635249177545e-05,
+      "loss": 1.1484,
+      "step": 3613
+    },
+    {
+      "epoch": 0.13143251991126303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946525530259653e-05,
+      "loss": 1.1797,
+      "step": 3614
+    },
+    {
+      "epoch": 0.13146888751500163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9464875224698524e-05,
+      "loss": 1.0859,
+      "step": 3615
+    },
+    {
+      "epoch": 0.13150525511874023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9464495015488803e-05,
+      "loss": 1.37,
+      "step": 3616
+    },
+    {
+      "epoch": 0.13154162272247882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9464114674972645e-05,
+      "loss": 1.0964,
+      "step": 3617
+    },
+    {
+      "epoch": 0.1315779903262174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946373420315532e-05,
+      "loss": 1.1467,
+      "step": 3618
+    },
+    {
+      "epoch": 0.131614357929956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9463353600042113e-05,
+      "loss": 1.222,
+      "step": 3619
+    },
+    {
+      "epoch": 0.13165072553369458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94629728656383e-05,
+      "loss": 1.1378,
+      "step": 3620
+    },
+    {
+      "epoch": 0.13168709313743318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9462591999949168e-05,
+      "loss": 1.1268,
+      "step": 3621
+    },
+    {
+      "epoch": 0.13172346074117178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9462211002979992e-05,
+      "loss": 1.1494,
+      "step": 3622
+    },
+    {
+      "epoch": 0.13175982834491035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9461829874736068e-05,
+      "loss": 1.2181,
+      "step": 3623
+    },
+    {
+      "epoch": 0.13179619594864894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9461448615222677e-05,
+      "loss": 1.1952,
+      "step": 3624
+    },
+    {
+      "epoch": 0.13183256355238754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9461067224445112e-05,
+      "loss": 1.1617,
+      "step": 3625
+    },
+    {
+      "epoch": 0.13186893115612613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9460685702408662e-05,
+      "loss": 1.1938,
+      "step": 3626
+    },
+    {
+      "epoch": 0.1319052987598647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946030404911862e-05,
+      "loss": 1.2438,
+      "step": 3627
+    },
+    {
+      "epoch": 0.1319416663636033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9459922264580282e-05,
+      "loss": 1.263,
+      "step": 3628
+    },
+    {
+      "epoch": 0.1319780339673419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9459540348798947e-05,
+      "loss": 1.0865,
+      "step": 3629
+    },
+    {
+      "epoch": 0.1320144015710805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945915830177991e-05,
+      "loss": 1.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.13205076917481906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9458776123528468e-05,
+      "loss": 1.1553,
+      "step": 3631
+    },
+    {
+      "epoch": 0.13208713677855766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945839381404993e-05,
+      "loss": 1.1743,
+      "step": 3632
+    },
+    {
+      "epoch": 0.13212350438229625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9458011373349598e-05,
+      "loss": 1.1496,
+      "step": 3633
+    },
+    {
+      "epoch": 0.13215987198603485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9457628801432778e-05,
+      "loss": 1.105,
+      "step": 3634
+    },
+    {
+      "epoch": 0.13219623958977342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9457246098304777e-05,
+      "loss": 1.1584,
+      "step": 3635
+    },
+    {
+      "epoch": 0.13223260719351201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9456863263970905e-05,
+      "loss": 1.069,
+      "step": 3636
+    },
+    {
+      "epoch": 0.1322689747972506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945648029843647e-05,
+      "loss": 1.1961,
+      "step": 3637
+    },
+    {
+      "epoch": 0.1323053424009892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9456097201706788e-05,
+      "loss": 1.1158,
+      "step": 3638
+    },
+    {
+      "epoch": 0.13234171000472778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945571397378718e-05,
+      "loss": 1.218,
+      "step": 3639
+    },
+    {
+      "epoch": 0.13237807760846637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945533061468295e-05,
+      "loss": 1.2609,
+      "step": 3640
+    },
+    {
+      "epoch": 0.13241444521220497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9454947124399424e-05,
+      "loss": 1.2303,
+      "step": 3641
+    },
+    {
+      "epoch": 0.13245081281594356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9454563502941924e-05,
+      "loss": 1.1549,
+      "step": 3642
+    },
+    {
+      "epoch": 0.13248718041968216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9454179750315768e-05,
+      "loss": 1.1511,
+      "step": 3643
+    },
+    {
+      "epoch": 0.13252354802342073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9453795866526282e-05,
+      "loss": 1.1894,
+      "step": 3644
+    },
+    {
+      "epoch": 0.13255991562715932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9453411851578793e-05,
+      "loss": 1.1418,
+      "step": 3645
+    },
+    {
+      "epoch": 0.13259628323089792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9453027705478623e-05,
+      "loss": 1.1435,
+      "step": 3646
+    },
+    {
+      "epoch": 0.13263265083463652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9452643428231108e-05,
+      "loss": 1.168,
+      "step": 3647
+    },
+    {
+      "epoch": 0.13266901843837509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945225901984158e-05,
+      "loss": 1.151,
+      "step": 3648
+    },
+    {
+      "epoch": 0.13270538604211368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9451874480315364e-05,
+      "loss": 1.1355,
+      "step": 3649
+    },
+    {
+      "epoch": 0.13274175364585228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9451489809657803e-05,
+      "loss": 0.9921,
+      "step": 3650
+    },
+    {
+      "epoch": 0.13277812124959087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945110500787423e-05,
+      "loss": 1.1116,
+      "step": 3651
+    },
+    {
+      "epoch": 0.13281448885332944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9450720074969983e-05,
+      "loss": 1.2815,
+      "step": 3652
+    },
+    {
+      "epoch": 0.13285085645706804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9450335010950406e-05,
+      "loss": 1.1582,
+      "step": 3653
+    },
+    {
+      "epoch": 0.13288722406080664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9449949815820843e-05,
+      "loss": 1.1399,
+      "step": 3654
+    },
+    {
+      "epoch": 0.13292359166454523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9449564489586625e-05,
+      "loss": 1.3527,
+      "step": 3655
+    },
+    {
+      "epoch": 0.1329599592682838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9449179032253113e-05,
+      "loss": 1.1212,
+      "step": 3656
+    },
+    {
+      "epoch": 0.1329963268720224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9448793443825647e-05,
+      "loss": 1.1946,
+      "step": 3657
+    },
+    {
+      "epoch": 0.133032694475761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9448407724309577e-05,
+      "loss": 1.1177,
+      "step": 3658
+    },
+    {
+      "epoch": 0.1330690620794996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944802187371026e-05,
+      "loss": 1.2117,
+      "step": 3659
+    },
+    {
+      "epoch": 0.13310542968323816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9447635892033035e-05,
+      "loss": 1.2034,
+      "step": 3660
+    },
+    {
+      "epoch": 0.13314179728697675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9447249779283274e-05,
+      "loss": 1.2153,
+      "step": 3661
+    },
+    {
+      "epoch": 0.13317816489071535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9446863535466325e-05,
+      "loss": 1.1215,
+      "step": 3662
+    },
+    {
+      "epoch": 0.13321453249445395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9446477160587548e-05,
+      "loss": 1.2443,
+      "step": 3663
+    },
+    {
+      "epoch": 0.13325090009819254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94460906546523e-05,
+      "loss": 1.252,
+      "step": 3664
+    },
+    {
+      "epoch": 0.1332872677019311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944570401766595e-05,
+      "loss": 1.1111,
+      "step": 3665
+    },
+    {
+      "epoch": 0.1333236353056697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944531724963386e-05,
+      "loss": 1.1423,
+      "step": 3666
+    },
+    {
+      "epoch": 0.1333600029094083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944493035056139e-05,
+      "loss": 1.1493,
+      "step": 3667
+    },
+    {
+      "epoch": 0.1333963705131469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9444543320453913e-05,
+      "loss": 1.1444,
+      "step": 3668
+    },
+    {
+      "epoch": 0.13343273811688547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94441561593168e-05,
+      "loss": 1.1472,
+      "step": 3669
+    },
+    {
+      "epoch": 0.13346910572062406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9443768867155415e-05,
+      "loss": 1.2239,
+      "step": 3670
+    },
+    {
+      "epoch": 0.13350547332436266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9443381443975137e-05,
+      "loss": 1.281,
+      "step": 3671
+    },
+    {
+      "epoch": 0.13354184092810126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9442993889781342e-05,
+      "loss": 1.1754,
+      "step": 3672
+    },
+    {
+      "epoch": 0.13357820853183983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9442606204579408e-05,
+      "loss": 1.1041,
+      "step": 3673
+    },
+    {
+      "epoch": 0.13361457613557842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9442218388374704e-05,
+      "loss": 1.1782,
+      "step": 3674
+    },
+    {
+      "epoch": 0.13365094373931702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9441830441172617e-05,
+      "loss": 1.1701,
+      "step": 3675
+    },
+    {
+      "epoch": 0.13368731134305561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9441442362978532e-05,
+      "loss": 1.0289,
+      "step": 3676
+    },
+    {
+      "epoch": 0.13372367894679418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.944105415379783e-05,
+      "loss": 1.069,
+      "step": 3677
+    },
+    {
+      "epoch": 0.13376004655053278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9440665813635894e-05,
+      "loss": 1.1988,
+      "step": 3678
+    },
+    {
+      "epoch": 0.13379641415427138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9440277342498116e-05,
+      "loss": 1.1547,
+      "step": 3679
+    },
+    {
+      "epoch": 0.13383278175800997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9439888740389884e-05,
+      "loss": 1.3192,
+      "step": 3680
+    },
+    {
+      "epoch": 0.13386914936174854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943950000731659e-05,
+      "loss": 1.0583,
+      "step": 3681
+    },
+    {
+      "epoch": 0.13390551696548714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9439111143283624e-05,
+      "loss": 1.1704,
+      "step": 3682
+    },
+    {
+      "epoch": 0.13394188456922573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9438722148296385e-05,
+      "loss": 1.2914,
+      "step": 3683
+    },
+    {
+      "epoch": 0.13397825217296433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9438333022360267e-05,
+      "loss": 1.1496,
+      "step": 3684
+    },
+    {
+      "epoch": 0.13401461977670293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9437943765480673e-05,
+      "loss": 1.1048,
+      "step": 3685
+    },
+    {
+      "epoch": 0.1340509873804415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9437554377662995e-05,
+      "loss": 1.1694,
+      "step": 3686
+    },
+    {
+      "epoch": 0.1340873549841801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943716485891264e-05,
+      "loss": 1.194,
+      "step": 3687
+    },
+    {
+      "epoch": 0.1341237225879187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9436775209235015e-05,
+      "loss": 1.2173,
+      "step": 3688
+    },
+    {
+      "epoch": 0.13416009019165728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9436385428635525e-05,
+      "loss": 1.0951,
+      "step": 3689
+    },
+    {
+      "epoch": 0.13419645779539585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9435995517119572e-05,
+      "loss": 1.1687,
+      "step": 3690
+    },
+    {
+      "epoch": 0.13423282539913445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943560547469257e-05,
+      "loss": 1.197,
+      "step": 3691
+    },
+    {
+      "epoch": 0.13426919300287304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9435215301359927e-05,
+      "loss": 1.1935,
+      "step": 3692
+    },
+    {
+      "epoch": 0.13430556060661164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9434824997127065e-05,
+      "loss": 1.1194,
+      "step": 3693
+    },
+    {
+      "epoch": 0.1343419282103502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9434434561999386e-05,
+      "loss": 1.1146,
+      "step": 3694
+    },
+    {
+      "epoch": 0.1343782958140888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9434043995982317e-05,
+      "loss": 1.2062,
+      "step": 3695
+    },
+    {
+      "epoch": 0.1344146634178274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943365329908127e-05,
+      "loss": 1.2099,
+      "step": 3696
+    },
+    {
+      "epoch": 0.134451031021566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943326247130167e-05,
+      "loss": 1.0101,
+      "step": 3697
+    },
+    {
+      "epoch": 0.13448739862530457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9432871512648936e-05,
+      "loss": 1.1753,
+      "step": 3698
+    },
+    {
+      "epoch": 0.13452376622904316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9432480423128495e-05,
+      "loss": 1.1191,
+      "step": 3699
+    },
+    {
+      "epoch": 0.13456013383278176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943208920274577e-05,
+      "loss": 1.2291,
+      "step": 3700
+    },
+    {
+      "epoch": 0.13459650143652035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9431697851506187e-05,
+      "loss": 1.1267,
+      "step": 3701
+    },
+    {
+      "epoch": 0.13463286904025895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9431306369415183e-05,
+      "loss": 1.1257,
+      "step": 3702
+    },
+    {
+      "epoch": 0.13466923664399752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9430914756478182e-05,
+      "loss": 1.0444,
+      "step": 3703
+    },
+    {
+      "epoch": 0.13470560424773612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9430523012700618e-05,
+      "loss": 1.232,
+      "step": 3704
+    },
+    {
+      "epoch": 0.1347419718514747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.943013113808793e-05,
+      "loss": 1.2623,
+      "step": 3705
+    },
+    {
+      "epoch": 0.1347783394552133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942973913264555e-05,
+      "loss": 1.1472,
+      "step": 3706
+    },
+    {
+      "epoch": 0.13481470705895188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9429346996378923e-05,
+      "loss": 1.2989,
+      "step": 3707
+    },
+    {
+      "epoch": 0.13485107466269047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9428954729293478e-05,
+      "loss": 1.0917,
+      "step": 3708
+    },
+    {
+      "epoch": 0.13488744226642907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942856233139467e-05,
+      "loss": 1.0623,
+      "step": 3709
+    },
+    {
+      "epoch": 0.13492380987016767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9428169802687937e-05,
+      "loss": 1.2678,
+      "step": 3710
+    },
+    {
+      "epoch": 0.13496017747390623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942777714317872e-05,
+      "loss": 1.153,
+      "step": 3711
+    },
+    {
+      "epoch": 0.13499654507764483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9427384352872475e-05,
+      "loss": 1.0875,
+      "step": 3712
+    },
+    {
+      "epoch": 0.13503291268138343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9426991431774646e-05,
+      "loss": 1.2821,
+      "step": 3713
+    },
+    {
+      "epoch": 0.13506928028512202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9426598379890685e-05,
+      "loss": 1.1774,
+      "step": 3714
+    },
+    {
+      "epoch": 0.1351056478888606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942620519722605e-05,
+      "loss": 1.1216,
+      "step": 3715
+    },
+    {
+      "epoch": 0.1351420154925992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9425811883786188e-05,
+      "loss": 1.2829,
+      "step": 3716
+    },
+    {
+      "epoch": 0.13517838309633778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9425418439576562e-05,
+      "loss": 1.1694,
+      "step": 3717
+    },
+    {
+      "epoch": 0.13521475070007638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9425024864602627e-05,
+      "loss": 1.0956,
+      "step": 3718
+    },
+    {
+      "epoch": 0.13525111830381495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9424631158869847e-05,
+      "loss": 1.1686,
+      "step": 3719
+    },
+    {
+      "epoch": 0.13528748590755355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9424237322383678e-05,
+      "loss": 1.0993,
+      "step": 3720
+    },
+    {
+      "epoch": 0.13532385351129214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9423843355149588e-05,
+      "loss": 1.1783,
+      "step": 3721
+    },
+    {
+      "epoch": 0.13536022111503074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9423449257173042e-05,
+      "loss": 1.2109,
+      "step": 3722
+    },
+    {
+      "epoch": 0.13539658871876933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942305502845951e-05,
+      "loss": 1.1288,
+      "step": 3723
+    },
+    {
+      "epoch": 0.1354329563225079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9422660669014455e-05,
+      "loss": 1.1747,
+      "step": 3724
+    },
+    {
+      "epoch": 0.1354693239262465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9422266178843355e-05,
+      "loss": 1.0774,
+      "step": 3725
+    },
+    {
+      "epoch": 0.1355056915299851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942187155795168e-05,
+      "loss": 1.2126,
+      "step": 3726
+    },
+    {
+      "epoch": 0.1355420591337237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9421476806344904e-05,
+      "loss": 1.1151,
+      "step": 3727
+    },
+    {
+      "epoch": 0.13557842673746226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9421081924028507e-05,
+      "loss": 1.0594,
+      "step": 3728
+    },
+    {
+      "epoch": 0.13561479434120086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9420686911007964e-05,
+      "loss": 1.2482,
+      "step": 3729
+    },
+    {
+      "epoch": 0.13565116194493945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9420291767288756e-05,
+      "loss": 1.1089,
+      "step": 3730
+    },
+    {
+      "epoch": 0.13568752954867805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9419896492876365e-05,
+      "loss": 1.1233,
+      "step": 3731
+    },
+    {
+      "epoch": 0.13572389715241662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9419501087776278e-05,
+      "loss": 1.1625,
+      "step": 3732
+    },
+    {
+      "epoch": 0.1357602647561552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9419105551993972e-05,
+      "loss": 1.2528,
+      "step": 3733
+    },
+    {
+      "epoch": 0.1357966323598938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9418709885534944e-05,
+      "loss": 1.2428,
+      "step": 3734
+    },
+    {
+      "epoch": 0.1358329999636324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941831408840468e-05,
+      "loss": 1.1845,
+      "step": 3735
+    },
+    {
+      "epoch": 0.13586936756737097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9417918160608673e-05,
+      "loss": 1.1002,
+      "step": 3736
+    },
+    {
+      "epoch": 0.13590573517110957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941752210215241e-05,
+      "loss": 1.0629,
+      "step": 3737
+    },
+    {
+      "epoch": 0.13594210277484817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941712591304139e-05,
+      "loss": 1.3148,
+      "step": 3738
+    },
+    {
+      "epoch": 0.13597847037858676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9416729593281113e-05,
+      "loss": 1.0731,
+      "step": 3739
+    },
+    {
+      "epoch": 0.13601483798232533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941633314287707e-05,
+      "loss": 1.1609,
+      "step": 3740
+    },
+    {
+      "epoch": 0.13605120558606393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9415936561834765e-05,
+      "loss": 1.1762,
+      "step": 3741
+    },
+    {
+      "epoch": 0.13608757318980252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94155398501597e-05,
+      "loss": 1.0633,
+      "step": 3742
+    },
+    {
+      "epoch": 0.13612394079354112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9415143007857377e-05,
+      "loss": 1.129,
+      "step": 3743
+    },
+    {
+      "epoch": 0.13616030839727972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9414746034933304e-05,
+      "loss": 1.1412,
+      "step": 3744
+    },
+    {
+      "epoch": 0.13619667600101829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941434893139299e-05,
+      "loss": 1.1866,
+      "step": 3745
+    },
+    {
+      "epoch": 0.13623304360475688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941395169724194e-05,
+      "loss": 1.1625,
+      "step": 3746
+    },
+    {
+      "epoch": 0.13626941120849548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9413554332485667e-05,
+      "loss": 1.0508,
+      "step": 3747
+    },
+    {
+      "epoch": 0.13630577881223407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941315683712968e-05,
+      "loss": 0.856,
+      "step": 3748
+    },
+    {
+      "epoch": 0.13634214641597264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9412759211179503e-05,
+      "loss": 1.0516,
+      "step": 3749
+    },
+    {
+      "epoch": 0.13637851401971124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9412361454640646e-05,
+      "loss": 1.2448,
+      "step": 3750
+    },
+    {
+      "epoch": 0.13641488162344984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9411963567518626e-05,
+      "loss": 1.0521,
+      "step": 3751
+    },
+    {
+      "epoch": 0.13645124922718843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9411565549818966e-05,
+      "loss": 1.2021,
+      "step": 3752
+    },
+    {
+      "epoch": 0.136487616830927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9411167401547188e-05,
+      "loss": 1.1344,
+      "step": 3753
+    },
+    {
+      "epoch": 0.1365239844346656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9410769122708815e-05,
+      "loss": 1.1746,
+      "step": 3754
+    },
+    {
+      "epoch": 0.1365603520384042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.941037071330937e-05,
+      "loss": 1.1218,
+      "step": 3755
+    },
+    {
+      "epoch": 0.1365967196421428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9409972173354385e-05,
+      "loss": 1.2643,
+      "step": 3756
+    },
+    {
+      "epoch": 0.13663308724588136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9409573502849387e-05,
+      "loss": 1.0695,
+      "step": 3757
+    },
+    {
+      "epoch": 0.13666945484961995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940917470179991e-05,
+      "loss": 1.2331,
+      "step": 3758
+    },
+    {
+      "epoch": 0.13670582245335855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9408775770211478e-05,
+      "loss": 1.1981,
+      "step": 3759
+    },
+    {
+      "epoch": 0.13674219005709715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9408376708089637e-05,
+      "loss": 1.1892,
+      "step": 3760
+    },
+    {
+      "epoch": 0.13677855766083571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9407977515439916e-05,
+      "loss": 1.1154,
+      "step": 3761
+    },
+    {
+      "epoch": 0.1368149252645743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9407578192267853e-05,
+      "loss": 1.1832,
+      "step": 3762
+    },
+    {
+      "epoch": 0.1368512928683129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940717873857899e-05,
+      "loss": 1.1603,
+      "step": 3763
+    },
+    {
+      "epoch": 0.1368876604720515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940677915437887e-05,
+      "loss": 1.1109,
+      "step": 3764
+    },
+    {
+      "epoch": 0.1369240280757901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9406379439673038e-05,
+      "loss": 1.2609,
+      "step": 3765
+    },
+    {
+      "epoch": 0.13696039567952867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9405979594467034e-05,
+      "loss": 1.166,
+      "step": 3766
+    },
+    {
+      "epoch": 0.13699676328326726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9405579618766413e-05,
+      "loss": 1.2481,
+      "step": 3767
+    },
+    {
+      "epoch": 0.13703313088700586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9405179512576717e-05,
+      "loss": 1.0705,
+      "step": 3768
+    },
+    {
+      "epoch": 0.13706949849074446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94047792759035e-05,
+      "loss": 1.1822,
+      "step": 3769
+    },
+    {
+      "epoch": 0.13710586609448303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940437890875231e-05,
+      "loss": 1.1437,
+      "step": 3770
+    },
+    {
+      "epoch": 0.13714223369822162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940397841112871e-05,
+      "loss": 1.1151,
+      "step": 3771
+    },
+    {
+      "epoch": 0.13717860130196022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940357778303825e-05,
+      "loss": 1.1602,
+      "step": 3772
+    },
+    {
+      "epoch": 0.13721496890569881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9403177024486494e-05,
+      "loss": 1.2784,
+      "step": 3773
+    },
+    {
+      "epoch": 0.13725133650943738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9402776135478994e-05,
+      "loss": 1.1114,
+      "step": 3774
+    },
+    {
+      "epoch": 0.13728770411317598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940237511602132e-05,
+      "loss": 1.2918,
+      "step": 3775
+    },
+    {
+      "epoch": 0.13732407171691458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9401973966119028e-05,
+      "loss": 1.1118,
+      "step": 3776
+    },
+    {
+      "epoch": 0.13736043932065317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940157268577769e-05,
+      "loss": 1.1508,
+      "step": 3777
+    },
+    {
+      "epoch": 0.13739680692439174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9401171275002864e-05,
+      "loss": 1.0126,
+      "step": 3778
+    },
+    {
+      "epoch": 0.13743317452813034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940076973380013e-05,
+      "loss": 1.0363,
+      "step": 3779
+    },
+    {
+      "epoch": 0.13746954213186893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940036806217505e-05,
+      "loss": 1.1114,
+      "step": 3780
+    },
+    {
+      "epoch": 0.13750590973560753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9399966260133208e-05,
+      "loss": 1.2084,
+      "step": 3781
+    },
+    {
+      "epoch": 0.1375422773393461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9399564327680165e-05,
+      "loss": 1.1358,
+      "step": 3782
+    },
+    {
+      "epoch": 0.1375786449430847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9399162264821502e-05,
+      "loss": 1.2065,
+      "step": 3783
+    },
+    {
+      "epoch": 0.1376150125468233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93987600715628e-05,
+      "loss": 1.0585,
+      "step": 3784
+    },
+    {
+      "epoch": 0.1376513801505619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939835774790964e-05,
+      "loss": 1.3329,
+      "step": 3785
+    },
+    {
+      "epoch": 0.13768774775430048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93979552938676e-05,
+      "loss": 1.2236,
+      "step": 3786
+    },
+    {
+      "epoch": 0.13772411535803905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939755270944226e-05,
+      "loss": 1.2469,
+      "step": 3787
+    },
+    {
+      "epoch": 0.13776048296177765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9397149994639214e-05,
+      "loss": 1.0808,
+      "step": 3788
+    },
+    {
+      "epoch": 0.13779685056551624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9396747149464044e-05,
+      "loss": 1.1525,
+      "step": 3789
+    },
+    {
+      "epoch": 0.13783321816925484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9396344173922335e-05,
+      "loss": 1.2263,
+      "step": 3790
+    },
+    {
+      "epoch": 0.1378695857729934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9395941068019687e-05,
+      "loss": 0.9765,
+      "step": 3791
+    },
+    {
+      "epoch": 0.137905953376732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9395537831761687e-05,
+      "loss": 1.183,
+      "step": 3792
+    },
+    {
+      "epoch": 0.1379423209804706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939513446515393e-05,
+      "loss": 1.1216,
+      "step": 3793
+    },
+    {
+      "epoch": 0.1379786885842092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9394730968202014e-05,
+      "loss": 1.2238,
+      "step": 3794
+    },
+    {
+      "epoch": 0.13801505618794777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9394327340911532e-05,
+      "loss": 1.1766,
+      "step": 3795
+    },
+    {
+      "epoch": 0.13805142379168636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9393923583288087e-05,
+      "loss": 1.1266,
+      "step": 3796
+    },
+    {
+      "epoch": 0.13808779139542496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9393519695337283e-05,
+      "loss": 1.1508,
+      "step": 3797
+    },
+    {
+      "epoch": 0.13812415899916355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9393115677064717e-05,
+      "loss": 1.1728,
+      "step": 3798
+    },
+    {
+      "epoch": 0.13816052660290212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9392711528476004e-05,
+      "loss": 0.9807,
+      "step": 3799
+    },
+    {
+      "epoch": 0.13819689420664072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939230724957674e-05,
+      "loss": 1.0307,
+      "step": 3800
+    },
+    {
+      "epoch": 0.13823326181037932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939190284037254e-05,
+      "loss": 1.0523,
+      "step": 3801
+    },
+    {
+      "epoch": 0.1382696294141179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9391498300869017e-05,
+      "loss": 1.1439,
+      "step": 3802
+    },
+    {
+      "epoch": 0.13830599701785648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9391093631071776e-05,
+      "loss": 1.1476,
+      "step": 3803
+    },
+    {
+      "epoch": 0.13834236462159508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9390688830986434e-05,
+      "loss": 1.2255,
+      "step": 3804
+    },
+    {
+      "epoch": 0.13837873222533367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9390283900618613e-05,
+      "loss": 1.1319,
+      "step": 3805
+    },
+    {
+      "epoch": 0.13841509982907227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938987883997392e-05,
+      "loss": 1.1992,
+      "step": 3806
+    },
+    {
+      "epoch": 0.13845146743281087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9389473649057984e-05,
+      "loss": 1.197,
+      "step": 3807
+    },
+    {
+      "epoch": 0.13848783503654943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9389068327876425e-05,
+      "loss": 1.1937,
+      "step": 3808
+    },
+    {
+      "epoch": 0.13852420264028803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938866287643486e-05,
+      "loss": 1.1147,
+      "step": 3809
+    },
+    {
+      "epoch": 0.13856057024402663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9388257294738922e-05,
+      "loss": 1.1429,
+      "step": 3810
+    },
+    {
+      "epoch": 0.13859693784776522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938785158279423e-05,
+      "loss": 1.1782,
+      "step": 3811
+    },
+    {
+      "epoch": 0.1386333054515038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9387445740606422e-05,
+      "loss": 1.119,
+      "step": 3812
+    },
+    {
+      "epoch": 0.1386696730552424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9387039768181118e-05,
+      "loss": 1.2203,
+      "step": 3813
+    },
+    {
+      "epoch": 0.13870604065898098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938663366552396e-05,
+      "loss": 1.2529,
+      "step": 3814
+    },
+    {
+      "epoch": 0.13874240826271958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9386227432640575e-05,
+      "loss": 1.157,
+      "step": 3815
+    },
+    {
+      "epoch": 0.13877877586645815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9385821069536605e-05,
+      "loss": 1.1172,
+      "step": 3816
+    },
+    {
+      "epoch": 0.13881514347019674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938541457621768e-05,
+      "loss": 1.1707,
+      "step": 3817
+    },
+    {
+      "epoch": 0.13885151107393534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9385007952689446e-05,
+      "loss": 1.223,
+      "step": 3818
+    },
+    {
+      "epoch": 0.13888787867767394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9384601198957543e-05,
+      "loss": 1.2101,
+      "step": 3819
+    },
+    {
+      "epoch": 0.1389242462814125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9384194315027613e-05,
+      "loss": 1.1846,
+      "step": 3820
+    },
+    {
+      "epoch": 0.1389606138851511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93837873009053e-05,
+      "loss": 1.1569,
+      "step": 3821
+    },
+    {
+      "epoch": 0.1389969814888897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9383380156596253e-05,
+      "loss": 1.2137,
+      "step": 3822
+    },
+    {
+      "epoch": 0.1390333490926283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9382972882106122e-05,
+      "loss": 1.2173,
+      "step": 3823
+    },
+    {
+      "epoch": 0.13906971669636686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938256547744055e-05,
+      "loss": 1.1147,
+      "step": 3824
+    },
+    {
+      "epoch": 0.13910608430010546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9382157942605197e-05,
+      "loss": 1.0499,
+      "step": 3825
+    },
+    {
+      "epoch": 0.13914245190384406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9381750277605715e-05,
+      "loss": 1.0857,
+      "step": 3826
+    },
+    {
+      "epoch": 0.13917881950758265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938134248244776e-05,
+      "loss": 1.0827,
+      "step": 3827
+    },
+    {
+      "epoch": 0.13921518711132125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9380934557136984e-05,
+      "loss": 1.1615,
+      "step": 3828
+    },
+    {
+      "epoch": 0.13925155471505982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9380526501679054e-05,
+      "loss": 1.084,
+      "step": 3829
+    },
+    {
+      "epoch": 0.1392879223187984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9380118316079628e-05,
+      "loss": 1.0886,
+      "step": 3830
+    },
+    {
+      "epoch": 0.139324289922537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9379710000344367e-05,
+      "loss": 1.0961,
+      "step": 3831
+    },
+    {
+      "epoch": 0.1393606575262756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937930155447894e-05,
+      "loss": 1.2534,
+      "step": 3832
+    },
+    {
+      "epoch": 0.13939702513001417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937889297848901e-05,
+      "loss": 1.1617,
+      "step": 3833
+    },
+    {
+      "epoch": 0.13943339273375277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9378484272380246e-05,
+      "loss": 1.0909,
+      "step": 3834
+    },
+    {
+      "epoch": 0.13946976033749137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937807543615832e-05,
+      "loss": 1.1733,
+      "step": 3835
+    },
+    {
+      "epoch": 0.13950612794122996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9377666469828904e-05,
+      "loss": 1.0342,
+      "step": 3836
+    },
+    {
+      "epoch": 0.13954249554496853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937725737339767e-05,
+      "loss": 1.1411,
+      "step": 3837
+    },
+    {
+      "epoch": 0.13957886314870713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9376848146870294e-05,
+      "loss": 1.1394,
+      "step": 3838
+    },
+    {
+      "epoch": 0.13961523075244572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9376438790252456e-05,
+      "loss": 1.203,
+      "step": 3839
+    },
+    {
+      "epoch": 0.13965159835618432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9376029303549833e-05,
+      "loss": 1.1638,
+      "step": 3840
+    },
+    {
+      "epoch": 0.1396879659599229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9375619686768107e-05,
+      "loss": 1.177,
+      "step": 3841
+    },
+    {
+      "epoch": 0.13972433356366148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9375209939912954e-05,
+      "loss": 1.2087,
+      "step": 3842
+    },
+    {
+      "epoch": 0.13976070116740008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9374800062990072e-05,
+      "loss": 1.0585,
+      "step": 3843
+    },
+    {
+      "epoch": 0.13979706877113868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9374390056005135e-05,
+      "loss": 1.1201,
+      "step": 3844
+    },
+    {
+      "epoch": 0.13983343637487725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9373979918963843e-05,
+      "loss": 1.134,
+      "step": 3845
+    },
+    {
+      "epoch": 0.13986980397861584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9373569651871873e-05,
+      "loss": 1.1107,
+      "step": 3846
+    },
+    {
+      "epoch": 0.13990617158235444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9373159254734924e-05,
+      "loss": 1.0773,
+      "step": 3847
+    },
+    {
+      "epoch": 0.13994253918609303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9372748727558692e-05,
+      "loss": 1.1485,
+      "step": 3848
+    },
+    {
+      "epoch": 0.13997890678983163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937233807034887e-05,
+      "loss": 1.0907,
+      "step": 3849
+    },
+    {
+      "epoch": 0.1400152743935702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937192728311115e-05,
+      "loss": 1.0777,
+      "step": 3850
+    },
+    {
+      "epoch": 0.1400516419973088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9371516365851243e-05,
+      "loss": 1.1052,
+      "step": 3851
+    },
+    {
+      "epoch": 0.1400880096010474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9371105318574836e-05,
+      "loss": 1.2875,
+      "step": 3852
+    },
+    {
+      "epoch": 0.140124377204786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9370694141287644e-05,
+      "loss": 1.2867,
+      "step": 3853
+    },
+    {
+      "epoch": 0.14016074480852456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9370282833995363e-05,
+      "loss": 1.1107,
+      "step": 3854
+    },
+    {
+      "epoch": 0.14019711241226315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9369871396703703e-05,
+      "loss": 1.143,
+      "step": 3855
+    },
+    {
+      "epoch": 0.14023348001600175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936945982941837e-05,
+      "loss": 1.0889,
+      "step": 3856
+    },
+    {
+      "epoch": 0.14026984761974035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936904813214508e-05,
+      "loss": 1.1469,
+      "step": 3857
+    },
+    {
+      "epoch": 0.14030621522347891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9368636304889538e-05,
+      "loss": 1.1426,
+      "step": 3858
+    },
+    {
+      "epoch": 0.1403425828272175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9368224347657457e-05,
+      "loss": 1.1854,
+      "step": 3859
+    },
+    {
+      "epoch": 0.1403789504309561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9367812260454558e-05,
+      "loss": 1.058,
+      "step": 3860
+    },
+    {
+      "epoch": 0.1404153180346947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9367400043286552e-05,
+      "loss": 1.1564,
+      "step": 3861
+    },
+    {
+      "epoch": 0.14045168563843327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936698769615916e-05,
+      "loss": 1.0931,
+      "step": 3862
+    },
+    {
+      "epoch": 0.14048805324217187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9366575219078113e-05,
+      "loss": 1.2643,
+      "step": 3863
+    },
+    {
+      "epoch": 0.14052442084591046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9366162612049115e-05,
+      "loss": 1.1591,
+      "step": 3864
+    },
+    {
+      "epoch": 0.14056078844964906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93657498750779e-05,
+      "loss": 1.0605,
+      "step": 3865
+    },
+    {
+      "epoch": 0.14059715605338763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93653370081702e-05,
+      "loss": 1.2825,
+      "step": 3866
+    },
+    {
+      "epoch": 0.14063352365712622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9364924011331733e-05,
+      "loss": 1.2519,
+      "step": 3867
+    },
+    {
+      "epoch": 0.14066989126086482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936451088456823e-05,
+      "loss": 1.1213,
+      "step": 3868
+    },
+    {
+      "epoch": 0.14070625886460342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936409762788543e-05,
+      "loss": 1.0247,
+      "step": 3869
+    },
+    {
+      "epoch": 0.140742626468342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9363684241289056e-05,
+      "loss": 1.1376,
+      "step": 3870
+    },
+    {
+      "epoch": 0.14077899407208058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9363270724784852e-05,
+      "loss": 1.19,
+      "step": 3871
+    },
+    {
+      "epoch": 0.14081536167581918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936285707837855e-05,
+      "loss": 1.1454,
+      "step": 3872
+    },
+    {
+      "epoch": 0.14085172927955777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9362443302075893e-05,
+      "loss": 1.0822,
+      "step": 3873
+    },
+    {
+      "epoch": 0.14088809688329637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9362029395882615e-05,
+      "loss": 1.2218,
+      "step": 3874
+    },
+    {
+      "epoch": 0.14092446448703494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9361615359804463e-05,
+      "loss": 1.037,
+      "step": 3875
+    },
+    {
+      "epoch": 0.14096083209077354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9361201193847183e-05,
+      "loss": 1.1058,
+      "step": 3876
+    },
+    {
+      "epoch": 0.14099719969451213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9360786898016515e-05,
+      "loss": 1.0561,
+      "step": 3877
+    },
+    {
+      "epoch": 0.14103356729825073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936037247231821e-05,
+      "loss": 1.1964,
+      "step": 3878
+    },
+    {
+      "epoch": 0.1410699349019893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935995791675802e-05,
+      "loss": 1.1127,
+      "step": 3879
+    },
+    {
+      "epoch": 0.1411063025057279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9359543231341694e-05,
+      "loss": 1.1159,
+      "step": 3880
+    },
+    {
+      "epoch": 0.1411426701094665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9359128416074983e-05,
+      "loss": 1.2022,
+      "step": 3881
+    },
+    {
+      "epoch": 0.14117903771320509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9358713470963645e-05,
+      "loss": 1.255,
+      "step": 3882
+    },
+    {
+      "epoch": 0.14121540531694365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9358298396013434e-05,
+      "loss": 1.1398,
+      "step": 3883
+    },
+    {
+      "epoch": 0.14125177292068225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9357883191230112e-05,
+      "loss": 1.0298,
+      "step": 3884
+    },
+    {
+      "epoch": 0.14128814052442085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935746785661944e-05,
+      "loss": 1.1644,
+      "step": 3885
+    },
+    {
+      "epoch": 0.14132450812815944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9357052392187175e-05,
+      "loss": 1.1293,
+      "step": 3886
+    },
+    {
+      "epoch": 0.141360875731898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9356636797939086e-05,
+      "loss": 1.1676,
+      "step": 3887
+    },
+    {
+      "epoch": 0.1413972433356366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9356221073880938e-05,
+      "loss": 1.1443,
+      "step": 3888
+    },
+    {
+      "epoch": 0.1414336109393752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9355805220018494e-05,
+      "loss": 1.1594,
+      "step": 3889
+    },
+    {
+      "epoch": 0.1414699785431138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9355389236357528e-05,
+      "loss": 1.192,
+      "step": 3890
+    },
+    {
+      "epoch": 0.1415063461468524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9354973122903806e-05,
+      "loss": 1.1749,
+      "step": 3891
+    },
+    {
+      "epoch": 0.14154271375059096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935455687966311e-05,
+      "loss": 1.0503,
+      "step": 3892
+    },
+    {
+      "epoch": 0.14157908135432956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9354140506641208e-05,
+      "loss": 1.2576,
+      "step": 3893
+    },
+    {
+      "epoch": 0.14161544895806816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935372400384388e-05,
+      "loss": 1.2042,
+      "step": 3894
+    },
+    {
+      "epoch": 0.14165181656180675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93533073712769e-05,
+      "loss": 1.1997,
+      "step": 3895
+    },
+    {
+      "epoch": 0.14168818416554532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935289060894605e-05,
+      "loss": 1.287,
+      "step": 3896
+    },
+    {
+      "epoch": 0.14172455176928392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9352473716857117e-05,
+      "loss": 1.1288,
+      "step": 3897
+    },
+    {
+      "epoch": 0.14176091937302251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935205669501588e-05,
+      "loss": 1.1735,
+      "step": 3898
+    },
+    {
+      "epoch": 0.1417972869767611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9351639543428122e-05,
+      "loss": 1.2455,
+      "step": 3899
+    },
+    {
+      "epoch": 0.14183365458049968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9351222262099636e-05,
+      "loss": 1.2313,
+      "step": 3900
+    },
+    {
+      "epoch": 0.14187002218423828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9350804851036207e-05,
+      "loss": 1.1269,
+      "step": 3901
+    },
+    {
+      "epoch": 0.14190638978797687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.935038731024363e-05,
+      "loss": 1.0497,
+      "step": 3902
+    },
+    {
+      "epoch": 0.14194275739171547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9349969639727693e-05,
+      "loss": 1.1077,
+      "step": 3903
+    },
+    {
+      "epoch": 0.14197912499545404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9349551839494193e-05,
+      "loss": 1.1325,
+      "step": 3904
+    },
+    {
+      "epoch": 0.14201549259919263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934913390954893e-05,
+      "loss": 1.1641,
+      "step": 3905
+    },
+    {
+      "epoch": 0.14205186020293123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93487158498977e-05,
+      "loss": 1.1396,
+      "step": 3906
+    },
+    {
+      "epoch": 0.14208822780666983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9348297660546295e-05,
+      "loss": 1.026,
+      "step": 3907
+    },
+    {
+      "epoch": 0.14212459541040842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9347879341500527e-05,
+      "loss": 1.2358,
+      "step": 3908
+    },
+    {
+      "epoch": 0.142160963014147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9347460892766195e-05,
+      "loss": 1.1134,
+      "step": 3909
+    },
+    {
+      "epoch": 0.1421973306178856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9347042314349107e-05,
+      "loss": 1.186,
+      "step": 3910
+    },
+    {
+      "epoch": 0.14223369822162418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934662360625507e-05,
+      "loss": 1.2308,
+      "step": 3911
+    },
+    {
+      "epoch": 0.14227006582536278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9346204768489888e-05,
+      "loss": 1.1623,
+      "step": 3912
+    },
+    {
+      "epoch": 0.14230643342910135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9345785801059376e-05,
+      "loss": 1.2925,
+      "step": 3913
+    },
+    {
+      "epoch": 0.14234280103283994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9345366703969347e-05,
+      "loss": 1.1494,
+      "step": 3914
+    },
+    {
+      "epoch": 0.14237916863657854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9344947477225613e-05,
+      "loss": 1.1921,
+      "step": 3915
+    },
+    {
+      "epoch": 0.14241553624031714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934452812083399e-05,
+      "loss": 1.2077,
+      "step": 3916
+    },
+    {
+      "epoch": 0.1424519038440557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9344108634800303e-05,
+      "loss": 1.0677,
+      "step": 3917
+    },
+    {
+      "epoch": 0.1424882714477943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9343689019130364e-05,
+      "loss": 1.1317,
+      "step": 3918
+    },
+    {
+      "epoch": 0.1425246390515329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9343269273829994e-05,
+      "loss": 1.1081,
+      "step": 3919
+    },
+    {
+      "epoch": 0.1425610066552715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9342849398905025e-05,
+      "loss": 1.1929,
+      "step": 3920
+    },
+    {
+      "epoch": 0.14259737425901006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934242939436127e-05,
+      "loss": 1.1368,
+      "step": 3921
+    },
+    {
+      "epoch": 0.14263374186274866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9342009260204568e-05,
+      "loss": 1.1596,
+      "step": 3922
+    },
+    {
+      "epoch": 0.14267010946648725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9341588996440738e-05,
+      "loss": 1.0009,
+      "step": 3923
+    },
+    {
+      "epoch": 0.14270647707022585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9341168603075613e-05,
+      "loss": 1.1442,
+      "step": 3924
+    },
+    {
+      "epoch": 0.14274284467396442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934074808011503e-05,
+      "loss": 1.1531,
+      "step": 3925
+    },
+    {
+      "epoch": 0.14277921227770302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934032742756482e-05,
+      "loss": 1.1445,
+      "step": 3926
+    },
+    {
+      "epoch": 0.1428155798814416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933990664543082e-05,
+      "loss": 1.1878,
+      "step": 3927
+    },
+    {
+      "epoch": 0.1428519474851802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9339485733718864e-05,
+      "loss": 1.1882,
+      "step": 3928
+    },
+    {
+      "epoch": 0.1428883150889188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93390646924348e-05,
+      "loss": 1.1062,
+      "step": 3929
+    },
+    {
+      "epoch": 0.14292468269265737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9338643521584457e-05,
+      "loss": 1.2104,
+      "step": 3930
+    },
+    {
+      "epoch": 0.14296105029639597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9338222221173686e-05,
+      "loss": 1.1122,
+      "step": 3931
+    },
+    {
+      "epoch": 0.14299741790013457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9337800791208336e-05,
+      "loss": 1.056,
+      "step": 3932
+    },
+    {
+      "epoch": 0.14303378550387316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9337379231694243e-05,
+      "loss": 1.3241,
+      "step": 3933
+    },
+    {
+      "epoch": 0.14307015310761173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9336957542637265e-05,
+      "loss": 1.2354,
+      "step": 3934
+    },
+    {
+      "epoch": 0.14310652071135033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9336535724043248e-05,
+      "loss": 1.056,
+      "step": 3935
+    },
+    {
+      "epoch": 0.14314288831508892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9336113775918045e-05,
+      "loss": 1.1429,
+      "step": 3936
+    },
+    {
+      "epoch": 0.14317925591882752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9335691698267508e-05,
+      "loss": 1.3387,
+      "step": 3937
+    },
+    {
+      "epoch": 0.1432156235225661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9335269491097495e-05,
+      "loss": 1.0499,
+      "step": 3938
+    },
+    {
+      "epoch": 0.14325199112630468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9334847154413864e-05,
+      "loss": 1.231,
+      "step": 3939
+    },
+    {
+      "epoch": 0.14328835873004328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9334424688222467e-05,
+      "loss": 1.0084,
+      "step": 3940
+    },
+    {
+      "epoch": 0.14332472633378188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933400209252918e-05,
+      "loss": 1.1516,
+      "step": 3941
+    },
+    {
+      "epoch": 0.14336109393752045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9333579367339853e-05,
+      "loss": 1.1882,
+      "step": 3942
+    },
+    {
+      "epoch": 0.14339746154125904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9333156512660355e-05,
+      "loss": 1.1551,
+      "step": 3943
+    },
+    {
+      "epoch": 0.14343382914499764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9332733528496554e-05,
+      "loss": 1.2337,
+      "step": 3944
+    },
+    {
+      "epoch": 0.14347019674873623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9332310414854313e-05,
+      "loss": 1.1927,
+      "step": 3945
+    },
+    {
+      "epoch": 0.1435065643524748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933188717173951e-05,
+      "loss": 1.2025,
+      "step": 3946
+    },
+    {
+      "epoch": 0.1435429319562134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933146379915801e-05,
+      "loss": 1.1623,
+      "step": 3947
+    },
+    {
+      "epoch": 0.143579299559952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933104029711569e-05,
+      "loss": 1.1866,
+      "step": 3948
+    },
+    {
+      "epoch": 0.1436156671636906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9330616665618424e-05,
+      "loss": 1.2144,
+      "step": 3949
+    },
+    {
+      "epoch": 0.1436520347674292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933019290467209e-05,
+      "loss": 1.1648,
+      "step": 3950
+    },
+    {
+      "epoch": 0.14368840237116776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9329769014282572e-05,
+      "loss": 1.1906,
+      "step": 3951
+    },
+    {
+      "epoch": 0.14372476997490635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932934499445574e-05,
+      "loss": 1.1656,
+      "step": 3952
+    },
+    {
+      "epoch": 0.14376113757864495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9328920845197486e-05,
+      "loss": 1.1558,
+      "step": 3953
+    },
+    {
+      "epoch": 0.14379750518238354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932849656651369e-05,
+      "loss": 1.2243,
+      "step": 3954
+    },
+    {
+      "epoch": 0.1438338727861221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932807215841024e-05,
+      "loss": 1.2445,
+      "step": 3955
+    },
+    {
+      "epoch": 0.1438702403898607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9327647620893026e-05,
+      "loss": 1.2556,
+      "step": 3956
+    },
+    {
+      "epoch": 0.1439066079935993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932722295396793e-05,
+      "loss": 1.0715,
+      "step": 3957
+    },
+    {
+      "epoch": 0.1439429755973379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9326798157640855e-05,
+      "loss": 1.1458,
+      "step": 3958
+    },
+    {
+      "epoch": 0.14397934320107647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9326373231917687e-05,
+      "loss": 1.201,
+      "step": 3959
+    },
+    {
+      "epoch": 0.14401571080481507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932594817680432e-05,
+      "loss": 1.1064,
+      "step": 3960
+    },
+    {
+      "epoch": 0.14405207840855366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9325522992306657e-05,
+      "loss": 1.1678,
+      "step": 3961
+    },
+    {
+      "epoch": 0.14408844601229226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9325097678430593e-05,
+      "loss": 1.1378,
+      "step": 3962
+    },
+    {
+      "epoch": 0.14412481361603083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9324672235182027e-05,
+      "loss": 1.2142,
+      "step": 3963
+    },
+    {
+      "epoch": 0.14416118121976942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9324246662566867e-05,
+      "loss": 1.2281,
+      "step": 3964
+    },
+    {
+      "epoch": 0.14419754882350802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932382096059101e-05,
+      "loss": 1.2205,
+      "step": 3965
+    },
+    {
+      "epoch": 0.14423391642724662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9323395129260367e-05,
+      "loss": 1.2504,
+      "step": 3966
+    },
+    {
+      "epoch": 0.14427028403098519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9322969168580847e-05,
+      "loss": 1.1585,
+      "step": 3967
+    },
+    {
+      "epoch": 0.14430665163472378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9322543078558356e-05,
+      "loss": 1.0101,
+      "step": 3968
+    },
+    {
+      "epoch": 0.14434301923846238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9322116859198807e-05,
+      "loss": 1.2624,
+      "step": 3969
+    },
+    {
+      "epoch": 0.14437938684220097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9321690510508114e-05,
+      "loss": 1.176,
+      "step": 3970
+    },
+    {
+      "epoch": 0.14441575444593957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932126403249219e-05,
+      "loss": 1.1373,
+      "step": 3971
+    },
+    {
+      "epoch": 0.14445212204967814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9320837425156954e-05,
+      "loss": 1.2823,
+      "step": 3972
+    },
+    {
+      "epoch": 0.14448848965341674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932041068850832e-05,
+      "loss": 1.1401,
+      "step": 3973
+    },
+    {
+      "epoch": 0.14452485725715533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9319983822552213e-05,
+      "loss": 1.0131,
+      "step": 3974
+    },
+    {
+      "epoch": 0.14456122486089393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9319556827294556e-05,
+      "loss": 1.1368,
+      "step": 3975
+    },
+    {
+      "epoch": 0.1445975924646325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9319129702741266e-05,
+      "loss": 1.1208,
+      "step": 3976
+    },
+    {
+      "epoch": 0.1446339600683711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9318702448898277e-05,
+      "loss": 1.1087,
+      "step": 3977
+    },
+    {
+      "epoch": 0.1446703276721097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931827506577151e-05,
+      "loss": 1.097,
+      "step": 3978
+    },
+    {
+      "epoch": 0.14470669527584828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93178475533669e-05,
+      "loss": 1.1886,
+      "step": 3979
+    },
+    {
+      "epoch": 0.14474306287958685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9317419911690378e-05,
+      "loss": 1.0865,
+      "step": 3980
+    },
+    {
+      "epoch": 0.14477943048332545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9316992140747873e-05,
+      "loss": 1.1998,
+      "step": 3981
+    },
+    {
+      "epoch": 0.14481579808706405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9316564240545318e-05,
+      "loss": 1.0439,
+      "step": 3982
+    },
+    {
+      "epoch": 0.14485216569080264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9316136211088655e-05,
+      "loss": 1.009,
+      "step": 3983
+    },
+    {
+      "epoch": 0.1448885332945412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931570805238382e-05,
+      "loss": 1.0984,
+      "step": 3984
+    },
+    {
+      "epoch": 0.1449249008982798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9315279764436752e-05,
+      "loss": 1.2307,
+      "step": 3985
+    },
+    {
+      "epoch": 0.1449612685020184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9314851347253394e-05,
+      "loss": 1.0509,
+      "step": 3986
+    },
+    {
+      "epoch": 0.144997636105757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9314422800839693e-05,
+      "loss": 1.1955,
+      "step": 3987
+    },
+    {
+      "epoch": 0.14503400370949557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9313994125201585e-05,
+      "loss": 1.2089,
+      "step": 3988
+    },
+    {
+      "epoch": 0.14507037131323416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931356532034503e-05,
+      "loss": 1.2735,
+      "step": 3989
+    },
+    {
+      "epoch": 0.14510673891697276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9313136386275967e-05,
+      "loss": 1.1395,
+      "step": 3990
+    },
+    {
+      "epoch": 0.14514310652071136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931270732300035e-05,
+      "loss": 1.1544,
+      "step": 3991
+    },
+    {
+      "epoch": 0.14517947412444995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931227813052413e-05,
+      "loss": 1.1334,
+      "step": 3992
+    },
+    {
+      "epoch": 0.14521584172818852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9311848808853267e-05,
+      "loss": 1.0897,
+      "step": 3993
+    },
+    {
+      "epoch": 0.14525220933192712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931141935799371e-05,
+      "loss": 1.0639,
+      "step": 3994
+    },
+    {
+      "epoch": 0.14528857693566571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9310989777951422e-05,
+      "loss": 1.1732,
+      "step": 3995
+    },
+    {
+      "epoch": 0.1453249445394043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931056006873236e-05,
+      "loss": 1.0663,
+      "step": 3996
+    },
+    {
+      "epoch": 0.14536131214314288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.931013023034249e-05,
+      "loss": 1.3115,
+      "step": 3997
+    },
+    {
+      "epoch": 0.14539767974688148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9309700262787773e-05,
+      "loss": 1.1481,
+      "step": 3998
+    },
+    {
+      "epoch": 0.14543404735062007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9309270166074168e-05,
+      "loss": 1.3845,
+      "step": 3999
+    },
+    {
+      "epoch": 0.14547041495435867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9308839940207656e-05,
+      "loss": 1.1092,
+      "step": 4000
+    },
+    {
+      "epoch": 0.14550678255809724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9308409585194193e-05,
+      "loss": 1.1525,
+      "step": 4001
+    },
+    {
+      "epoch": 0.14554315016183583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930797910103975e-05,
+      "loss": 1.1056,
+      "step": 4002
+    },
+    {
+      "epoch": 0.14557951776557443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9307548487750308e-05,
+      "loss": 1.0678,
+      "step": 4003
+    },
+    {
+      "epoch": 0.14561588536931303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9307117745331836e-05,
+      "loss": 1.1351,
+      "step": 4004
+    },
+    {
+      "epoch": 0.1456522529730516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930668687379031e-05,
+      "loss": 1.1139,
+      "step": 4005
+    },
+    {
+      "epoch": 0.1456886205767902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9306255873131708e-05,
+      "loss": 1.0134,
+      "step": 4006
+    },
+    {
+      "epoch": 0.14572498818052879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930582474336201e-05,
+      "loss": 1.1678,
+      "step": 4007
+    },
+    {
+      "epoch": 0.14576135578426738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9305393484487197e-05,
+      "loss": 1.1381,
+      "step": 4008
+    },
+    {
+      "epoch": 0.14579772338800595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930496209651325e-05,
+      "loss": 1.1909,
+      "step": 4009
+    },
+    {
+      "epoch": 0.14583409099174455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930453057944616e-05,
+      "loss": 1.0371,
+      "step": 4010
+    },
+    {
+      "epoch": 0.14587045859548314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9304098933291904e-05,
+      "loss": 1.0197,
+      "step": 4011
+    },
+    {
+      "epoch": 0.14590682619922174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930366715805648e-05,
+      "loss": 1.1325,
+      "step": 4012
+    },
+    {
+      "epoch": 0.14594319380296034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9303235253745874e-05,
+      "loss": 1.0068,
+      "step": 4013
+    },
+    {
+      "epoch": 0.1459795614066989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9302803220366076e-05,
+      "loss": 1.2667,
+      "step": 4014
+    },
+    {
+      "epoch": 0.1460159290104375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9302371057923085e-05,
+      "loss": 1.1839,
+      "step": 4015
+    },
+    {
+      "epoch": 0.1460522966141761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930193876642289e-05,
+      "loss": 1.2072,
+      "step": 4016
+    },
+    {
+      "epoch": 0.1460886642179147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93015063458715e-05,
+      "loss": 1.0592,
+      "step": 4017
+    },
+    {
+      "epoch": 0.14612503182165326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93010737962749e-05,
+      "loss": 1.0403,
+      "step": 4018
+    },
+    {
+      "epoch": 0.14616139942539186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9300641117639102e-05,
+      "loss": 1.1691,
+      "step": 4019
+    },
+    {
+      "epoch": 0.14619776702913045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9300208309970098e-05,
+      "loss": 1.2479,
+      "step": 4020
+    },
+    {
+      "epoch": 0.14623413463286905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9299775373273904e-05,
+      "loss": 1.2033,
+      "step": 4021
+    },
+    {
+      "epoch": 0.14627050223660762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929934230755652e-05,
+      "loss": 1.2496,
+      "step": 4022
+    },
+    {
+      "epoch": 0.14630686984034622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929890911282396e-05,
+      "loss": 1.1788,
+      "step": 4023
+    },
+    {
+      "epoch": 0.1463432374440848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929847578908222e-05,
+      "loss": 1.0661,
+      "step": 4024
+    },
+    {
+      "epoch": 0.1463796050478234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9298042336337327e-05,
+      "loss": 1.1926,
+      "step": 4025
+    },
+    {
+      "epoch": 0.14641597265156198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9297608754595288e-05,
+      "loss": 1.072,
+      "step": 4026
+    },
+    {
+      "epoch": 0.14645234025530057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929717504386212e-05,
+      "loss": 1.2581,
+      "step": 4027
+    },
+    {
+      "epoch": 0.14648870785903917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9296741204143837e-05,
+      "loss": 1.318,
+      "step": 4028
+    },
+    {
+      "epoch": 0.14652507546277777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9296307235446465e-05,
+      "loss": 1.0459,
+      "step": 4029
+    },
+    {
+      "epoch": 0.14656144306651633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9295873137776016e-05,
+      "loss": 1.142,
+      "step": 4030
+    },
+    {
+      "epoch": 0.14659781067025493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9295438911138514e-05,
+      "loss": 1.2839,
+      "step": 4031
+    },
+    {
+      "epoch": 0.14663417827399353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929500455553999e-05,
+      "loss": 1.1415,
+      "step": 4032
+    },
+    {
+      "epoch": 0.14667054587773212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9294570070986465e-05,
+      "loss": 1.2214,
+      "step": 4033
+    },
+    {
+      "epoch": 0.14670691348147072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9294135457483966e-05,
+      "loss": 1.1907,
+      "step": 4034
+    },
+    {
+      "epoch": 0.1467432810852093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9293700715038527e-05,
+      "loss": 1.1241,
+      "step": 4035
+    },
+    {
+      "epoch": 0.14677964868894788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9293265843656174e-05,
+      "loss": 1.0967,
+      "step": 4036
+    },
+    {
+      "epoch": 0.14681601629268648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929283084334294e-05,
+      "loss": 1.2078,
+      "step": 4037
+    },
+    {
+      "epoch": 0.14685238389642508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929239571410487e-05,
+      "loss": 1.0905,
+      "step": 4038
+    },
+    {
+      "epoch": 0.14688875150016364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9291960455947988e-05,
+      "loss": 1.2138,
+      "step": 4039
+    },
+    {
+      "epoch": 0.14692511910390224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.929152506887834e-05,
+      "loss": 1.0963,
+      "step": 4040
+    },
+    {
+      "epoch": 0.14696148670764084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9291089552901965e-05,
+      "loss": 1.3713,
+      "step": 4041
+    },
+    {
+      "epoch": 0.14699785431137943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9290653908024906e-05,
+      "loss": 1.102,
+      "step": 4042
+    },
+    {
+      "epoch": 0.147034221915118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9290218134253204e-05,
+      "loss": 1.0936,
+      "step": 4043
+    },
+    {
+      "epoch": 0.1470705895188566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9289782231592906e-05,
+      "loss": 1.0013,
+      "step": 4044
+    },
+    {
+      "epoch": 0.1471069571225952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928934620005006e-05,
+      "loss": 1.2243,
+      "step": 4045
+    },
+    {
+      "epoch": 0.1471433247263338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9288910039630715e-05,
+      "loss": 1.1818,
+      "step": 4046
+    },
+    {
+      "epoch": 0.14717969233007236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9288473750340927e-05,
+      "loss": 1.1878,
+      "step": 4047
+    },
+    {
+      "epoch": 0.14721605993381096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928803733218674e-05,
+      "loss": 1.0679,
+      "step": 4048
+    },
+    {
+      "epoch": 0.14725242753754955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928760078517421e-05,
+      "loss": 1.0925,
+      "step": 4049
+    },
+    {
+      "epoch": 0.14728879514128815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92871641093094e-05,
+      "loss": 1.1838,
+      "step": 4050
+    },
+    {
+      "epoch": 0.14732516274502672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9286727304598367e-05,
+      "loss": 1.1553,
+      "step": 4051
+    },
+    {
+      "epoch": 0.1473615303487653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9286290371047168e-05,
+      "loss": 1.1085,
+      "step": 4052
+    },
+    {
+      "epoch": 0.1473978979525039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9285853308661865e-05,
+      "loss": 1.2061,
+      "step": 4053
+    },
+    {
+      "epoch": 0.1474342655562425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928541611744852e-05,
+      "loss": 1.1537,
+      "step": 4054
+    },
+    {
+      "epoch": 0.1474706331599811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9284978797413204e-05,
+      "loss": 1.027,
+      "step": 4055
+    },
+    {
+      "epoch": 0.14750700076371967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928454134856198e-05,
+      "loss": 1.0497,
+      "step": 4056
+    },
+    {
+      "epoch": 0.14754336836745827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9284103770900917e-05,
+      "loss": 1.1246,
+      "step": 4057
+    },
+    {
+      "epoch": 0.14757973597119686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9283666064436083e-05,
+      "loss": 1.1595,
+      "step": 4058
+    },
+    {
+      "epoch": 0.14761610357493546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9283228229173558e-05,
+      "loss": 1.139,
+      "step": 4059
+    },
+    {
+      "epoch": 0.14765247117867403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9282790265119416e-05,
+      "loss": 1.1131,
+      "step": 4060
+    },
+    {
+      "epoch": 0.14768883878241262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9282352172279724e-05,
+      "loss": 1.0406,
+      "step": 4061
+    },
+    {
+      "epoch": 0.14772520638615122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928191395066057e-05,
+      "loss": 1.1716,
+      "step": 4062
+    },
+    {
+      "epoch": 0.14776157398988982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9281475600268024e-05,
+      "loss": 1.0689,
+      "step": 4063
+    },
+    {
+      "epoch": 0.14779794159362838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9281037121108176e-05,
+      "loss": 1.0719,
+      "step": 4064
+    },
+    {
+      "epoch": 0.14783430919736698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9280598513187106e-05,
+      "loss": 1.1343,
+      "step": 4065
+    },
+    {
+      "epoch": 0.14787067680110558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92801597765109e-05,
+      "loss": 1.1643,
+      "step": 4066
+    },
+    {
+      "epoch": 0.14790704440484417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927972091108564e-05,
+      "loss": 1.1904,
+      "step": 4067
+    },
+    {
+      "epoch": 0.14794341200858274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9279281916917422e-05,
+      "loss": 1.1353,
+      "step": 4068
+    },
+    {
+      "epoch": 0.14797977961232134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927884279401233e-05,
+      "loss": 1.0833,
+      "step": 4069
+    },
+    {
+      "epoch": 0.14801614721605993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9278403542376463e-05,
+      "loss": 1.1396,
+      "step": 4070
+    },
+    {
+      "epoch": 0.14805251481979853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927796416201591e-05,
+      "loss": 1.1636,
+      "step": 4071
+    },
+    {
+      "epoch": 0.1480888824235371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9277524652936766e-05,
+      "loss": 1.0564,
+      "step": 4072
+    },
+    {
+      "epoch": 0.1481252500272757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9277085015145133e-05,
+      "loss": 1.114,
+      "step": 4073
+    },
+    {
+      "epoch": 0.1481616176310143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9276645248647105e-05,
+      "loss": 1.1987,
+      "step": 4074
+    },
+    {
+      "epoch": 0.1481979852347529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927620535344879e-05,
+      "loss": 1.1452,
+      "step": 4075
+    },
+    {
+      "epoch": 0.14823435283849148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9275765329556285e-05,
+      "loss": 1.1786,
+      "step": 4076
+    },
+    {
+      "epoch": 0.14827072044223005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9275325176975693e-05,
+      "loss": 1.1548,
+      "step": 4077
+    },
+    {
+      "epoch": 0.14830708804596865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9274884895713127e-05,
+      "loss": 1.226,
+      "step": 4078
+    },
+    {
+      "epoch": 0.14834345564970725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9274444485774696e-05,
+      "loss": 1.1967,
+      "step": 4079
+    },
+    {
+      "epoch": 0.14837982325344584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9274003947166503e-05,
+      "loss": 1.1757,
+      "step": 4080
+    },
+    {
+      "epoch": 0.1484161908571844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9273563279894665e-05,
+      "loss": 1.1392,
+      "step": 4081
+    },
+    {
+      "epoch": 0.148452558460923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927312248396529e-05,
+      "loss": 1.1193,
+      "step": 4082
+    },
+    {
+      "epoch": 0.1484889260646616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92726815593845e-05,
+      "loss": 1.1469,
+      "step": 4083
+    },
+    {
+      "epoch": 0.1485252936684002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927224050615841e-05,
+      "loss": 1.0963,
+      "step": 4084
+    },
+    {
+      "epoch": 0.14856166127213877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9271799324293138e-05,
+      "loss": 1.2517,
+      "step": 4085
+    },
+    {
+      "epoch": 0.14859802887587736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9271358013794804e-05,
+      "loss": 1.1302,
+      "step": 4086
+    },
+    {
+      "epoch": 0.14863439647961596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9270916574669536e-05,
+      "loss": 1.1849,
+      "step": 4087
+    },
+    {
+      "epoch": 0.14867076408335456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927047500692345e-05,
+      "loss": 1.0553,
+      "step": 4088
+    },
+    {
+      "epoch": 0.14870713168709312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927003331056268e-05,
+      "loss": 1.1945,
+      "step": 4089
+    },
+    {
+      "epoch": 0.14874349929083172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9269591485593348e-05,
+      "loss": 1.1405,
+      "step": 4090
+    },
+    {
+      "epoch": 0.14877986689457032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9269149532021586e-05,
+      "loss": 1.1825,
+      "step": 4091
+    },
+    {
+      "epoch": 0.1488162344983089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9268707449853526e-05,
+      "loss": 1.1003,
+      "step": 4092
+    },
+    {
+      "epoch": 0.1488526021020475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9268265239095298e-05,
+      "loss": 1.1866,
+      "step": 4093
+    },
+    {
+      "epoch": 0.14888896970578608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9267822899753043e-05,
+      "loss": 1.0967,
+      "step": 4094
+    },
+    {
+      "epoch": 0.14892533730952467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9267380431832894e-05,
+      "loss": 1.1704,
+      "step": 4095
+    },
+    {
+      "epoch": 0.14896170491326327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926693783534099e-05,
+      "loss": 1.1694,
+      "step": 4096
+    },
+    {
+      "epoch": 0.14899807251700187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9266495110283472e-05,
+      "loss": 1.1732,
+      "step": 4097
+    },
+    {
+      "epoch": 0.14903444012074044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926605225666648e-05,
+      "loss": 1.1017,
+      "step": 4098
+    },
+    {
+      "epoch": 0.14907080772447903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9265609274496163e-05,
+      "loss": 1.1616,
+      "step": 4099
+    },
+    {
+      "epoch": 0.14910717532821763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9265166163778662e-05,
+      "loss": 1.0829,
+      "step": 4100
+    },
+    {
+      "epoch": 0.14914354293195622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9264722924520127e-05,
+      "loss": 1.1993,
+      "step": 4101
+    },
+    {
+      "epoch": 0.1491799105356948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9264279556726703e-05,
+      "loss": 1.0923,
+      "step": 4102
+    },
+    {
+      "epoch": 0.1492162781394334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9263836060404547e-05,
+      "loss": 1.1285,
+      "step": 4103
+    },
+    {
+      "epoch": 0.14925264574317199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926339243555981e-05,
+      "loss": 1.1963,
+      "step": 4104
+    },
+    {
+      "epoch": 0.14928901334691058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9262948682198645e-05,
+      "loss": 1.1884,
+      "step": 4105
+    },
+    {
+      "epoch": 0.14932538095064915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9262504800327207e-05,
+      "loss": 1.1768,
+      "step": 4106
+    },
+    {
+      "epoch": 0.14936174855438775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9262060789951658e-05,
+      "loss": 1.0805,
+      "step": 4107
+    },
+    {
+      "epoch": 0.14939811615812634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9261616651078156e-05,
+      "loss": 1.1349,
+      "step": 4108
+    },
+    {
+      "epoch": 0.14943448376186494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926117238371287e-05,
+      "loss": 1.0958,
+      "step": 4109
+    },
+    {
+      "epoch": 0.1494708513656035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926072798786195e-05,
+      "loss": 1.1807,
+      "step": 4110
+    },
+    {
+      "epoch": 0.1495072189693421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926028346353157e-05,
+      "loss": 1.1215,
+      "step": 4111
+    },
+    {
+      "epoch": 0.1495435865730807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9259838810727894e-05,
+      "loss": 1.079,
+      "step": 4112
+    },
+    {
+      "epoch": 0.1495799541768193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9259394029457096e-05,
+      "loss": 1.1256,
+      "step": 4113
+    },
+    {
+      "epoch": 0.1496163217805579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925894911972534e-05,
+      "loss": 1.1241,
+      "step": 4114
+    },
+    {
+      "epoch": 0.14965268938429646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9258504081538806e-05,
+      "loss": 1.0611,
+      "step": 4115
+    },
+    {
+      "epoch": 0.14968905698803506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925805891490366e-05,
+      "loss": 1.1637,
+      "step": 4116
+    },
+    {
+      "epoch": 0.14972542459177365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9257613619826085e-05,
+      "loss": 1.1572,
+      "step": 4117
+    },
+    {
+      "epoch": 0.14976179219551225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9257168196312254e-05,
+      "loss": 1.2265,
+      "step": 4118
+    },
+    {
+      "epoch": 0.14979815979925082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925672264436835e-05,
+      "loss": 1.0969,
+      "step": 4119
+    },
+    {
+      "epoch": 0.14983452740298941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9256276964000552e-05,
+      "loss": 1.051,
+      "step": 4120
+    },
+    {
+      "epoch": 0.149870895006728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925583115521504e-05,
+      "loss": 1.3474,
+      "step": 4121
+    },
+    {
+      "epoch": 0.1499072626104666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9255385218018008e-05,
+      "loss": 1.2619,
+      "step": 4122
+    },
+    {
+      "epoch": 0.14994363021420518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9254939152415638e-05,
+      "loss": 1.0491,
+      "step": 4123
+    },
+    {
+      "epoch": 0.14997999781794377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9254492958414115e-05,
+      "loss": 1.2267,
+      "step": 4124
+    },
+    {
+      "epoch": 0.15001636542168237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9254046636019637e-05,
+      "loss": 1.0493,
+      "step": 4125
+    },
+    {
+      "epoch": 0.15005273302542096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9253600185238385e-05,
+      "loss": 1.1299,
+      "step": 4126
+    },
+    {
+      "epoch": 0.15008910062915953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9253153606076565e-05,
+      "loss": 1.0066,
+      "step": 4127
+    },
+    {
+      "epoch": 0.15012546823289813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9252706898540366e-05,
+      "loss": 0.9399,
+      "step": 4128
+    },
+    {
+      "epoch": 0.15016183583663673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925226006263599e-05,
+      "loss": 1.2188,
+      "step": 4129
+    },
+    {
+      "epoch": 0.15019820344037532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9251813098369626e-05,
+      "loss": 1.088,
+      "step": 4130
+    },
+    {
+      "epoch": 0.1502345710441139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9251366005747485e-05,
+      "loss": 1.0818,
+      "step": 4131
+    },
+    {
+      "epoch": 0.1502709386478525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9250918784775766e-05,
+      "loss": 1.1102,
+      "step": 4132
+    },
+    {
+      "epoch": 0.15030730625159108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9250471435460677e-05,
+      "loss": 1.1101,
+      "step": 4133
+    },
+    {
+      "epoch": 0.15034367385532968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925002395780842e-05,
+      "loss": 1.0585,
+      "step": 4134
+    },
+    {
+      "epoch": 0.15038004145906828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9249576351825205e-05,
+      "loss": 1.0746,
+      "step": 4135
+    },
+    {
+      "epoch": 0.15041640906280684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924912861751724e-05,
+      "loss": 1.0767,
+      "step": 4136
+    },
+    {
+      "epoch": 0.15045277666654544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9248680754890738e-05,
+      "loss": 1.201,
+      "step": 4137
+    },
+    {
+      "epoch": 0.15048914427028404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9248232763951914e-05,
+      "loss": 1.1957,
+      "step": 4138
+    },
+    {
+      "epoch": 0.15052551187402263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9247784644706983e-05,
+      "loss": 1.1309,
+      "step": 4139
+    },
+    {
+      "epoch": 0.1505618794777612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924733639716216e-05,
+      "loss": 1.1767,
+      "step": 4140
+    },
+    {
+      "epoch": 0.1505982470814998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9246888021323666e-05,
+      "loss": 1.2338,
+      "step": 4141
+    },
+    {
+      "epoch": 0.1506346146852384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9246439517197717e-05,
+      "loss": 1.0732,
+      "step": 4142
+    },
+    {
+      "epoch": 0.150670982288977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9245990884790543e-05,
+      "loss": 1.0181,
+      "step": 4143
+    },
+    {
+      "epoch": 0.15070734989271556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9245542124108357e-05,
+      "loss": 1.0793,
+      "step": 4144
+    },
+    {
+      "epoch": 0.15074371749645415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9245093235157397e-05,
+      "loss": 1.1669,
+      "step": 4145
+    },
+    {
+      "epoch": 0.15078008510019275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9244644217943884e-05,
+      "loss": 1.055,
+      "step": 4146
+    },
+    {
+      "epoch": 0.15081645270393135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924419507247405e-05,
+      "loss": 1.1906,
+      "step": 4147
+    },
+    {
+      "epoch": 0.15085282030766992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9243745798754126e-05,
+      "loss": 1.0144,
+      "step": 4148
+    },
+    {
+      "epoch": 0.1508891879114085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924329639679034e-05,
+      "loss": 1.1483,
+      "step": 4149
+    },
+    {
+      "epoch": 0.1509255555151471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9242846866588936e-05,
+      "loss": 1.1192,
+      "step": 4150
+    },
+    {
+      "epoch": 0.1509619231188857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9242397208156144e-05,
+      "loss": 1.1738,
+      "step": 4151
+    },
+    {
+      "epoch": 0.15099829072262427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9241947421498202e-05,
+      "loss": 1.1835,
+      "step": 4152
+    },
+    {
+      "epoch": 0.15103465832636287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9241497506621353e-05,
+      "loss": 1.1362,
+      "step": 4153
+    },
+    {
+      "epoch": 0.15107102593010147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9241047463531838e-05,
+      "loss": 1.1061,
+      "step": 4154
+    },
+    {
+      "epoch": 0.15110739353384006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9240597292235903e-05,
+      "loss": 1.0737,
+      "step": 4155
+    },
+    {
+      "epoch": 0.15114376113757866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924014699273979e-05,
+      "loss": 1.131,
+      "step": 4156
+    },
+    {
+      "epoch": 0.15118012874131723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9239696565049747e-05,
+      "loss": 1.2211,
+      "step": 4157
+    },
+    {
+      "epoch": 0.15121649634505582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9239246009172025e-05,
+      "loss": 1.1332,
+      "step": 4158
+    },
+    {
+      "epoch": 0.15125286394879442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 1.2758,
+      "step": 4159
+    },
+    {
+      "epoch": 0.15128923155253302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923834451287854e-05,
+      "loss": 1.1746,
+      "step": 4160
+    },
+    {
+      "epoch": 0.15132559915627158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9237893572475286e-05,
+      "loss": 1.1542,
+      "step": 4161
+    },
+    {
+      "epoch": 0.15136196676001018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9237442503909368e-05,
+      "loss": 1.1993,
+      "step": 4162
+    },
+    {
+      "epoch": 0.15139833436374878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923699130718704e-05,
+      "loss": 1.1803,
+      "step": 4163
+    },
+    {
+      "epoch": 0.15143470196748737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9236539982314563e-05,
+      "loss": 1.232,
+      "step": 4164
+    },
+    {
+      "epoch": 0.15147106957122594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92360885292982e-05,
+      "loss": 1.1719,
+      "step": 4165
+    },
+    {
+      "epoch": 0.15150743717496454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9235636948144213e-05,
+      "loss": 1.1171,
+      "step": 4166
+    },
+    {
+      "epoch": 0.15154380477870313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9235185238858864e-05,
+      "loss": 1.1453,
+      "step": 4167
+    },
+    {
+      "epoch": 0.15158017238244173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9234733401448427e-05,
+      "loss": 1.0423,
+      "step": 4168
+    },
+    {
+      "epoch": 0.1516165399861803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9234281435919165e-05,
+      "loss": 1.0711,
+      "step": 4169
+    },
+    {
+      "epoch": 0.1516529075899189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923382934227735e-05,
+      "loss": 1.1267,
+      "step": 4170
+    },
+    {
+      "epoch": 0.1516892751936575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9233377120529253e-05,
+      "loss": 1.1305,
+      "step": 4171
+    },
+    {
+      "epoch": 0.1517256427973961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923292477068115e-05,
+      "loss": 1.1825,
+      "step": 4172
+    },
+    {
+      "epoch": 0.15176201040113466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9232472292739312e-05,
+      "loss": 1.1725,
+      "step": 4173
+    },
+    {
+      "epoch": 0.15179837800487325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9232019686710024e-05,
+      "loss": 1.1495,
+      "step": 4174
+    },
+    {
+      "epoch": 0.15183474560861185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9231566952599562e-05,
+      "loss": 1.2328,
+      "step": 4175
+    },
+    {
+      "epoch": 0.15187111321235044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9231114090414202e-05,
+      "loss": 1.1818,
+      "step": 4176
+    },
+    {
+      "epoch": 0.15190748081608904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9230661100160235e-05,
+      "loss": 1.0452,
+      "step": 4177
+    },
+    {
+      "epoch": 0.1519438484198276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923020798184394e-05,
+      "loss": 1.1201,
+      "step": 4178
+    },
+    {
+      "epoch": 0.1519802160235662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9229754735471608e-05,
+      "loss": 1.151,
+      "step": 4179
+    },
+    {
+      "epoch": 0.1520165836273048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922930136104952e-05,
+      "loss": 1.2315,
+      "step": 4180
+    },
+    {
+      "epoch": 0.1520529512310434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9228847858583976e-05,
+      "loss": 1.1043,
+      "step": 4181
+    },
+    {
+      "epoch": 0.15208931883478197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9228394228081254e-05,
+      "loss": 1.141,
+      "step": 4182
+    },
+    {
+      "epoch": 0.15212568643852056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922794046954766e-05,
+      "loss": 1.215,
+      "step": 4183
+    },
+    {
+      "epoch": 0.15216205404225916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9227486582989488e-05,
+      "loss": 1.0877,
+      "step": 4184
+    },
+    {
+      "epoch": 0.15219842164599776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9227032568413028e-05,
+      "loss": 1.2203,
+      "step": 4185
+    },
+    {
+      "epoch": 0.15223478924973632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9226578425824584e-05,
+      "loss": 1.0554,
+      "step": 4186
+    },
+    {
+      "epoch": 0.15227115685347492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922612415523045e-05,
+      "loss": 1.1578,
+      "step": 4187
+    },
+    {
+      "epoch": 0.15230752445721352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922566975663694e-05,
+      "loss": 1.0983,
+      "step": 4188
+    },
+    {
+      "epoch": 0.1523438920609521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9225215230050348e-05,
+      "loss": 1.0146,
+      "step": 4189
+    },
+    {
+      "epoch": 0.15238025966469068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9224760575476986e-05,
+      "loss": 1.0337,
+      "step": 4190
+    },
+    {
+      "epoch": 0.15241662726842928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922430579292316e-05,
+      "loss": 1.1169,
+      "step": 4191
+    },
+    {
+      "epoch": 0.15245299487216787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9223850882395174e-05,
+      "loss": 1.2264,
+      "step": 4192
+    },
+    {
+      "epoch": 0.15248936247590647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9223395843899348e-05,
+      "loss": 1.1146,
+      "step": 4193
+    },
+    {
+      "epoch": 0.15252573007964504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922294067744199e-05,
+      "loss": 0.8886,
+      "step": 4194
+    },
+    {
+      "epoch": 0.15256209768338364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9222485383029417e-05,
+      "loss": 1.1074,
+      "step": 4195
+    },
+    {
+      "epoch": 0.15259846528712223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922202996066794e-05,
+      "loss": 1.181,
+      "step": 4196
+    },
+    {
+      "epoch": 0.15263483289086083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9221574410363886e-05,
+      "loss": 0.953,
+      "step": 4197
+    },
+    {
+      "epoch": 0.15267120049459942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922111873212357e-05,
+      "loss": 1.2589,
+      "step": 4198
+    },
+    {
+      "epoch": 0.152707568098338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9220662925953314e-05,
+      "loss": 1.0783,
+      "step": 4199
+    },
+    {
+      "epoch": 0.1527439357020766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922020699185944e-05,
+      "loss": 1.2291,
+      "step": 4200
+    },
+    {
+      "epoch": 0.15278030330581518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921975092984828e-05,
+      "loss": 1.3477,
+      "step": 4201
+    },
+    {
+      "epoch": 0.15281667090955378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9219294739926154e-05,
+      "loss": 1.0444,
+      "step": 4202
+    },
+    {
+      "epoch": 0.15285303851329235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9218838422099395e-05,
+      "loss": 1.1823,
+      "step": 4203
+    },
+    {
+      "epoch": 0.15288940611703095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9218381976374333e-05,
+      "loss": 1.0818,
+      "step": 4204
+    },
+    {
+      "epoch": 0.15292577372076954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92179254027573e-05,
+      "loss": 1.2513,
+      "step": 4205
+    },
+    {
+      "epoch": 0.15296214132450814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921746870125463e-05,
+      "loss": 1.0833,
+      "step": 4206
+    },
+    {
+      "epoch": 0.1529985089282467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921701187187266e-05,
+      "loss": 1.0955,
+      "step": 4207
+    },
+    {
+      "epoch": 0.1530348765319853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9216554914617726e-05,
+      "loss": 1.1124,
+      "step": 4208
+    },
+    {
+      "epoch": 0.1530712441357239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921609782949617e-05,
+      "loss": 1.0531,
+      "step": 4209
+    },
+    {
+      "epoch": 0.1531076117394625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9215640616514336e-05,
+      "loss": 1.1611,
+      "step": 4210
+    },
+    {
+      "epoch": 0.15314397934320106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921518327567856e-05,
+      "loss": 1.163,
+      "step": 4211
+    },
+    {
+      "epoch": 0.15318034694693966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921472580699519e-05,
+      "loss": 1.0774,
+      "step": 4212
+    },
+    {
+      "epoch": 0.15321671455067826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9214268210470573e-05,
+      "loss": 1.1043,
+      "step": 4213
+    },
+    {
+      "epoch": 0.15325308215441685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9213810486111063e-05,
+      "loss": 1.1813,
+      "step": 4214
+    },
+    {
+      "epoch": 0.15328944975815542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9213352633923002e-05,
+      "loss": 1.1205,
+      "step": 4215
+    },
+    {
+      "epoch": 0.15332581736189402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9212894653912746e-05,
+      "loss": 1.0617,
+      "step": 4216
+    },
+    {
+      "epoch": 0.15336218496563261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921243654608665e-05,
+      "loss": 1.0559,
+      "step": 4217
+    },
+    {
+      "epoch": 0.1533985525693712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921197831045106e-05,
+      "loss": 1.0856,
+      "step": 4218
+    },
+    {
+      "epoch": 0.1534349201731098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9211519947012348e-05,
+      "loss": 1.1959,
+      "step": 4219
+    },
+    {
+      "epoch": 0.15347128777684838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9211061455776865e-05,
+      "loss": 1.1593,
+      "step": 4220
+    },
+    {
+      "epoch": 0.15350765538058697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921060283675097e-05,
+      "loss": 1.1724,
+      "step": 4221
+    },
+    {
+      "epoch": 0.15354402298432557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9210144089941033e-05,
+      "loss": 1.2666,
+      "step": 4222
+    },
+    {
+      "epoch": 0.15358039058806416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9209685215353413e-05,
+      "loss": 1.1807,
+      "step": 4223
+    },
+    {
+      "epoch": 0.15361675819180273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9209226212994475e-05,
+      "loss": 1.2269,
+      "step": 4224
+    },
+    {
+      "epoch": 0.15365312579554133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920876708287059e-05,
+      "loss": 1.1266,
+      "step": 4225
+    },
+    {
+      "epoch": 0.15368949339927993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920830782498813e-05,
+      "loss": 1.0621,
+      "step": 4226
+    },
+    {
+      "epoch": 0.15372586100301852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9207848439353466e-05,
+      "loss": 1.0587,
+      "step": 4227
+    },
+    {
+      "epoch": 0.1537622286067571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9207388925972964e-05,
+      "loss": 1.1589,
+      "step": 4228
+    },
+    {
+      "epoch": 0.15379859621049569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9206929284853006e-05,
+      "loss": 1.0521,
+      "step": 4229
+    },
+    {
+      "epoch": 0.15383496381423428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9206469515999964e-05,
+      "loss": 1.1334,
+      "step": 4230
+    },
+    {
+      "epoch": 0.15387133141797288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9206009619420226e-05,
+      "loss": 1.1113,
+      "step": 4231
+    },
+    {
+      "epoch": 0.15390769902171145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9205549595120163e-05,
+      "loss": 1.1039,
+      "step": 4232
+    },
+    {
+      "epoch": 0.15394406662545004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920508944310616e-05,
+      "loss": 1.1234,
+      "step": 4233
+    },
+    {
+      "epoch": 0.15398043422918864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.92046291633846e-05,
+      "loss": 1.1815,
+      "step": 4234
+    },
+    {
+      "epoch": 0.15401680183292724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9204168755961873e-05,
+      "loss": 1.1582,
+      "step": 4235
+    },
+    {
+      "epoch": 0.1540531694366658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9203708220844358e-05,
+      "loss": 1.0921,
+      "step": 4236
+    },
+    {
+      "epoch": 0.1540895370404044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9203247558038456e-05,
+      "loss": 1.1345,
+      "step": 4237
+    },
+    {
+      "epoch": 0.154125904644143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920278676755055e-05,
+      "loss": 1.2227,
+      "step": 4238
+    },
+    {
+      "epoch": 0.1541622722478816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9202325849387034e-05,
+      "loss": 1.0819,
+      "step": 4239
+    },
+    {
+      "epoch": 0.1541986398516202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9201864803554304e-05,
+      "loss": 1.1856,
+      "step": 4240
+    },
+    {
+      "epoch": 0.15423500745535876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9201403630058753e-05,
+      "loss": 1.0755,
+      "step": 4241
+    },
+    {
+      "epoch": 0.15427137505909735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9200942328906786e-05,
+      "loss": 1.1472,
+      "step": 4242
+    },
+    {
+      "epoch": 0.15430774266283595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9200480900104795e-05,
+      "loss": 1.144,
+      "step": 4243
+    },
+    {
+      "epoch": 0.15434411026657455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9200019343659188e-05,
+      "loss": 1.0811,
+      "step": 4244
+    },
+    {
+      "epoch": 0.15438047787031312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9199557659576364e-05,
+      "loss": 1.0508,
+      "step": 4245
+    },
+    {
+      "epoch": 0.1544168454740517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919909584786273e-05,
+      "loss": 1.2061,
+      "step": 4246
+    },
+    {
+      "epoch": 0.1544532130777903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9198633908524693e-05,
+      "loss": 1.1318,
+      "step": 4247
+    },
+    {
+      "epoch": 0.1544895806815289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919817184156866e-05,
+      "loss": 1.0422,
+      "step": 4248
+    },
+    {
+      "epoch": 0.15452594828526747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9197709647001047e-05,
+      "loss": 1.2459,
+      "step": 4249
+    },
+    {
+      "epoch": 0.15456231588900607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919724732482826e-05,
+      "loss": 1.0653,
+      "step": 4250
+    },
+    {
+      "epoch": 0.15459868349274467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919678487505672e-05,
+      "loss": 1.1404,
+      "step": 4251
+    },
+    {
+      "epoch": 0.15463505109648326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919632229769283e-05,
+      "loss": 1.1683,
+      "step": 4252
+    },
+    {
+      "epoch": 0.15467141870022183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919585959274302e-05,
+      "loss": 1.1057,
+      "step": 4253
+    },
+    {
+      "epoch": 0.15470778630396043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9195396760213708e-05,
+      "loss": 1.1359,
+      "step": 4254
+    },
+    {
+      "epoch": 0.15474415390769902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9194933800111308e-05,
+      "loss": 1.2108,
+      "step": 4255
+    },
+    {
+      "epoch": 0.15478052151143762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9194470712442253e-05,
+      "loss": 1.2219,
+      "step": 4256
+    },
+    {
+      "epoch": 0.1548168891151762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919400749721296e-05,
+      "loss": 1.1371,
+      "step": 4257
+    },
+    {
+      "epoch": 0.15485325671891478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9193544154429853e-05,
+      "loss": 1.0967,
+      "step": 4258
+    },
+    {
+      "epoch": 0.15488962432265338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919308068409937e-05,
+      "loss": 1.1858,
+      "step": 4259
+    },
+    {
+      "epoch": 0.15492599192639198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9192617086227933e-05,
+      "loss": 1.0568,
+      "step": 4260
+    },
+    {
+      "epoch": 0.15496235953013057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9192153360821975e-05,
+      "loss": 1.2333,
+      "step": 4261
+    },
+    {
+      "epoch": 0.15499872713386914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9191689507887936e-05,
+      "loss": 1.1237,
+      "step": 4262
+    },
+    {
+      "epoch": 0.15503509473760774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9191225527432246e-05,
+      "loss": 0.9942,
+      "step": 4263
+    },
+    {
+      "epoch": 0.15507146234134633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919076141946134e-05,
+      "loss": 1.1286,
+      "step": 4264
+    },
+    {
+      "epoch": 0.15510782994508493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919029718398166e-05,
+      "loss": 1.1821,
+      "step": 4265
+    },
+    {
+      "epoch": 0.1551441975488235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9189832820999645e-05,
+      "loss": 1.1072,
+      "step": 4266
+    },
+    {
+      "epoch": 0.1551805651525621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918936833052174e-05,
+      "loss": 1.2089,
+      "step": 4267
+    },
+    {
+      "epoch": 0.1552169327563007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9188903712554385e-05,
+      "loss": 1.1613,
+      "step": 4268
+    },
+    {
+      "epoch": 0.1552533003600393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918843896710403e-05,
+      "loss": 1.1465,
+      "step": 4269
+    },
+    {
+      "epoch": 0.15528966796377786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918797409417712e-05,
+      "loss": 1.1104,
+      "step": 4270
+    },
+    {
+      "epoch": 0.15532603556751645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9187509093780107e-05,
+      "loss": 1.1638,
+      "step": 4271
+    },
+    {
+      "epoch": 0.15536240317125505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918704396591944e-05,
+      "loss": 1.1605,
+      "step": 4272
+    },
+    {
+      "epoch": 0.15539877077499364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9186578710601574e-05,
+      "loss": 1.1665,
+      "step": 4273
+    },
+    {
+      "epoch": 0.1554351383787322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918611332783296e-05,
+      "loss": 1.2733,
+      "step": 4274
+    },
+    {
+      "epoch": 0.1554715059824708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918564781762006e-05,
+      "loss": 1.1747,
+      "step": 4275
+    },
+    {
+      "epoch": 0.1555078735862094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9185182179969325e-05,
+      "loss": 1.1265,
+      "step": 4276
+    },
+    {
+      "epoch": 0.155544241189948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918471641488722e-05,
+      "loss": 1.2595,
+      "step": 4277
+    },
+    {
+      "epoch": 0.1555806087936866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918425052238021e-05,
+      "loss": 1.1203,
+      "step": 4278
+    },
+    {
+      "epoch": 0.15561697639742517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9183784502454754e-05,
+      "loss": 1.1372,
+      "step": 4279
+    },
+    {
+      "epoch": 0.15565334400116376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9183318355117315e-05,
+      "loss": 1.2285,
+      "step": 4280
+    },
+    {
+      "epoch": 0.15568971160490236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9182852080374364e-05,
+      "loss": 1.2141,
+      "step": 4281
+    },
+    {
+      "epoch": 0.15572607920864096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9182385678232373e-05,
+      "loss": 1.1724,
+      "step": 4282
+    },
+    {
+      "epoch": 0.15576244681237952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91819191486978e-05,
+      "loss": 1.1219,
+      "step": 4283
+    },
+    {
+      "epoch": 0.15579881441611812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9181452491777136e-05,
+      "loss": 1.1753,
+      "step": 4284
+    },
+    {
+      "epoch": 0.15583518201985672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918098570747684e-05,
+      "loss": 1.0517,
+      "step": 4285
+    },
+    {
+      "epoch": 0.1558715496235953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9180518795803395e-05,
+      "loss": 1.1447,
+      "step": 4286
+    },
+    {
+      "epoch": 0.15590791722733388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918005175676328e-05,
+      "loss": 1.0156,
+      "step": 4287
+    },
+    {
+      "epoch": 0.15594428483107248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917958459036297e-05,
+      "loss": 1.0348,
+      "step": 4288
+    },
+    {
+      "epoch": 0.15598065243481107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9179117296608943e-05,
+      "loss": 1.1486,
+      "step": 4289
+    },
+    {
+      "epoch": 0.15601702003854967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917864987550769e-05,
+      "loss": 1.1093,
+      "step": 4290
+    },
+    {
+      "epoch": 0.15605338764228824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9178182327065695e-05,
+      "loss": 1.1221,
+      "step": 4291
+    },
+    {
+      "epoch": 0.15608975524602683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917771465128944e-05,
+      "loss": 1.1855,
+      "step": 4292
+    },
+    {
+      "epoch": 0.15612612284976543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917724684818542e-05,
+      "loss": 1.0632,
+      "step": 4293
+    },
+    {
+      "epoch": 0.15616249045350403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9176778917760116e-05,
+      "loss": 1.0277,
+      "step": 4294
+    },
+    {
+      "epoch": 0.1561988580572426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9176310860020028e-05,
+      "loss": 1.1642,
+      "step": 4295
+    },
+    {
+      "epoch": 0.1562352256609812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917584267497164e-05,
+      "loss": 1.1313,
+      "step": 4296
+    },
+    {
+      "epoch": 0.1562715932647198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917537436262146e-05,
+      "loss": 1.1293,
+      "step": 4297
+    },
+    {
+      "epoch": 0.15630796086845838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917490592297598e-05,
+      "loss": 1.0471,
+      "step": 4298
+    },
+    {
+      "epoch": 0.15634432847219698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9174437356041692e-05,
+      "loss": 1.2119,
+      "step": 4299
+    },
+    {
+      "epoch": 0.15638069607593555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9173968661825103e-05,
+      "loss": 0.984,
+      "step": 4300
+    },
+    {
+      "epoch": 0.15641706367967415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917349984033272e-05,
+      "loss": 1.1031,
+      "step": 4301
+    },
+    {
+      "epoch": 0.15645343128341274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9173030891571037e-05,
+      "loss": 1.057,
+      "step": 4302
+    },
+    {
+      "epoch": 0.15648979888715134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9172561815546568e-05,
+      "loss": 1.1391,
+      "step": 4303
+    },
+    {
+      "epoch": 0.1565261664908899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9172092612265816e-05,
+      "loss": 1.1457,
+      "step": 4304
+    },
+    {
+      "epoch": 0.1565625340946285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9171623281735295e-05,
+      "loss": 1.0655,
+      "step": 4305
+    },
+    {
+      "epoch": 0.1565989016983671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9171153823961513e-05,
+      "loss": 1.2361,
+      "step": 4306
+    },
+    {
+      "epoch": 0.1566352693021057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917068423895098e-05,
+      "loss": 1.1815,
+      "step": 4307
+    },
+    {
+      "epoch": 0.15667163690584426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9170214526710223e-05,
+      "loss": 0.9803,
+      "step": 4308
+    },
+    {
+      "epoch": 0.15670800450958286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9169744687245746e-05,
+      "loss": 1.1142,
+      "step": 4309
+    },
+    {
+      "epoch": 0.15674437211332146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916927472056407e-05,
+      "loss": 1.1215,
+      "step": 4310
+    },
+    {
+      "epoch": 0.15678073971706005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9168804626671718e-05,
+      "loss": 1.0417,
+      "step": 4311
+    },
+    {
+      "epoch": 0.15681710732079862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9168334405575212e-05,
+      "loss": 1.1554,
+      "step": 4312
+    },
+    {
+      "epoch": 0.15685347492453722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9167864057281072e-05,
+      "loss": 1.1325,
+      "step": 4313
+    },
+    {
+      "epoch": 0.1568898425282758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9167393581795828e-05,
+      "loss": 1.0245,
+      "step": 4314
+    },
+    {
+      "epoch": 0.1569262101320144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9166922979126005e-05,
+      "loss": 1.2504,
+      "step": 4315
+    },
+    {
+      "epoch": 0.15696257773575298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916645224927813e-05,
+      "loss": 1.1333,
+      "step": 4316
+    },
+    {
+      "epoch": 0.15699894533949157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9165981392258736e-05,
+      "loss": 1.0356,
+      "step": 4317
+    },
+    {
+      "epoch": 0.15703531294323017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9165510408074354e-05,
+      "loss": 1.1117,
+      "step": 4318
+    },
+    {
+      "epoch": 0.15707168054696877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916503929673152e-05,
+      "loss": 1.1609,
+      "step": 4319
+    },
+    {
+      "epoch": 0.15710804815070736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916456805823677e-05,
+      "loss": 1.1302,
+      "step": 4320
+    },
+    {
+      "epoch": 0.15714441575444593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916409669259664e-05,
+      "loss": 1.2049,
+      "step": 4321
+    },
+    {
+      "epoch": 0.15718078335818453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916362519981767e-05,
+      "loss": 1.1601,
+      "step": 4322
+    },
+    {
+      "epoch": 0.15721715096192312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9163153579906404e-05,
+      "loss": 1.0982,
+      "step": 4323
+    },
+    {
+      "epoch": 0.15725351856566172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916268183286938e-05,
+      "loss": 1.0026,
+      "step": 4324
+    },
+    {
+      "epoch": 0.1572898861694003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916220995871315e-05,
+      "loss": 1.1374,
+      "step": 4325
+    },
+    {
+      "epoch": 0.15732625377313889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9161737957444252e-05,
+      "loss": 1.1748,
+      "step": 4326
+    },
+    {
+      "epoch": 0.15736262137687748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.916126582906924e-05,
+      "loss": 1.0126,
+      "step": 4327
+    },
+    {
+      "epoch": 0.15739898898061608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9160793573594664e-05,
+      "loss": 1.1658,
+      "step": 4328
+    },
+    {
+      "epoch": 0.15743535658435465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9160321191027073e-05,
+      "loss": 1.2108,
+      "step": 4329
+    },
+    {
+      "epoch": 0.15747172418809324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9159848681373022e-05,
+      "loss": 1.1614,
+      "step": 4330
+    },
+    {
+      "epoch": 0.15750809179183184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915937604463907e-05,
+      "loss": 1.17,
+      "step": 4331
+    },
+    {
+      "epoch": 0.15754445939557044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915890328083177e-05,
+      "loss": 1.0936,
+      "step": 4332
+    },
+    {
+      "epoch": 0.157580826999309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9158430389957677e-05,
+      "loss": 1.2059,
+      "step": 4333
+    },
+    {
+      "epoch": 0.1576171946030476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915795737202336e-05,
+      "loss": 1.2056,
+      "step": 4334
+    },
+    {
+      "epoch": 0.1576535622067862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9157484227035378e-05,
+      "loss": 1.1329,
+      "step": 4335
+    },
+    {
+      "epoch": 0.1576899298105248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9157010955000294e-05,
+      "loss": 1.0653,
+      "step": 4336
+    },
+    {
+      "epoch": 0.15772629741426336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9156537555924677e-05,
+      "loss": 1.1821,
+      "step": 4337
+    },
+    {
+      "epoch": 0.15776266501800196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915606402981509e-05,
+      "loss": 1.2272,
+      "step": 4338
+    },
+    {
+      "epoch": 0.15779903262174055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915559037667811e-05,
+      "loss": 1.1625,
+      "step": 4339
+    },
+    {
+      "epoch": 0.15783540022547915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91551165965203e-05,
+      "loss": 1.2119,
+      "step": 4340
+    },
+    {
+      "epoch": 0.15787176782921775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9154642689348236e-05,
+      "loss": 1.1329,
+      "step": 4341
+    },
+    {
+      "epoch": 0.15790813543295631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9154168655168495e-05,
+      "loss": 1.1053,
+      "step": 4342
+    },
+    {
+      "epoch": 0.1579445030366949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9153694493987653e-05,
+      "loss": 1.0517,
+      "step": 4343
+    },
+    {
+      "epoch": 0.1579808706404335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9153220205812285e-05,
+      "loss": 1.3071,
+      "step": 4344
+    },
+    {
+      "epoch": 0.1580172382441721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9152745790648973e-05,
+      "loss": 1.145,
+      "step": 4345
+    },
+    {
+      "epoch": 0.15805360584791067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9152271248504302e-05,
+      "loss": 1.2153,
+      "step": 4346
+    },
+    {
+      "epoch": 0.15808997345164927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915179657938485e-05,
+      "loss": 1.2274,
+      "step": 4347
+    },
+    {
+      "epoch": 0.15812634105538786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9151321783297206e-05,
+      "loss": 1.2175,
+      "step": 4348
+    },
+    {
+      "epoch": 0.15816270865912646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9150846860247958e-05,
+      "loss": 1.129,
+      "step": 4349
+    },
+    {
+      "epoch": 0.15819907626286503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9150371810243694e-05,
+      "loss": 1.0594,
+      "step": 4350
+    },
+    {
+      "epoch": 0.15823544386660363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9149896633291e-05,
+      "loss": 1.1028,
+      "step": 4351
+    },
+    {
+      "epoch": 0.15827181147034222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9149421329396474e-05,
+      "loss": 1.1061,
+      "step": 4352
+    },
+    {
+      "epoch": 0.15830817907408082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9148945898566708e-05,
+      "loss": 1.2775,
+      "step": 4353
+    },
+    {
+      "epoch": 0.1583445466778194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91484703408083e-05,
+      "loss": 1.1516,
+      "step": 4354
+    },
+    {
+      "epoch": 0.15838091428155798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9147994656127846e-05,
+      "loss": 1.223,
+      "step": 4355
+    },
+    {
+      "epoch": 0.15841728188529658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9147518844531942e-05,
+      "loss": 1.1017,
+      "step": 4356
+    },
+    {
+      "epoch": 0.15845364948903518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9147042906027193e-05,
+      "loss": 1.1984,
+      "step": 4357
+    },
+    {
+      "epoch": 0.15849001709277374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9146566840620204e-05,
+      "loss": 1.2969,
+      "step": 4358
+    },
+    {
+      "epoch": 0.15852638469651234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914609064831758e-05,
+      "loss": 1.1323,
+      "step": 4359
+    },
+    {
+      "epoch": 0.15856275230025094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914561432912592e-05,
+      "loss": 1.1827,
+      "step": 4360
+    },
+    {
+      "epoch": 0.15859911990398953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914513788305184e-05,
+      "loss": 1.1977,
+      "step": 4361
+    },
+    {
+      "epoch": 0.15863548750772813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9144661310101943e-05,
+      "loss": 1.1714,
+      "step": 4362
+    },
+    {
+      "epoch": 0.1586718551114667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914418461028285e-05,
+      "loss": 1.1206,
+      "step": 4363
+    },
+    {
+      "epoch": 0.1587082227152053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9143707783601167e-05,
+      "loss": 1.1041,
+      "step": 4364
+    },
+    {
+      "epoch": 0.1587445903189439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914323083006351e-05,
+      "loss": 1.2485,
+      "step": 4365
+    },
+    {
+      "epoch": 0.1587809579226825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9142753749676502e-05,
+      "loss": 1.1115,
+      "step": 4366
+    },
+    {
+      "epoch": 0.15881732552642105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9142276542446753e-05,
+      "loss": 1.1631,
+      "step": 4367
+    },
+    {
+      "epoch": 0.15885369313015965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9141799208380894e-05,
+      "loss": 1.1599,
+      "step": 4368
+    },
+    {
+      "epoch": 0.15889006073389825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9141321747485536e-05,
+      "loss": 0.9864,
+      "step": 4369
+    },
+    {
+      "epoch": 0.15892642833763684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9140844159767312e-05,
+      "loss": 1.1875,
+      "step": 4370
+    },
+    {
+      "epoch": 0.1589627959413754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.914036644523284e-05,
+      "loss": 1.0275,
+      "step": 4371
+    },
+    {
+      "epoch": 0.158999163545114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9139888603888755e-05,
+      "loss": 1.228,
+      "step": 4372
+    },
+    {
+      "epoch": 0.1590355311488526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9139410635741686e-05,
+      "loss": 1.0948,
+      "step": 4373
+    },
+    {
+      "epoch": 0.1590718987525912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913893254079826e-05,
+      "loss": 1.1142,
+      "step": 4374
+    },
+    {
+      "epoch": 0.15910826635632977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913845431906511e-05,
+      "loss": 1.0983,
+      "step": 4375
+    },
+    {
+      "epoch": 0.15914463396006837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913797597054887e-05,
+      "loss": 1.2149,
+      "step": 4376
+    },
+    {
+      "epoch": 0.15918100156380696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9137497495256183e-05,
+      "loss": 1.1167,
+      "step": 4377
+    },
+    {
+      "epoch": 0.15921736916754556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913701889319368e-05,
+      "loss": 1.0697,
+      "step": 4378
+    },
+    {
+      "epoch": 0.15925373677128413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913654016436801e-05,
+      "loss": 1.2118,
+      "step": 4379
+    },
+    {
+      "epoch": 0.15929010437502272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9136061308785798e-05,
+      "loss": 1.0566,
+      "step": 4380
+    },
+    {
+      "epoch": 0.15932647197876132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9135582326453702e-05,
+      "loss": 1.2206,
+      "step": 4381
+    },
+    {
+      "epoch": 0.15936283958249992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9135103217378362e-05,
+      "loss": 1.271,
+      "step": 4382
+    },
+    {
+      "epoch": 0.1593992071862385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913462398156643e-05,
+      "loss": 1.2188,
+      "step": 4383
+    },
+    {
+      "epoch": 0.15943557478997708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9134144619024547e-05,
+      "loss": 1.0818,
+      "step": 4384
+    },
+    {
+      "epoch": 0.15947194239371568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913366512975937e-05,
+      "loss": 1.058,
+      "step": 4385
+    },
+    {
+      "epoch": 0.15950830999745427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9133185513777542e-05,
+      "loss": 1.0919,
+      "step": 4386
+    },
+    {
+      "epoch": 0.15954467760119287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9132705771085728e-05,
+      "loss": 1.2704,
+      "step": 4387
+    },
+    {
+      "epoch": 0.15958104520493144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913222590169058e-05,
+      "loss": 1.1756,
+      "step": 4388
+    },
+    {
+      "epoch": 0.15961741280867003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913174590559875e-05,
+      "loss": 1.0677,
+      "step": 4389
+    },
+    {
+      "epoch": 0.15965378041240863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9131265782816908e-05,
+      "loss": 1.2156,
+      "step": 4390
+    },
+    {
+      "epoch": 0.15969014801614723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9130785533351706e-05,
+      "loss": 1.1008,
+      "step": 4391
+    },
+    {
+      "epoch": 0.1597265156198858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913030515720981e-05,
+      "loss": 1.1826,
+      "step": 4392
+    },
+    {
+      "epoch": 0.1597628832236244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9129824654397882e-05,
+      "loss": 1.1168,
+      "step": 4393
+    },
+    {
+      "epoch": 0.159799250827363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9129344024922593e-05,
+      "loss": 1.1986,
+      "step": 4394
+    },
+    {
+      "epoch": 0.15983561843110158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9128863268790606e-05,
+      "loss": 1.188,
+      "step": 4395
+    },
+    {
+      "epoch": 0.15987198603484015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9128382386008596e-05,
+      "loss": 1.2542,
+      "step": 4396
+    },
+    {
+      "epoch": 0.15990835363857875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9127901376583228e-05,
+      "loss": 1.1179,
+      "step": 4397
+    },
+    {
+      "epoch": 0.15994472124231734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9127420240521183e-05,
+      "loss": 1.0628,
+      "step": 4398
+    },
+    {
+      "epoch": 0.15998108884605594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912693897782913e-05,
+      "loss": 1.1769,
+      "step": 4399
+    },
+    {
+      "epoch": 0.1600174564497945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912645758851375e-05,
+      "loss": 1.0464,
+      "step": 4400
+    },
+    {
+      "epoch": 0.1600538240535331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912597607258172e-05,
+      "loss": 1.0491,
+      "step": 4401
+    },
+    {
+      "epoch": 0.1600901916572717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9125494430039717e-05,
+      "loss": 1.1482,
+      "step": 4402
+    },
+    {
+      "epoch": 0.1601265592610103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9125012660894428e-05,
+      "loss": 1.0349,
+      "step": 4403
+    },
+    {
+      "epoch": 0.1601629268647489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9124530765152537e-05,
+      "loss": 1.0928,
+      "step": 4404
+    },
+    {
+      "epoch": 0.16019929446848746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9124048742820725e-05,
+      "loss": 1.0984,
+      "step": 4405
+    },
+    {
+      "epoch": 0.16023566207222606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9123566593905684e-05,
+      "loss": 1.0859,
+      "step": 4406
+    },
+    {
+      "epoch": 0.16027202967596466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9123084318414097e-05,
+      "loss": 1.151,
+      "step": 4407
+    },
+    {
+      "epoch": 0.16030839727970325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912260191635266e-05,
+      "loss": 1.2917,
+      "step": 4408
+    },
+    {
+      "epoch": 0.16034476488344182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912211938772807e-05,
+      "loss": 1.1969,
+      "step": 4409
+    },
+    {
+      "epoch": 0.16038113248718042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9121636732547014e-05,
+      "loss": 1.1254,
+      "step": 4410
+    },
+    {
+      "epoch": 0.160417500090919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9121153950816188e-05,
+      "loss": 1.0487,
+      "step": 4411
+    },
+    {
+      "epoch": 0.1604538676946576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9120671042542293e-05,
+      "loss": 1.1128,
+      "step": 4412
+    },
+    {
+      "epoch": 0.16049023529839618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.912018800773203e-05,
+      "loss": 1.2099,
+      "step": 4413
+    },
+    {
+      "epoch": 0.16052660290213477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9119704846392096e-05,
+      "loss": 1.1332,
+      "step": 4414
+    },
+    {
+      "epoch": 0.16056297050587337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9119221558529198e-05,
+      "loss": 1.0602,
+      "step": 4415
+    },
+    {
+      "epoch": 0.16059933810961197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9118738144150037e-05,
+      "loss": 1.0951,
+      "step": 4416
+    },
+    {
+      "epoch": 0.16063570571335054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9118254603261324e-05,
+      "loss": 1.1464,
+      "step": 4417
+    },
+    {
+      "epoch": 0.16067207331708913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9117770935869763e-05,
+      "loss": 1.08,
+      "step": 4418
+    },
+    {
+      "epoch": 0.16070844092082773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9117287141982067e-05,
+      "loss": 1.0966,
+      "step": 4419
+    },
+    {
+      "epoch": 0.16074480852456632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911680322160495e-05,
+      "loss": 1.0715,
+      "step": 4420
+    },
+    {
+      "epoch": 0.1607811761283049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911631917474512e-05,
+      "loss": 1.2045,
+      "step": 4421
+    },
+    {
+      "epoch": 0.1608175437320435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91158350014093e-05,
+      "loss": 1.1091,
+      "step": 4422
+    },
+    {
+      "epoch": 0.16085391133578208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.91153507016042e-05,
+      "loss": 1.1432,
+      "step": 4423
+    },
+    {
+      "epoch": 0.16089027893952068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9114866275336546e-05,
+      "loss": 1.0886,
+      "step": 4424
+    },
+    {
+      "epoch": 0.16092664654325928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911438172261305e-05,
+      "loss": 1.1412,
+      "step": 4425
+    },
+    {
+      "epoch": 0.16096301414699785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9113897043440442e-05,
+      "loss": 1.2511,
+      "step": 4426
+    },
+    {
+      "epoch": 0.16099938175073644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911341223782544e-05,
+      "loss": 1.1592,
+      "step": 4427
+    },
+    {
+      "epoch": 0.16103574935447504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911292730577478e-05,
+      "loss": 1.1779,
+      "step": 4428
+    },
+    {
+      "epoch": 0.16107211695821363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9112442247295177e-05,
+      "loss": 1.131,
+      "step": 4429
+    },
+    {
+      "epoch": 0.1611084845619522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.911195706239337e-05,
+      "loss": 1.0486,
+      "step": 4430
+    },
+    {
+      "epoch": 0.1611448521656908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9111471751076083e-05,
+      "loss": 0.9729,
+      "step": 4431
+    },
+    {
+      "epoch": 0.1611812197694294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9110986313350056e-05,
+      "loss": 1.1917,
+      "step": 4432
+    },
+    {
+      "epoch": 0.161217587373168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9110500749222024e-05,
+      "loss": 1.0209,
+      "step": 4433
+    },
+    {
+      "epoch": 0.16125395497690656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9110015058698715e-05,
+      "loss": 1.1515,
+      "step": 4434
+    },
+    {
+      "epoch": 0.16129032258064516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9109529241786876e-05,
+      "loss": 0.9921,
+      "step": 4435
+    },
+    {
+      "epoch": 0.16132669018438375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910904329849324e-05,
+      "loss": 1.1212,
+      "step": 4436
+    },
+    {
+      "epoch": 0.16136305778812235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9108557228824552e-05,
+      "loss": 1.088,
+      "step": 4437
+    },
+    {
+      "epoch": 0.16139942539186092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910807103278756e-05,
+      "loss": 1.1534,
+      "step": 4438
+    },
+    {
+      "epoch": 0.16143579299559951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9107584710389e-05,
+      "loss": 1.0503,
+      "step": 4439
+    },
+    {
+      "epoch": 0.1614721605993381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910709826163563e-05,
+      "loss": 1.2065,
+      "step": 4440
+    },
+    {
+      "epoch": 0.1615085282030767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9106611686534188e-05,
+      "loss": 1.1027,
+      "step": 4441
+    },
+    {
+      "epoch": 0.16154489580681528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9106124985091432e-05,
+      "loss": 1.0861,
+      "step": 4442
+    },
+    {
+      "epoch": 0.16158126341055387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910563815731411e-05,
+      "loss": 1.1908,
+      "step": 4443
+    },
+    {
+      "epoch": 0.16161763101429247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910515120320898e-05,
+      "loss": 1.1253,
+      "step": 4444
+    },
+    {
+      "epoch": 0.16165399861803106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910466412278279e-05,
+      "loss": 1.1187,
+      "step": 4445
+    },
+    {
+      "epoch": 0.16169036622176966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910417691604231e-05,
+      "loss": 1.1157,
+      "step": 4446
+    },
+    {
+      "epoch": 0.16172673382550823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910368958299429e-05,
+      "loss": 1.0648,
+      "step": 4447
+    },
+    {
+      "epoch": 0.16176310142924683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.910320212364549e-05,
+      "loss": 1.0911,
+      "step": 4448
+    },
+    {
+      "epoch": 0.16179946903298542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9102714538002684e-05,
+      "loss": 1.1031,
+      "step": 4449
+    },
+    {
+      "epoch": 0.16183583663672402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9102226826072625e-05,
+      "loss": 1.2833,
+      "step": 4450
+    },
+    {
+      "epoch": 0.16187220424046259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9101738987862083e-05,
+      "loss": 1.119,
+      "step": 4451
+    },
+    {
+      "epoch": 0.16190857184420118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9101251023377824e-05,
+      "loss": 1.2259,
+      "step": 4452
+    },
+    {
+      "epoch": 0.16194493944793978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9100762932626624e-05,
+      "loss": 1.0793,
+      "step": 4453
+    },
+    {
+      "epoch": 0.16198130705167837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9100274715615246e-05,
+      "loss": 1.2395,
+      "step": 4454
+    },
+    {
+      "epoch": 0.16201767465541694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9099786372350474e-05,
+      "loss": 1.2209,
+      "step": 4455
+    },
+    {
+      "epoch": 0.16205404225915554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909929790283907e-05,
+      "loss": 1.0816,
+      "step": 4456
+    },
+    {
+      "epoch": 0.16209040986289414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9098809307087823e-05,
+      "loss": 1.1996,
+      "step": 4457
+    },
+    {
+      "epoch": 0.16212677746663273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9098320585103503e-05,
+      "loss": 0.995,
+      "step": 4458
+    },
+    {
+      "epoch": 0.1621631450703713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9097831736892894e-05,
+      "loss": 1.0223,
+      "step": 4459
+    },
+    {
+      "epoch": 0.1621995126741099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909734276246278e-05,
+      "loss": 1.12,
+      "step": 4460
+    },
+    {
+      "epoch": 0.1622358802778485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9096853661819942e-05,
+      "loss": 1.1347,
+      "step": 4461
+    },
+    {
+      "epoch": 0.1622722478815871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909636443497117e-05,
+      "loss": 1.1138,
+      "step": 4462
+    },
+    {
+      "epoch": 0.16230861548532569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909587508192324e-05,
+      "loss": 1.2147,
+      "step": 4463
+    },
+    {
+      "epoch": 0.16234498308906425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9095385602682952e-05,
+      "loss": 1.0549,
+      "step": 4464
+    },
+    {
+      "epoch": 0.16238135069280285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9094895997257095e-05,
+      "loss": 1.0841,
+      "step": 4465
+    },
+    {
+      "epoch": 0.16241771829654145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9094406265652457e-05,
+      "loss": 1.1934,
+      "step": 4466
+    },
+    {
+      "epoch": 0.16245408590028004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9093916407875838e-05,
+      "loss": 1.072,
+      "step": 4467
+    },
+    {
+      "epoch": 0.1624904535040186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909342642393403e-05,
+      "loss": 1.0229,
+      "step": 4468
+    },
+    {
+      "epoch": 0.1625268211077572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909293631383383e-05,
+      "loss": 1.1742,
+      "step": 4469
+    },
+    {
+      "epoch": 0.1625631887114958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9092446077582044e-05,
+      "loss": 1.0446,
+      "step": 4470
+    },
+    {
+      "epoch": 0.1625995563152344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909195571518547e-05,
+      "loss": 1.1708,
+      "step": 4471
+    },
+    {
+      "epoch": 0.16263592391897297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909146522665091e-05,
+      "loss": 1.1389,
+      "step": 4472
+    },
+    {
+      "epoch": 0.16267229152271157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9090974611985163e-05,
+      "loss": 1.1674,
+      "step": 4473
+    },
+    {
+      "epoch": 0.16270865912645016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9090483871195046e-05,
+      "loss": 1.1993,
+      "step": 4474
+    },
+    {
+      "epoch": 0.16274502673018876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908999300428736e-05,
+      "loss": 1.0525,
+      "step": 4475
+    },
+    {
+      "epoch": 0.16278139433392733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9089502011268923e-05,
+      "loss": 1.0334,
+      "step": 4476
+    },
+    {
+      "epoch": 0.16281776193766592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908901089214654e-05,
+      "loss": 1.2375,
+      "step": 4477
+    },
+    {
+      "epoch": 0.16285412954140452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9088519646927025e-05,
+      "loss": 1.0632,
+      "step": 4478
+    },
+    {
+      "epoch": 0.16289049714514311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9088028275617193e-05,
+      "loss": 1.0612,
+      "step": 4479
+    },
+    {
+      "epoch": 0.16292686474888168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9087536778223868e-05,
+      "loss": 1.1504,
+      "step": 4480
+    },
+    {
+      "epoch": 0.16296323235262028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9087045154753863e-05,
+      "loss": 1.0489,
+      "step": 4481
+    },
+    {
+      "epoch": 0.16299959995635888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9086553405213995e-05,
+      "loss": 1.1566,
+      "step": 4482
+    },
+    {
+      "epoch": 0.16303596756009747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9086061529611094e-05,
+      "loss": 1.0669,
+      "step": 4483
+    },
+    {
+      "epoch": 0.16307233516383607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9085569527951978e-05,
+      "loss": 1.2115,
+      "step": 4484
+    },
+    {
+      "epoch": 0.16310870276757464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908507740024348e-05,
+      "loss": 1.0755,
+      "step": 4485
+    },
+    {
+      "epoch": 0.16314507037131323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908458514649242e-05,
+      "loss": 1.0919,
+      "step": 4486
+    },
+    {
+      "epoch": 0.16318143797505183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9084092766705632e-05,
+      "loss": 1.1898,
+      "step": 4487
+    },
+    {
+      "epoch": 0.16321780557879043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9083600260889944e-05,
+      "loss": 1.1617,
+      "step": 4488
+    },
+    {
+      "epoch": 0.163254173182529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9083107629052192e-05,
+      "loss": 1.1776,
+      "step": 4489
+    },
+    {
+      "epoch": 0.1632905407862676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9082614871199208e-05,
+      "loss": 1.176,
+      "step": 4490
+    },
+    {
+      "epoch": 0.1633269083900062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9082121987337828e-05,
+      "loss": 1.13,
+      "step": 4491
+    },
+    {
+      "epoch": 0.16336327599374478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9081628977474893e-05,
+      "loss": 1.0741,
+      "step": 4492
+    },
+    {
+      "epoch": 0.16339964359748335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908113584161724e-05,
+      "loss": 1.1664,
+      "step": 4493
+    },
+    {
+      "epoch": 0.16343601120122195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9080642579771714e-05,
+      "loss": 1.1839,
+      "step": 4494
+    },
+    {
+      "epoch": 0.16347237880496054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.908014919194515e-05,
+      "loss": 1.287,
+      "step": 4495
+    },
+    {
+      "epoch": 0.16350874640869914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9079655678144403e-05,
+      "loss": 1.0881,
+      "step": 4496
+    },
+    {
+      "epoch": 0.1635451140124377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9079162038376316e-05,
+      "loss": 1.1391,
+      "step": 4497
+    },
+    {
+      "epoch": 0.1635814816161763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907866827264774e-05,
+      "loss": 1.202,
+      "step": 4498
+    },
+    {
+      "epoch": 0.1636178492199149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9078174380965518e-05,
+      "loss": 1.1404,
+      "step": 4499
+    },
+    {
+      "epoch": 0.1636542168236535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907768036333651e-05,
+      "loss": 1.1523,
+      "step": 4500
+    },
+    {
+      "epoch": 0.16369058442739207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9077186219767566e-05,
+      "loss": 1.1388,
+      "step": 4501
+    },
+    {
+      "epoch": 0.16372695203113066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907669195026554e-05,
+      "loss": 1.0463,
+      "step": 4502
+    },
+    {
+      "epoch": 0.16376331963486926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907619755483729e-05,
+      "loss": 1.1374,
+      "step": 4503
+    },
+    {
+      "epoch": 0.16379968723860786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907570303348968e-05,
+      "loss": 1.2156,
+      "step": 4504
+    },
+    {
+      "epoch": 0.16383605484234645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9075208386229567e-05,
+      "loss": 1.1658,
+      "step": 4505
+    },
+    {
+      "epoch": 0.16387242244608502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9074713613063814e-05,
+      "loss": 1.1392,
+      "step": 4506
+    },
+    {
+      "epoch": 0.16390879004982362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907421871399928e-05,
+      "loss": 0.9568,
+      "step": 4507
+    },
+    {
+      "epoch": 0.1639451576535622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9073723689042844e-05,
+      "loss": 1.1672,
+      "step": 4508
+    },
+    {
+      "epoch": 0.1639815252573008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907322853820136e-05,
+      "loss": 1.1143,
+      "step": 4509
+    },
+    {
+      "epoch": 0.16401789286103938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9072733261481705e-05,
+      "loss": 1.2398,
+      "step": 4510
+    },
+    {
+      "epoch": 0.16405426046477797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907223785889075e-05,
+      "loss": 1.186,
+      "step": 4511
+    },
+    {
+      "epoch": 0.16409062806851657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907174233043536e-05,
+      "loss": 1.1324,
+      "step": 4512
+    },
+    {
+      "epoch": 0.16412699567225517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907124667612242e-05,
+      "loss": 1.0567,
+      "step": 4513
+    },
+    {
+      "epoch": 0.16416336327599373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9070750895958804e-05,
+      "loss": 1.1297,
+      "step": 4514
+    },
+    {
+      "epoch": 0.16419973087973233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.907025498995139e-05,
+      "loss": 1.1308,
+      "step": 4515
+    },
+    {
+      "epoch": 0.16423609848347093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9069758958107058e-05,
+      "loss": 1.2936,
+      "step": 4516
+    },
+    {
+      "epoch": 0.16427246608720952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9069262800432688e-05,
+      "loss": 1.1464,
+      "step": 4517
+    },
+    {
+      "epoch": 0.1643088336909481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906876651693516e-05,
+      "loss": 1.0564,
+      "step": 4518
+    },
+    {
+      "epoch": 0.1643452012946867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9068270107621367e-05,
+      "loss": 1.1962,
+      "step": 4519
+    },
+    {
+      "epoch": 0.16438156889842528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9067773572498192e-05,
+      "loss": 1.2247,
+      "step": 4520
+    },
+    {
+      "epoch": 0.16441793650216388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906727691157252e-05,
+      "loss": 1.1581,
+      "step": 4521
+    },
+    {
+      "epoch": 0.16445430410590245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9066780124851254e-05,
+      "loss": 1.1116,
+      "step": 4522
+    },
+    {
+      "epoch": 0.16449067170964105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906628321234127e-05,
+      "loss": 1.1677,
+      "step": 4523
+    },
+    {
+      "epoch": 0.16452703931337964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9065786174049473e-05,
+      "loss": 1.0518,
+      "step": 4524
+    },
+    {
+      "epoch": 0.16456340691711824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906528900998276e-05,
+      "loss": 1.2879,
+      "step": 4525
+    },
+    {
+      "epoch": 0.16459977452085683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9064791720148018e-05,
+      "loss": 1.1714,
+      "step": 4526
+    },
+    {
+      "epoch": 0.1646361421245954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9064294304552153e-05,
+      "loss": 1.1281,
+      "step": 4527
+    },
+    {
+      "epoch": 0.164672509728334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9063796763202064e-05,
+      "loss": 1.0025,
+      "step": 4528
+    },
+    {
+      "epoch": 0.1647088773320726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9063299096104658e-05,
+      "loss": 0.9933,
+      "step": 4529
+    },
+    {
+      "epoch": 0.1647452449358112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9062801303266837e-05,
+      "loss": 1.1066,
+      "step": 4530
+    },
+    {
+      "epoch": 0.16478161253954976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90623033846955e-05,
+      "loss": 1.1684,
+      "step": 4531
+    },
+    {
+      "epoch": 0.16481798014328836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906180534039757e-05,
+      "loss": 1.1375,
+      "step": 4532
+    },
+    {
+      "epoch": 0.16485434774702695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906130717037994e-05,
+      "loss": 1.1707,
+      "step": 4533
+    },
+    {
+      "epoch": 0.16489071535076555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9060808874649536e-05,
+      "loss": 1.2376,
+      "step": 4534
+    },
+    {
+      "epoch": 0.16492708295450412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.906031045321326e-05,
+      "loss": 1.1206,
+      "step": 4535
+    },
+    {
+      "epoch": 0.1649634505582427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9059811906078032e-05,
+      "loss": 1.148,
+      "step": 4536
+    },
+    {
+      "epoch": 0.1649998181619813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9059313233250767e-05,
+      "loss": 1.0827,
+      "step": 4537
+    },
+    {
+      "epoch": 0.1650361857657199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9058814434738386e-05,
+      "loss": 1.2635,
+      "step": 4538
+    },
+    {
+      "epoch": 0.16507255336945847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9058315510547806e-05,
+      "loss": 1.0953,
+      "step": 4539
+    },
+    {
+      "epoch": 0.16510892097319707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9057816460685953e-05,
+      "loss": 1.2062,
+      "step": 4540
+    },
+    {
+      "epoch": 0.16514528857693567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9057317285159745e-05,
+      "loss": 1.1129,
+      "step": 4541
+    },
+    {
+      "epoch": 0.16518165618067426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905681798397611e-05,
+      "loss": 1.1834,
+      "step": 4542
+    },
+    {
+      "epoch": 0.16521802378441283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905631855714198e-05,
+      "loss": 1.1694,
+      "step": 4543
+    },
+    {
+      "epoch": 0.16525439138815143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905581900466428e-05,
+      "loss": 1.058,
+      "step": 4544
+    },
+    {
+      "epoch": 0.16529075899189002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9055319326549933e-05,
+      "loss": 1.2615,
+      "step": 4545
+    },
+    {
+      "epoch": 0.16532712659562862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9054819522805884e-05,
+      "loss": 1.0686,
+      "step": 4546
+    },
+    {
+      "epoch": 0.16536349419936722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905431959343906e-05,
+      "loss": 1.1944,
+      "step": 4547
+    },
+    {
+      "epoch": 0.16539986180310579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9053819538456394e-05,
+      "loss": 1.25,
+      "step": 4548
+    },
+    {
+      "epoch": 0.16543622940684438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905331935786483e-05,
+      "loss": 1.1799,
+      "step": 4549
+    },
+    {
+      "epoch": 0.16547259701058298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9052819051671307e-05,
+      "loss": 1.0622,
+      "step": 4550
+    },
+    {
+      "epoch": 0.16550896461432157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9052318619882763e-05,
+      "loss": 1.2256,
+      "step": 4551
+    },
+    {
+      "epoch": 0.16554533221806014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.905181806250614e-05,
+      "loss": 1.2516,
+      "step": 4552
+    },
+    {
+      "epoch": 0.16558169982179874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9051317379548387e-05,
+      "loss": 1.0022,
+      "step": 4553
+    },
+    {
+      "epoch": 0.16561806742553734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9050816571016444e-05,
+      "loss": 1.1942,
+      "step": 4554
+    },
+    {
+      "epoch": 0.16565443502927593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9050315636917264e-05,
+      "loss": 1.191,
+      "step": 4555
+    },
+    {
+      "epoch": 0.1656908026330145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9049814577257795e-05,
+      "loss": 1.1351,
+      "step": 4556
+    },
+    {
+      "epoch": 0.1657271702367531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904931339204499e-05,
+      "loss": 1.197,
+      "step": 4557
+    },
+    {
+      "epoch": 0.1657635378404917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90488120812858e-05,
+      "loss": 1.1904,
+      "step": 4558
+    },
+    {
+      "epoch": 0.1657999054442303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9048310644987182e-05,
+      "loss": 1.0124,
+      "step": 4559
+    },
+    {
+      "epoch": 0.16583627304796886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904780908315609e-05,
+      "loss": 1.2158,
+      "step": 4560
+    },
+    {
+      "epoch": 0.16587264065170745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9047307395799484e-05,
+      "loss": 1.2448,
+      "step": 4561
+    },
+    {
+      "epoch": 0.16590900825544605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904680558292433e-05,
+      "loss": 1.1591,
+      "step": 4562
+    },
+    {
+      "epoch": 0.16594537585918465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9046303644537577e-05,
+      "loss": 1.1682,
+      "step": 4563
+    },
+    {
+      "epoch": 0.16598174346292321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90458015806462e-05,
+      "loss": 1.1451,
+      "step": 4564
+    },
+    {
+      "epoch": 0.1660181110666618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9045299391257155e-05,
+      "loss": 1.208,
+      "step": 4565
+    },
+    {
+      "epoch": 0.1660544786704004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904479707637742e-05,
+      "loss": 0.9993,
+      "step": 4566
+    },
+    {
+      "epoch": 0.166090846274139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9044294636013957e-05,
+      "loss": 1.0767,
+      "step": 4567
+    },
+    {
+      "epoch": 0.1661272138778776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9043792070173736e-05,
+      "loss": 1.1575,
+      "step": 4568
+    },
+    {
+      "epoch": 0.16616358148161617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9043289378863737e-05,
+      "loss": 1.3073,
+      "step": 4569
+    },
+    {
+      "epoch": 0.16619994908535476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9042786562090923e-05,
+      "loss": 1.1233,
+      "step": 4570
+    },
+    {
+      "epoch": 0.16623631668909336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9042283619862278e-05,
+      "loss": 1.1107,
+      "step": 4571
+    },
+    {
+      "epoch": 0.16627268429283196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904178055218478e-05,
+      "loss": 1.2272,
+      "step": 4572
+    },
+    {
+      "epoch": 0.16630905189657053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90412773590654e-05,
+      "loss": 1.187,
+      "step": 4573
+    },
+    {
+      "epoch": 0.16634541950030912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9040774040511127e-05,
+      "loss": 1.1256,
+      "step": 4574
+    },
+    {
+      "epoch": 0.16638178710404772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904027059652894e-05,
+      "loss": 1.1676,
+      "step": 4575
+    },
+    {
+      "epoch": 0.16641815470778631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9039767027125827e-05,
+      "loss": 1.0616,
+      "step": 4576
+    },
+    {
+      "epoch": 0.16645452231152488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903926333230877e-05,
+      "loss": 1.1587,
+      "step": 4577
+    },
+    {
+      "epoch": 0.16649088991526348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9038759512084762e-05,
+      "loss": 1.1774,
+      "step": 4578
+    },
+    {
+      "epoch": 0.16652725751900208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903825556646079e-05,
+      "loss": 1.0204,
+      "step": 4579
+    },
+    {
+      "epoch": 0.16656362512274067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903775149544384e-05,
+      "loss": 1.2206,
+      "step": 4580
+    },
+    {
+      "epoch": 0.16659999272647924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9037247299040916e-05,
+      "loss": 1.1559,
+      "step": 4581
+    },
+    {
+      "epoch": 0.16663636033021784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903674297725901e-05,
+      "loss": 1.2061,
+      "step": 4582
+    },
+    {
+      "epoch": 0.16667272793395643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9036238530105114e-05,
+      "loss": 1.187,
+      "step": 4583
+    },
+    {
+      "epoch": 0.16670909553769503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9035733957586226e-05,
+      "loss": 1.313,
+      "step": 4584
+    },
+    {
+      "epoch": 0.1667454631414336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9035229259709354e-05,
+      "loss": 1.2024,
+      "step": 4585
+    },
+    {
+      "epoch": 0.1667818307451722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9034724436481494e-05,
+      "loss": 1.0871,
+      "step": 4586
+    },
+    {
+      "epoch": 0.1668181983489108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903421948790965e-05,
+      "loss": 1.1235,
+      "step": 4587
+    },
+    {
+      "epoch": 0.1668545659526494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9033714414000826e-05,
+      "loss": 1.2007,
+      "step": 4588
+    },
+    {
+      "epoch": 0.16689093355638798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903320921476204e-05,
+      "loss": 1.1243,
+      "step": 4589
+    },
+    {
+      "epoch": 0.16692730116012655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9032703890200284e-05,
+      "loss": 1.1165,
+      "step": 4590
+    },
+    {
+      "epoch": 0.16696366876386515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903219844032258e-05,
+      "loss": 1.0206,
+      "step": 4591
+    },
+    {
+      "epoch": 0.16700003636760374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.903169286513594e-05,
+      "loss": 1.1747,
+      "step": 4592
+    },
+    {
+      "epoch": 0.16703640397134234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9031187164647375e-05,
+      "loss": 1.1185,
+      "step": 4593
+    },
+    {
+      "epoch": 0.1670727715750809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9030681338863898e-05,
+      "loss": 1.1992,
+      "step": 4594
+    },
+    {
+      "epoch": 0.1671091391788195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9030175387792535e-05,
+      "loss": 1.1344,
+      "step": 4595
+    },
+    {
+      "epoch": 0.1671455067825581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9029669311440296e-05,
+      "loss": 1.1482,
+      "step": 4596
+    },
+    {
+      "epoch": 0.1671818743862967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9029163109814213e-05,
+      "loss": 1.1033,
+      "step": 4597
+    },
+    {
+      "epoch": 0.16721824199003527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90286567829213e-05,
+      "loss": 1.0999,
+      "step": 4598
+    },
+    {
+      "epoch": 0.16725460959377386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9028150330768584e-05,
+      "loss": 0.9603,
+      "step": 4599
+    },
+    {
+      "epoch": 0.16729097719751246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902764375336309e-05,
+      "loss": 1.2068,
+      "step": 4600
+    },
+    {
+      "epoch": 0.16732734480125105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9027137050711846e-05,
+      "loss": 1.1105,
+      "step": 4601
+    },
+    {
+      "epoch": 0.16736371240498962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9026630222821883e-05,
+      "loss": 1.0303,
+      "step": 4602
+    },
+    {
+      "epoch": 0.16740008000872822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9026123269700237e-05,
+      "loss": 1.0585,
+      "step": 4603
+    },
+    {
+      "epoch": 0.16743644761246682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9025616191353935e-05,
+      "loss": 1.0678,
+      "step": 4604
+    },
+    {
+      "epoch": 0.1674728152162054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9025108987790016e-05,
+      "loss": 1.3096,
+      "step": 4605
+    },
+    {
+      "epoch": 0.16750918281994398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9024601659015513e-05,
+      "loss": 1.1369,
+      "step": 4606
+    },
+    {
+      "epoch": 0.16754555042368258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9024094205037464e-05,
+      "loss": 1.1937,
+      "step": 4607
+    },
+    {
+      "epoch": 0.16758191802742117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9023586625862914e-05,
+      "loss": 1.0224,
+      "step": 4608
+    },
+    {
+      "epoch": 0.16761828563115977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90230789214989e-05,
+      "loss": 1.1854,
+      "step": 4609
+    },
+    {
+      "epoch": 0.16765465323489837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902257109195247e-05,
+      "loss": 1.0921,
+      "step": 4610
+    },
+    {
+      "epoch": 0.16769102083863693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9022063137230668e-05,
+      "loss": 1.1082,
+      "step": 4611
+    },
+    {
+      "epoch": 0.16772738844237553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.902155505734054e-05,
+      "loss": 1.2037,
+      "step": 4612
+    },
+    {
+      "epoch": 0.16776375604611413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9021046852289134e-05,
+      "loss": 1.2429,
+      "step": 4613
+    },
+    {
+      "epoch": 0.16780012364985272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9020538522083505e-05,
+      "loss": 1.1115,
+      "step": 4614
+    },
+    {
+      "epoch": 0.1678364912535913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9020030066730702e-05,
+      "loss": 1.1935,
+      "step": 4615
+    },
+    {
+      "epoch": 0.1678728588573299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9019521486237777e-05,
+      "loss": 1.1982,
+      "step": 4616
+    },
+    {
+      "epoch": 0.16790922646106848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9019012780611793e-05,
+      "loss": 1.1504,
+      "step": 4617
+    },
+    {
+      "epoch": 0.16794559406480708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9018503949859795e-05,
+      "loss": 1.1318,
+      "step": 4618
+    },
+    {
+      "epoch": 0.16798196166854565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9017994993988857e-05,
+      "loss": 1.1448,
+      "step": 4619
+    },
+    {
+      "epoch": 0.16801832927228424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9017485913006034e-05,
+      "loss": 1.0237,
+      "step": 4620
+    },
+    {
+      "epoch": 0.16805469687602284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9016976706918386e-05,
+      "loss": 1.2263,
+      "step": 4621
+    },
+    {
+      "epoch": 0.16809106447976144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9016467375732976e-05,
+      "loss": 1.0717,
+      "step": 4622
+    },
+    {
+      "epoch": 0.1681274320835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9015957919456878e-05,
+      "loss": 1.0141,
+      "step": 4623
+    },
+    {
+      "epoch": 0.1681637996872386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9015448338097157e-05,
+      "loss": 1.2363,
+      "step": 4624
+    },
+    {
+      "epoch": 0.1682001672909772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.901493863166088e-05,
+      "loss": 1.1765,
+      "step": 4625
+    },
+    {
+      "epoch": 0.1682365348947158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9014428800155117e-05,
+      "loss": 1.0832,
+      "step": 4626
+    },
+    {
+      "epoch": 0.16827290249845436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9013918843586946e-05,
+      "loss": 1.1371,
+      "step": 4627
+    },
+    {
+      "epoch": 0.16830927010219296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9013408761963445e-05,
+      "loss": 1.0775,
+      "step": 4628
+    },
+    {
+      "epoch": 0.16834563770593156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.901289855529168e-05,
+      "loss": 1.1642,
+      "step": 4629
+    },
+    {
+      "epoch": 0.16838200530967015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9012388223578736e-05,
+      "loss": 1.2954,
+      "step": 4630
+    },
+    {
+      "epoch": 0.16841837291340875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.901187776683169e-05,
+      "loss": 1.1539,
+      "step": 4631
+    },
+    {
+      "epoch": 0.16845474051714732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9011367185057633e-05,
+      "loss": 1.1977,
+      "step": 4632
+    },
+    {
+      "epoch": 0.1684911081208859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9010856478263636e-05,
+      "loss": 1.2178,
+      "step": 4633
+    },
+    {
+      "epoch": 0.1685274757246245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9010345646456795e-05,
+      "loss": 1.1043,
+      "step": 4634
+    },
+    {
+      "epoch": 0.1685638433283631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9009834689644185e-05,
+      "loss": 1.084,
+      "step": 4635
+    },
+    {
+      "epoch": 0.16860021093210167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900932360783291e-05,
+      "loss": 1.0801,
+      "step": 4636
+    },
+    {
+      "epoch": 0.16863657853584027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900881240103005e-05,
+      "loss": 1.1222,
+      "step": 4637
+    },
+    {
+      "epoch": 0.16867294613957887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9008301069242695e-05,
+      "loss": 1.1383,
+      "step": 4638
+    },
+    {
+      "epoch": 0.16870931374331746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9007789612477948e-05,
+      "loss": 1.0668,
+      "step": 4639
+    },
+    {
+      "epoch": 0.16874568134705603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9007278030742898e-05,
+      "loss": 1.0636,
+      "step": 4640
+    },
+    {
+      "epoch": 0.16878204895079463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9006766324044647e-05,
+      "loss": 1.1337,
+      "step": 4641
+    },
+    {
+      "epoch": 0.16881841655453322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9006254492390293e-05,
+      "loss": 1.1628,
+      "step": 4642
+    },
+    {
+      "epoch": 0.16885478415827182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9005742535786935e-05,
+      "loss": 1.2064,
+      "step": 4643
+    },
+    {
+      "epoch": 0.1688911517620104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900523045424168e-05,
+      "loss": 1.2026,
+      "step": 4644
+    },
+    {
+      "epoch": 0.16892751936574898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9004718247761623e-05,
+      "loss": 1.1382,
+      "step": 4645
+    },
+    {
+      "epoch": 0.16896388696948758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.900420591635388e-05,
+      "loss": 1.2161,
+      "step": 4646
+    },
+    {
+      "epoch": 0.16900025457322618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9003693460025555e-05,
+      "loss": 1.0768,
+      "step": 4647
+    },
+    {
+      "epoch": 0.16903662217696477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9003180878783757e-05,
+      "loss": 0.9899,
+      "step": 4648
+    },
+    {
+      "epoch": 0.16907298978070334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90026681726356e-05,
+      "loss": 1.165,
+      "step": 4649
+    },
+    {
+      "epoch": 0.16910935738444194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9002155341588194e-05,
+      "loss": 1.1692,
+      "step": 4650
+    },
+    {
+      "epoch": 0.16914572498818053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9001642385648652e-05,
+      "loss": 1.1381,
+      "step": 4651
+    },
+    {
+      "epoch": 0.16918209259191913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9001129304824097e-05,
+      "loss": 1.1688,
+      "step": 4652
+    },
+    {
+      "epoch": 0.1692184601956577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9000616099121645e-05,
+      "loss": 1.0907,
+      "step": 4653
+    },
+    {
+      "epoch": 0.1692548277993963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9000102768548413e-05,
+      "loss": 1.2372,
+      "step": 4654
+    },
+    {
+      "epoch": 0.1692911954031349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8999589313111523e-05,
+      "loss": 1.2217,
+      "step": 4655
+    },
+    {
+      "epoch": 0.1693275630068735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8999075732818104e-05,
+      "loss": 1.0403,
+      "step": 4656
+    },
+    {
+      "epoch": 0.16936393061061206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8998562027675274e-05,
+      "loss": 1.1057,
+      "step": 4657
+    },
+    {
+      "epoch": 0.16940029821435065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8998048197690163e-05,
+      "loss": 1.2176,
+      "step": 4658
+    },
+    {
+      "epoch": 0.16943666581808925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89975342428699e-05,
+      "loss": 1.1496,
+      "step": 4659
+    },
+    {
+      "epoch": 0.16947303342182785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.899702016322162e-05,
+      "loss": 1.0711,
+      "step": 4660
+    },
+    {
+      "epoch": 0.16950940102556641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8996505958752443e-05,
+      "loss": 1.1674,
+      "step": 4661
+    },
+    {
+      "epoch": 0.169545768629305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8995991629469514e-05,
+      "loss": 1.1418,
+      "step": 4662
+    },
+    {
+      "epoch": 0.1695821362330436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8995477175379962e-05,
+      "loss": 1.1571,
+      "step": 4663
+    },
+    {
+      "epoch": 0.1696185038367822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.899496259649093e-05,
+      "loss": 1.12,
+      "step": 4664
+    },
+    {
+      "epoch": 0.16965487144052077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8994447892809556e-05,
+      "loss": 1.1871,
+      "step": 4665
+    },
+    {
+      "epoch": 0.16969123904425937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8993933064342976e-05,
+      "loss": 1.0802,
+      "step": 4666
+    },
+    {
+      "epoch": 0.16972760664799796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8993418111098335e-05,
+      "loss": 1.1024,
+      "step": 4667
+    },
+    {
+      "epoch": 0.16976397425173656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.899290303308278e-05,
+      "loss": 1.2552,
+      "step": 4668
+    },
+    {
+      "epoch": 0.16980034185547516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8992387830303453e-05,
+      "loss": 1.1284,
+      "step": 4669
+    },
+    {
+      "epoch": 0.16983670945921372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8991872502767506e-05,
+      "loss": 1.1197,
+      "step": 4670
+    },
+    {
+      "epoch": 0.16987307706295232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.899135705048208e-05,
+      "loss": 1.2847,
+      "step": 4671
+    },
+    {
+      "epoch": 0.16990944466669092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8990841473454338e-05,
+      "loss": 1.1653,
+      "step": 4672
+    },
+    {
+      "epoch": 0.1699458122704295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8990325771691425e-05,
+      "loss": 1.0834,
+      "step": 4673
+    },
+    {
+      "epoch": 0.16998217987416808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8989809945200495e-05,
+      "loss": 1.0296,
+      "step": 4674
+    },
+    {
+      "epoch": 0.17001854747790668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898929399398871e-05,
+      "loss": 1.1808,
+      "step": 4675
+    },
+    {
+      "epoch": 0.17005491508164527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8988777918063223e-05,
+      "loss": 1.1214,
+      "step": 4676
+    },
+    {
+      "epoch": 0.17009128268538387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8988261717431196e-05,
+      "loss": 1.1008,
+      "step": 4677
+    },
+    {
+      "epoch": 0.17012765028912244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8987745392099792e-05,
+      "loss": 1.195,
+      "step": 4678
+    },
+    {
+      "epoch": 0.17016401789286104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8987228942076173e-05,
+      "loss": 1.2273,
+      "step": 4679
+    },
+    {
+      "epoch": 0.17020038549659963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89867123673675e-05,
+      "loss": 1.1727,
+      "step": 4680
+    },
+    {
+      "epoch": 0.17023675310033823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8986195667980947e-05,
+      "loss": 1.1916,
+      "step": 4681
+    },
+    {
+      "epoch": 0.1702731207040768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898567884392368e-05,
+      "loss": 1.0816,
+      "step": 4682
+    },
+    {
+      "epoch": 0.1703094883078154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8985161895202863e-05,
+      "loss": 1.0743,
+      "step": 4683
+    },
+    {
+      "epoch": 0.170345855911554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8984644821825675e-05,
+      "loss": 1.0666,
+      "step": 4684
+    },
+    {
+      "epoch": 0.17038222351529259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8984127623799288e-05,
+      "loss": 1.1502,
+      "step": 4685
+    },
+    {
+      "epoch": 0.17041859111903115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8983610301130878e-05,
+      "loss": 1.016,
+      "step": 4686
+    },
+    {
+      "epoch": 0.17045495872276975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898309285382762e-05,
+      "loss": 1.1746,
+      "step": 4687
+    },
+    {
+      "epoch": 0.17049132632650835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8982575281896697e-05,
+      "loss": 1.0677,
+      "step": 4688
+    },
+    {
+      "epoch": 0.17052769393024694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898205758534528e-05,
+      "loss": 1.246,
+      "step": 4689
+    },
+    {
+      "epoch": 0.17056406153398554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8981539764180563e-05,
+      "loss": 1.181,
+      "step": 4690
+    },
+    {
+      "epoch": 0.1706004291377241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8981021818409723e-05,
+      "loss": 1.0937,
+      "step": 4691
+    },
+    {
+      "epoch": 0.1706367967414627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898050374803995e-05,
+      "loss": 1.1214,
+      "step": 4692
+    },
+    {
+      "epoch": 0.1706731643452013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8979985553078427e-05,
+      "loss": 1.0887,
+      "step": 4693
+    },
+    {
+      "epoch": 0.1707095319489399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8979467233532343e-05,
+      "loss": 1.0953,
+      "step": 4694
+    },
+    {
+      "epoch": 0.17074589955267847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8978948789408896e-05,
+      "loss": 1.1529,
+      "step": 4695
+    },
+    {
+      "epoch": 0.17078226715641706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8978430220715275e-05,
+      "loss": 1.0393,
+      "step": 4696
+    },
+    {
+      "epoch": 0.17081863476015566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897791152745867e-05,
+      "loss": 1.0968,
+      "step": 4697
+    },
+    {
+      "epoch": 0.17085500236389425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897739270964628e-05,
+      "loss": 1.1809,
+      "step": 4698
+    },
+    {
+      "epoch": 0.17089136996763282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8976873767285308e-05,
+      "loss": 1.1612,
+      "step": 4699
+    },
+    {
+      "epoch": 0.17092773757137142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8976354700382942e-05,
+      "loss": 1.1622,
+      "step": 4700
+    },
+    {
+      "epoch": 0.17096410517511001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8975835508946395e-05,
+      "loss": 1.2578,
+      "step": 4701
+    },
+    {
+      "epoch": 0.1710004727788486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8975316192982865e-05,
+      "loss": 1.1585,
+      "step": 4702
+    },
+    {
+      "epoch": 0.17103684038258718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8974796752499556e-05,
+      "loss": 1.3037,
+      "step": 4703
+    },
+    {
+      "epoch": 0.17107320798632578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8974277187503675e-05,
+      "loss": 0.9969,
+      "step": 4704
+    },
+    {
+      "epoch": 0.17110957559006437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897375749800243e-05,
+      "loss": 1.1656,
+      "step": 4705
+    },
+    {
+      "epoch": 0.17114594319380297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8973237684003036e-05,
+      "loss": 1.2079,
+      "step": 4706
+    },
+    {
+      "epoch": 0.17118231079754154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89727177455127e-05,
+      "loss": 1.1146,
+      "step": 4707
+    },
+    {
+      "epoch": 0.17121867840128013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897219768253863e-05,
+      "loss": 1.135,
+      "step": 4708
+    },
+    {
+      "epoch": 0.17125504600501873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8971677495088054e-05,
+      "loss": 1.0917,
+      "step": 4709
+    },
+    {
+      "epoch": 0.17129141360875733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.897115718316818e-05,
+      "loss": 1.0796,
+      "step": 4710
+    },
+    {
+      "epoch": 0.17132778121249592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8970636746786225e-05,
+      "loss": 1.0244,
+      "step": 4711
+    },
+    {
+      "epoch": 0.1713641488162345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8970116185949418e-05,
+      "loss": 1.1901,
+      "step": 4712
+    },
+    {
+      "epoch": 0.1714005164199731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896959550066497e-05,
+      "loss": 1.177,
+      "step": 4713
+    },
+    {
+      "epoch": 0.17143688402371168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8969074690940113e-05,
+      "loss": 1.0904,
+      "step": 4714
+    },
+    {
+      "epoch": 0.17147325162745028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8968553756782073e-05,
+      "loss": 1.1578,
+      "step": 4715
+    },
+    {
+      "epoch": 0.17150961923118885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8968032698198074e-05,
+      "loss": 1.215,
+      "step": 4716
+    },
+    {
+      "epoch": 0.17154598683492744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8967511515195343e-05,
+      "loss": 1.1349,
+      "step": 4717
+    },
+    {
+      "epoch": 0.17158235443866604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896699020778111e-05,
+      "loss": 1.1064,
+      "step": 4718
+    },
+    {
+      "epoch": 0.17161872204240464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8966468775962615e-05,
+      "loss": 1.0645,
+      "step": 4719
+    },
+    {
+      "epoch": 0.1716550896461432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896594721974709e-05,
+      "loss": 0.9885,
+      "step": 4720
+    },
+    {
+      "epoch": 0.1716914572498818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896542553914176e-05,
+      "loss": 1.189,
+      "step": 4721
+    },
+    {
+      "epoch": 0.1717278248536204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8964903734153878e-05,
+      "loss": 1.1198,
+      "step": 4722
+    },
+    {
+      "epoch": 0.171764192457359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8964381804790674e-05,
+      "loss": 1.0573,
+      "step": 4723
+    },
+    {
+      "epoch": 0.17180056006109756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8963859751059392e-05,
+      "loss": 0.9943,
+      "step": 4724
+    },
+    {
+      "epoch": 0.17183692766483616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.896333757296727e-05,
+      "loss": 1.1259,
+      "step": 4725
+    },
+    {
+      "epoch": 0.17187329526857476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8962815270521563e-05,
+      "loss": 1.207,
+      "step": 4726
+    },
+    {
+      "epoch": 0.17190966287231335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8962292843729506e-05,
+      "loss": 1.162,
+      "step": 4727
+    },
+    {
+      "epoch": 0.17194603047605192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8961770292598353e-05,
+      "loss": 1.1479,
+      "step": 4728
+    },
+    {
+      "epoch": 0.17198239807979052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8961247617135353e-05,
+      "loss": 1.0562,
+      "step": 4729
+    },
+    {
+      "epoch": 0.1720187656835291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8960724817347755e-05,
+      "loss": 1.144,
+      "step": 4730
+    },
+    {
+      "epoch": 0.1720551332872677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8960201893242814e-05,
+      "loss": 0.961,
+      "step": 4731
+    },
+    {
+      "epoch": 0.1720915008910063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8959678844827786e-05,
+      "loss": 1.0685,
+      "step": 4732
+    },
+    {
+      "epoch": 0.17212786849474487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8959155672109927e-05,
+      "loss": 1.0796,
+      "step": 4733
+    },
+    {
+      "epoch": 0.17216423609848347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8958632375096492e-05,
+      "loss": 1.2544,
+      "step": 4734
+    },
+    {
+      "epoch": 0.17220060370222207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8958108953794744e-05,
+      "loss": 1.1794,
+      "step": 4735
+    },
+    {
+      "epoch": 0.17223697130596066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8957585408211945e-05,
+      "loss": 0.9912,
+      "step": 4736
+    },
+    {
+      "epoch": 0.17227333890969923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8957061738355356e-05,
+      "loss": 1.1286,
+      "step": 4737
+    },
+    {
+      "epoch": 0.17230970651343783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8956537944232246e-05,
+      "loss": 1.0132,
+      "step": 4738
+    },
+    {
+      "epoch": 0.17234607411717642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8956014025849878e-05,
+      "loss": 1.1176,
+      "step": 4739
+    },
+    {
+      "epoch": 0.17238244172091502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8955489983215525e-05,
+      "loss": 1.1015,
+      "step": 4740
+    },
+    {
+      "epoch": 0.1724188093246536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895496581633645e-05,
+      "loss": 1.114,
+      "step": 4741
+    },
+    {
+      "epoch": 0.17245517692839218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8954441525219935e-05,
+      "loss": 1.0983,
+      "step": 4742
+    },
+    {
+      "epoch": 0.17249154453213078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8953917109873243e-05,
+      "loss": 1.0962,
+      "step": 4743
+    },
+    {
+      "epoch": 0.17252791213586938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895339257030366e-05,
+      "loss": 1.2906,
+      "step": 4744
+    },
+    {
+      "epoch": 0.17256427973960795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8952867906518454e-05,
+      "loss": 1.1953,
+      "step": 4745
+    },
+    {
+      "epoch": 0.17260064734334654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.895234311852491e-05,
+      "loss": 1.1255,
+      "step": 4746
+    },
+    {
+      "epoch": 0.17263701494708514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8951818206330307e-05,
+      "loss": 0.9971,
+      "step": 4747
+    },
+    {
+      "epoch": 0.17267338255082373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8951293169941928e-05,
+      "loss": 1.1331,
+      "step": 4748
+    },
+    {
+      "epoch": 0.1727097501545623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8950768009367057e-05,
+      "loss": 1.1838,
+      "step": 4749
+    },
+    {
+      "epoch": 0.1727461177583009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8950242724612978e-05,
+      "loss": 1.1927,
+      "step": 4750
+    },
+    {
+      "epoch": 0.1727824853620395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8949717315686983e-05,
+      "loss": 1.2387,
+      "step": 4751
+    },
+    {
+      "epoch": 0.1728188529657781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8949191782596356e-05,
+      "loss": 1.0387,
+      "step": 4752
+    },
+    {
+      "epoch": 0.1728552205695167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8948666125348388e-05,
+      "loss": 1.0522,
+      "step": 4753
+    },
+    {
+      "epoch": 0.17289158817325526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8948140343950375e-05,
+      "loss": 1.2056,
+      "step": 4754
+    },
+    {
+      "epoch": 0.17292795577699385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8947614438409613e-05,
+      "loss": 0.9774,
+      "step": 4755
+    },
+    {
+      "epoch": 0.17296432338073245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.894708840873339e-05,
+      "loss": 1.1613,
+      "step": 4756
+    },
+    {
+      "epoch": 0.17300069098447104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8946562254929018e-05,
+      "loss": 1.0525,
+      "step": 4757
+    },
+    {
+      "epoch": 0.1730370585882096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8946035977003784e-05,
+      "loss": 1.1172,
+      "step": 4758
+    },
+    {
+      "epoch": 0.1730734261919482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8945509574964994e-05,
+      "loss": 1.1426,
+      "step": 4759
+    },
+    {
+      "epoch": 0.1731097937956868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.894498304881995e-05,
+      "loss": 1.1406,
+      "step": 4760
+    },
+    {
+      "epoch": 0.1731461613994254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8944456398575955e-05,
+      "loss": 1.1447,
+      "step": 4761
+    },
+    {
+      "epoch": 0.17318252900316397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.894392962424032e-05,
+      "loss": 1.0718,
+      "step": 4762
+    },
+    {
+      "epoch": 0.17321889660690257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.894340272582035e-05,
+      "loss": 1.245,
+      "step": 4763
+    },
+    {
+      "epoch": 0.17325526421064116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.894287570332336e-05,
+      "loss": 1.1826,
+      "step": 4764
+    },
+    {
+      "epoch": 0.17329163181437976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8942348556756655e-05,
+      "loss": 1.1455,
+      "step": 4765
+    },
+    {
+      "epoch": 0.17332799941811833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8941821286127556e-05,
+      "loss": 1.2181,
+      "step": 4766
+    },
+    {
+      "epoch": 0.17336436702185692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8941293891443368e-05,
+      "loss": 1.1172,
+      "step": 4767
+    },
+    {
+      "epoch": 0.17340073462559552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8940766372711414e-05,
+      "loss": 1.1617,
+      "step": 4768
+    },
+    {
+      "epoch": 0.17343710222933412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8940238729939014e-05,
+      "loss": 1.1468,
+      "step": 4769
+    },
+    {
+      "epoch": 0.17347346983307269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8939710963133485e-05,
+      "loss": 1.0274,
+      "step": 4770
+    },
+    {
+      "epoch": 0.17350983743681128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893918307230215e-05,
+      "loss": 1.2141,
+      "step": 4771
+    },
+    {
+      "epoch": 0.17354620504054988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8938655057452333e-05,
+      "loss": 1.0416,
+      "step": 4772
+    },
+    {
+      "epoch": 0.17358257264428847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893812691859136e-05,
+      "loss": 1.1345,
+      "step": 4773
+    },
+    {
+      "epoch": 0.17361894024802707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8937598655726556e-05,
+      "loss": 1.088,
+      "step": 4774
+    },
+    {
+      "epoch": 0.17365530785176564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8937070268865252e-05,
+      "loss": 1.1281,
+      "step": 4775
+    },
+    {
+      "epoch": 0.17369167545550424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8936541758014778e-05,
+      "loss": 1.0688,
+      "step": 4776
+    },
+    {
+      "epoch": 0.17372804305924283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893601312318247e-05,
+      "loss": 1.1427,
+      "step": 4777
+    },
+    {
+      "epoch": 0.17376441066298143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8935484364375652e-05,
+      "loss": 1.2842,
+      "step": 4778
+    },
+    {
+      "epoch": 0.17380077826672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893495548160167e-05,
+      "loss": 1.1807,
+      "step": 4779
+    },
+    {
+      "epoch": 0.1738371458704586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893442647486786e-05,
+      "loss": 1.0698,
+      "step": 4780
+    },
+    {
+      "epoch": 0.1738735134741972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8933897344181552e-05,
+      "loss": 1.0788,
+      "step": 4781
+    },
+    {
+      "epoch": 0.17390988107793579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89333680895501e-05,
+      "loss": 1.1796,
+      "step": 4782
+    },
+    {
+      "epoch": 0.17394624868167435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8932838710980838e-05,
+      "loss": 0.9999,
+      "step": 4783
+    },
+    {
+      "epoch": 0.17398261628541295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8932309208481117e-05,
+      "loss": 0.995,
+      "step": 4784
+    },
+    {
+      "epoch": 0.17401898388915155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8931779582058273e-05,
+      "loss": 1.1308,
+      "step": 4785
+    },
+    {
+      "epoch": 0.17405535149289014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8931249831719666e-05,
+      "loss": 1.09,
+      "step": 4786
+    },
+    {
+      "epoch": 0.1740917190966287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8930719957472634e-05,
+      "loss": 1.1642,
+      "step": 4787
+    },
+    {
+      "epoch": 0.1741280867003673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8930189959324536e-05,
+      "loss": 1.1168,
+      "step": 4788
+    },
+    {
+      "epoch": 0.1741644543041059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892965983728272e-05,
+      "loss": 1.1512,
+      "step": 4789
+    },
+    {
+      "epoch": 0.1742008219078445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892912959135455e-05,
+      "loss": 1.1416,
+      "step": 4790
+    },
+    {
+      "epoch": 0.17423718951158307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8928599221547367e-05,
+      "loss": 1.1265,
+      "step": 4791
+    },
+    {
+      "epoch": 0.17427355711532166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8928068727868542e-05,
+      "loss": 1.1573,
+      "step": 4792
+    },
+    {
+      "epoch": 0.17430992471906026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892753811032543e-05,
+      "loss": 1.1682,
+      "step": 4793
+    },
+    {
+      "epoch": 0.17434629232279886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8927007368925396e-05,
+      "loss": 1.2344,
+      "step": 4794
+    },
+    {
+      "epoch": 0.17438265992653745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8926476503675796e-05,
+      "loss": 1.1691,
+      "step": 4795
+    },
+    {
+      "epoch": 0.17441902753027602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8925945514584002e-05,
+      "loss": 1.094,
+      "step": 4796
+    },
+    {
+      "epoch": 0.17445539513401462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8925414401657377e-05,
+      "loss": 1.0506,
+      "step": 4797
+    },
+    {
+      "epoch": 0.17449176273775321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8924883164903294e-05,
+      "loss": 1.1452,
+      "step": 4798
+    },
+    {
+      "epoch": 0.1745281303414918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8924351804329118e-05,
+      "loss": 1.1855,
+      "step": 4799
+    },
+    {
+      "epoch": 0.17456449794523038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892382031994222e-05,
+      "loss": 1.0401,
+      "step": 4800
+    },
+    {
+      "epoch": 0.17460086554896898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892328871174998e-05,
+      "loss": 0.9945,
+      "step": 4801
+    },
+    {
+      "epoch": 0.17463723315270757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.892275697975977e-05,
+      "loss": 1.1398,
+      "step": 4802
+    },
+    {
+      "epoch": 0.17467360075644617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8922225123978967e-05,
+      "loss": 1.1801,
+      "step": 4803
+    },
+    {
+      "epoch": 0.17470996836018474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8921693144414943e-05,
+      "loss": 1.0741,
+      "step": 4804
+    },
+    {
+      "epoch": 0.17474633596392333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8921161041075092e-05,
+      "loss": 1.1779,
+      "step": 4805
+    },
+    {
+      "epoch": 0.17478270356766193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8920628813966786e-05,
+      "loss": 1.1606,
+      "step": 4806
+    },
+    {
+      "epoch": 0.17481907117140053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8920096463097413e-05,
+      "loss": 1.0751,
+      "step": 4807
+    },
+    {
+      "epoch": 0.1748554387751391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8919563988474354e-05,
+      "loss": 1.1543,
+      "step": 4808
+    },
+    {
+      "epoch": 0.1748918063788777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8919031390105003e-05,
+      "loss": 1.2487,
+      "step": 4809
+    },
+    {
+      "epoch": 0.1749281739826163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8918498667996745e-05,
+      "loss": 1.1494,
+      "step": 4810
+    },
+    {
+      "epoch": 0.17496454158635488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8917965822156974e-05,
+      "loss": 1.079,
+      "step": 4811
+    },
+    {
+      "epoch": 0.17500090919009345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.891743285259308e-05,
+      "loss": 1.1329,
+      "step": 4812
+    },
+    {
+      "epoch": 0.17503727679383205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8916899759312452e-05,
+      "loss": 1.0722,
+      "step": 4813
+    },
+    {
+      "epoch": 0.17507364439757064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8916366542322495e-05,
+      "loss": 1.287,
+      "step": 4814
+    },
+    {
+      "epoch": 0.17511001200130924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8915833201630606e-05,
+      "loss": 1.1601,
+      "step": 4815
+    },
+    {
+      "epoch": 0.17514637960504784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8915299737244174e-05,
+      "loss": 1.2368,
+      "step": 4816
+    },
+    {
+      "epoch": 0.1751827472087864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8914766149170614e-05,
+      "loss": 1.1931,
+      "step": 4817
+    },
+    {
+      "epoch": 0.175219114812525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8914232437417318e-05,
+      "loss": 1.0785,
+      "step": 4818
+    },
+    {
+      "epoch": 0.1752554824162636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8913698601991695e-05,
+      "loss": 1.1387,
+      "step": 4819
+    },
+    {
+      "epoch": 0.1752918500200022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8913164642901153e-05,
+      "loss": 1.3524,
+      "step": 4820
+    },
+    {
+      "epoch": 0.17532821762374076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8912630560153094e-05,
+      "loss": 1.3063,
+      "step": 4821
+    },
+    {
+      "epoch": 0.17536458522747936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8912096353754935e-05,
+      "loss": 1.1525,
+      "step": 4822
+    },
+    {
+      "epoch": 0.17540095283121795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8911562023714083e-05,
+      "loss": 1.0641,
+      "step": 4823
+    },
+    {
+      "epoch": 0.17543732043495655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8911027570037954e-05,
+      "loss": 1.092,
+      "step": 4824
+    },
+    {
+      "epoch": 0.17547368803869512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8910492992733957e-05,
+      "loss": 1.0556,
+      "step": 4825
+    },
+    {
+      "epoch": 0.17551005564243372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890995829180951e-05,
+      "loss": 1.1987,
+      "step": 4826
+    },
+    {
+      "epoch": 0.1755464232461723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890942346727204e-05,
+      "loss": 1.295,
+      "step": 4827
+    },
+    {
+      "epoch": 0.1755827908499109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8908888519128955e-05,
+      "loss": 1.0959,
+      "step": 4828
+    },
+    {
+      "epoch": 0.17561915845364948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8908353447387683e-05,
+      "loss": 1.1839,
+      "step": 4829
+    },
+    {
+      "epoch": 0.17565552605738807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8907818252055644e-05,
+      "loss": 1.0651,
+      "step": 4830
+    },
+    {
+      "epoch": 0.17569189366112667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890728293314027e-05,
+      "loss": 1.0689,
+      "step": 4831
+    },
+    {
+      "epoch": 0.17572826126486527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890674749064898e-05,
+      "loss": 1.1855,
+      "step": 4832
+    },
+    {
+      "epoch": 0.17576462886860383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8906211924589207e-05,
+      "loss": 1.1637,
+      "step": 4833
+    },
+    {
+      "epoch": 0.17580099647234243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8905676234968378e-05,
+      "loss": 1.1179,
+      "step": 4834
+    },
+    {
+      "epoch": 0.17583736407608103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890514042179393e-05,
+      "loss": 1.205,
+      "step": 4835
+    },
+    {
+      "epoch": 0.17587373167981962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890460448507329e-05,
+      "loss": 1.0019,
+      "step": 4836
+    },
+    {
+      "epoch": 0.17591009928355822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.89040684248139e-05,
+      "loss": 1.1265,
+      "step": 4837
+    },
+    {
+      "epoch": 0.1759464668872968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890353224102319e-05,
+      "loss": 1.0664,
+      "step": 4838
+    },
+    {
+      "epoch": 0.17598283449103538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8902995933708607e-05,
+      "loss": 1.194,
+      "step": 4839
+    },
+    {
+      "epoch": 0.17601920209477398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890245950287758e-05,
+      "loss": 0.9997,
+      "step": 4840
+    },
+    {
+      "epoch": 0.17605556969851258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8901922948537567e-05,
+      "loss": 1.1919,
+      "step": 4841
+    },
+    {
+      "epoch": 0.17609193730225114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8901386270696e-05,
+      "loss": 1.0933,
+      "step": 4842
+    },
+    {
+      "epoch": 0.17612830490598974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8900849469360324e-05,
+      "loss": 1.1199,
+      "step": 4843
+    },
+    {
+      "epoch": 0.17616467250972834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8900312544537995e-05,
+      "loss": 1.2101,
+      "step": 4844
+    },
+    {
+      "epoch": 0.17620104011346693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8899775496236458e-05,
+      "loss": 1.0449,
+      "step": 4845
+    },
+    {
+      "epoch": 0.1762374077172055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889923832446316e-05,
+      "loss": 1.0647,
+      "step": 4846
+    },
+    {
+      "epoch": 0.1762737753209441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8898701029225556e-05,
+      "loss": 1.2687,
+      "step": 4847
+    },
+    {
+      "epoch": 0.1763101429246827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8898163610531103e-05,
+      "loss": 1.0936,
+      "step": 4848
+    },
+    {
+      "epoch": 0.1763465105284213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8897626068387253e-05,
+      "loss": 1.187,
+      "step": 4849
+    },
+    {
+      "epoch": 0.17638287813215986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8897088402801466e-05,
+      "loss": 1.1812,
+      "step": 4850
+    },
+    {
+      "epoch": 0.17641924573589846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8896550613781203e-05,
+      "loss": 1.068,
+      "step": 4851
+    },
+    {
+      "epoch": 0.17645561333963705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8896012701333922e-05,
+      "loss": 1.0703,
+      "step": 4852
+    },
+    {
+      "epoch": 0.17649198094337565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8895474665467082e-05,
+      "loss": 1.1506,
+      "step": 4853
+    },
+    {
+      "epoch": 0.17652834854711424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8894936506188156e-05,
+      "loss": 1.0433,
+      "step": 4854
+    },
+    {
+      "epoch": 0.1765647161508528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8894398223504606e-05,
+      "loss": 1.1581,
+      "step": 4855
+    },
+    {
+      "epoch": 0.1766010837545914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88938598174239e-05,
+      "loss": 1.03,
+      "step": 4856
+    },
+    {
+      "epoch": 0.17663745135833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889332128795351e-05,
+      "loss": 1.2371,
+      "step": 4857
+    },
+    {
+      "epoch": 0.1766738189620686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8892782635100903e-05,
+      "loss": 1.1083,
+      "step": 4858
+    },
+    {
+      "epoch": 0.17671018656580717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8892243858873552e-05,
+      "loss": 1.0522,
+      "step": 4859
+    },
+    {
+      "epoch": 0.17674655416954577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8891704959278936e-05,
+      "loss": 1.1445,
+      "step": 4860
+    },
+    {
+      "epoch": 0.17678292177328436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8891165936324532e-05,
+      "loss": 1.0822,
+      "step": 4861
+    },
+    {
+      "epoch": 0.17681928937702296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889062679001781e-05,
+      "loss": 1.0872,
+      "step": 4862
+    },
+    {
+      "epoch": 0.17685565698076153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.889008752036626e-05,
+      "loss": 1.1465,
+      "step": 4863
+    },
+    {
+      "epoch": 0.17689202458450012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8889548127377357e-05,
+      "loss": 1.2572,
+      "step": 4864
+    },
+    {
+      "epoch": 0.17692839218823872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8889008611058587e-05,
+      "loss": 1.02,
+      "step": 4865
+    },
+    {
+      "epoch": 0.17696475979197732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8888468971417435e-05,
+      "loss": 1.1181,
+      "step": 4866
+    },
+    {
+      "epoch": 0.17700112739571588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8887929208461387e-05,
+      "loss": 1.277,
+      "step": 4867
+    },
+    {
+      "epoch": 0.17703749499945448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8887389322197928e-05,
+      "loss": 1.1096,
+      "step": 4868
+    },
+    {
+      "epoch": 0.17707386260319308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8886849312634556e-05,
+      "loss": 1.0779,
+      "step": 4869
+    },
+    {
+      "epoch": 0.17711023020693167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8886309179778756e-05,
+      "loss": 1.1583,
+      "step": 4870
+    },
+    {
+      "epoch": 0.17714659781067024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8885768923638025e-05,
+      "loss": 1.3121,
+      "step": 4871
+    },
+    {
+      "epoch": 0.17718296541440884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8885228544219858e-05,
+      "loss": 1.1253,
+      "step": 4872
+    },
+    {
+      "epoch": 0.17721933301814743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8884688041531753e-05,
+      "loss": 1.1085,
+      "step": 4873
+    },
+    {
+      "epoch": 0.17725570062188603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8884147415581204e-05,
+      "loss": 1.1652,
+      "step": 4874
+    },
+    {
+      "epoch": 0.17729206822562463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8883606666375717e-05,
+      "loss": 1.0972,
+      "step": 4875
+    },
+    {
+      "epoch": 0.1773284358293632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.888306579392279e-05,
+      "loss": 1.1227,
+      "step": 4876
+    },
+    {
+      "epoch": 0.1773648034331018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8882524798229928e-05,
+      "loss": 1.0196,
+      "step": 4877
+    },
+    {
+      "epoch": 0.1774011710368404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8881983679304635e-05,
+      "loss": 1.2443,
+      "step": 4878
+    },
+    {
+      "epoch": 0.17743753864057898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8881442437154426e-05,
+      "loss": 1.1079,
+      "step": 4879
+    },
+    {
+      "epoch": 0.17747390624431755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88809010717868e-05,
+      "loss": 1.1707,
+      "step": 4880
+    },
+    {
+      "epoch": 0.17751027384805615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8880359583209277e-05,
+      "loss": 1.1313,
+      "step": 4881
+    },
+    {
+      "epoch": 0.17754664145179475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887981797142936e-05,
+      "loss": 1.1074,
+      "step": 4882
+    },
+    {
+      "epoch": 0.17758300905553334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8879276236454568e-05,
+      "loss": 0.947,
+      "step": 4883
+    },
+    {
+      "epoch": 0.1776193766592719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8878734378292415e-05,
+      "loss": 1.1414,
+      "step": 4884
+    },
+    {
+      "epoch": 0.1776557442630105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8878192396950423e-05,
+      "loss": 1.0312,
+      "step": 4885
+    },
+    {
+      "epoch": 0.1776921118667491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8877650292436106e-05,
+      "loss": 1.0885,
+      "step": 4886
+    },
+    {
+      "epoch": 0.1777284794704877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8877108064756988e-05,
+      "loss": 1.1394,
+      "step": 4887
+    },
+    {
+      "epoch": 0.17776484707422627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887656571392059e-05,
+      "loss": 1.0958,
+      "step": 4888
+    },
+    {
+      "epoch": 0.17780121467796486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8876023239934435e-05,
+      "loss": 1.1008,
+      "step": 4889
+    },
+    {
+      "epoch": 0.17783758228170346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8875480642806054e-05,
+      "loss": 1.1721,
+      "step": 4890
+    },
+    {
+      "epoch": 0.17787394988544206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8874937922542972e-05,
+      "loss": 1.1942,
+      "step": 4891
+    },
+    {
+      "epoch": 0.17791031748918062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8874395079152718e-05,
+      "loss": 1.0555,
+      "step": 4892
+    },
+    {
+      "epoch": 0.17794668509291922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887385211264282e-05,
+      "loss": 1.1827,
+      "step": 4893
+    },
+    {
+      "epoch": 0.17798305269665782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8873309023020816e-05,
+      "loss": 1.1212,
+      "step": 4894
+    },
+    {
+      "epoch": 0.1780194203003964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8872765810294243e-05,
+      "loss": 1.0693,
+      "step": 4895
+    },
+    {
+      "epoch": 0.178055787904135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8872222474470627e-05,
+      "loss": 1.0413,
+      "step": 4896
+    },
+    {
+      "epoch": 0.17809215550787358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8871679015557515e-05,
+      "loss": 1.0888,
+      "step": 4897
+    },
+    {
+      "epoch": 0.17812852311161217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8871135433562446e-05,
+      "loss": 1.1162,
+      "step": 4898
+    },
+    {
+      "epoch": 0.17816489071535077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8870591728492954e-05,
+      "loss": 1.2375,
+      "step": 4899
+    },
+    {
+      "epoch": 0.17820125831908937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.887004790035659e-05,
+      "loss": 1.2235,
+      "step": 4900
+    },
+    {
+      "epoch": 0.17823762592282794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88695039491609e-05,
+      "loss": 1.1413,
+      "step": 4901
+    },
+    {
+      "epoch": 0.17827399352656653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8868959874913423e-05,
+      "loss": 1.1422,
+      "step": 4902
+    },
+    {
+      "epoch": 0.17831036113030513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886841567762171e-05,
+      "loss": 1.0678,
+      "step": 4903
+    },
+    {
+      "epoch": 0.17834672873404372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8867871357293314e-05,
+      "loss": 1.1138,
+      "step": 4904
+    },
+    {
+      "epoch": 0.1783830963377823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8867326913935786e-05,
+      "loss": 0.9869,
+      "step": 4905
+    },
+    {
+      "epoch": 0.1784194639415209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8866782347556678e-05,
+      "loss": 1.002,
+      "step": 4906
+    },
+    {
+      "epoch": 0.17845583154525949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8866237658163543e-05,
+      "loss": 1.1696,
+      "step": 4907
+    },
+    {
+      "epoch": 0.17849219914899808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8865692845763942e-05,
+      "loss": 1.132,
+      "step": 4908
+    },
+    {
+      "epoch": 0.17852856675273665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886514791036543e-05,
+      "loss": 1.1895,
+      "step": 4909
+    },
+    {
+      "epoch": 0.17856493435647525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8864602851975572e-05,
+      "loss": 1.0251,
+      "step": 4910
+    },
+    {
+      "epoch": 0.17860130196021384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8864057670601924e-05,
+      "loss": 1.2323,
+      "step": 4911
+    },
+    {
+      "epoch": 0.17863766956395244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886351236625205e-05,
+      "loss": 1.0956,
+      "step": 4912
+    },
+    {
+      "epoch": 0.178674037167691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886296693893352e-05,
+      "loss": 1.1571,
+      "step": 4913
+    },
+    {
+      "epoch": 0.1787104047714296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8862421388653904e-05,
+      "loss": 1.1944,
+      "step": 4914
+    },
+    {
+      "epoch": 0.1787467723751682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8861875715420756e-05,
+      "loss": 1.1994,
+      "step": 4915
+    },
+    {
+      "epoch": 0.1787831399789068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.886132991924166e-05,
+      "loss": 1.0311,
+      "step": 4916
+    },
+    {
+      "epoch": 0.1788195075826454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8860784000124182e-05,
+      "loss": 1.2057,
+      "step": 4917
+    },
+    {
+      "epoch": 0.17885587518638396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88602379580759e-05,
+      "loss": 1.1664,
+      "step": 4918
+    },
+    {
+      "epoch": 0.17889224279012256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885969179310439e-05,
+      "loss": 1.2106,
+      "step": 4919
+    },
+    {
+      "epoch": 0.17892861039386115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8859145505217223e-05,
+      "loss": 1.3309,
+      "step": 4920
+    },
+    {
+      "epoch": 0.17896497799759975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885859909442198e-05,
+      "loss": 1.3114,
+      "step": 4921
+    },
+    {
+      "epoch": 0.17900134560133832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8858052560726247e-05,
+      "loss": 1.2111,
+      "step": 4922
+    },
+    {
+      "epoch": 0.17903771320507691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8857505904137602e-05,
+      "loss": 1.0559,
+      "step": 4923
+    },
+    {
+      "epoch": 0.1790740808088155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8856959124663626e-05,
+      "loss": 1.0106,
+      "step": 4924
+    },
+    {
+      "epoch": 0.1791104484125541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8856412222311916e-05,
+      "loss": 1.1428,
+      "step": 4925
+    },
+    {
+      "epoch": 0.17914681601629268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8855865197090046e-05,
+      "loss": 1.1567,
+      "step": 4926
+    },
+    {
+      "epoch": 0.17918318362003127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8855318049005614e-05,
+      "loss": 1.3311,
+      "step": 4927
+    },
+    {
+      "epoch": 0.17921955122376987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8854770778066206e-05,
+      "loss": 1.116,
+      "step": 4928
+    },
+    {
+      "epoch": 0.17925591882750846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8854223384279417e-05,
+      "loss": 1.1675,
+      "step": 4929
+    },
+    {
+      "epoch": 0.17929228643124703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885367586765284e-05,
+      "loss": 1.0324,
+      "step": 4930
+    },
+    {
+      "epoch": 0.17932865403498563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8853128228194075e-05,
+      "loss": 1.1648,
+      "step": 4931
+    },
+    {
+      "epoch": 0.17936502163872423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8852580465910715e-05,
+      "loss": 1.2068,
+      "step": 4932
+    },
+    {
+      "epoch": 0.17940138924246282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8852032580810362e-05,
+      "loss": 1.1464,
+      "step": 4933
+    },
+    {
+      "epoch": 0.1794377568462014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8851484572900617e-05,
+      "loss": 1.1001,
+      "step": 4934
+    },
+    {
+      "epoch": 0.17947412444994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885093644218908e-05,
+      "loss": 1.1444,
+      "step": 4935
+    },
+    {
+      "epoch": 0.17951049205367858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885038818868336e-05,
+      "loss": 1.1526,
+      "step": 4936
+    },
+    {
+      "epoch": 0.17954685965741718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884983981239106e-05,
+      "loss": 1.1569,
+      "step": 4937
+    },
+    {
+      "epoch": 0.17958322726115578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884929131331979e-05,
+      "loss": 1.0191,
+      "step": 4938
+    },
+    {
+      "epoch": 0.17961959486489434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8848742691477152e-05,
+      "loss": 1.2343,
+      "step": 4939
+    },
+    {
+      "epoch": 0.17965596246863294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8848193946870774e-05,
+      "loss": 1.1645,
+      "step": 4940
+    },
+    {
+      "epoch": 0.17969233007237154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884764507950825e-05,
+      "loss": 1.2078,
+      "step": 4941
+    },
+    {
+      "epoch": 0.17972869767611013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884709608939721e-05,
+      "loss": 1.1148,
+      "step": 4942
+    },
+    {
+      "epoch": 0.1797650652798487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884654697654526e-05,
+      "loss": 1.2294,
+      "step": 4943
+    },
+    {
+      "epoch": 0.1798014328835873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884599774096003e-05,
+      "loss": 1.0143,
+      "step": 4944
+    },
+    {
+      "epoch": 0.1798378004873259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884544838264912e-05,
+      "loss": 1.1021,
+      "step": 4945
+    },
+    {
+      "epoch": 0.1798741680910645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8844898901620173e-05,
+      "loss": 1.2552,
+      "step": 4946
+    },
+    {
+      "epoch": 0.17991053569480306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88443492978808e-05,
+      "loss": 0.9791,
+      "step": 4947
+    },
+    {
+      "epoch": 0.17994690329854166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884379957143863e-05,
+      "loss": 1.2215,
+      "step": 4948
+    },
+    {
+      "epoch": 0.17998327090228025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8843249722301285e-05,
+      "loss": 1.0958,
+      "step": 4949
+    },
+    {
+      "epoch": 0.18001963850601885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88426997504764e-05,
+      "loss": 1.1306,
+      "step": 4950
+    },
+    {
+      "epoch": 0.18005600610975742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8842149655971602e-05,
+      "loss": 1.0847,
+      "step": 4951
+    },
+    {
+      "epoch": 0.180092373713496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8841599438794525e-05,
+      "loss": 1.1516,
+      "step": 4952
+    },
+    {
+      "epoch": 0.1801287413172346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8841049098952797e-05,
+      "loss": 1.0849,
+      "step": 4953
+    },
+    {
+      "epoch": 0.1801651089209732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8840498636454056e-05,
+      "loss": 1.1074,
+      "step": 4954
+    },
+    {
+      "epoch": 0.18020147652471177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.883994805130594e-05,
+      "loss": 1.098,
+      "step": 4955
+    },
+    {
+      "epoch": 0.18023784412845037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8839397343516092e-05,
+      "loss": 1.1607,
+      "step": 4956
+    },
+    {
+      "epoch": 0.18027421173218897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.883884651309214e-05,
+      "loss": 1.0623,
+      "step": 4957
+    },
+    {
+      "epoch": 0.18031057933592756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.883829556004174e-05,
+      "loss": 1.2632,
+      "step": 4958
+    },
+    {
+      "epoch": 0.18034694693966616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8837744484372526e-05,
+      "loss": 1.2413,
+      "step": 4959
+    },
+    {
+      "epoch": 0.18038331454340473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8837193286092145e-05,
+      "loss": 1.1726,
+      "step": 4960
+    },
+    {
+      "epoch": 0.18041968214714332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8836641965208247e-05,
+      "loss": 1.0738,
+      "step": 4961
+    },
+    {
+      "epoch": 0.18045604975088192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8836090521728477e-05,
+      "loss": 1.145,
+      "step": 4962
+    },
+    {
+      "epoch": 0.18049241735462052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.883553895566049e-05,
+      "loss": 1.0678,
+      "step": 4963
+    },
+    {
+      "epoch": 0.18052878495835908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8834987267011937e-05,
+      "loss": 1.0358,
+      "step": 4964
+    },
+    {
+      "epoch": 0.18056515256209768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8834435455790468e-05,
+      "loss": 1.0975,
+      "step": 4965
+    },
+    {
+      "epoch": 0.18060152016583628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8833883522003745e-05,
+      "loss": 1.0499,
+      "step": 4966
+    },
+    {
+      "epoch": 0.18063788776957487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8833331465659422e-05,
+      "loss": 1.1627,
+      "step": 4967
+    },
+    {
+      "epoch": 0.18067425537331344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8832779286765155e-05,
+      "loss": 1.2712,
+      "step": 4968
+    },
+    {
+      "epoch": 0.18071062297705204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8832226985328606e-05,
+      "loss": 1.0654,
+      "step": 4969
+    },
+    {
+      "epoch": 0.18074699058079063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8831674561357442e-05,
+      "loss": 1.1707,
+      "step": 4970
+    },
+    {
+      "epoch": 0.18078335818452923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8831122014859323e-05,
+      "loss": 1.1517,
+      "step": 4971
+    },
+    {
+      "epoch": 0.1808197257882678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8830569345841917e-05,
+      "loss": 1.0677,
+      "step": 4972
+    },
+    {
+      "epoch": 0.1808560933920064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8830016554312887e-05,
+      "loss": 1.163,
+      "step": 4973
+    },
+    {
+      "epoch": 0.180892460995745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8829463640279908e-05,
+      "loss": 1.1038,
+      "step": 4974
+    },
+    {
+      "epoch": 0.1809288285994836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8828910603750647e-05,
+      "loss": 1.3255,
+      "step": 4975
+    },
+    {
+      "epoch": 0.18096519620322216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.882835744473278e-05,
+      "loss": 1.1389,
+      "step": 4976
+    },
+    {
+      "epoch": 0.18100156380696075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.882780416323398e-05,
+      "loss": 1.0747,
+      "step": 4977
+    },
+    {
+      "epoch": 0.18103793141069935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8827250759261918e-05,
+      "loss": 1.0571,
+      "step": 4978
+    },
+    {
+      "epoch": 0.18107429901443794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8826697232824277e-05,
+      "loss": 1.1962,
+      "step": 4979
+    },
+    {
+      "epoch": 0.18111066661817654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.882614358392874e-05,
+      "loss": 1.1366,
+      "step": 4980
+    },
+    {
+      "epoch": 0.1811470342219151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8825589812582977e-05,
+      "loss": 1.1448,
+      "step": 4981
+    },
+    {
+      "epoch": 0.1811834018256537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8825035918794682e-05,
+      "loss": 1.2206,
+      "step": 4982
+    },
+    {
+      "epoch": 0.1812197694293923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8824481902571533e-05,
+      "loss": 1.1521,
+      "step": 4983
+    },
+    {
+      "epoch": 0.1812561370331309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8823927763921216e-05,
+      "loss": 1.1964,
+      "step": 4984
+    },
+    {
+      "epoch": 0.18129250463686947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8823373502851422e-05,
+      "loss": 1.2281,
+      "step": 4985
+    },
+    {
+      "epoch": 0.18132887224060806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8822819119369843e-05,
+      "loss": 1.1161,
+      "step": 4986
+    },
+    {
+      "epoch": 0.18136523984434666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8822264613484164e-05,
+      "loss": 1.1089,
+      "step": 4987
+    },
+    {
+      "epoch": 0.18140160744808526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.882170998520208e-05,
+      "loss": 1.071,
+      "step": 4988
+    },
+    {
+      "epoch": 0.18143797505182382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8821155234531286e-05,
+      "loss": 1.0089,
+      "step": 4989
+    },
+    {
+      "epoch": 0.18147434265556242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.882060036147948e-05,
+      "loss": 1.0753,
+      "step": 4990
+    },
+    {
+      "epoch": 0.18151071025930102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8820045366054357e-05,
+      "loss": 1.056,
+      "step": 4991
+    },
+    {
+      "epoch": 0.1815470778630396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881949024826362e-05,
+      "loss": 1.0644,
+      "step": 4992
+    },
+    {
+      "epoch": 0.18158344546677818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8818935008114968e-05,
+      "loss": 1.0856,
+      "step": 4993
+    },
+    {
+      "epoch": 0.18161981307051678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8818379645616107e-05,
+      "loss": 1.2349,
+      "step": 4994
+    },
+    {
+      "epoch": 0.18165618067425537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8817824160774738e-05,
+      "loss": 1.0352,
+      "step": 4995
+    },
+    {
+      "epoch": 0.18169254827799397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8817268553598572e-05,
+      "loss": 1.168,
+      "step": 4996
+    },
+    {
+      "epoch": 0.18172891588173254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8816712824095314e-05,
+      "loss": 1.0761,
+      "step": 4997
+    },
+    {
+      "epoch": 0.18176528348547114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881615697227267e-05,
+      "loss": 1.1355,
+      "step": 4998
+    },
+    {
+      "epoch": 0.18180165108920973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8815600998138363e-05,
+      "loss": 1.0432,
+      "step": 4999
+    },
+    {
+      "epoch": 0.18183801869294833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8815044901700097e-05,
+      "loss": 1.0746,
+      "step": 5000
+    },
+    {
+      "epoch": 0.18187438629668692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881448868296559e-05,
+      "loss": 1.1108,
+      "step": 5001
+    },
+    {
+      "epoch": 0.1819107539004255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8813932341942563e-05,
+      "loss": 0.9964,
+      "step": 5002
+    },
+    {
+      "epoch": 0.1819471215041641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8813375878638725e-05,
+      "loss": 1.1789,
+      "step": 5003
+    },
+    {
+      "epoch": 0.18198348910790269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.88128192930618e-05,
+      "loss": 1.1125,
+      "step": 5004
+    },
+    {
+      "epoch": 0.18201985671164128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8812262585219514e-05,
+      "loss": 1.2223,
+      "step": 5005
+    },
+    {
+      "epoch": 0.18205622431537985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8811705755119588e-05,
+      "loss": 1.1376,
+      "step": 5006
+    },
+    {
+      "epoch": 0.18209259191911845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8811148802769748e-05,
+      "loss": 1.1895,
+      "step": 5007
+    },
+    {
+      "epoch": 0.18212895952285704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.881059172817772e-05,
+      "loss": 1.1695,
+      "step": 5008
+    },
+    {
+      "epoch": 0.18216532712659564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8810034531351226e-05,
+      "loss": 1.2383,
+      "step": 5009
+    },
+    {
+      "epoch": 0.1822016947303342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8809477212298007e-05,
+      "loss": 1.1886,
+      "step": 5010
+    },
+    {
+      "epoch": 0.1822380623340728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880891977102579e-05,
+      "loss": 1.0849,
+      "step": 5011
+    },
+    {
+      "epoch": 0.1822744299378114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880836220754231e-05,
+      "loss": 1.0853,
+      "step": 5012
+    },
+    {
+      "epoch": 0.18231079754155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8807804521855302e-05,
+      "loss": 1.0341,
+      "step": 5013
+    },
+    {
+      "epoch": 0.18234716514528856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8807246713972502e-05,
+      "loss": 1.2264,
+      "step": 5014
+    },
+    {
+      "epoch": 0.18238353274902716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880668878390165e-05,
+      "loss": 1.0881,
+      "step": 5015
+    },
+    {
+      "epoch": 0.18241990035276576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8806130731650487e-05,
+      "loss": 1.1213,
+      "step": 5016
+    },
+    {
+      "epoch": 0.18245626795650435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8805572557226755e-05,
+      "loss": 1.0722,
+      "step": 5017
+    },
+    {
+      "epoch": 0.18249263556024292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8805014260638196e-05,
+      "loss": 1.1935,
+      "step": 5018
+    },
+    {
+      "epoch": 0.18252900316398152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8804455841892558e-05,
+      "loss": 1.0284,
+      "step": 5019
+    },
+    {
+      "epoch": 0.18256537076772011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8803897300997586e-05,
+      "loss": 1.1868,
+      "step": 5020
+    },
+    {
+      "epoch": 0.1826017383714587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8803338637961032e-05,
+      "loss": 1.1694,
+      "step": 5021
+    },
+    {
+      "epoch": 0.1826381059751973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880277985279064e-05,
+      "loss": 1.1716,
+      "step": 5022
+    },
+    {
+      "epoch": 0.18267447357893588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.880222094549417e-05,
+      "loss": 1.203,
+      "step": 5023
+    },
+    {
+      "epoch": 0.18271084118267447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8801661916079373e-05,
+      "loss": 1.1691,
+      "step": 5024
+    },
+    {
+      "epoch": 0.18274720878641307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8801102764554006e-05,
+      "loss": 1.2661,
+      "step": 5025
+    },
+    {
+      "epoch": 0.18278357639015166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8800543490925824e-05,
+      "loss": 1.2117,
+      "step": 5026
+    },
+    {
+      "epoch": 0.18281994399389023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8799984095202586e-05,
+      "loss": 1.0711,
+      "step": 5027
+    },
+    {
+      "epoch": 0.18285631159762883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8799424577392057e-05,
+      "loss": 1.1086,
+      "step": 5028
+    },
+    {
+      "epoch": 0.18289267920136743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8798864937501994e-05,
+      "loss": 1.0122,
+      "step": 5029
+    },
+    {
+      "epoch": 0.18292904680510602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8798305175540165e-05,
+      "loss": 1.1974,
+      "step": 5030
+    },
+    {
+      "epoch": 0.1829654144088446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8797745291514335e-05,
+      "loss": 1.2245,
+      "step": 5031
+    },
+    {
+      "epoch": 0.1830017820125832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.879718528543227e-05,
+      "loss": 1.2227,
+      "step": 5032
+    },
+    {
+      "epoch": 0.18303814961632178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8796625157301744e-05,
+      "loss": 1.1056,
+      "step": 5033
+    },
+    {
+      "epoch": 0.18307451722006038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8796064907130524e-05,
+      "loss": 1.1414,
+      "step": 5034
+    },
+    {
+      "epoch": 0.18311088482379895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8795504534926383e-05,
+      "loss": 1.2507,
+      "step": 5035
+    },
+    {
+      "epoch": 0.18314725242753754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8794944040697096e-05,
+      "loss": 1.1668,
+      "step": 5036
+    },
+    {
+      "epoch": 0.18318362003127614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8794383424450437e-05,
+      "loss": 1.178,
+      "step": 5037
+    },
+    {
+      "epoch": 0.18321998763501474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8793822686194184e-05,
+      "loss": 1.1058,
+      "step": 5038
+    },
+    {
+      "epoch": 0.18325635523875333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8793261825936123e-05,
+      "loss": 1.0535,
+      "step": 5039
+    },
+    {
+      "epoch": 0.1832927228424919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8792700843684027e-05,
+      "loss": 1.0893,
+      "step": 5040
+    },
+    {
+      "epoch": 0.1833290904462305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8792139739445684e-05,
+      "loss": 1.1303,
+      "step": 5041
+    },
+    {
+      "epoch": 0.1833654580499691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8791578513228875e-05,
+      "loss": 1.149,
+      "step": 5042
+    },
+    {
+      "epoch": 0.1834018256537077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8791017165041388e-05,
+      "loss": 1.133,
+      "step": 5043
+    },
+    {
+      "epoch": 0.18343819325744626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.879045569489101e-05,
+      "loss": 1.0593,
+      "step": 5044
+    },
+    {
+      "epoch": 0.18347456086118485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8789894102785532e-05,
+      "loss": 1.097,
+      "step": 5045
+    },
+    {
+      "epoch": 0.18351092846492345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8789332388732744e-05,
+      "loss": 1.2284,
+      "step": 5046
+    },
+    {
+      "epoch": 0.18354729606866205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8788770552740443e-05,
+      "loss": 1.0491,
+      "step": 5047
+    },
+    {
+      "epoch": 0.18358366367240062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8788208594816418e-05,
+      "loss": 1.2124,
+      "step": 5048
+    },
+    {
+      "epoch": 0.1836200312761392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8787646514968465e-05,
+      "loss": 1.0427,
+      "step": 5049
+    },
+    {
+      "epoch": 0.1836563988798778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8787084313204386e-05,
+      "loss": 1.1956,
+      "step": 5050
+    },
+    {
+      "epoch": 0.1836927664836164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878652198953198e-05,
+      "loss": 1.0792,
+      "step": 5051
+    },
+    {
+      "epoch": 0.18372913408735497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878595954395905e-05,
+      "loss": 1.1147,
+      "step": 5052
+    },
+    {
+      "epoch": 0.18376550169109357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8785396976493392e-05,
+      "loss": 0.9355,
+      "step": 5053
+    },
+    {
+      "epoch": 0.18380186929483217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878483428714282e-05,
+      "loss": 1.1672,
+      "step": 5054
+    },
+    {
+      "epoch": 0.18383823689857076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8784271475915138e-05,
+      "loss": 1.0707,
+      "step": 5055
+    },
+    {
+      "epoch": 0.18387460450230933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8783708542818147e-05,
+      "loss": 1.1501,
+      "step": 5056
+    },
+    {
+      "epoch": 0.18391097210604793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8783145487859664e-05,
+      "loss": 1.121,
+      "step": 5057
+    },
+    {
+      "epoch": 0.18394733970978652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87825823110475e-05,
+      "loss": 1.2458,
+      "step": 5058
+    },
+    {
+      "epoch": 0.18398370731352512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878201901238947e-05,
+      "loss": 1.2181,
+      "step": 5059
+    },
+    {
+      "epoch": 0.18402007491726372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878145559189338e-05,
+      "loss": 1.0328,
+      "step": 5060
+    },
+    {
+      "epoch": 0.18405644252100228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8780892049567057e-05,
+      "loss": 1.0323,
+      "step": 5061
+    },
+    {
+      "epoch": 0.18409281012474088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8780328385418314e-05,
+      "loss": 1.0334,
+      "step": 5062
+    },
+    {
+      "epoch": 0.18412917772847948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8779764599454972e-05,
+      "loss": 1.1539,
+      "step": 5063
+    },
+    {
+      "epoch": 0.18416554533221807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8779200691684856e-05,
+      "loss": 1.1647,
+      "step": 5064
+    },
+    {
+      "epoch": 0.18420191293595664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8778636662115787e-05,
+      "loss": 1.1252,
+      "step": 5065
+    },
+    {
+      "epoch": 0.18423828053969524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8778072510755584e-05,
+      "loss": 1.1056,
+      "step": 5066
+    },
+    {
+      "epoch": 0.18427464814343383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877750823761208e-05,
+      "loss": 1.0559,
+      "step": 5067
+    },
+    {
+      "epoch": 0.18431101574717243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8776943842693107e-05,
+      "loss": 1.0961,
+      "step": 5068
+    },
+    {
+      "epoch": 0.184347383350911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877637932600649e-05,
+      "loss": 1.1649,
+      "step": 5069
+    },
+    {
+      "epoch": 0.1843837509546496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877581468756006e-05,
+      "loss": 1.1282,
+      "step": 5070
+    },
+    {
+      "epoch": 0.1844201185583882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8775249927361652e-05,
+      "loss": 0.9858,
+      "step": 5071
+    },
+    {
+      "epoch": 0.1844564861621268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8774685045419107e-05,
+      "loss": 1.1961,
+      "step": 5072
+    },
+    {
+      "epoch": 0.18449285376586536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8774120041740254e-05,
+      "loss": 1.1907,
+      "step": 5073
+    },
+    {
+      "epoch": 0.18452922136960395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.877355491633293e-05,
+      "loss": 1.1599,
+      "step": 5074
+    },
+    {
+      "epoch": 0.18456558897334255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8772989669204985e-05,
+      "loss": 1.19,
+      "step": 5075
+    },
+    {
+      "epoch": 0.18460195657708114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8772424300364252e-05,
+      "loss": 1.1008,
+      "step": 5076
+    },
+    {
+      "epoch": 0.1846383241808197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8771858809818583e-05,
+      "loss": 1.0867,
+      "step": 5077
+    },
+    {
+      "epoch": 0.1846746917845583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8771293197575814e-05,
+      "loss": 1.2052,
+      "step": 5078
+    },
+    {
+      "epoch": 0.1847110593882969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8770727463643796e-05,
+      "loss": 1.2479,
+      "step": 5079
+    },
+    {
+      "epoch": 0.1847474269920355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8770161608030382e-05,
+      "loss": 1.1056,
+      "step": 5080
+    },
+    {
+      "epoch": 0.1847837945957741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8769595630743417e-05,
+      "loss": 1.1765,
+      "step": 5081
+    },
+    {
+      "epoch": 0.18482016219951267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8769029531790756e-05,
+      "loss": 1.0771,
+      "step": 5082
+    },
+    {
+      "epoch": 0.18485652980325126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876846331118025e-05,
+      "loss": 1.0724,
+      "step": 5083
+    },
+    {
+      "epoch": 0.18489289740698986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8767896968919758e-05,
+      "loss": 1.0258,
+      "step": 5084
+    },
+    {
+      "epoch": 0.18492926501072846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8767330505017135e-05,
+      "loss": 1.0333,
+      "step": 5085
+    },
+    {
+      "epoch": 0.18496563261446702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876676391948024e-05,
+      "loss": 1.0333,
+      "step": 5086
+    },
+    {
+      "epoch": 0.18500200021820562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8766197212316938e-05,
+      "loss": 1.2312,
+      "step": 5087
+    },
+    {
+      "epoch": 0.18503836782194422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876563038353508e-05,
+      "loss": 1.0696,
+      "step": 5088
+    },
+    {
+      "epoch": 0.1850747354256828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8765063433142545e-05,
+      "loss": 1.1658,
+      "step": 5089
+    },
+    {
+      "epoch": 0.18511110302942138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8764496361147183e-05,
+      "loss": 1.191,
+      "step": 5090
+    },
+    {
+      "epoch": 0.18514747063315998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8763929167556877e-05,
+      "loss": 1.0156,
+      "step": 5091
+    },
+    {
+      "epoch": 0.18518383823689857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8763361852379484e-05,
+      "loss": 1.0257,
+      "step": 5092
+    },
+    {
+      "epoch": 0.18522020584063717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876279441562288e-05,
+      "loss": 1.1452,
+      "step": 5093
+    },
+    {
+      "epoch": 0.18525657344437574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8762226857294937e-05,
+      "loss": 1.0583,
+      "step": 5094
+    },
+    {
+      "epoch": 0.18529294104811433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8761659177403527e-05,
+      "loss": 1.0887,
+      "step": 5095
+    },
+    {
+      "epoch": 0.18532930865185293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8761091375956523e-05,
+      "loss": 1.1722,
+      "step": 5096
+    },
+    {
+      "epoch": 0.18536567625559153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8760523452961813e-05,
+      "loss": 1.0703,
+      "step": 5097
+    },
+    {
+      "epoch": 0.1854020438593301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875995540842727e-05,
+      "loss": 1.2811,
+      "step": 5098
+    },
+    {
+      "epoch": 0.1854384114630687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8759387242360768e-05,
+      "loss": 1.1376,
+      "step": 5099
+    },
+    {
+      "epoch": 0.1854747790668073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87588189547702e-05,
+      "loss": 1.0569,
+      "step": 5100
+    },
+    {
+      "epoch": 0.18551114667054588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875825054566345e-05,
+      "loss": 1.0095,
+      "step": 5101
+    },
+    {
+      "epoch": 0.18554751427428448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875768201504839e-05,
+      "loss": 1.1114,
+      "step": 5102
+    },
+    {
+      "epoch": 0.18558388187802305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8757113362932926e-05,
+      "loss": 1.1547,
+      "step": 5103
+    },
+    {
+      "epoch": 0.18562024948176165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8756544589324938e-05,
+      "loss": 1.0847,
+      "step": 5104
+    },
+    {
+      "epoch": 0.18565661708550024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8755975694232313e-05,
+      "loss": 1.1068,
+      "step": 5105
+    },
+    {
+      "epoch": 0.18569298468923884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8755406677662953e-05,
+      "loss": 1.0531,
+      "step": 5106
+    },
+    {
+      "epoch": 0.1857293522929774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8754837539624745e-05,
+      "loss": 1.1335,
+      "step": 5107
+    },
+    {
+      "epoch": 0.185765719896716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875426828012559e-05,
+      "loss": 0.9995,
+      "step": 5108
+    },
+    {
+      "epoch": 0.1858020875004546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8753698899173383e-05,
+      "loss": 1.1653,
+      "step": 5109
+    },
+    {
+      "epoch": 0.1858384551041932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875312939677602e-05,
+      "loss": 1.0385,
+      "step": 5110
+    },
+    {
+      "epoch": 0.18587482270793176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.875255977294141e-05,
+      "loss": 1.1415,
+      "step": 5111
+    },
+    {
+      "epoch": 0.18591119031167036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8751990027677447e-05,
+      "loss": 1.0086,
+      "step": 5112
+    },
+    {
+      "epoch": 0.18594755791540896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8751420160992045e-05,
+      "loss": 1.0199,
+      "step": 5113
+    },
+    {
+      "epoch": 0.18598392551914755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87508501728931e-05,
+      "loss": 1.0596,
+      "step": 5114
+    },
+    {
+      "epoch": 0.18602029312288612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8750280063388528e-05,
+      "loss": 1.093,
+      "step": 5115
+    },
+    {
+      "epoch": 0.18605666072662472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8749709832486233e-05,
+      "loss": 1.1155,
+      "step": 5116
+    },
+    {
+      "epoch": 0.1860930283303633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.874913948019413e-05,
+      "loss": 0.9554,
+      "step": 5117
+    },
+    {
+      "epoch": 0.1861293959341019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8748569006520128e-05,
+      "loss": 1.1983,
+      "step": 5118
+    },
+    {
+      "epoch": 0.18616576353784048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8747998411472148e-05,
+      "loss": 1.1321,
+      "step": 5119
+    },
+    {
+      "epoch": 0.18620213114157907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87474276950581e-05,
+      "loss": 1.079,
+      "step": 5120
+    },
+    {
+      "epoch": 0.18623849874531767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87468568572859e-05,
+      "loss": 1.1683,
+      "step": 5121
+    },
+    {
+      "epoch": 0.18627486634905627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8746285898163474e-05,
+      "loss": 1.1269,
+      "step": 5122
+    },
+    {
+      "epoch": 0.18631123395279486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8745714817698744e-05,
+      "loss": 1.1584,
+      "step": 5123
+    },
+    {
+      "epoch": 0.18634760155653343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8745143615899623e-05,
+      "loss": 1.1135,
+      "step": 5124
+    },
+    {
+      "epoch": 0.18638396916027203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8744572292774048e-05,
+      "loss": 0.9654,
+      "step": 5125
+    },
+    {
+      "epoch": 0.18642033676401062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8744000848329936e-05,
+      "loss": 1.1199,
+      "step": 5126
+    },
+    {
+      "epoch": 0.18645670436774922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8743429282575216e-05,
+      "loss": 1.1285,
+      "step": 5127
+    },
+    {
+      "epoch": 0.1864930719714878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8742857595517825e-05,
+      "loss": 1.0074,
+      "step": 5128
+    },
+    {
+      "epoch": 0.18652943957522639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8742285787165686e-05,
+      "loss": 1.2844,
+      "step": 5129
+    },
+    {
+      "epoch": 0.18656580717896498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8741713857526736e-05,
+      "loss": 1.1046,
+      "step": 5130
+    },
+    {
+      "epoch": 0.18660217478270358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8741141806608908e-05,
+      "loss": 1.1152,
+      "step": 5131
+    },
+    {
+      "epoch": 0.18663854238644215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8740569634420142e-05,
+      "loss": 1.018,
+      "step": 5132
+    },
+    {
+      "epoch": 0.18667490999018074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8739997340968375e-05,
+      "loss": 1.2545,
+      "step": 5133
+    },
+    {
+      "epoch": 0.18671127759391934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8739424926261544e-05,
+      "loss": 1.2541,
+      "step": 5134
+    },
+    {
+      "epoch": 0.18674764519765794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8738852390307588e-05,
+      "loss": 1.022,
+      "step": 5135
+    },
+    {
+      "epoch": 0.1867840128013965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873827973311446e-05,
+      "loss": 1.1393,
+      "step": 5136
+    },
+    {
+      "epoch": 0.1868203804051351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8737706954690094e-05,
+      "loss": 1.1124,
+      "step": 5137
+    },
+    {
+      "epoch": 0.1868567480088737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873713405504244e-05,
+      "loss": 1.1011,
+      "step": 5138
+    },
+    {
+      "epoch": 0.1868931156126123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873656103417945e-05,
+      "loss": 1.0359,
+      "step": 5139
+    },
+    {
+      "epoch": 0.18692948321635086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8735987892109072e-05,
+      "loss": 1.1198,
+      "step": 5140
+    },
+    {
+      "epoch": 0.18696585082008946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8735414628839258e-05,
+      "loss": 1.1624,
+      "step": 5141
+    },
+    {
+      "epoch": 0.18700221842382805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8734841244377958e-05,
+      "loss": 1.084,
+      "step": 5142
+    },
+    {
+      "epoch": 0.18703858602756665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8734267738733128e-05,
+      "loss": 1.1318,
+      "step": 5143
+    },
+    {
+      "epoch": 0.18707495363130525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8733694111912727e-05,
+      "loss": 1.1555,
+      "step": 5144
+    },
+    {
+      "epoch": 0.18711132123504381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8733120363924707e-05,
+      "loss": 1.1895,
+      "step": 5145
+    },
+    {
+      "epoch": 0.1871476888387824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873254649477704e-05,
+      "loss": 1.1072,
+      "step": 5146
+    },
+    {
+      "epoch": 0.187184056442521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8731972504477675e-05,
+      "loss": 1.0888,
+      "step": 5147
+    },
+    {
+      "epoch": 0.1872204240462596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873139839303458e-05,
+      "loss": 1.1885,
+      "step": 5148
+    },
+    {
+      "epoch": 0.18725679164999817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8730824160455724e-05,
+      "loss": 1.1224,
+      "step": 5149
+    },
+    {
+      "epoch": 0.18729315925373677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873024980674907e-05,
+      "loss": 1.2518,
+      "step": 5150
+    },
+    {
+      "epoch": 0.18732952685747536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8729675331922586e-05,
+      "loss": 1.1093,
+      "step": 5151
+    },
+    {
+      "epoch": 0.18736589446121396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872910073598424e-05,
+      "loss": 0.9946,
+      "step": 5152
+    },
+    {
+      "epoch": 0.18740226206495253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872852601894201e-05,
+      "loss": 1.1627,
+      "step": 5153
+    },
+    {
+      "epoch": 0.18743862966869113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8727951180803865e-05,
+      "loss": 1.1318,
+      "step": 5154
+    },
+    {
+      "epoch": 0.18747499727242972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872737622157778e-05,
+      "loss": 1.067,
+      "step": 5155
+    },
+    {
+      "epoch": 0.18751136487616832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872680114127173e-05,
+      "loss": 1.0955,
+      "step": 5156
+    },
+    {
+      "epoch": 0.1875477324799069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.87262259398937e-05,
+      "loss": 1.1155,
+      "step": 5157
+    },
+    {
+      "epoch": 0.18758410008364548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8725650617451663e-05,
+      "loss": 1.1918,
+      "step": 5158
+    },
+    {
+      "epoch": 0.18762046768738408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8725075173953604e-05,
+      "loss": 1.0866,
+      "step": 5159
+    },
+    {
+      "epoch": 0.18765683529112268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8724499609407507e-05,
+      "loss": 1.1014,
+      "step": 5160
+    },
+    {
+      "epoch": 0.18769320289486124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8723923923821358e-05,
+      "loss": 1.0112,
+      "step": 5161
+    },
+    {
+      "epoch": 0.18772957049859984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8723348117203138e-05,
+      "loss": 1.1604,
+      "step": 5162
+    },
+    {
+      "epoch": 0.18776593810233844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8722772189560844e-05,
+      "loss": 1.1494,
+      "step": 5163
+    },
+    {
+      "epoch": 0.18780230570607703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872219614090246e-05,
+      "loss": 1.1738,
+      "step": 5164
+    },
+    {
+      "epoch": 0.18783867330981563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872161997123598e-05,
+      "loss": 1.1803,
+      "step": 5165
+    },
+    {
+      "epoch": 0.1878750409135542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8721043680569394e-05,
+      "loss": 1.1224,
+      "step": 5166
+    },
+    {
+      "epoch": 0.1879114085172928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.872046726891071e-05,
+      "loss": 1.1473,
+      "step": 5167
+    },
+    {
+      "epoch": 0.1879477761210314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8719890736267905e-05,
+      "loss": 1.1253,
+      "step": 5168
+    },
+    {
+      "epoch": 0.18798414372477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8719314082648993e-05,
+      "loss": 1.1677,
+      "step": 5169
+    },
+    {
+      "epoch": 0.18802051132850855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.871873730806197e-05,
+      "loss": 1.2332,
+      "step": 5170
+    },
+    {
+      "epoch": 0.18805687893224715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.871816041251483e-05,
+      "loss": 1.0757,
+      "step": 5171
+    },
+    {
+      "epoch": 0.18809324653598575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.871758339601559e-05,
+      "loss": 1.092,
+      "step": 5172
+    },
+    {
+      "epoch": 0.18812961413972434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.871700625857225e-05,
+      "loss": 1.1664,
+      "step": 5173
+    },
+    {
+      "epoch": 0.1881659817434629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8716429000192814e-05,
+      "loss": 1.1925,
+      "step": 5174
+    },
+    {
+      "epoch": 0.1882023493472015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8715851620885292e-05,
+      "loss": 1.0496,
+      "step": 5175
+    },
+    {
+      "epoch": 0.1882387169509401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8715274120657694e-05,
+      "loss": 1.1102,
+      "step": 5176
+    },
+    {
+      "epoch": 0.1882750845546787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8714696499518034e-05,
+      "loss": 1.0826,
+      "step": 5177
+    },
+    {
+      "epoch": 0.18831145215841727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8714118757474328e-05,
+      "loss": 1.1297,
+      "step": 5178
+    },
+    {
+      "epoch": 0.18834781976215587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8713540894534584e-05,
+      "loss": 1.1909,
+      "step": 5179
+    },
+    {
+      "epoch": 0.18838418736589446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8712962910706824e-05,
+      "loss": 1.095,
+      "step": 5180
+    },
+    {
+      "epoch": 0.18842055496963306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8712384805999063e-05,
+      "loss": 1.1916,
+      "step": 5181
+    },
+    {
+      "epoch": 0.18845692257337163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8711806580419328e-05,
+      "loss": 1.0451,
+      "step": 5182
+    },
+    {
+      "epoch": 0.18849329017711022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8711228233975634e-05,
+      "loss": 1.0658,
+      "step": 5183
+    },
+    {
+      "epoch": 0.18852965778084882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8710649766676013e-05,
+      "loss": 1.162,
+      "step": 5184
+    },
+    {
+      "epoch": 0.18856602538458742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.871007117852848e-05,
+      "loss": 1.0439,
+      "step": 5185
+    },
+    {
+      "epoch": 0.188602392988326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8709492469541072e-05,
+      "loss": 1.1145,
+      "step": 5186
+    },
+    {
+      "epoch": 0.18863876059206458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8708913639721808e-05,
+      "loss": 1.1172,
+      "step": 5187
+    },
+    {
+      "epoch": 0.18867512819580318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8708334689078726e-05,
+      "loss": 1.1492,
+      "step": 5188
+    },
+    {
+      "epoch": 0.18871149579954177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8707755617619855e-05,
+      "loss": 1.1966,
+      "step": 5189
+    },
+    {
+      "epoch": 0.18874786340328037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870717642535323e-05,
+      "loss": 1.0921,
+      "step": 5190
+    },
+    {
+      "epoch": 0.18878423100701894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870659711228689e-05,
+      "loss": 1.2008,
+      "step": 5191
+    },
+    {
+      "epoch": 0.18882059861075753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870601767842886e-05,
+      "loss": 1.1004,
+      "step": 5192
+    },
+    {
+      "epoch": 0.18885696621449613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8705438123787193e-05,
+      "loss": 1.0584,
+      "step": 5193
+    },
+    {
+      "epoch": 0.18889333381823473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870485844836992e-05,
+      "loss": 1.1284,
+      "step": 5194
+    },
+    {
+      "epoch": 0.1889297014219733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870427865218509e-05,
+      "loss": 0.9666,
+      "step": 5195
+    },
+    {
+      "epoch": 0.1889660690257119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8703698735240742e-05,
+      "loss": 1.1732,
+      "step": 5196
+    },
+    {
+      "epoch": 0.1890024366294505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8703118697544923e-05,
+      "loss": 1.0802,
+      "step": 5197
+    },
+    {
+      "epoch": 0.18903880423318908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870253853910568e-05,
+      "loss": 1.1126,
+      "step": 5198
+    },
+    {
+      "epoch": 0.18907517183692765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870195825993106e-05,
+      "loss": 1.1198,
+      "step": 5199
+    },
+    {
+      "epoch": 0.18911153944066625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870137786002912e-05,
+      "loss": 1.0773,
+      "step": 5200
+    },
+    {
+      "epoch": 0.18914790704440484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8700797339407905e-05,
+      "loss": 1.2615,
+      "step": 5201
+    },
+    {
+      "epoch": 0.18918427464814344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870021669807547e-05,
+      "loss": 1.0916,
+      "step": 5202
+    },
+    {
+      "epoch": 0.189220642251882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8699635936039876e-05,
+      "loss": 1.0452,
+      "step": 5203
+    },
+    {
+      "epoch": 0.1892570098556206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8699055053309173e-05,
+      "loss": 1.1369,
+      "step": 5204
+    },
+    {
+      "epoch": 0.1892933774593592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869847404989143e-05,
+      "loss": 1.0919,
+      "step": 5205
+    },
+    {
+      "epoch": 0.1893297450630978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8697892925794693e-05,
+      "loss": 1.0989,
+      "step": 5206
+    },
+    {
+      "epoch": 0.1893661126668364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8697311681027037e-05,
+      "loss": 1.0097,
+      "step": 5207
+    },
+    {
+      "epoch": 0.18940248027057496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869673031559652e-05,
+      "loss": 1.132,
+      "step": 5208
+    },
+    {
+      "epoch": 0.18943884787431356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869614882951121e-05,
+      "loss": 1.2484,
+      "step": 5209
+    },
+    {
+      "epoch": 0.18947521547805216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8695567222779173e-05,
+      "loss": 1.0603,
+      "step": 5210
+    },
+    {
+      "epoch": 0.18951158308179075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8694985495408475e-05,
+      "loss": 1.0769,
+      "step": 5211
+    },
+    {
+      "epoch": 0.18954795068552932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8694403647407194e-05,
+      "loss": 1.148,
+      "step": 5212
+    },
+    {
+      "epoch": 0.18958431828926792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8693821678783396e-05,
+      "loss": 1.0738,
+      "step": 5213
+    },
+    {
+      "epoch": 0.1896206858930065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8693239589545156e-05,
+      "loss": 1.0974,
+      "step": 5214
+    },
+    {
+      "epoch": 0.1896570534967451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8692657379700554e-05,
+      "loss": 1.0868,
+      "step": 5215
+    },
+    {
+      "epoch": 0.18969342110048368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.869207504925766e-05,
+      "loss": 1.1801,
+      "step": 5216
+    },
+    {
+      "epoch": 0.18972978870422227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8691492598224565e-05,
+      "loss": 1.1482,
+      "step": 5217
+    },
+    {
+      "epoch": 0.18976615630796087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8690910026609335e-05,
+      "loss": 1.1226,
+      "step": 5218
+    },
+    {
+      "epoch": 0.18980252391169947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8690327334420056e-05,
+      "loss": 1.1834,
+      "step": 5219
+    },
+    {
+      "epoch": 0.18983889151543804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8689744521664823e-05,
+      "loss": 1.0955,
+      "step": 5220
+    },
+    {
+      "epoch": 0.18987525911917663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8689161588351708e-05,
+      "loss": 1.2736,
+      "step": 5221
+    },
+    {
+      "epoch": 0.18991162672291523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8688578534488807e-05,
+      "loss": 1.0868,
+      "step": 5222
+    },
+    {
+      "epoch": 0.18994799432665382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868799536008421e-05,
+      "loss": 1.039,
+      "step": 5223
+    },
+    {
+      "epoch": 0.18998436193039242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8687412065145998e-05,
+      "loss": 1.0218,
+      "step": 5224
+    },
+    {
+      "epoch": 0.190020729534131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8686828649682272e-05,
+      "loss": 1.08,
+      "step": 5225
+    },
+    {
+      "epoch": 0.19005709713786959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8686245113701123e-05,
+      "loss": 1.2225,
+      "step": 5226
+    },
+    {
+      "epoch": 0.19009346474160818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8685661457210648e-05,
+      "loss": 1.0523,
+      "step": 5227
+    },
+    {
+      "epoch": 0.19012983234534678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8685077680218945e-05,
+      "loss": 1.1647,
+      "step": 5228
+    },
+    {
+      "epoch": 0.19016619994908535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868449378273411e-05,
+      "loss": 1.106,
+      "step": 5229
+    },
+    {
+      "epoch": 0.19020256755282394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8683909764764246e-05,
+      "loss": 1.2828,
+      "step": 5230
+    },
+    {
+      "epoch": 0.19023893515656254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8683325626317452e-05,
+      "loss": 1.0161,
+      "step": 5231
+    },
+    {
+      "epoch": 0.19027530276030113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8682741367401837e-05,
+      "loss": 1.0432,
+      "step": 5232
+    },
+    {
+      "epoch": 0.1903116703640397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8682156988025503e-05,
+      "loss": 1.1195,
+      "step": 5233
+    },
+    {
+      "epoch": 0.1903480379677783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868157248819656e-05,
+      "loss": 1.1447,
+      "step": 5234
+    },
+    {
+      "epoch": 0.1903844055715169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868098786792312e-05,
+      "loss": 1.083,
+      "step": 5235
+    },
+    {
+      "epoch": 0.1904207731752555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8680403127213286e-05,
+      "loss": 0.9906,
+      "step": 5236
+    },
+    {
+      "epoch": 0.19045714077899406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8679818266075176e-05,
+      "loss": 1.3308,
+      "step": 5237
+    },
+    {
+      "epoch": 0.19049350838273266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8679233284516903e-05,
+      "loss": 1.2302,
+      "step": 5238
+    },
+    {
+      "epoch": 0.19052987598647125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8678648182546583e-05,
+      "loss": 1.2066,
+      "step": 5239
+    },
+    {
+      "epoch": 0.19056624359020985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867806296017233e-05,
+      "loss": 1.1639,
+      "step": 5240
+    },
+    {
+      "epoch": 0.19060261119394842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867747761740227e-05,
+      "loss": 1.0957,
+      "step": 5241
+    },
+    {
+      "epoch": 0.19063897879768701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8676892154244516e-05,
+      "loss": 1.0455,
+      "step": 5242
+    },
+    {
+      "epoch": 0.1906753464014256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8676306570707194e-05,
+      "loss": 1.2819,
+      "step": 5243
+    },
+    {
+      "epoch": 0.1907117140051642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867572086679843e-05,
+      "loss": 1.1262,
+      "step": 5244
+    },
+    {
+      "epoch": 0.1907480816089028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8675135042526352e-05,
+      "loss": 1.0464,
+      "step": 5245
+    },
+    {
+      "epoch": 0.19078444921264137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8674549097899078e-05,
+      "loss": 1.1586,
+      "step": 5246
+    },
+    {
+      "epoch": 0.19082081681637997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8673963032924743e-05,
+      "loss": 1.0948,
+      "step": 5247
+    },
+    {
+      "epoch": 0.19085718442011856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8673376847611482e-05,
+      "loss": 1.2441,
+      "step": 5248
+    },
+    {
+      "epoch": 0.19089355202385716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867279054196742e-05,
+      "loss": 1.0458,
+      "step": 5249
+    },
+    {
+      "epoch": 0.19092991962759573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8672204116000693e-05,
+      "loss": 1.2038,
+      "step": 5250
+    },
+    {
+      "epoch": 0.19096628723133433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.867161756971944e-05,
+      "loss": 1.109,
+      "step": 5251
+    },
+    {
+      "epoch": 0.19100265483507292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.86710309031318e-05,
+      "loss": 0.9305,
+      "step": 5252
+    },
+    {
+      "epoch": 0.19103902243881152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8670444116245904e-05,
+      "loss": 1.0338,
+      "step": 5253
+    },
+    {
+      "epoch": 0.1910753900425501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8669857209069898e-05,
+      "loss": 1.1356,
+      "step": 5254
+    },
+    {
+      "epoch": 0.19111175764628868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8669270181611927e-05,
+      "loss": 1.1046,
+      "step": 5255
+    },
+    {
+      "epoch": 0.19114812525002728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8668683033880132e-05,
+      "loss": 1.1407,
+      "step": 5256
+    },
+    {
+      "epoch": 0.19118449285376587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8668095765882656e-05,
+      "loss": 1.1088,
+      "step": 5257
+    },
+    {
+      "epoch": 0.19122086045750444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8667508377627653e-05,
+      "loss": 1.1348,
+      "step": 5258
+    },
+    {
+      "epoch": 0.19125722806124304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8666920869123265e-05,
+      "loss": 1.1389,
+      "step": 5259
+    },
+    {
+      "epoch": 0.19129359566498164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866633324037765e-05,
+      "loss": 1.1144,
+      "step": 5260
+    },
+    {
+      "epoch": 0.19132996326872023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8665745491398954e-05,
+      "loss": 1.1079,
+      "step": 5261
+    },
+    {
+      "epoch": 0.1913663308724588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8665157622195337e-05,
+      "loss": 1.1933,
+      "step": 5262
+    },
+    {
+      "epoch": 0.1914026984761974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8664569632774953e-05,
+      "loss": 1.1202,
+      "step": 5263
+    },
+    {
+      "epoch": 0.191439066079936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8663981523145957e-05,
+      "loss": 1.0846,
+      "step": 5264
+    },
+    {
+      "epoch": 0.1914754336836746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866339329331651e-05,
+      "loss": 1.0646,
+      "step": 5265
+    },
+    {
+      "epoch": 0.19151180128741319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8662804943294773e-05,
+      "loss": 1.2369,
+      "step": 5266
+    },
+    {
+      "epoch": 0.19154816889115175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.866221647308891e-05,
+      "loss": 0.9437,
+      "step": 5267
+    },
+    {
+      "epoch": 0.19158453649489035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8661627882707083e-05,
+      "loss": 1.1469,
+      "step": 5268
+    },
+    {
+      "epoch": 0.19162090409862895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8661039172157457e-05,
+      "loss": 1.1521,
+      "step": 5269
+    },
+    {
+      "epoch": 0.19165727170236754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8660450341448202e-05,
+      "loss": 1.215,
+      "step": 5270
+    },
+    {
+      "epoch": 0.1916936393061061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8659861390587486e-05,
+      "loss": 1.1055,
+      "step": 5271
+    },
+    {
+      "epoch": 0.1917300069098447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865927231958348e-05,
+      "loss": 1.1453,
+      "step": 5272
+    },
+    {
+      "epoch": 0.1917663745135833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8658683128444352e-05,
+      "loss": 1.1526,
+      "step": 5273
+    },
+    {
+      "epoch": 0.1918027421173219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8658093817178286e-05,
+      "loss": 1.049,
+      "step": 5274
+    },
+    {
+      "epoch": 0.19183910972106047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865750438579345e-05,
+      "loss": 1.141,
+      "step": 5275
+    },
+    {
+      "epoch": 0.19187547732479907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8656914834298026e-05,
+      "loss": 1.1343,
+      "step": 5276
+    },
+    {
+      "epoch": 0.19191184492853766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865632516270019e-05,
+      "loss": 1.1956,
+      "step": 5277
+    },
+    {
+      "epoch": 0.19194821253227626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8655735371008126e-05,
+      "loss": 1.0777,
+      "step": 5278
+    },
+    {
+      "epoch": 0.19198458013601483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8655145459230012e-05,
+      "loss": 1.1823,
+      "step": 5279
+    },
+    {
+      "epoch": 0.19202094773975342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8654555427374037e-05,
+      "loss": 1.1636,
+      "step": 5280
+    },
+    {
+      "epoch": 0.19205731534349202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865396527544838e-05,
+      "loss": 1.0264,
+      "step": 5281
+    },
+    {
+      "epoch": 0.19209368294723062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865337500346124e-05,
+      "loss": 1.1217,
+      "step": 5282
+    },
+    {
+      "epoch": 0.19213005055096918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8652784611420797e-05,
+      "loss": 1.1107,
+      "step": 5283
+    },
+    {
+      "epoch": 0.19216641815470778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8652194099335244e-05,
+      "loss": 1.2229,
+      "step": 5284
+    },
+    {
+      "epoch": 0.19220278575844638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865160346721277e-05,
+      "loss": 1.2194,
+      "step": 5285
+    },
+    {
+      "epoch": 0.19223915336218497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865101271506158e-05,
+      "loss": 1.1933,
+      "step": 5286
+    },
+    {
+      "epoch": 0.19227552096592357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8650421842889853e-05,
+      "loss": 1.1614,
+      "step": 5287
+    },
+    {
+      "epoch": 0.19231188856966214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8649830850705805e-05,
+      "loss": 1.0942,
+      "step": 5288
+    },
+    {
+      "epoch": 0.19234825617340073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8649239738517623e-05,
+      "loss": 1.2429,
+      "step": 5289
+    },
+    {
+      "epoch": 0.19238462377713933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.864864850633351e-05,
+      "loss": 1.2345,
+      "step": 5290
+    },
+    {
+      "epoch": 0.19242099138087793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8648057154161674e-05,
+      "loss": 1.1735,
+      "step": 5291
+    },
+    {
+      "epoch": 0.1924573589846165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.864746568201031e-05,
+      "loss": 1.0744,
+      "step": 5292
+    },
+    {
+      "epoch": 0.1924937265883551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8646874089887636e-05,
+      "loss": 1.0741,
+      "step": 5293
+    },
+    {
+      "epoch": 0.1925300941920937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8646282377801843e-05,
+      "loss": 1.0957,
+      "step": 5294
+    },
+    {
+      "epoch": 0.19256646179583228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8645690545761156e-05,
+      "loss": 1.0785,
+      "step": 5295
+    },
+    {
+      "epoch": 0.19260282939957085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.864509859377378e-05,
+      "loss": 1.1806,
+      "step": 5296
+    },
+    {
+      "epoch": 0.19263919700330945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8644506521847923e-05,
+      "loss": 1.0981,
+      "step": 5297
+    },
+    {
+      "epoch": 0.19267556460704804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8643914329991803e-05,
+      "loss": 1.1078,
+      "step": 5298
+    },
+    {
+      "epoch": 0.19271193221078664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8643322018213636e-05,
+      "loss": 1.2036,
+      "step": 5299
+    },
+    {
+      "epoch": 0.1927482998145252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.864272958652164e-05,
+      "loss": 0.9926,
+      "step": 5300
+    },
+    {
+      "epoch": 0.1927846674182638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8642137034924036e-05,
+      "loss": 1.0881,
+      "step": 5301
+    },
+    {
+      "epoch": 0.1928210350220024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.864154436342904e-05,
+      "loss": 1.153,
+      "step": 5302
+    },
+    {
+      "epoch": 0.192857402625741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8640951572044876e-05,
+      "loss": 1.0585,
+      "step": 5303
+    },
+    {
+      "epoch": 0.19289377022947957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.864035866077977e-05,
+      "loss": 1.1105,
+      "step": 5304
+    },
+    {
+      "epoch": 0.19293013783321816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8639765629641943e-05,
+      "loss": 1.1021,
+      "step": 5305
+    },
+    {
+      "epoch": 0.19296650543695676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8639172478639632e-05,
+      "loss": 1.1683,
+      "step": 5306
+    },
+    {
+      "epoch": 0.19300287304069536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8638579207781055e-05,
+      "loss": 1.0798,
+      "step": 5307
+    },
+    {
+      "epoch": 0.19303924064443395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.863798581707445e-05,
+      "loss": 1.2211,
+      "step": 5308
+    },
+    {
+      "epoch": 0.19307560824817252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8637392306528047e-05,
+      "loss": 1.0851,
+      "step": 5309
+    },
+    {
+      "epoch": 0.19311197585191112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8636798676150083e-05,
+      "loss": 1.1265,
+      "step": 5310
+    },
+    {
+      "epoch": 0.1931483434556497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8636204925948788e-05,
+      "loss": 1.1296,
+      "step": 5311
+    },
+    {
+      "epoch": 0.1931847110593883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8635611055932406e-05,
+      "loss": 1.0385,
+      "step": 5312
+    },
+    {
+      "epoch": 0.19322107866312688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8635017066109173e-05,
+      "loss": 1.006,
+      "step": 5313
+    },
+    {
+      "epoch": 0.19325744626686547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8634422956487327e-05,
+      "loss": 1.1212,
+      "step": 5314
+    },
+    {
+      "epoch": 0.19329381387060407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8633828727075112e-05,
+      "loss": 1.159,
+      "step": 5315
+    },
+    {
+      "epoch": 0.19333018147434267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8633234377880777e-05,
+      "loss": 1.0161,
+      "step": 5316
+    },
+    {
+      "epoch": 0.19336654907808123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8632639908912565e-05,
+      "loss": 1.0716,
+      "step": 5317
+    },
+    {
+      "epoch": 0.19340291668181983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.863204532017872e-05,
+      "loss": 1.2042,
+      "step": 5318
+    },
+    {
+      "epoch": 0.19343928428555843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8631450611687495e-05,
+      "loss": 1.1148,
+      "step": 5319
+    },
+    {
+      "epoch": 0.19347565188929702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8630855783447134e-05,
+      "loss": 1.231,
+      "step": 5320
+    },
+    {
+      "epoch": 0.1935120194930356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.86302608354659e-05,
+      "loss": 1.1479,
+      "step": 5321
+    },
+    {
+      "epoch": 0.1935483870967742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862966576775204e-05,
+      "loss": 1.1523,
+      "step": 5322
+    },
+    {
+      "epoch": 0.19358475470051278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862907058031381e-05,
+      "loss": 1.1377,
+      "step": 5323
+    },
+    {
+      "epoch": 0.19362112230425138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8628475273159473e-05,
+      "loss": 1.2698,
+      "step": 5324
+    },
+    {
+      "epoch": 0.19365748990798995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862787984629728e-05,
+      "loss": 1.1836,
+      "step": 5325
+    },
+    {
+      "epoch": 0.19369385751172855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8627284299735497e-05,
+      "loss": 1.1939,
+      "step": 5326
+    },
+    {
+      "epoch": 0.19373022511546714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862668863348238e-05,
+      "loss": 1.1375,
+      "step": 5327
+    },
+    {
+      "epoch": 0.19376659271920574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8626092847546204e-05,
+      "loss": 1.1342,
+      "step": 5328
+    },
+    {
+      "epoch": 0.19380296032294433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862549694193523e-05,
+      "loss": 1.1786,
+      "step": 5329
+    },
+    {
+      "epoch": 0.1938393279266829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862490091665772e-05,
+      "loss": 1.0958,
+      "step": 5330
+    },
+    {
+      "epoch": 0.1938756955304215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8624304771721945e-05,
+      "loss": 1.1409,
+      "step": 5331
+    },
+    {
+      "epoch": 0.1939120631341601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862370850713618e-05,
+      "loss": 1.1281,
+      "step": 5332
+    },
+    {
+      "epoch": 0.1939484307378987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8623112122908696e-05,
+      "loss": 1.0327,
+      "step": 5333
+    },
+    {
+      "epoch": 0.19398479834163726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8622515619047763e-05,
+      "loss": 1.14,
+      "step": 5334
+    },
+    {
+      "epoch": 0.19402116594537586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8621918995561663e-05,
+      "loss": 1.2129,
+      "step": 5335
+    },
+    {
+      "epoch": 0.19405753354911445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862132225245867e-05,
+      "loss": 1.1245,
+      "step": 5336
+    },
+    {
+      "epoch": 0.19409390115285305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.862072538974706e-05,
+      "loss": 1.1445,
+      "step": 5337
+    },
+    {
+      "epoch": 0.19413026875659162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8620128407435114e-05,
+      "loss": 1.0057,
+      "step": 5338
+    },
+    {
+      "epoch": 0.1941666363603302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861953130553112e-05,
+      "loss": 1.3096,
+      "step": 5339
+    },
+    {
+      "epoch": 0.1942030039640688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861893408404336e-05,
+      "loss": 1.183,
+      "step": 5340
+    },
+    {
+      "epoch": 0.1942393715678074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8618336742980116e-05,
+      "loss": 1.0591,
+      "step": 5341
+    },
+    {
+      "epoch": 0.19427573917154597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861773928234968e-05,
+      "loss": 1.2123,
+      "step": 5342
+    },
+    {
+      "epoch": 0.19431210677528457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861714170216034e-05,
+      "loss": 1.2698,
+      "step": 5343
+    },
+    {
+      "epoch": 0.19434847437902317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861654400242038e-05,
+      "loss": 1.0601,
+      "step": 5344
+    },
+    {
+      "epoch": 0.19438484198276176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.86159461831381e-05,
+      "loss": 1.0454,
+      "step": 5345
+    },
+    {
+      "epoch": 0.19442120958650033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861534824432179e-05,
+      "loss": 1.1006,
+      "step": 5346
+    },
+    {
+      "epoch": 0.19445757719023893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8614750185979744e-05,
+      "loss": 1.1024,
+      "step": 5347
+    },
+    {
+      "epoch": 0.19449394479397752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861415200812027e-05,
+      "loss": 1.1029,
+      "step": 5348
+    },
+    {
+      "epoch": 0.19453031239771612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861355371075165e-05,
+      "loss": 1.1287,
+      "step": 5349
+    },
+    {
+      "epoch": 0.19456668000145472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8612955293882197e-05,
+      "loss": 1.0985,
+      "step": 5350
+    },
+    {
+      "epoch": 0.19460304760519329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8612356757520207e-05,
+      "loss": 1.1834,
+      "step": 5351
+    },
+    {
+      "epoch": 0.19463941520893188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.861175810167399e-05,
+      "loss": 1.0642,
+      "step": 5352
+    },
+    {
+      "epoch": 0.19467578281267048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8611159326351847e-05,
+      "loss": 1.1698,
+      "step": 5353
+    },
+    {
+      "epoch": 0.19471215041640907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8610560431562085e-05,
+      "loss": 0.9714,
+      "step": 5354
+    },
+    {
+      "epoch": 0.19474851802014764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8609961417313017e-05,
+      "loss": 1.164,
+      "step": 5355
+    },
+    {
+      "epoch": 0.19478488562388624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8609362283612948e-05,
+      "loss": 1.1421,
+      "step": 5356
+    },
+    {
+      "epoch": 0.19482125322762484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860876303047019e-05,
+      "loss": 1.0923,
+      "step": 5357
+    },
+    {
+      "epoch": 0.19485762083136343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8608163657893063e-05,
+      "loss": 1.1782,
+      "step": 5358
+    },
+    {
+      "epoch": 0.194893988435102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8607564165889877e-05,
+      "loss": 1.127,
+      "step": 5359
+    },
+    {
+      "epoch": 0.1949303560388406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8606964554468948e-05,
+      "loss": 1.0329,
+      "step": 5360
+    },
+    {
+      "epoch": 0.1949667236425792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8606364823638602e-05,
+      "loss": 1.1258,
+      "step": 5361
+    },
+    {
+      "epoch": 0.1950030912463178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8605764973407155e-05,
+      "loss": 1.0202,
+      "step": 5362
+    },
+    {
+      "epoch": 0.19503945885005636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8605165003782927e-05,
+      "loss": 1.1017,
+      "step": 5363
+    },
+    {
+      "epoch": 0.19507582645379495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860456491477424e-05,
+      "loss": 1.0422,
+      "step": 5364
+    },
+    {
+      "epoch": 0.19511219405753355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8603964706389427e-05,
+      "loss": 1.0742,
+      "step": 5365
+    },
+    {
+      "epoch": 0.19514856166127215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8603364378636813e-05,
+      "loss": 1.1787,
+      "step": 5366
+    },
+    {
+      "epoch": 0.19518492926501071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8602763931524724e-05,
+      "loss": 1.1478,
+      "step": 5367
+    },
+    {
+      "epoch": 0.1952212968687493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8602163365061488e-05,
+      "loss": 1.0501,
+      "step": 5368
+    },
+    {
+      "epoch": 0.1952576644724879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860156267925544e-05,
+      "loss": 1.1664,
+      "step": 5369
+    },
+    {
+      "epoch": 0.1952940320762265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860096187411492e-05,
+      "loss": 1.0936,
+      "step": 5370
+    },
+    {
+      "epoch": 0.1953303996799651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860036094964825e-05,
+      "loss": 1.1499,
+      "step": 5371
+    },
+    {
+      "epoch": 0.19536676728370367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859975990586378e-05,
+      "loss": 1.1314,
+      "step": 5372
+    },
+    {
+      "epoch": 0.19540313488744226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8599158742769837e-05,
+      "loss": 1.104,
+      "step": 5373
+    },
+    {
+      "epoch": 0.19543950249118086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8598557460374774e-05,
+      "loss": 1.2267,
+      "step": 5374
+    },
+    {
+      "epoch": 0.19547587009491946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859795605868692e-05,
+      "loss": 1.2442,
+      "step": 5375
+    },
+    {
+      "epoch": 0.19551223769865803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8597354537714628e-05,
+      "loss": 1.1242,
+      "step": 5376
+    },
+    {
+      "epoch": 0.19554860530239662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8596752897466238e-05,
+      "loss": 1.1776,
+      "step": 5377
+    },
+    {
+      "epoch": 0.19558497290613522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.85961511379501e-05,
+      "loss": 1.2025,
+      "step": 5378
+    },
+    {
+      "epoch": 0.19562134050987381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8595549259174558e-05,
+      "loss": 1.2042,
+      "step": 5379
+    },
+    {
+      "epoch": 0.19565770811361238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8594947261147968e-05,
+      "loss": 1.0438,
+      "step": 5380
+    },
+    {
+      "epoch": 0.19569407571735098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859434514387868e-05,
+      "loss": 1.1275,
+      "step": 5381
+    },
+    {
+      "epoch": 0.19573044332108958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859374290737505e-05,
+      "loss": 1.2029,
+      "step": 5382
+    },
+    {
+      "epoch": 0.19576681092482817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8593140551645422e-05,
+      "loss": 1.1088,
+      "step": 5383
+    },
+    {
+      "epoch": 0.19580317852856674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859253807669817e-05,
+      "loss": 1.1426,
+      "step": 5384
+    },
+    {
+      "epoch": 0.19583954613230534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8591935482541636e-05,
+      "loss": 1.1869,
+      "step": 5385
+    },
+    {
+      "epoch": 0.19587591373604393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859133276918419e-05,
+      "loss": 1.0729,
+      "step": 5386
+    },
+    {
+      "epoch": 0.19591228133978253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.859072993663419e-05,
+      "loss": 1.0867,
+      "step": 5387
+    },
+    {
+      "epoch": 0.1959486489435211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8590126984900005e-05,
+      "loss": 1.1736,
+      "step": 5388
+    },
+    {
+      "epoch": 0.1959850165472597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8589523913989992e-05,
+      "loss": 1.1681,
+      "step": 5389
+    },
+    {
+      "epoch": 0.1960213841509983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8588920723912523e-05,
+      "loss": 1.1648,
+      "step": 5390
+    },
+    {
+      "epoch": 0.1960577517547369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8588317414675966e-05,
+      "loss": 0.9637,
+      "step": 5391
+    },
+    {
+      "epoch": 0.19609411935847548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8587713986288688e-05,
+      "loss": 0.9927,
+      "step": 5392
+    },
+    {
+      "epoch": 0.19613048696221405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8587110438759063e-05,
+      "loss": 1.1711,
+      "step": 5393
+    },
+    {
+      "epoch": 0.19616685456595265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8586506772095467e-05,
+      "loss": 1.1301,
+      "step": 5394
+    },
+    {
+      "epoch": 0.19620322216969124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.858590298630627e-05,
+      "loss": 1.1546,
+      "step": 5395
+    },
+    {
+      "epoch": 0.19623958977342984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.858529908139985e-05,
+      "loss": 1.0211,
+      "step": 5396
+    },
+    {
+      "epoch": 0.1962759573771684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8584695057384586e-05,
+      "loss": 1.0465,
+      "step": 5397
+    },
+    {
+      "epoch": 0.196312324980907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.858409091426886e-05,
+      "loss": 1.1263,
+      "step": 5398
+    },
+    {
+      "epoch": 0.1963486925846456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8583486652061054e-05,
+      "loss": 1.22,
+      "step": 5399
+    },
+    {
+      "epoch": 0.1963850601883842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8582882270769547e-05,
+      "loss": 1.1585,
+      "step": 5400
+    },
+    {
+      "epoch": 0.19642142779212277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8582277770402727e-05,
+      "loss": 1.1194,
+      "step": 5401
+    },
+    {
+      "epoch": 0.19645779539586136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8581673150968974e-05,
+      "loss": 0.8636,
+      "step": 5402
+    },
+    {
+      "epoch": 0.19649416299959996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.858106841247669e-05,
+      "loss": 1.0281,
+      "step": 5403
+    },
+    {
+      "epoch": 0.19653053060333855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.858046355493425e-05,
+      "loss": 1.1404,
+      "step": 5404
+    },
+    {
+      "epoch": 0.19656689820707712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8579858578350054e-05,
+      "loss": 1.122,
+      "step": 5405
+    },
+    {
+      "epoch": 0.19660326581081572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8579253482732494e-05,
+      "loss": 1.1899,
+      "step": 5406
+    },
+    {
+      "epoch": 0.19663963341455432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8578648268089966e-05,
+      "loss": 1.1205,
+      "step": 5407
+    },
+    {
+      "epoch": 0.1966760010182929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8578042934430863e-05,
+      "loss": 1.1051,
+      "step": 5408
+    },
+    {
+      "epoch": 0.1967123686220315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8577437481763586e-05,
+      "loss": 1.1491,
+      "step": 5409
+    },
+    {
+      "epoch": 0.19674873622577008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8576831910096532e-05,
+      "loss": 1.2579,
+      "step": 5410
+    },
+    {
+      "epoch": 0.19678510382950867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8576226219438106e-05,
+      "loss": 1.1181,
+      "step": 5411
+    },
+    {
+      "epoch": 0.19682147143324727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8575620409796705e-05,
+      "loss": 1.0103,
+      "step": 5412
+    },
+    {
+      "epoch": 0.19685783903698587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8575014481180744e-05,
+      "loss": 1.16,
+      "step": 5413
+    },
+    {
+      "epoch": 0.19689420664072443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.857440843359862e-05,
+      "loss": 1.0907,
+      "step": 5414
+    },
+    {
+      "epoch": 0.19693057424446303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.857380226705875e-05,
+      "loss": 1.061,
+      "step": 5415
+    },
+    {
+      "epoch": 0.19696694184820163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8573195981569528e-05,
+      "loss": 1.0828,
+      "step": 5416
+    },
+    {
+      "epoch": 0.19700330945194022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8572589577139382e-05,
+      "loss": 1.1281,
+      "step": 5417
+    },
+    {
+      "epoch": 0.1970396770556788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.857198305377672e-05,
+      "loss": 1.2654,
+      "step": 5418
+    },
+    {
+      "epoch": 0.1970760446594174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.857137641148995e-05,
+      "loss": 1.0238,
+      "step": 5419
+    },
+    {
+      "epoch": 0.19711241226315598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8570769650287496e-05,
+      "loss": 1.1401,
+      "step": 5420
+    },
+    {
+      "epoch": 0.19714877986689458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8570162770177775e-05,
+      "loss": 1.1139,
+      "step": 5421
+    },
+    {
+      "epoch": 0.19718514747063315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8569555771169203e-05,
+      "loss": 1.1436,
+      "step": 5422
+    },
+    {
+      "epoch": 0.19722151507437174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8568948653270206e-05,
+      "loss": 1.1987,
+      "step": 5423
+    },
+    {
+      "epoch": 0.19725788267811034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8568341416489202e-05,
+      "loss": 1.1867,
+      "step": 5424
+    },
+    {
+      "epoch": 0.19729425028184894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8567734060834616e-05,
+      "loss": 1.1204,
+      "step": 5425
+    },
+    {
+      "epoch": 0.1973306178855875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856712658631488e-05,
+      "loss": 1.1304,
+      "step": 5426
+    },
+    {
+      "epoch": 0.1973669854893261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8566518992938416e-05,
+      "loss": 1.0815,
+      "step": 5427
+    },
+    {
+      "epoch": 0.1974033530930647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8565911280713656e-05,
+      "loss": 1.0024,
+      "step": 5428
+    },
+    {
+      "epoch": 0.1974397206968033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8565303449649027e-05,
+      "loss": 1.0837,
+      "step": 5429
+    },
+    {
+      "epoch": 0.1974760883005419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8564695499752968e-05,
+      "loss": 1.1703,
+      "step": 5430
+    },
+    {
+      "epoch": 0.19751245590428046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856408743103391e-05,
+      "loss": 1.0482,
+      "step": 5431
+    },
+    {
+      "epoch": 0.19754882350801906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856347924350029e-05,
+      "loss": 1.1185,
+      "step": 5432
+    },
+    {
+      "epoch": 0.19758519111175765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8562870937160545e-05,
+      "loss": 1.2,
+      "step": 5433
+    },
+    {
+      "epoch": 0.19762155871549625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8562262512023115e-05,
+      "loss": 1.0961,
+      "step": 5434
+    },
+    {
+      "epoch": 0.19765792631923482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8561653968096443e-05,
+      "loss": 1.0772,
+      "step": 5435
+    },
+    {
+      "epoch": 0.1976942939229734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8561045305388968e-05,
+      "loss": 1.0919,
+      "step": 5436
+    },
+    {
+      "epoch": 0.197730661526712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8560436523909137e-05,
+      "loss": 1.094,
+      "step": 5437
+    },
+    {
+      "epoch": 0.1977670291304506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855982762366539e-05,
+      "loss": 1.1071,
+      "step": 5438
+    },
+    {
+      "epoch": 0.19780339673418917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8559218604666185e-05,
+      "loss": 1.3112,
+      "step": 5439
+    },
+    {
+      "epoch": 0.19783976433792777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8558609466919963e-05,
+      "loss": 1.1266,
+      "step": 5440
+    },
+    {
+      "epoch": 0.19787613194166637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8558000210435182e-05,
+      "loss": 1.0998,
+      "step": 5441
+    },
+    {
+      "epoch": 0.19791249954540496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8557390835220282e-05,
+      "loss": 1.1455,
+      "step": 5442
+    },
+    {
+      "epoch": 0.19794886714914353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8556781341283733e-05,
+      "loss": 1.098,
+      "step": 5443
+    },
+    {
+      "epoch": 0.19798523475288213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855617172863398e-05,
+      "loss": 1.1938,
+      "step": 5444
+    },
+    {
+      "epoch": 0.19802160235662072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8555561997279484e-05,
+      "loss": 1.1169,
+      "step": 5445
+    },
+    {
+      "epoch": 0.19805796996035932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8554952147228705e-05,
+      "loss": 1.0421,
+      "step": 5446
+    },
+    {
+      "epoch": 0.1980943375640979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8554342178490104e-05,
+      "loss": 1.1658,
+      "step": 5447
+    },
+    {
+      "epoch": 0.19813070516783649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855373209107214e-05,
+      "loss": 1.0638,
+      "step": 5448
+    },
+    {
+      "epoch": 0.19816707277157508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8553121884983284e-05,
+      "loss": 1.0264,
+      "step": 5449
+    },
+    {
+      "epoch": 0.19820344037531368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855251156023199e-05,
+      "loss": 1.1374,
+      "step": 5450
+    },
+    {
+      "epoch": 0.19823980797905227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855190111682674e-05,
+      "loss": 1.2054,
+      "step": 5451
+    },
+    {
+      "epoch": 0.19827617558279084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8551290554775988e-05,
+      "loss": 1.1921,
+      "step": 5452
+    },
+    {
+      "epoch": 0.19831254318652944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8550679874088217e-05,
+      "loss": 1.1855,
+      "step": 5453
+    },
+    {
+      "epoch": 0.19834891079026803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8550069074771893e-05,
+      "loss": 1.0933,
+      "step": 5454
+    },
+    {
+      "epoch": 0.19838527839400663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854945815683549e-05,
+      "loss": 1.259,
+      "step": 5455
+    },
+    {
+      "epoch": 0.1984216459977452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854884712028749e-05,
+      "loss": 1.1418,
+      "step": 5456
+    },
+    {
+      "epoch": 0.1984580136014838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854823596513636e-05,
+      "loss": 1.2034,
+      "step": 5457
+    },
+    {
+      "epoch": 0.1984943812052224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854762469139059e-05,
+      "loss": 1.1639,
+      "step": 5458
+    },
+    {
+      "epoch": 0.198530748808961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8547013299058652e-05,
+      "loss": 1.1486,
+      "step": 5459
+    },
+    {
+      "epoch": 0.19856711641269956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854640178814903e-05,
+      "loss": 1.1034,
+      "step": 5460
+    },
+    {
+      "epoch": 0.19860348401643815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854579015867021e-05,
+      "loss": 1.2169,
+      "step": 5461
+    },
+    {
+      "epoch": 0.19863985162017675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8545178410630676e-05,
+      "loss": 1.0337,
+      "step": 5462
+    },
+    {
+      "epoch": 0.19867621922391535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8544566544038917e-05,
+      "loss": 1.1592,
+      "step": 5463
+    },
+    {
+      "epoch": 0.19871258682765391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854395455890342e-05,
+      "loss": 1.1367,
+      "step": 5464
+    },
+    {
+      "epoch": 0.1987489544313925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8543342455232674e-05,
+      "loss": 1.2001,
+      "step": 5465
+    },
+    {
+      "epoch": 0.1987853220351311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8542730233035176e-05,
+      "loss": 1.22,
+      "step": 5466
+    },
+    {
+      "epoch": 0.1988216896388697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854211789231941e-05,
+      "loss": 1.2047,
+      "step": 5467
+    },
+    {
+      "epoch": 0.19885805724260827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8541505433093884e-05,
+      "loss": 1.0125,
+      "step": 5468
+    },
+    {
+      "epoch": 0.19889442484634687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8540892855367086e-05,
+      "loss": 1.0357,
+      "step": 5469
+    },
+    {
+      "epoch": 0.19893079245008546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854028015914752e-05,
+      "loss": 1.2987,
+      "step": 5470
+    },
+    {
+      "epoch": 0.19896716005382406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8539667344443682e-05,
+      "loss": 1.2107,
+      "step": 5471
+    },
+    {
+      "epoch": 0.19900352765756266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8539054411264075e-05,
+      "loss": 1.0462,
+      "step": 5472
+    },
+    {
+      "epoch": 0.19903989526130123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8538441359617207e-05,
+      "loss": 1.0346,
+      "step": 5473
+    },
+    {
+      "epoch": 0.19907626286503982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853782818951158e-05,
+      "loss": 1.0166,
+      "step": 5474
+    },
+    {
+      "epoch": 0.19911263046877842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.85372149009557e-05,
+      "loss": 1.1713,
+      "step": 5475
+    },
+    {
+      "epoch": 0.19914899807251701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8536601493958072e-05,
+      "loss": 1.2401,
+      "step": 5476
+    },
+    {
+      "epoch": 0.19918536567625558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853598796852721e-05,
+      "loss": 1.0355,
+      "step": 5477
+    },
+    {
+      "epoch": 0.19922173327999418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853537432467163e-05,
+      "loss": 1.0723,
+      "step": 5478
+    },
+    {
+      "epoch": 0.19925810088373277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853476056239984e-05,
+      "loss": 1.0406,
+      "step": 5479
+    },
+    {
+      "epoch": 0.19929446848747137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8534146681720354e-05,
+      "loss": 1.1287,
+      "step": 5480
+    },
+    {
+      "epoch": 0.19933083609120994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8533532682641693e-05,
+      "loss": 1.0438,
+      "step": 5481
+    },
+    {
+      "epoch": 0.19936720369494854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8532918565172372e-05,
+      "loss": 1.1444,
+      "step": 5482
+    },
+    {
+      "epoch": 0.19940357129868713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8532304329320917e-05,
+      "loss": 1.1105,
+      "step": 5483
+    },
+    {
+      "epoch": 0.19943993890242573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8531689975095842e-05,
+      "loss": 1.1036,
+      "step": 5484
+    },
+    {
+      "epoch": 0.1994763065061643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853107550250567e-05,
+      "loss": 1.217,
+      "step": 5485
+    },
+    {
+      "epoch": 0.1995126741099029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8530460911558937e-05,
+      "loss": 1.1215,
+      "step": 5486
+    },
+    {
+      "epoch": 0.1995490417136415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8529846202264155e-05,
+      "loss": 1.2986,
+      "step": 5487
+    },
+    {
+      "epoch": 0.19958540931738009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8529231374629862e-05,
+      "loss": 0.9511,
+      "step": 5488
+    },
+    {
+      "epoch": 0.19962177692111865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852861642866459e-05,
+      "loss": 1.0707,
+      "step": 5489
+    },
+    {
+      "epoch": 0.19965814452485725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852800136437686e-05,
+      "loss": 1.137,
+      "step": 5490
+    },
+    {
+      "epoch": 0.19969451212859585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852738618177521e-05,
+      "loss": 1.0848,
+      "step": 5491
+    },
+    {
+      "epoch": 0.19973087973233444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8526770880868178e-05,
+      "loss": 1.1469,
+      "step": 5492
+    },
+    {
+      "epoch": 0.19976724733607304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8526155461664296e-05,
+      "loss": 1.1128,
+      "step": 5493
+    },
+    {
+      "epoch": 0.1998036149398116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8525539924172106e-05,
+      "loss": 1.1122,
+      "step": 5494
+    },
+    {
+      "epoch": 0.1998399825435502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8524924268400145e-05,
+      "loss": 1.1074,
+      "step": 5495
+    },
+    {
+      "epoch": 0.1998763501472888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8524308494356953e-05,
+      "loss": 1.1342,
+      "step": 5496
+    },
+    {
+      "epoch": 0.1999127177510274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8523692602051078e-05,
+      "loss": 1.1249,
+      "step": 5497
+    },
+    {
+      "epoch": 0.19994908535476597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8523076591491062e-05,
+      "loss": 1.1198,
+      "step": 5498
+    },
+    {
+      "epoch": 0.19998545295850456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8522460462685447e-05,
+      "loss": 1.162,
+      "step": 5499
+    },
+    {
+      "epoch": 0.20002182056224316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852184421564279e-05,
+      "loss": 1.2435,
+      "step": 5500
+    },
+    {
+      "epoch": 0.20005818816598175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852122785037163e-05,
+      "loss": 1.2007,
+      "step": 5501
+    },
+    {
+      "epoch": 0.20009455576972032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.852061136688053e-05,
+      "loss": 1.1292,
+      "step": 5502
+    },
+    {
+      "epoch": 0.20013092337345892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851999476517803e-05,
+      "loss": 1.1618,
+      "step": 5503
+    },
+    {
+      "epoch": 0.20016729097719752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8519378045272692e-05,
+      "loss": 1.0608,
+      "step": 5504
+    },
+    {
+      "epoch": 0.2002036585809361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8518761207173074e-05,
+      "loss": 1.0998,
+      "step": 5505
+    },
+    {
+      "epoch": 0.20024002618467468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8518144250887732e-05,
+      "loss": 1.0885,
+      "step": 5506
+    },
+    {
+      "epoch": 0.20027639378841328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8517527176425224e-05,
+      "loss": 1.1309,
+      "step": 5507
+    },
+    {
+      "epoch": 0.20031276139215187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8516909983794108e-05,
+      "loss": 1.0032,
+      "step": 5508
+    },
+    {
+      "epoch": 0.20034912899589047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851629267300295e-05,
+      "loss": 1.0387,
+      "step": 5509
+    },
+    {
+      "epoch": 0.20038549659962904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8515675244060314e-05,
+      "loss": 1.082,
+      "step": 5510
+    },
+    {
+      "epoch": 0.20042186420336763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851505769697477e-05,
+      "loss": 1.1867,
+      "step": 5511
+    },
+    {
+      "epoch": 0.20045823180710623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8514440031754875e-05,
+      "loss": 1.133,
+      "step": 5512
+    },
+    {
+      "epoch": 0.20049459941084483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8513822248409213e-05,
+      "loss": 1.1489,
+      "step": 5513
+    },
+    {
+      "epoch": 0.20053096701458342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851320434694634e-05,
+      "loss": 1.0525,
+      "step": 5514
+    },
+    {
+      "epoch": 0.200567334618322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851258632737484e-05,
+      "loss": 1.0739,
+      "step": 5515
+    },
+    {
+      "epoch": 0.2006037022220606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851196818970328e-05,
+      "loss": 1.0891,
+      "step": 5516
+    },
+    {
+      "epoch": 0.20064006982579918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.851134993394024e-05,
+      "loss": 1.1608,
+      "step": 5517
+    },
+    {
+      "epoch": 0.20067643742953778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8510731560094296e-05,
+      "loss": 1.1436,
+      "step": 5518
+    },
+    {
+      "epoch": 0.20071280503327635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8510113068174026e-05,
+      "loss": 1.093,
+      "step": 5519
+    },
+    {
+      "epoch": 0.20074917263701494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8509494458188005e-05,
+      "loss": 1.149,
+      "step": 5520
+    },
+    {
+      "epoch": 0.20078554024075354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850887573014483e-05,
+      "loss": 1.0566,
+      "step": 5521
+    },
+    {
+      "epoch": 0.20082190784449214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8508256884053074e-05,
+      "loss": 1.09,
+      "step": 5522
+    },
+    {
+      "epoch": 0.2008582754482307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8507637919921325e-05,
+      "loss": 1.1687,
+      "step": 5523
+    },
+    {
+      "epoch": 0.2008946430519693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8507018837758174e-05,
+      "loss": 1.0403,
+      "step": 5524
+    },
+    {
+      "epoch": 0.2009310106557079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8506399637572203e-05,
+      "loss": 1.0727,
+      "step": 5525
+    },
+    {
+      "epoch": 0.2009673782594465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8505780319372008e-05,
+      "loss": 1.0288,
+      "step": 5526
+    },
+    {
+      "epoch": 0.20100374586318506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8505160883166178e-05,
+      "loss": 1.1905,
+      "step": 5527
+    },
+    {
+      "epoch": 0.20104011346692366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8504541328963308e-05,
+      "loss": 1.201,
+      "step": 5528
+    },
+    {
+      "epoch": 0.20107648107066226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8503921656771995e-05,
+      "loss": 1.1727,
+      "step": 5529
+    },
+    {
+      "epoch": 0.20111284867440085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850330186660083e-05,
+      "loss": 1.1836,
+      "step": 5530
+    },
+    {
+      "epoch": 0.20114921627813942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850268195845842e-05,
+      "loss": 1.1371,
+      "step": 5531
+    },
+    {
+      "epoch": 0.20118558388187802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850206193235336e-05,
+      "loss": 1.1362,
+      "step": 5532
+    },
+    {
+      "epoch": 0.2012219514856166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8501441788294256e-05,
+      "loss": 1.2166,
+      "step": 5533
+    },
+    {
+      "epoch": 0.2012583190893552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.850082152628971e-05,
+      "loss": 1.0461,
+      "step": 5534
+    },
+    {
+      "epoch": 0.2012946866930938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8500201146348322e-05,
+      "loss": 1.1344,
+      "step": 5535
+    },
+    {
+      "epoch": 0.20133105429683237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8499580648478704e-05,
+      "loss": 1.0011,
+      "step": 5536
+    },
+    {
+      "epoch": 0.20136742190057097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8498960032689465e-05,
+      "loss": 1.0449,
+      "step": 5537
+    },
+    {
+      "epoch": 0.20140378950430957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8498339298989216e-05,
+      "loss": 1.0832,
+      "step": 5538
+    },
+    {
+      "epoch": 0.20144015710804816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8497718447386563e-05,
+      "loss": 1.1389,
+      "step": 5539
+    },
+    {
+      "epoch": 0.20147652471178673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8497097477890125e-05,
+      "loss": 1.1078,
+      "step": 5540
+    },
+    {
+      "epoch": 0.20151289231552533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8496476390508517e-05,
+      "loss": 1.2285,
+      "step": 5541
+    },
+    {
+      "epoch": 0.20154925991926392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8495855185250355e-05,
+      "loss": 1.0693,
+      "step": 5542
+    },
+    {
+      "epoch": 0.20158562752300252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8495233862124248e-05,
+      "loss": 1.2805,
+      "step": 5543
+    },
+    {
+      "epoch": 0.2016219951267411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8494612421138832e-05,
+      "loss": 1.1583,
+      "step": 5544
+    },
+    {
+      "epoch": 0.20165836273047968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.849399086230272e-05,
+      "loss": 1.1124,
+      "step": 5545
+    },
+    {
+      "epoch": 0.20169473033421828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8493369185624536e-05,
+      "loss": 1.142,
+      "step": 5546
+    },
+    {
+      "epoch": 0.20173109793795688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8492747391112905e-05,
+      "loss": 1.1982,
+      "step": 5547
+    },
+    {
+      "epoch": 0.20176746554169545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.849212547877645e-05,
+      "loss": 1.0953,
+      "step": 5548
+    },
+    {
+      "epoch": 0.20180383314543404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8491503448623806e-05,
+      "loss": 1.1701,
+      "step": 5549
+    },
+    {
+      "epoch": 0.20184020074917264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8490881300663602e-05,
+      "loss": 1.1225,
+      "step": 5550
+    },
+    {
+      "epoch": 0.20187656835291123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8490259034904462e-05,
+      "loss": 1.0729,
+      "step": 5551
+    },
+    {
+      "epoch": 0.2019129359566498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8489636651355026e-05,
+      "loss": 1.0682,
+      "step": 5552
+    },
+    {
+      "epoch": 0.2019493035603884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8489014150023928e-05,
+      "loss": 0.9314,
+      "step": 5553
+    },
+    {
+      "epoch": 0.201985671164127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.84883915309198e-05,
+      "loss": 1.1323,
+      "step": 5554
+    },
+    {
+      "epoch": 0.2020220387678656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8487768794051286e-05,
+      "loss": 1.0116,
+      "step": 5555
+    },
+    {
+      "epoch": 0.2020584063716042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848714593942702e-05,
+      "loss": 1.2571,
+      "step": 5556
+    },
+    {
+      "epoch": 0.20209477397534276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8486522967055646e-05,
+      "loss": 1.1535,
+      "step": 5557
+    },
+    {
+      "epoch": 0.20213114157908135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848589987694581e-05,
+      "loss": 1.0583,
+      "step": 5558
+    },
+    {
+      "epoch": 0.20216750918281995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848527666910615e-05,
+      "loss": 0.9909,
+      "step": 5559
+    },
+    {
+      "epoch": 0.20220387678655855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8484653343545312e-05,
+      "loss": 1.1471,
+      "step": 5560
+    },
+    {
+      "epoch": 0.2022402443902971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8484029900271953e-05,
+      "loss": 1.195,
+      "step": 5561
+    },
+    {
+      "epoch": 0.2022766119940357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8483406339294717e-05,
+      "loss": 1.1165,
+      "step": 5562
+    },
+    {
+      "epoch": 0.2023129795977743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.848278266062225e-05,
+      "loss": 0.95,
+      "step": 5563
+    },
+    {
+      "epoch": 0.2023493472015129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8482158864263213e-05,
+      "loss": 1.2432,
+      "step": 5564
+    },
+    {
+      "epoch": 0.20238571480525147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8481534950226252e-05,
+      "loss": 1.1507,
+      "step": 5565
+    },
+    {
+      "epoch": 0.20242208240899007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8480910918520033e-05,
+      "loss": 1.1967,
+      "step": 5566
+    },
+    {
+      "epoch": 0.20245845001272866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8480286769153203e-05,
+      "loss": 1.1477,
+      "step": 5567
+    },
+    {
+      "epoch": 0.20249481761646726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8479662502134424e-05,
+      "loss": 1.0499,
+      "step": 5568
+    },
+    {
+      "epoch": 0.20253118522020583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8479038117472365e-05,
+      "loss": 1.0771,
+      "step": 5569
+    },
+    {
+      "epoch": 0.20256755282394442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8478413615175677e-05,
+      "loss": 1.1628,
+      "step": 5570
+    },
+    {
+      "epoch": 0.20260392042768302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8477788995253035e-05,
+      "loss": 1.0768,
+      "step": 5571
+    },
+    {
+      "epoch": 0.20264028803142162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8477164257713096e-05,
+      "loss": 1.2113,
+      "step": 5572
+    },
+    {
+      "epoch": 0.20267665563516019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.847653940256453e-05,
+      "loss": 1.1345,
+      "step": 5573
+    },
+    {
+      "epoch": 0.20271302323889878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8475914429816005e-05,
+      "loss": 1.0631,
+      "step": 5574
+    },
+    {
+      "epoch": 0.20274939084263738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8475289339476198e-05,
+      "loss": 1.0954,
+      "step": 5575
+    },
+    {
+      "epoch": 0.20278575844637597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.847466413155377e-05,
+      "loss": 1.0894,
+      "step": 5576
+    },
+    {
+      "epoch": 0.20282212605011457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8474038806057407e-05,
+      "loss": 1.1704,
+      "step": 5577
+    },
+    {
+      "epoch": 0.20285849365385314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8473413362995777e-05,
+      "loss": 1.0927,
+      "step": 5578
+    },
+    {
+      "epoch": 0.20289486125759174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8472787802377555e-05,
+      "loss": 1.1507,
+      "step": 5579
+    },
+    {
+      "epoch": 0.20293122886133033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8472162124211427e-05,
+      "loss": 1.2408,
+      "step": 5580
+    },
+    {
+      "epoch": 0.20296759646506893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.847153632850607e-05,
+      "loss": 1.1834,
+      "step": 5581
+    },
+    {
+      "epoch": 0.2030039640688075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8470910415270164e-05,
+      "loss": 1.164,
+      "step": 5582
+    },
+    {
+      "epoch": 0.2030403316725461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8470284384512397e-05,
+      "loss": 0.9454,
+      "step": 5583
+    },
+    {
+      "epoch": 0.2030766992762847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846965823624145e-05,
+      "loss": 1.0842,
+      "step": 5584
+    },
+    {
+      "epoch": 0.20311306688002329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8469031970466014e-05,
+      "loss": 1.1455,
+      "step": 5585
+    },
+    {
+      "epoch": 0.20314943448376185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8468405587194774e-05,
+      "loss": 1.1165,
+      "step": 5586
+    },
+    {
+      "epoch": 0.20318580208750045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846777908643642e-05,
+      "loss": 1.1073,
+      "step": 5587
+    },
+    {
+      "epoch": 0.20322216969123905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8467152468199647e-05,
+      "loss": 1.1454,
+      "step": 5588
+    },
+    {
+      "epoch": 0.20325853729497764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846652573249315e-05,
+      "loss": 1.2477,
+      "step": 5589
+    },
+    {
+      "epoch": 0.2032949048987162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8465898879325618e-05,
+      "loss": 1.2153,
+      "step": 5590
+    },
+    {
+      "epoch": 0.2033312725024548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846527190870575e-05,
+      "loss": 1.1414,
+      "step": 5591
+    },
+    {
+      "epoch": 0.2033676401061934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8464644820642243e-05,
+      "loss": 1.1126,
+      "step": 5592
+    },
+    {
+      "epoch": 0.203404007709932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8464017615143805e-05,
+      "loss": 1.2361,
+      "step": 5593
+    },
+    {
+      "epoch": 0.20344037531367057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8463390292219126e-05,
+      "loss": 1.1257,
+      "step": 5594
+    },
+    {
+      "epoch": 0.20347674291740916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8462762851876913e-05,
+      "loss": 0.8981,
+      "step": 5595
+    },
+    {
+      "epoch": 0.20351311052114776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846213529412588e-05,
+      "loss": 1.0283,
+      "step": 5596
+    },
+    {
+      "epoch": 0.20354947812488636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8461507618974722e-05,
+      "loss": 1.1192,
+      "step": 5597
+    },
+    {
+      "epoch": 0.20358584572862495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8460879826432148e-05,
+      "loss": 1.1602,
+      "step": 5598
+    },
+    {
+      "epoch": 0.20362221333236352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8460251916506874e-05,
+      "loss": 1.1374,
+      "step": 5599
+    },
+    {
+      "epoch": 0.20365858093610212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8459623889207607e-05,
+      "loss": 1.2007,
+      "step": 5600
+    },
+    {
+      "epoch": 0.20369494853984071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8458995744543065e-05,
+      "loss": 1.1846,
+      "step": 5601
+    },
+    {
+      "epoch": 0.2037313161435793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8458367482521953e-05,
+      "loss": 1.1401,
+      "step": 5602
+    },
+    {
+      "epoch": 0.20376768374731788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8457739103152998e-05,
+      "loss": 1.0346,
+      "step": 5603
+    },
+    {
+      "epoch": 0.20380405135105648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.845711060644491e-05,
+      "loss": 1.1217,
+      "step": 5604
+    },
+    {
+      "epoch": 0.20384041895479507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8456481992406413e-05,
+      "loss": 1.1109,
+      "step": 5605
+    },
+    {
+      "epoch": 0.20387678655853367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8455853261046225e-05,
+      "loss": 1.119,
+      "step": 5606
+    },
+    {
+      "epoch": 0.20391315416227224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8455224412373072e-05,
+      "loss": 1.1576,
+      "step": 5607
+    },
+    {
+      "epoch": 0.20394952176601083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8454595446395675e-05,
+      "loss": 1.2258,
+      "step": 5608
+    },
+    {
+      "epoch": 0.20398588936974943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8453966363122764e-05,
+      "loss": 1.1453,
+      "step": 5609
+    },
+    {
+      "epoch": 0.20402225697348803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.845333716256306e-05,
+      "loss": 1.1607,
+      "step": 5610
+    },
+    {
+      "epoch": 0.2040586245772266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.84527078447253e-05,
+      "loss": 1.1492,
+      "step": 5611
+    },
+    {
+      "epoch": 0.2040949921809652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8452078409618214e-05,
+      "loss": 1.149,
+      "step": 5612
+    },
+    {
+      "epoch": 0.2041313597847038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.845144885725053e-05,
+      "loss": 1.0619,
+      "step": 5613
+    },
+    {
+      "epoch": 0.20416772738844238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8450819187630983e-05,
+      "loss": 1.1865,
+      "step": 5614
+    },
+    {
+      "epoch": 0.20420409499218098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8450189400768308e-05,
+      "loss": 1.0581,
+      "step": 5615
+    },
+    {
+      "epoch": 0.20424046259591955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8449559496671245e-05,
+      "loss": 1.2326,
+      "step": 5616
+    },
+    {
+      "epoch": 0.20427683019965814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8448929475348535e-05,
+      "loss": 1.1229,
+      "step": 5617
+    },
+    {
+      "epoch": 0.20431319780339674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8448299336808914e-05,
+      "loss": 1.0869,
+      "step": 5618
+    },
+    {
+      "epoch": 0.20434956540713534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8447669081061126e-05,
+      "loss": 1.0858,
+      "step": 5619
+    },
+    {
+      "epoch": 0.2043859330108739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8447038708113915e-05,
+      "loss": 1.1466,
+      "step": 5620
+    },
+    {
+      "epoch": 0.2044223006146125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844640821797603e-05,
+      "loss": 1.1411,
+      "step": 5621
+    },
+    {
+      "epoch": 0.2044586682183511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844577761065621e-05,
+      "loss": 1.0743,
+      "step": 5622
+    },
+    {
+      "epoch": 0.2044950358220897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8445146886163213e-05,
+      "loss": 1.1742,
+      "step": 5623
+    },
+    {
+      "epoch": 0.20453140342582826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8444516044505784e-05,
+      "loss": 0.9768,
+      "step": 5624
+    },
+    {
+      "epoch": 0.20456777102956686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8443885085692674e-05,
+      "loss": 1.1612,
+      "step": 5625
+    },
+    {
+      "epoch": 0.20460413863330545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844325400973264e-05,
+      "loss": 1.1557,
+      "step": 5626
+    },
+    {
+      "epoch": 0.20464050623704405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8442622816634434e-05,
+      "loss": 1.0507,
+      "step": 5627
+    },
+    {
+      "epoch": 0.20467687384078262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844199150640682e-05,
+      "loss": 0.9476,
+      "step": 5628
+    },
+    {
+      "epoch": 0.20471324144452122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844136007905855e-05,
+      "loss": 1.2479,
+      "step": 5629
+    },
+    {
+      "epoch": 0.2047496090482598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.844072853459838e-05,
+      "loss": 1.0664,
+      "step": 5630
+    },
+    {
+      "epoch": 0.2047859766519984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8440096873035082e-05,
+      "loss": 1.1104,
+      "step": 5631
+    },
+    {
+      "epoch": 0.20482234425573698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8439465094377417e-05,
+      "loss": 1.1589,
+      "step": 5632
+    },
+    {
+      "epoch": 0.20485871185947557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8438833198634146e-05,
+      "loss": 1.2192,
+      "step": 5633
+    },
+    {
+      "epoch": 0.20489507946321417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8438201185814036e-05,
+      "loss": 1.1119,
+      "step": 5634
+    },
+    {
+      "epoch": 0.20493144706695277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8437569055925855e-05,
+      "loss": 1.2167,
+      "step": 5635
+    },
+    {
+      "epoch": 0.20496781467069136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8436936808978377e-05,
+      "loss": 1.1258,
+      "step": 5636
+    },
+    {
+      "epoch": 0.20500418227442993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8436304444980372e-05,
+      "loss": 0.8944,
+      "step": 5637
+    },
+    {
+      "epoch": 0.20504054987816853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8435671963940612e-05,
+      "loss": 1.157,
+      "step": 5638
+    },
+    {
+      "epoch": 0.20507691748190712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.843503936586787e-05,
+      "loss": 1.0986,
+      "step": 5639
+    },
+    {
+      "epoch": 0.20511328508564572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8434406650770928e-05,
+      "loss": 1.0925,
+      "step": 5640
+    },
+    {
+      "epoch": 0.2051496526893843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.843377381865856e-05,
+      "loss": 1.0905,
+      "step": 5641
+    },
+    {
+      "epoch": 0.20518602029312288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8433140869539545e-05,
+      "loss": 0.9214,
+      "step": 5642
+    },
+    {
+      "epoch": 0.20522238789686148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8432507803422666e-05,
+      "loss": 1.0411,
+      "step": 5643
+    },
+    {
+      "epoch": 0.20525875550060008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8431874620316706e-05,
+      "loss": 1.1194,
+      "step": 5644
+    },
+    {
+      "epoch": 0.20529512310433864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8431241320230448e-05,
+      "loss": 1.0893,
+      "step": 5645
+    },
+    {
+      "epoch": 0.20533149070807724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8430607903172677e-05,
+      "loss": 1.0708,
+      "step": 5646
+    },
+    {
+      "epoch": 0.20536785831181584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8429974369152185e-05,
+      "loss": 1.0375,
+      "step": 5647
+    },
+    {
+      "epoch": 0.20540422591555443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8429340718177758e-05,
+      "loss": 1.1708,
+      "step": 5648
+    },
+    {
+      "epoch": 0.205440593519293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842870695025819e-05,
+      "loss": 1.1578,
+      "step": 5649
+    },
+    {
+      "epoch": 0.2054769611230316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8428073065402274e-05,
+      "loss": 1.1545,
+      "step": 5650
+    },
+    {
+      "epoch": 0.2055133287267702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.84274390636188e-05,
+      "loss": 1.0709,
+      "step": 5651
+    },
+    {
+      "epoch": 0.2055496963305088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8426804944916565e-05,
+      "loss": 1.1964,
+      "step": 5652
+    },
+    {
+      "epoch": 0.20558606393424736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8426170709304365e-05,
+      "loss": 1.1315,
+      "step": 5653
+    },
+    {
+      "epoch": 0.20562243153798596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8425536356791006e-05,
+      "loss": 1.0497,
+      "step": 5654
+    },
+    {
+      "epoch": 0.20565879914172455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842490188738528e-05,
+      "loss": 1.2019,
+      "step": 5655
+    },
+    {
+      "epoch": 0.20569516674546315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8424267301095997e-05,
+      "loss": 1.1651,
+      "step": 5656
+    },
+    {
+      "epoch": 0.20573153434920174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8423632597931957e-05,
+      "loss": 1.0927,
+      "step": 5657
+    },
+    {
+      "epoch": 0.2057679019529403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842299777790197e-05,
+      "loss": 1.1336,
+      "step": 5658
+    },
+    {
+      "epoch": 0.2058042695566789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8422362841014833e-05,
+      "loss": 1.1683,
+      "step": 5659
+    },
+    {
+      "epoch": 0.2058406371604175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8421727787279364e-05,
+      "loss": 1.1823,
+      "step": 5660
+    },
+    {
+      "epoch": 0.2058770047641561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8421092616704374e-05,
+      "loss": 1.0795,
+      "step": 5661
+    },
+    {
+      "epoch": 0.20591337236789467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8420457329298668e-05,
+      "loss": 1.1152,
+      "step": 5662
+    },
+    {
+      "epoch": 0.20594973997163327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8419821925071062e-05,
+      "loss": 1.0871,
+      "step": 5663
+    },
+    {
+      "epoch": 0.20598610757537186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8419186404030375e-05,
+      "loss": 1.1524,
+      "step": 5664
+    },
+    {
+      "epoch": 0.20602247517911046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8418550766185424e-05,
+      "loss": 1.106,
+      "step": 5665
+    },
+    {
+      "epoch": 0.20605884278284903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841791501154502e-05,
+      "loss": 1.139,
+      "step": 5666
+    },
+    {
+      "epoch": 0.20609521038658762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8417279140117993e-05,
+      "loss": 1.1014,
+      "step": 5667
+    },
+    {
+      "epoch": 0.20613157799032622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8416643151913164e-05,
+      "loss": 1.1326,
+      "step": 5668
+    },
+    {
+      "epoch": 0.20616794559406482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8416007046939348e-05,
+      "loss": 1.1796,
+      "step": 5669
+    },
+    {
+      "epoch": 0.20620431319780338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8415370825205374e-05,
+      "loss": 1.2063,
+      "step": 5670
+    },
+    {
+      "epoch": 0.20624068080154198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8414734486720075e-05,
+      "loss": 1.157,
+      "step": 5671
+    },
+    {
+      "epoch": 0.20627704840528058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841409803149227e-05,
+      "loss": 1.1317,
+      "step": 5672
+    },
+    {
+      "epoch": 0.20631341600901917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8413461459530792e-05,
+      "loss": 1.1899,
+      "step": 5673
+    },
+    {
+      "epoch": 0.20634978361275774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841282477084447e-05,
+      "loss": 1.0158,
+      "step": 5674
+    },
+    {
+      "epoch": 0.20638615121649634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841218796544215e-05,
+      "loss": 1.0164,
+      "step": 5675
+    },
+    {
+      "epoch": 0.20642251882023493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841155104333265e-05,
+      "loss": 1.0465,
+      "step": 5676
+    },
+    {
+      "epoch": 0.20645888642397353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8410914004524818e-05,
+      "loss": 1.1618,
+      "step": 5677
+    },
+    {
+      "epoch": 0.20649525402771213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8410276849027486e-05,
+      "loss": 1.1378,
+      "step": 5678
+    },
+    {
+      "epoch": 0.2065316216314507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8409639576849496e-05,
+      "loss": 1.2141,
+      "step": 5679
+    },
+    {
+      "epoch": 0.2065679892351893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8409002187999686e-05,
+      "loss": 1.0082,
+      "step": 5680
+    },
+    {
+      "epoch": 0.2066043568389279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8408364682486907e-05,
+      "loss": 1.0505,
+      "step": 5681
+    },
+    {
+      "epoch": 0.20664072444266648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840772706031999e-05,
+      "loss": 1.2225,
+      "step": 5682
+    },
+    {
+      "epoch": 0.20667709204640505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8407089321507796e-05,
+      "loss": 1.0931,
+      "step": 5683
+    },
+    {
+      "epoch": 0.20671345965014365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8406451466059164e-05,
+      "loss": 1.1691,
+      "step": 5684
+    },
+    {
+      "epoch": 0.20674982725388225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840581349398294e-05,
+      "loss": 1.0284,
+      "step": 5685
+    },
+    {
+      "epoch": 0.20678619485762084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840517540528799e-05,
+      "loss": 1.0938,
+      "step": 5686
+    },
+    {
+      "epoch": 0.2068225624613594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8404537199983148e-05,
+      "loss": 1.0451,
+      "step": 5687
+    },
+    {
+      "epoch": 0.206858930065098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840389887807728e-05,
+      "loss": 1.027,
+      "step": 5688
+    },
+    {
+      "epoch": 0.2068952976688366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8403260439579238e-05,
+      "loss": 1.0052,
+      "step": 5689
+    },
+    {
+      "epoch": 0.2069316652725752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840262188449788e-05,
+      "loss": 1.2453,
+      "step": 5690
+    },
+    {
+      "epoch": 0.20696803287631377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8401983212842064e-05,
+      "loss": 1.1534,
+      "step": 5691
+    },
+    {
+      "epoch": 0.20700440048005236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8401344424620655e-05,
+      "loss": 1.0814,
+      "step": 5692
+    },
+    {
+      "epoch": 0.20704076808379096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8400705519842512e-05,
+      "loss": 1.2436,
+      "step": 5693
+    },
+    {
+      "epoch": 0.20707713568752956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8400066498516498e-05,
+      "loss": 1.11,
+      "step": 5694
+    },
+    {
+      "epoch": 0.20711350329126813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839942736065148e-05,
+      "loss": 1.1253,
+      "step": 5695
+    },
+    {
+      "epoch": 0.20714987089500672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839878810625632e-05,
+      "loss": 1.1565,
+      "step": 5696
+    },
+    {
+      "epoch": 0.20718623849874532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8398148735339894e-05,
+      "loss": 1.1444,
+      "step": 5697
+    },
+    {
+      "epoch": 0.20722260610248391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8397509247911073e-05,
+      "loss": 1.1088,
+      "step": 5698
+    },
+    {
+      "epoch": 0.2072589737062225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8396869643978725e-05,
+      "loss": 1.1535,
+      "step": 5699
+    },
+    {
+      "epoch": 0.20729534130996108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839622992355172e-05,
+      "loss": 1.0789,
+      "step": 5700
+    },
+    {
+      "epoch": 0.20733170891369967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8395590086638938e-05,
+      "loss": 1.0524,
+      "step": 5701
+    },
+    {
+      "epoch": 0.20736807651743827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839495013324926e-05,
+      "loss": 1.0761,
+      "step": 5702
+    },
+    {
+      "epoch": 0.20740444412117687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8394310063391555e-05,
+      "loss": 1.1209,
+      "step": 5703
+    },
+    {
+      "epoch": 0.20744081172491544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839366987707471e-05,
+      "loss": 1.0935,
+      "step": 5704
+    },
+    {
+      "epoch": 0.20747717932865403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8393029574307607e-05,
+      "loss": 1.0765,
+      "step": 5705
+    },
+    {
+      "epoch": 0.20751354693239263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8392389155099122e-05,
+      "loss": 1.1681,
+      "step": 5706
+    },
+    {
+      "epoch": 0.20754991453613122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839174861945815e-05,
+      "loss": 1.131,
+      "step": 5707
+    },
+    {
+      "epoch": 0.2075862821398698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8391107967393568e-05,
+      "loss": 1.002,
+      "step": 5708
+    },
+    {
+      "epoch": 0.2076226497436084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.839046719891427e-05,
+      "loss": 1.1396,
+      "step": 5709
+    },
+    {
+      "epoch": 0.20765901734734699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8389826314029143e-05,
+      "loss": 1.0357,
+      "step": 5710
+    },
+    {
+      "epoch": 0.20769538495108558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.838918531274708e-05,
+      "loss": 1.0963,
+      "step": 5711
+    },
+    {
+      "epoch": 0.20773175255482415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8388544195076972e-05,
+      "loss": 1.2126,
+      "step": 5712
+    },
+    {
+      "epoch": 0.20776812015856275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8387902961027718e-05,
+      "loss": 1.2151,
+      "step": 5713
+    },
+    {
+      "epoch": 0.20780448776230134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.838726161060821e-05,
+      "loss": 1.1025,
+      "step": 5714
+    },
+    {
+      "epoch": 0.20784085536603994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8386620143827346e-05,
+      "loss": 1.1716,
+      "step": 5715
+    },
+    {
+      "epoch": 0.2078772229697785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8385978560694028e-05,
+      "loss": 1.1224,
+      "step": 5716
+    },
+    {
+      "epoch": 0.2079135905735171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8385336861217154e-05,
+      "loss": 1.176,
+      "step": 5717
+    },
+    {
+      "epoch": 0.2079499581772557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.838469504540563e-05,
+      "loss": 1.1028,
+      "step": 5718
+    },
+    {
+      "epoch": 0.2079863257809943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.838405311326836e-05,
+      "loss": 1.093,
+      "step": 5719
+    },
+    {
+      "epoch": 0.2080226933847329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8383411064814244e-05,
+      "loss": 0.9783,
+      "step": 5720
+    },
+    {
+      "epoch": 0.20805906098847146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.83827689000522e-05,
+      "loss": 1.1121,
+      "step": 5721
+    },
+    {
+      "epoch": 0.20809542859221006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8382126618991125e-05,
+      "loss": 1.0049,
+      "step": 5722
+    },
+    {
+      "epoch": 0.20813179619594865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.838148422163994e-05,
+      "loss": 0.9775,
+      "step": 5723
+    },
+    {
+      "epoch": 0.20816816379968725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.838084170800755e-05,
+      "loss": 1.0752,
+      "step": 5724
+    },
+    {
+      "epoch": 0.20820453140342582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8380199078102876e-05,
+      "loss": 1.0379,
+      "step": 5725
+    },
+    {
+      "epoch": 0.20824089900716442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8379556331934826e-05,
+      "loss": 1.1449,
+      "step": 5726
+    },
+    {
+      "epoch": 0.208277266610903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8378913469512324e-05,
+      "loss": 1.2156,
+      "step": 5727
+    },
+    {
+      "epoch": 0.2083136342146416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8378270490844285e-05,
+      "loss": 1.1226,
+      "step": 5728
+    },
+    {
+      "epoch": 0.20835000181838018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8377627395939628e-05,
+      "loss": 1.1116,
+      "step": 5729
+    },
+    {
+      "epoch": 0.20838636942211877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.837698418480728e-05,
+      "loss": 1.0714,
+      "step": 5730
+    },
+    {
+      "epoch": 0.20842273702585737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.837634085745616e-05,
+      "loss": 1.0646,
+      "step": 5731
+    },
+    {
+      "epoch": 0.20845910462959596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8375697413895195e-05,
+      "loss": 1.1788,
+      "step": 5732
+    },
+    {
+      "epoch": 0.20849547223333453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8375053854133314e-05,
+      "loss": 1.0933,
+      "step": 5733
+    },
+    {
+      "epoch": 0.20853183983707313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.837441017817944e-05,
+      "loss": 1.1649,
+      "step": 5734
+    },
+    {
+      "epoch": 0.20856820744081173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.837376638604251e-05,
+      "loss": 1.1562,
+      "step": 5735
+    },
+    {
+      "epoch": 0.20860457504455032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.837312247773145e-05,
+      "loss": 1.1332,
+      "step": 5736
+    },
+    {
+      "epoch": 0.2086409426482889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8372478453255194e-05,
+      "loss": 1.0677,
+      "step": 5737
+    },
+    {
+      "epoch": 0.2086773102520275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8371834312622682e-05,
+      "loss": 1.1291,
+      "step": 5738
+    },
+    {
+      "epoch": 0.20871367785576608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8371190055842846e-05,
+      "loss": 1.2472,
+      "step": 5739
+    },
+    {
+      "epoch": 0.20875004545950468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8370545682924626e-05,
+      "loss": 1.1987,
+      "step": 5740
+    },
+    {
+      "epoch": 0.20878641306324328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.836990119387696e-05,
+      "loss": 1.083,
+      "step": 5741
+    },
+    {
+      "epoch": 0.20882278066698184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.836925658870879e-05,
+      "loss": 1.0517,
+      "step": 5742
+    },
+    {
+      "epoch": 0.20885914827072044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.836861186742906e-05,
+      "loss": 1.1395,
+      "step": 5743
+    },
+    {
+      "epoch": 0.20889551587445904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8367967030046712e-05,
+      "loss": 1.0614,
+      "step": 5744
+    },
+    {
+      "epoch": 0.20893188347819763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8367322076570694e-05,
+      "loss": 1.1541,
+      "step": 5745
+    },
+    {
+      "epoch": 0.2089682510819362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8366677007009955e-05,
+      "loss": 1.0667,
+      "step": 5746
+    },
+    {
+      "epoch": 0.2090046186856748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.836603182137344e-05,
+      "loss": 1.1272,
+      "step": 5747
+    },
+    {
+      "epoch": 0.2090409862894134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.836538651967011e-05,
+      "loss": 1.0557,
+      "step": 5748
+    },
+    {
+      "epoch": 0.209077353893152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8364741101908908e-05,
+      "loss": 1.0861,
+      "step": 5749
+    },
+    {
+      "epoch": 0.20911372149689056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.836409556809879e-05,
+      "loss": 1.1312,
+      "step": 5750
+    },
+    {
+      "epoch": 0.20915008910062916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8363449918248715e-05,
+      "loss": 1.0963,
+      "step": 5751
+    },
+    {
+      "epoch": 0.20918645670436775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8362804152367638e-05,
+      "loss": 1.1309,
+      "step": 5752
+    },
+    {
+      "epoch": 0.20922282430810635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8362158270464515e-05,
+      "loss": 1.1348,
+      "step": 5753
+    },
+    {
+      "epoch": 0.20925919191184492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8361512272548317e-05,
+      "loss": 1.1222,
+      "step": 5754
+    },
+    {
+      "epoch": 0.2092955595155835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8360866158627996e-05,
+      "loss": 1.1163,
+      "step": 5755
+    },
+    {
+      "epoch": 0.2093319271193221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8360219928712517e-05,
+      "loss": 1.062,
+      "step": 5756
+    },
+    {
+      "epoch": 0.2093682947230607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835957358281085e-05,
+      "loss": 1.1231,
+      "step": 5757
+    },
+    {
+      "epoch": 0.20940466232679927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8358927120931963e-05,
+      "loss": 1.1892,
+      "step": 5758
+    },
+    {
+      "epoch": 0.20944102993053787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8358280543084822e-05,
+      "loss": 1.1069,
+      "step": 5759
+    },
+    {
+      "epoch": 0.20947739753427647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8357633849278393e-05,
+      "loss": 1.098,
+      "step": 5760
+    },
+    {
+      "epoch": 0.20951376513801506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8356987039521654e-05,
+      "loss": 1.0552,
+      "step": 5761
+    },
+    {
+      "epoch": 0.20955013274175366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8356340113823578e-05,
+      "loss": 1.1201,
+      "step": 5762
+    },
+    {
+      "epoch": 0.20958650034549223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835569307219314e-05,
+      "loss": 1.0527,
+      "step": 5763
+    },
+    {
+      "epoch": 0.20962286794923082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8355045914639315e-05,
+      "loss": 1.1414,
+      "step": 5764
+    },
+    {
+      "epoch": 0.20965923555296942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835439864117108e-05,
+      "loss": 1.0906,
+      "step": 5765
+    },
+    {
+      "epoch": 0.20969560315670802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835375125179742e-05,
+      "loss": 1.118,
+      "step": 5766
+    },
+    {
+      "epoch": 0.20973197076044658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835310374652731e-05,
+      "loss": 1.1794,
+      "step": 5767
+    },
+    {
+      "epoch": 0.20976833836418518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835245612536974e-05,
+      "loss": 1.0729,
+      "step": 5768
+    },
+    {
+      "epoch": 0.20980470596792378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835180838833369e-05,
+      "loss": 1.1133,
+      "step": 5769
+    },
+    {
+      "epoch": 0.20984107357166237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835116053542815e-05,
+      "loss": 1.155,
+      "step": 5770
+    },
+    {
+      "epoch": 0.20987744117540094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8350512566662107e-05,
+      "loss": 1.1016,
+      "step": 5771
+    },
+    {
+      "epoch": 0.20991380877913954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834986448204455e-05,
+      "loss": 1.1406,
+      "step": 5772
+    },
+    {
+      "epoch": 0.20995017638287813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834921628158447e-05,
+      "loss": 1.053,
+      "step": 5773
+    },
+    {
+      "epoch": 0.20998654398661673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8348567965290858e-05,
+      "loss": 1.1626,
+      "step": 5774
+    },
+    {
+      "epoch": 0.2100229115903553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834791953317271e-05,
+      "loss": 1.0861,
+      "step": 5775
+    },
+    {
+      "epoch": 0.2100592791940939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8347270985239025e-05,
+      "loss": 1.0088,
+      "step": 5776
+    },
+    {
+      "epoch": 0.2100956467978325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8346622321498797e-05,
+      "loss": 1.1981,
+      "step": 5777
+    },
+    {
+      "epoch": 0.2101320144015711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8345973541961027e-05,
+      "loss": 1.187,
+      "step": 5778
+    },
+    {
+      "epoch": 0.21016838200530966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8345324646634718e-05,
+      "loss": 1.1393,
+      "step": 5779
+    },
+    {
+      "epoch": 0.21020474960904825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8344675635528863e-05,
+      "loss": 1.1696,
+      "step": 5780
+    },
+    {
+      "epoch": 0.21024111721278685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834402650865248e-05,
+      "loss": 1.0121,
+      "step": 5781
+    },
+    {
+      "epoch": 0.21027748481652545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8343377266014565e-05,
+      "loss": 1.0532,
+      "step": 5782
+    },
+    {
+      "epoch": 0.21031385242026404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8342727907624128e-05,
+      "loss": 1.0992,
+      "step": 5783
+    },
+    {
+      "epoch": 0.2103502200240026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834207843349018e-05,
+      "loss": 1.1322,
+      "step": 5784
+    },
+    {
+      "epoch": 0.2103865876277412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834142884362173e-05,
+      "loss": 1.0731,
+      "step": 5785
+    },
+    {
+      "epoch": 0.2104229552314798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8340779138027788e-05,
+      "loss": 0.9757,
+      "step": 5786
+    },
+    {
+      "epoch": 0.2104593228352184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.834012931671737e-05,
+      "loss": 1.1535,
+      "step": 5787
+    },
+    {
+      "epoch": 0.21049569043895697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.833947937969949e-05,
+      "loss": 1.1146,
+      "step": 5788
+    },
+    {
+      "epoch": 0.21053205804269556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8338829326983165e-05,
+      "loss": 1.0331,
+      "step": 5789
+    },
+    {
+      "epoch": 0.21056842564643416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8338179158577415e-05,
+      "loss": 1.1137,
+      "step": 5790
+    },
+    {
+      "epoch": 0.21060479325017276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.833752887449126e-05,
+      "loss": 1.1345,
+      "step": 5791
+    },
+    {
+      "epoch": 0.21064116085391132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8336878474733724e-05,
+      "loss": 1.1397,
+      "step": 5792
+    },
+    {
+      "epoch": 0.21067752845764992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8336227959313825e-05,
+      "loss": 1.2046,
+      "step": 5793
+    },
+    {
+      "epoch": 0.21071389606138852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8335577328240586e-05,
+      "loss": 1.1158,
+      "step": 5794
+    },
+    {
+      "epoch": 0.2107502636651271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8334926581523044e-05,
+      "loss": 1.1185,
+      "step": 5795
+    },
+    {
+      "epoch": 0.21078663126886568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8334275719170224e-05,
+      "loss": 1.1038,
+      "step": 5796
+    },
+    {
+      "epoch": 0.21082299887260428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8333624741191148e-05,
+      "loss": 1.2513,
+      "step": 5797
+    },
+    {
+      "epoch": 0.21085936647634287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8332973647594854e-05,
+      "loss": 1.1109,
+      "step": 5798
+    },
+    {
+      "epoch": 0.21089573408008147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8332322438390373e-05,
+      "loss": 0.9287,
+      "step": 5799
+    },
+    {
+      "epoch": 0.21093210168382007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.833167111358674e-05,
+      "loss": 1.1673,
+      "step": 5800
+    },
+    {
+      "epoch": 0.21096846928755864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.833101967319299e-05,
+      "loss": 1.1459,
+      "step": 5801
+    },
+    {
+      "epoch": 0.21100483689129723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8330368117218168e-05,
+      "loss": 1.1103,
+      "step": 5802
+    },
+    {
+      "epoch": 0.21104120449503583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.83297164456713e-05,
+      "loss": 1.0901,
+      "step": 5803
+    },
+    {
+      "epoch": 0.21107757209877442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832906465856144e-05,
+      "loss": 1.1526,
+      "step": 5804
+    },
+    {
+      "epoch": 0.211113939702513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8328412755897625e-05,
+      "loss": 1.0812,
+      "step": 5805
+    },
+    {
+      "epoch": 0.2111503073062516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8327760737688896e-05,
+      "loss": 0.9585,
+      "step": 5806
+    },
+    {
+      "epoch": 0.21118667490999019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8327108603944305e-05,
+      "loss": 1.2891,
+      "step": 5807
+    },
+    {
+      "epoch": 0.21122304251372878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.83264563546729e-05,
+      "loss": 1.1813,
+      "step": 5808
+    },
+    {
+      "epoch": 0.21125941011746735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8325803989883724e-05,
+      "loss": 1.3331,
+      "step": 5809
+    },
+    {
+      "epoch": 0.21129577772120595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8325151509585827e-05,
+      "loss": 1.1649,
+      "step": 5810
+    },
+    {
+      "epoch": 0.21133214532494454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832449891378827e-05,
+      "loss": 1.1065,
+      "step": 5811
+    },
+    {
+      "epoch": 0.21136851292868314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8323846202500094e-05,
+      "loss": 0.9074,
+      "step": 5812
+    },
+    {
+      "epoch": 0.2114048805324217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.832319337573037e-05,
+      "loss": 1.0996,
+      "step": 5813
+    },
+    {
+      "epoch": 0.2114412481361603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8322540433488144e-05,
+      "loss": 1.1226,
+      "step": 5814
+    },
+    {
+      "epoch": 0.2114776157398989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8321887375782477e-05,
+      "loss": 1.1075,
+      "step": 5815
+    },
+    {
+      "epoch": 0.2115139833436375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8321234202622427e-05,
+      "loss": 1.1193,
+      "step": 5816
+    },
+    {
+      "epoch": 0.21155035094737606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8320580914017062e-05,
+      "loss": 1.1169,
+      "step": 5817
+    },
+    {
+      "epoch": 0.21158671855111466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8319927509975445e-05,
+      "loss": 1.1844,
+      "step": 5818
+    },
+    {
+      "epoch": 0.21162308615485326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8319273990506636e-05,
+      "loss": 1.0531,
+      "step": 5819
+    },
+    {
+      "epoch": 0.21165945375859185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.83186203556197e-05,
+      "loss": 0.9907,
+      "step": 5820
+    },
+    {
+      "epoch": 0.21169582136233045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8317966605323714e-05,
+      "loss": 1.1044,
+      "step": 5821
+    },
+    {
+      "epoch": 0.21173218896606902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.831731273962774e-05,
+      "loss": 1.1979,
+      "step": 5822
+    },
+    {
+      "epoch": 0.21176855656980761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8316658758540856e-05,
+      "loss": 1.0901,
+      "step": 5823
+    },
+    {
+      "epoch": 0.2118049241735462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.831600466207213e-05,
+      "loss": 1.1447,
+      "step": 5824
+    },
+    {
+      "epoch": 0.2118412917772848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8315350450230637e-05,
+      "loss": 1.167,
+      "step": 5825
+    },
+    {
+      "epoch": 0.21187765938102338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 1.1173,
+      "step": 5826
+    },
+    {
+      "epoch": 0.21191402698476197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.831404168046566e-05,
+      "loss": 0.9725,
+      "step": 5827
+    },
+    {
+      "epoch": 0.21195039458850057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8313387122560336e-05,
+      "loss": 1.0794,
+      "step": 5828
+    },
+    {
+      "epoch": 0.21198676219223916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8312732449318557e-05,
+      "loss": 1.2314,
+      "step": 5829
+    },
+    {
+      "epoch": 0.21202312979597773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.831207766074941e-05,
+      "loss": 1.0366,
+      "step": 5830
+    },
+    {
+      "epoch": 0.21205949739971633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8311422756861983e-05,
+      "loss": 1.1633,
+      "step": 5831
+    },
+    {
+      "epoch": 0.21209586500345493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8310767737665355e-05,
+      "loss": 1.0609,
+      "step": 5832
+    },
+    {
+      "epoch": 0.21213223260719352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8310112603168613e-05,
+      "loss": 1.1498,
+      "step": 5833
+    },
+    {
+      "epoch": 0.2121686002109321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8309457353380852e-05,
+      "loss": 1.1243,
+      "step": 5834
+    },
+    {
+      "epoch": 0.2122049678146707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.830880198831116e-05,
+      "loss": 1.1266,
+      "step": 5835
+    },
+    {
+      "epoch": 0.21224133541840928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.830814650796863e-05,
+      "loss": 1.1248,
+      "step": 5836
+    },
+    {
+      "epoch": 0.21227770302214788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8307490912362352e-05,
+      "loss": 0.9911,
+      "step": 5837
+    },
+    {
+      "epoch": 0.21231407062588645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8306835201501428e-05,
+      "loss": 1.218,
+      "step": 5838
+    },
+    {
+      "epoch": 0.21235043822962504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8306179375394946e-05,
+      "loss": 1.0556,
+      "step": 5839
+    },
+    {
+      "epoch": 0.21238680583336364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8305523434052018e-05,
+      "loss": 1.0512,
+      "step": 5840
+    },
+    {
+      "epoch": 0.21242317343710224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.830486737748173e-05,
+      "loss": 1.2273,
+      "step": 5841
+    },
+    {
+      "epoch": 0.21245954104084083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8304211205693193e-05,
+      "loss": 1.0771,
+      "step": 5842
+    },
+    {
+      "epoch": 0.2124959086445794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8303554918695505e-05,
+      "loss": 1.1428,
+      "step": 5843
+    },
+    {
+      "epoch": 0.212532276248318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8302898516497776e-05,
+      "loss": 1.2077,
+      "step": 5844
+    },
+    {
+      "epoch": 0.2125686438520566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8302241999109108e-05,
+      "loss": 1.0851,
+      "step": 5845
+    },
+    {
+      "epoch": 0.2126050114557952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8301585366538614e-05,
+      "loss": 1.095,
+      "step": 5846
+    },
+    {
+      "epoch": 0.21264137905953376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8300928618795403e-05,
+      "loss": 1.0294,
+      "step": 5847
+    },
+    {
+      "epoch": 0.21267774666327235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8300271755888583e-05,
+      "loss": 1.019,
+      "step": 5848
+    },
+    {
+      "epoch": 0.21271411426701095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8299614777827267e-05,
+      "loss": 1.0512,
+      "step": 5849
+    },
+    {
+      "epoch": 0.21275048187074955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8298957684620574e-05,
+      "loss": 1.0104,
+      "step": 5850
+    },
+    {
+      "epoch": 0.21278684947448812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8298300476277618e-05,
+      "loss": 1.1738,
+      "step": 5851
+    },
+    {
+      "epoch": 0.2128232170782267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8297643152807515e-05,
+      "loss": 1.1172,
+      "step": 5852
+    },
+    {
+      "epoch": 0.2128595846819653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8296985714219387e-05,
+      "loss": 1.0048,
+      "step": 5853
+    },
+    {
+      "epoch": 0.2128959522857039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8296328160522355e-05,
+      "loss": 1.0912,
+      "step": 5854
+    },
+    {
+      "epoch": 0.21293231988944247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829567049172554e-05,
+      "loss": 1.1394,
+      "step": 5855
+    },
+    {
+      "epoch": 0.21296868749318107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8295012707838068e-05,
+      "loss": 0.985,
+      "step": 5856
+    },
+    {
+      "epoch": 0.21300505509691967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8294354808869058e-05,
+      "loss": 1.0478,
+      "step": 5857
+    },
+    {
+      "epoch": 0.21304142270065826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829369679482765e-05,
+      "loss": 1.2075,
+      "step": 5858
+    },
+    {
+      "epoch": 0.21307779030439683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8293038665722963e-05,
+      "loss": 1.126,
+      "step": 5859
+    },
+    {
+      "epoch": 0.21311415790813543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.829238042156413e-05,
+      "loss": 1.2112,
+      "step": 5860
+    },
+    {
+      "epoch": 0.21315052551187402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8291722062360287e-05,
+      "loss": 1.2517,
+      "step": 5861
+    },
+    {
+      "epoch": 0.21318689311561262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8291063588120563e-05,
+      "loss": 1.0025,
+      "step": 5862
+    },
+    {
+      "epoch": 0.21322326071935122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8290404998854097e-05,
+      "loss": 1.0137,
+      "step": 5863
+    },
+    {
+      "epoch": 0.21325962832308978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8289746294570024e-05,
+      "loss": 1.1805,
+      "step": 5864
+    },
+    {
+      "epoch": 0.21329599592682838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828908747527748e-05,
+      "loss": 1.161,
+      "step": 5865
+    },
+    {
+      "epoch": 0.21333236353056698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8288428540985607e-05,
+      "loss": 1.224,
+      "step": 5866
+    },
+    {
+      "epoch": 0.21336873113430557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828776949170355e-05,
+      "loss": 1.1949,
+      "step": 5867
+    },
+    {
+      "epoch": 0.21340509873804414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828711032744045e-05,
+      "loss": 1.1062,
+      "step": 5868
+    },
+    {
+      "epoch": 0.21344146634178274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8286451048205453e-05,
+      "loss": 1.1757,
+      "step": 5869
+    },
+    {
+      "epoch": 0.21347783394552133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8285791654007704e-05,
+      "loss": 1.0952,
+      "step": 5870
+    },
+    {
+      "epoch": 0.21351420154925993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8285132144856352e-05,
+      "loss": 1.0569,
+      "step": 5871
+    },
+    {
+      "epoch": 0.2135505691529985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828447252076055e-05,
+      "loss": 1.1005,
+      "step": 5872
+    },
+    {
+      "epoch": 0.2135869367567371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828381278172944e-05,
+      "loss": 1.0592,
+      "step": 5873
+    },
+    {
+      "epoch": 0.2136233043604757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8283152927772182e-05,
+      "loss": 1.0635,
+      "step": 5874
+    },
+    {
+      "epoch": 0.2136596719642143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828249295889793e-05,
+      "loss": 1.2028,
+      "step": 5875
+    },
+    {
+      "epoch": 0.21369603956795286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8281832875115843e-05,
+      "loss": 1.0761,
+      "step": 5876
+    },
+    {
+      "epoch": 0.21373240717169145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.828117267643507e-05,
+      "loss": 1.192,
+      "step": 5877
+    },
+    {
+      "epoch": 0.21376877477543005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8280512362864776e-05,
+      "loss": 1.1836,
+      "step": 5878
+    },
+    {
+      "epoch": 0.21380514237916864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8279851934414127e-05,
+      "loss": 1.0157,
+      "step": 5879
+    },
+    {
+      "epoch": 0.2138415099829072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8279191391092274e-05,
+      "loss": 1.1268,
+      "step": 5880
+    },
+    {
+      "epoch": 0.2138778775866458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827853073290839e-05,
+      "loss": 1.0753,
+      "step": 5881
+    },
+    {
+      "epoch": 0.2139142451903844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8277869959871636e-05,
+      "loss": 1.1959,
+      "step": 5882
+    },
+    {
+      "epoch": 0.213950612794123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8277209071991183e-05,
+      "loss": 1.1011,
+      "step": 5883
+    },
+    {
+      "epoch": 0.2139869803978616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8276548069276197e-05,
+      "loss": 1.09,
+      "step": 5884
+    },
+    {
+      "epoch": 0.21402334800160017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827588695173585e-05,
+      "loss": 1.0149,
+      "step": 5885
+    },
+    {
+      "epoch": 0.21405971560533876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827522571937931e-05,
+      "loss": 1.1065,
+      "step": 5886
+    },
+    {
+      "epoch": 0.21409608320907736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8274564372215758e-05,
+      "loss": 1.1561,
+      "step": 5887
+    },
+    {
+      "epoch": 0.21413245081281596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8273902910254365e-05,
+      "loss": 1.0536,
+      "step": 5888
+    },
+    {
+      "epoch": 0.21416881841655452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827324133350431e-05,
+      "loss": 1.0244,
+      "step": 5889
+    },
+    {
+      "epoch": 0.21420518602029312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827257964197477e-05,
+      "loss": 1.1545,
+      "step": 5890
+    },
+    {
+      "epoch": 0.21424155362403172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.827191783567492e-05,
+      "loss": 1.1113,
+      "step": 5891
+    },
+    {
+      "epoch": 0.2142779212277703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8271255914613948e-05,
+      "loss": 1.1274,
+      "step": 5892
+    },
+    {
+      "epoch": 0.21431428883150888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8270593878801035e-05,
+      "loss": 1.1556,
+      "step": 5893
+    },
+    {
+      "epoch": 0.21435065643524748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826993172824537e-05,
+      "loss": 1.0972,
+      "step": 5894
+    },
+    {
+      "epoch": 0.21438702403898607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8269269462956133e-05,
+      "loss": 0.9735,
+      "step": 5895
+    },
+    {
+      "epoch": 0.21442339164272467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8268607082942512e-05,
+      "loss": 1.0881,
+      "step": 5896
+    },
+    {
+      "epoch": 0.21445975924646324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8267944588213704e-05,
+      "loss": 1.0985,
+      "step": 5897
+    },
+    {
+      "epoch": 0.21449612685020183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8267281978778895e-05,
+      "loss": 1.1146,
+      "step": 5898
+    },
+    {
+      "epoch": 0.21453249445394043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8266619254647277e-05,
+      "loss": 1.1138,
+      "step": 5899
+    },
+    {
+      "epoch": 0.21456886205767903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8265956415828044e-05,
+      "loss": 1.0467,
+      "step": 5900
+    },
+    {
+      "epoch": 0.2146052296614176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8265293462330397e-05,
+      "loss": 1.0335,
+      "step": 5901
+    },
+    {
+      "epoch": 0.2146415972651562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826463039416353e-05,
+      "loss": 1.0874,
+      "step": 5902
+    },
+    {
+      "epoch": 0.2146779648688948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826396721133664e-05,
+      "loss": 1.0521,
+      "step": 5903
+    },
+    {
+      "epoch": 0.21471433247263338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826330391385893e-05,
+      "loss": 1.1002,
+      "step": 5904
+    },
+    {
+      "epoch": 0.21475070007637198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8262640501739603e-05,
+      "loss": 1.1557,
+      "step": 5905
+    },
+    {
+      "epoch": 0.21478706768011055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8261976974987862e-05,
+      "loss": 1.1354,
+      "step": 5906
+    },
+    {
+      "epoch": 0.21482343528384915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8261313333612915e-05,
+      "loss": 1.0082,
+      "step": 5907
+    },
+    {
+      "epoch": 0.21485980288758774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8260649577623964e-05,
+      "loss": 1.1314,
+      "step": 5908
+    },
+    {
+      "epoch": 0.21489617049132634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8259985707030223e-05,
+      "loss": 1.2533,
+      "step": 5909
+    },
+    {
+      "epoch": 0.2149325380950649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8259321721840896e-05,
+      "loss": 1.0551,
+      "step": 5910
+    },
+    {
+      "epoch": 0.2149689056988035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8258657622065202e-05,
+      "loss": 1.1046,
+      "step": 5911
+    },
+    {
+      "epoch": 0.2150052733025421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8257993407712348e-05,
+      "loss": 1.1711,
+      "step": 5912
+    },
+    {
+      "epoch": 0.2150416409062807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8257329078791555e-05,
+      "loss": 1.0617,
+      "step": 5913
+    },
+    {
+      "epoch": 0.21507800851001926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8256664635312033e-05,
+      "loss": 1.1014,
+      "step": 5914
+    },
+    {
+      "epoch": 0.21511437611375786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8256000077283004e-05,
+      "loss": 1.1904,
+      "step": 5915
+    },
+    {
+      "epoch": 0.21515074371749646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.825533540471369e-05,
+      "loss": 1.1739,
+      "step": 5916
+    },
+    {
+      "epoch": 0.21518711132123505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.825467061761331e-05,
+      "loss": 1.1026,
+      "step": 5917
+    },
+    {
+      "epoch": 0.21522347892497362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8254005715991088e-05,
+      "loss": 0.9518,
+      "step": 5918
+    },
+    {
+      "epoch": 0.21525984652871222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.825334069985625e-05,
+      "loss": 1.2024,
+      "step": 5919
+    },
+    {
+      "epoch": 0.2152962141324508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8252675569218013e-05,
+      "loss": 0.946,
+      "step": 5920
+    },
+    {
+      "epoch": 0.2153325817361894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8252010324085615e-05,
+      "loss": 1.0826,
+      "step": 5921
+    },
+    {
+      "epoch": 0.21536894933992798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.825134496446828e-05,
+      "loss": 1.046,
+      "step": 5922
+    },
+    {
+      "epoch": 0.21540531694366657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.825067949037524e-05,
+      "loss": 1.1422,
+      "step": 5923
+    },
+    {
+      "epoch": 0.21544168454740517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8250013901815733e-05,
+      "loss": 1.138,
+      "step": 5924
+    },
+    {
+      "epoch": 0.21547805215114377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8249348198798985e-05,
+      "loss": 1.0977,
+      "step": 5925
+    },
+    {
+      "epoch": 0.21551441975488236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8248682381334237e-05,
+      "loss": 1.0222,
+      "step": 5926
+    },
+    {
+      "epoch": 0.21555078735862093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8248016449430724e-05,
+      "loss": 1.1964,
+      "step": 5927
+    },
+    {
+      "epoch": 0.21558715496235953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8247350403097685e-05,
+      "loss": 1.164,
+      "step": 5928
+    },
+    {
+      "epoch": 0.21562352256609812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8246684242344358e-05,
+      "loss": 1.1553,
+      "step": 5929
+    },
+    {
+      "epoch": 0.21565989016983672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824601796717999e-05,
+      "loss": 1.1085,
+      "step": 5930
+    },
+    {
+      "epoch": 0.2156962577735753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8245351577613823e-05,
+      "loss": 1.1964,
+      "step": 5931
+    },
+    {
+      "epoch": 0.21573262537731389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82446850736551e-05,
+      "loss": 1.064,
+      "step": 5932
+    },
+    {
+      "epoch": 0.21576899298105248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824401845531307e-05,
+      "loss": 1.2008,
+      "step": 5933
+    },
+    {
+      "epoch": 0.21580536058479108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824335172259698e-05,
+      "loss": 1.1218,
+      "step": 5934
+    },
+    {
+      "epoch": 0.21584172818852965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824268487551608e-05,
+      "loss": 1.1529,
+      "step": 5935
+    },
+    {
+      "epoch": 0.21587809579226824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8242017914079622e-05,
+      "loss": 1.1799,
+      "step": 5936
+    },
+    {
+      "epoch": 0.21591446339600684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824135083829686e-05,
+      "loss": 1.0876,
+      "step": 5937
+    },
+    {
+      "epoch": 0.21595083099974544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824068364817705e-05,
+      "loss": 1.1194,
+      "step": 5938
+    },
+    {
+      "epoch": 0.215987198603484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8240016343729445e-05,
+      "loss": 1.0615,
+      "step": 5939
+    },
+    {
+      "epoch": 0.2160235662072226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8239348924963303e-05,
+      "loss": 1.0649,
+      "step": 5940
+    },
+    {
+      "epoch": 0.2160599338109612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8238681391887887e-05,
+      "loss": 1.019,
+      "step": 5941
+    },
+    {
+      "epoch": 0.2160963014146998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8238013744512453e-05,
+      "loss": 1.244,
+      "step": 5942
+    },
+    {
+      "epoch": 0.21613266901843836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.823734598284627e-05,
+      "loss": 1.0861,
+      "step": 5943
+    },
+    {
+      "epoch": 0.21616903662217696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8236678106898598e-05,
+      "loss": 1.1766,
+      "step": 5944
+    },
+    {
+      "epoch": 0.21620540422591555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82360101166787e-05,
+      "loss": 1.1101,
+      "step": 5945
+    },
+    {
+      "epoch": 0.21624177182965415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.823534201219585e-05,
+      "loss": 1.0068,
+      "step": 5946
+    },
+    {
+      "epoch": 0.21627813943339275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8234673793459316e-05,
+      "loss": 1.1524,
+      "step": 5947
+    },
+    {
+      "epoch": 0.21631450703713132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8234005460478365e-05,
+      "loss": 1.0742,
+      "step": 5948
+    },
+    {
+      "epoch": 0.2163508746408699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8233337013262268e-05,
+      "loss": 1.1033,
+      "step": 5949
+    },
+    {
+      "epoch": 0.2163872422446085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82326684518203e-05,
+      "loss": 1.0137,
+      "step": 5950
+    },
+    {
+      "epoch": 0.2164236098483471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8231999776161743e-05,
+      "loss": 1.0602,
+      "step": 5951
+    },
+    {
+      "epoch": 0.21645997745208567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.823133098629587e-05,
+      "loss": 1.1017,
+      "step": 5952
+    },
+    {
+      "epoch": 0.21649634505582427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8230662082231955e-05,
+      "loss": 1.1239,
+      "step": 5953
+    },
+    {
+      "epoch": 0.21653271265956286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822999306397928e-05,
+      "loss": 1.1341,
+      "step": 5954
+    },
+    {
+      "epoch": 0.21656908026330146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822932393154713e-05,
+      "loss": 1.2406,
+      "step": 5955
+    },
+    {
+      "epoch": 0.21660544786704003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8228654684944788e-05,
+      "loss": 1.2055,
+      "step": 5956
+    },
+    {
+      "epoch": 0.21664181547077863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8227985324181533e-05,
+      "loss": 1.1909,
+      "step": 5957
+    },
+    {
+      "epoch": 0.21667818307451722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822731584926666e-05,
+      "loss": 1.1369,
+      "step": 5958
+    },
+    {
+      "epoch": 0.21671455067825582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822664626020945e-05,
+      "loss": 1.0396,
+      "step": 5959
+    },
+    {
+      "epoch": 0.2167509182819944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8225976557019196e-05,
+      "loss": 1.0774,
+      "step": 5960
+    },
+    {
+      "epoch": 0.21678728588573298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8225306739705185e-05,
+      "loss": 1.0675,
+      "step": 5961
+    },
+    {
+      "epoch": 0.21682365348947158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8224636808276718e-05,
+      "loss": 1.0945,
+      "step": 5962
+    },
+    {
+      "epoch": 0.21686002109321018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8223966762743083e-05,
+      "loss": 1.2178,
+      "step": 5963
+    },
+    {
+      "epoch": 0.21689638869694874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822329660311358e-05,
+      "loss": 1.102,
+      "step": 5964
+    },
+    {
+      "epoch": 0.21693275630068734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8222626329397498e-05,
+      "loss": 0.9893,
+      "step": 5965
+    },
+    {
+      "epoch": 0.21696912390442594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8221955941604145e-05,
+      "loss": 1.0396,
+      "step": 5966
+    },
+    {
+      "epoch": 0.21700549150816453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822128543974282e-05,
+      "loss": 1.1983,
+      "step": 5967
+    },
+    {
+      "epoch": 0.21704185911190313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8220614823822823e-05,
+      "loss": 1.1649,
+      "step": 5968
+    },
+    {
+      "epoch": 0.2170782267156417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821994409385346e-05,
+      "loss": 1.1871,
+      "step": 5969
+    },
+    {
+      "epoch": 0.2171145943193803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8219273249844036e-05,
+      "loss": 0.975,
+      "step": 5970
+    },
+    {
+      "epoch": 0.2171509619231189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8218602291803854e-05,
+      "loss": 1.0706,
+      "step": 5971
+    },
+    {
+      "epoch": 0.2171873295268575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821793121974223e-05,
+      "loss": 1.0742,
+      "step": 5972
+    },
+    {
+      "epoch": 0.21722369713059606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8217260033668465e-05,
+      "loss": 1.1638,
+      "step": 5973
+    },
+    {
+      "epoch": 0.21726006473433465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821658873359188e-05,
+      "loss": 1.0436,
+      "step": 5974
+    },
+    {
+      "epoch": 0.21729643233807325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821591731952178e-05,
+      "loss": 0.9754,
+      "step": 5975
+    },
+    {
+      "epoch": 0.21733279994181184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8215245791467485e-05,
+      "loss": 1.0272,
+      "step": 5976
+    },
+    {
+      "epoch": 0.2173691675455504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8214574149438315e-05,
+      "loss": 1.0977,
+      "step": 5977
+    },
+    {
+      "epoch": 0.217405535149289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.821390239344358e-05,
+      "loss": 1.1079,
+      "step": 5978
+    },
+    {
+      "epoch": 0.2174419027530276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.82132305234926e-05,
+      "loss": 1.2108,
+      "step": 5979
+    },
+    {
+      "epoch": 0.2174782703567662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8212558539594706e-05,
+      "loss": 1.0891,
+      "step": 5980
+    },
+    {
+      "epoch": 0.21751463796050477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8211886441759215e-05,
+      "loss": 1.2148,
+      "step": 5981
+    },
+    {
+      "epoch": 0.21755100556424337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8211214229995447e-05,
+      "loss": 1.1285,
+      "step": 5982
+    },
+    {
+      "epoch": 0.21758737316798196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8210541904312734e-05,
+      "loss": 1.1619,
+      "step": 5983
+    },
+    {
+      "epoch": 0.21762374077172056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8209869464720397e-05,
+      "loss": 1.105,
+      "step": 5984
+    },
+    {
+      "epoch": 0.21766010837545915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8209196911227773e-05,
+      "loss": 1.0727,
+      "step": 5985
+    },
+    {
+      "epoch": 0.21769647597919772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8208524243844186e-05,
+      "loss": 1.106,
+      "step": 5986
+    },
+    {
+      "epoch": 0.21773284358293632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8207851462578975e-05,
+      "loss": 1.103,
+      "step": 5987
+    },
+    {
+      "epoch": 0.21776921118667492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.820717856744147e-05,
+      "loss": 1.0491,
+      "step": 5988
+    },
+    {
+      "epoch": 0.2178055787904135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8206505558441005e-05,
+      "loss": 1.1841,
+      "step": 5989
+    },
+    {
+      "epoch": 0.21784194639415208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8205832435586922e-05,
+      "loss": 1.25,
+      "step": 5990
+    },
+    {
+      "epoch": 0.21787831399789068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8205159198888555e-05,
+      "loss": 1.1302,
+      "step": 5991
+    },
+    {
+      "epoch": 0.21791468160162927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.820448584835524e-05,
+      "loss": 1.0582,
+      "step": 5992
+    },
+    {
+      "epoch": 0.21795104920536787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.820381238399633e-05,
+      "loss": 0.9937,
+      "step": 5993
+    },
+    {
+      "epoch": 0.21798741680910644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8203138805821165e-05,
+      "loss": 1.132,
+      "step": 5994
+    },
+    {
+      "epoch": 0.21802378441284503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8202465113839085e-05,
+      "loss": 1.1265,
+      "step": 5995
+    },
+    {
+      "epoch": 0.21806015201658363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.820179130805944e-05,
+      "loss": 1.0547,
+      "step": 5996
+    },
+    {
+      "epoch": 0.21809651962032223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8201117388491577e-05,
+      "loss": 1.2219,
+      "step": 5997
+    },
+    {
+      "epoch": 0.2181328872240608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8200443355144845e-05,
+      "loss": 1.1311,
+      "step": 5998
+    },
+    {
+      "epoch": 0.2181692548277994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8199769208028598e-05,
+      "loss": 1.2017,
+      "step": 5999
+    },
+    {
+      "epoch": 0.218205622431538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8199094947152188e-05,
+      "loss": 1.2434,
+      "step": 6000
+    },
+    {
+      "epoch": 0.21824199003527658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8198420572524968e-05,
+      "loss": 0.9975,
+      "step": 6001
+    },
+    {
+      "epoch": 0.21827835763901515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.819774608415629e-05,
+      "loss": 1.137,
+      "step": 6002
+    },
+    {
+      "epoch": 0.21831472524275375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8197071482055523e-05,
+      "loss": 1.128,
+      "step": 6003
+    },
+    {
+      "epoch": 0.21835109284649235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8196396766232018e-05,
+      "loss": 1.1397,
+      "step": 6004
+    },
+    {
+      "epoch": 0.21838746045023094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8195721936695135e-05,
+      "loss": 1.1,
+      "step": 6005
+    },
+    {
+      "epoch": 0.21842382805396954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8195046993454237e-05,
+      "loss": 1.0524,
+      "step": 6006
+    },
+    {
+      "epoch": 0.2184601956577081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8194371936518692e-05,
+      "loss": 1.236,
+      "step": 6007
+    },
+    {
+      "epoch": 0.2184965632614467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.819369676589786e-05,
+      "loss": 1.2095,
+      "step": 6008
+    },
+    {
+      "epoch": 0.2185329308651853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8193021481601112e-05,
+      "loss": 1.184,
+      "step": 6009
+    },
+    {
+      "epoch": 0.2185692984689239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8192346083637817e-05,
+      "loss": 1.1486,
+      "step": 6010
+    },
+    {
+      "epoch": 0.21860566607266246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.819167057201734e-05,
+      "loss": 0.9839,
+      "step": 6011
+    },
+    {
+      "epoch": 0.21864203367640106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.819099494674906e-05,
+      "loss": 1.1543,
+      "step": 6012
+    },
+    {
+      "epoch": 0.21867840128013966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8190319207842346e-05,
+      "loss": 1.1011,
+      "step": 6013
+    },
+    {
+      "epoch": 0.21871476888387825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818964335530657e-05,
+      "loss": 1.0897,
+      "step": 6014
+    },
+    {
+      "epoch": 0.21875113648761682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8188967389151115e-05,
+      "loss": 1.13,
+      "step": 6015
+    },
+    {
+      "epoch": 0.21878750409135542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818829130938536e-05,
+      "loss": 1.1432,
+      "step": 6016
+    },
+    {
+      "epoch": 0.218823871695094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818761511601867e-05,
+      "loss": 1.1572,
+      "step": 6017
+    },
+    {
+      "epoch": 0.2188602392988326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8186938809060448e-05,
+      "loss": 1.1748,
+      "step": 6018
+    },
+    {
+      "epoch": 0.21889660690257118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818626238852006e-05,
+      "loss": 1.1247,
+      "step": 6019
+    },
+    {
+      "epoch": 0.21893297450630977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8185585854406897e-05,
+      "loss": 1.13,
+      "step": 6020
+    },
+    {
+      "epoch": 0.21896934211004837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8184909206730345e-05,
+      "loss": 1.1788,
+      "step": 6021
+    },
+    {
+      "epoch": 0.21900570971378697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8184232445499786e-05,
+      "loss": 1.1341,
+      "step": 6022
+    },
+    {
+      "epoch": 0.21904207731752554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818355557072462e-05,
+      "loss": 1.0634,
+      "step": 6023
+    },
+    {
+      "epoch": 0.21907844492126413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8182878582414232e-05,
+      "loss": 1.1554,
+      "step": 6024
+    },
+    {
+      "epoch": 0.21911481252500273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818220148057801e-05,
+      "loss": 1.0979,
+      "step": 6025
+    },
+    {
+      "epoch": 0.21915118012874132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8181524265225353e-05,
+      "loss": 1.0482,
+      "step": 6026
+    },
+    {
+      "epoch": 0.21918754773247992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8180846936365653e-05,
+      "loss": 1.1656,
+      "step": 6027
+    },
+    {
+      "epoch": 0.2192239153362185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.818016949400831e-05,
+      "loss": 1.1248,
+      "step": 6028
+    },
+    {
+      "epoch": 0.21926028293995709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.817949193816272e-05,
+      "loss": 1.0168,
+      "step": 6029
+    },
+    {
+      "epoch": 0.21929665054369568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8178814268838287e-05,
+      "loss": 1.1001,
+      "step": 6030
+    },
+    {
+      "epoch": 0.21933301814743428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.817813648604441e-05,
+      "loss": 1.1027,
+      "step": 6031
+    },
+    {
+      "epoch": 0.21936938575117285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.817745858979049e-05,
+      "loss": 1.1991,
+      "step": 6032
+    },
+    {
+      "epoch": 0.21940575335491144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8176780580085937e-05,
+      "loss": 1.2109,
+      "step": 6033
+    },
+    {
+      "epoch": 0.21944212095865004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8176102456940156e-05,
+      "loss": 1.1318,
+      "step": 6034
+    },
+    {
+      "epoch": 0.21947848856238864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.817542422036255e-05,
+      "loss": 1.202,
+      "step": 6035
+    },
+    {
+      "epoch": 0.2195148561661272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8174745870362532e-05,
+      "loss": 1.0403,
+      "step": 6036
+    },
+    {
+      "epoch": 0.2195512237698658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8174067406949518e-05,
+      "loss": 1.0481,
+      "step": 6037
+    },
+    {
+      "epoch": 0.2195875913736044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8173388830132914e-05,
+      "loss": 1.1454,
+      "step": 6038
+    },
+    {
+      "epoch": 0.219623958977343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8172710139922135e-05,
+      "loss": 1.2097,
+      "step": 6039
+    },
+    {
+      "epoch": 0.21966032658108156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8172031336326598e-05,
+      "loss": 1.1288,
+      "step": 6040
+    },
+    {
+      "epoch": 0.21969669418482016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8171352419355725e-05,
+      "loss": 1.1541,
+      "step": 6041
+    },
+    {
+      "epoch": 0.21973306178855875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8170673389018927e-05,
+      "loss": 1.0834,
+      "step": 6042
+    },
+    {
+      "epoch": 0.21976942939229735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816999424532563e-05,
+      "loss": 1.1567,
+      "step": 6043
+    },
+    {
+      "epoch": 0.21980579699603592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8169314988285257e-05,
+      "loss": 1.2559,
+      "step": 6044
+    },
+    {
+      "epoch": 0.21984216459977451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8168635617907225e-05,
+      "loss": 1.0718,
+      "step": 6045
+    },
+    {
+      "epoch": 0.2198785322035131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8167956134200965e-05,
+      "loss": 1.1092,
+      "step": 6046
+    },
+    {
+      "epoch": 0.2199148998072517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81672765371759e-05,
+      "loss": 1.0325,
+      "step": 6047
+    },
+    {
+      "epoch": 0.2199512674109903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8166596826841466e-05,
+      "loss": 1.0473,
+      "step": 6048
+    },
+    {
+      "epoch": 0.21998763501472887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8165917003207085e-05,
+      "loss": 1.0429,
+      "step": 6049
+    },
+    {
+      "epoch": 0.22002400261846747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816523706628219e-05,
+      "loss": 1.1309,
+      "step": 6050
+    },
+    {
+      "epoch": 0.22006037022220606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816455701607622e-05,
+      "loss": 1.231,
+      "step": 6051
+    },
+    {
+      "epoch": 0.22009673782594466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8163876852598603e-05,
+      "loss": 1.0627,
+      "step": 6052
+    },
+    {
+      "epoch": 0.22013310542968323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8163196575858776e-05,
+      "loss": 1.2849,
+      "step": 6053
+    },
+    {
+      "epoch": 0.22016947303342183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816251618586618e-05,
+      "loss": 1.1778,
+      "step": 6054
+    },
+    {
+      "epoch": 0.22020584063716042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8161835682630254e-05,
+      "loss": 1.1866,
+      "step": 6055
+    },
+    {
+      "epoch": 0.22024220824089902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8161155066160437e-05,
+      "loss": 1.102,
+      "step": 6056
+    },
+    {
+      "epoch": 0.2202785758446376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.816047433646617e-05,
+      "loss": 1.0315,
+      "step": 6057
+    },
+    {
+      "epoch": 0.22031494344837618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8159793493556902e-05,
+      "loss": 1.1301,
+      "step": 6058
+    },
+    {
+      "epoch": 0.22035131105211478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.815911253744208e-05,
+      "loss": 1.0859,
+      "step": 6059
+    },
+    {
+      "epoch": 0.22038767865585338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8158431468131143e-05,
+      "loss": 1.1878,
+      "step": 6060
+    },
+    {
+      "epoch": 0.22042404625959194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8157750285633548e-05,
+      "loss": 1.2632,
+      "step": 6061
+    },
+    {
+      "epoch": 0.22046041386333054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.815706898995874e-05,
+      "loss": 1.1041,
+      "step": 6062
+    },
+    {
+      "epoch": 0.22049678146706914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8156387581116175e-05,
+      "loss": 0.9482,
+      "step": 6063
+    },
+    {
+      "epoch": 0.22053314907080773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8155706059115303e-05,
+      "loss": 1.0733,
+      "step": 6064
+    },
+    {
+      "epoch": 0.2205695166745463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.815502442396558e-05,
+      "loss": 1.0376,
+      "step": 6065
+    },
+    {
+      "epoch": 0.2206058842782849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8154342675676466e-05,
+      "loss": 1.1381,
+      "step": 6066
+    },
+    {
+      "epoch": 0.2206422518820235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8153660814257417e-05,
+      "loss": 1.1588,
+      "step": 6067
+    },
+    {
+      "epoch": 0.2206786194857621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8152978839717893e-05,
+      "loss": 1.1074,
+      "step": 6068
+    },
+    {
+      "epoch": 0.22071498708950069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8152296752067352e-05,
+      "loss": 1.2892,
+      "step": 6069
+    },
+    {
+      "epoch": 0.22075135469323925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8151614551315262e-05,
+      "loss": 1.0101,
+      "step": 6070
+    },
+    {
+      "epoch": 0.22078772229697785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8150932237471088e-05,
+      "loss": 1.1689,
+      "step": 6071
+    },
+    {
+      "epoch": 0.22082408990071645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8150249810544288e-05,
+      "loss": 1.1495,
+      "step": 6072
+    },
+    {
+      "epoch": 0.22086045750445504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8149567270544342e-05,
+      "loss": 1.1046,
+      "step": 6073
+    },
+    {
+      "epoch": 0.2208968251081936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8148884617480708e-05,
+      "loss": 1.1238,
+      "step": 6074
+    },
+    {
+      "epoch": 0.2209331927119322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8148201851362864e-05,
+      "loss": 1.074,
+      "step": 6075
+    },
+    {
+      "epoch": 0.2209695603156708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814751897220028e-05,
+      "loss": 1.0954,
+      "step": 6076
+    },
+    {
+      "epoch": 0.2210059279194094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814683598000243e-05,
+      "loss": 1.1595,
+      "step": 6077
+    },
+    {
+      "epoch": 0.22104229552314797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814615287477879e-05,
+      "loss": 1.2631,
+      "step": 6078
+    },
+    {
+      "epoch": 0.22107866312688657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8145469656538835e-05,
+      "loss": 1.1503,
+      "step": 6079
+    },
+    {
+      "epoch": 0.22111503073062516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8144786325292047e-05,
+      "loss": 1.1424,
+      "step": 6080
+    },
+    {
+      "epoch": 0.22115139833436376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8144102881047903e-05,
+      "loss": 1.2407,
+      "step": 6081
+    },
+    {
+      "epoch": 0.22118776593810233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8143419323815886e-05,
+      "loss": 0.9567,
+      "step": 6082
+    },
+    {
+      "epoch": 0.22122413354184092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8142735653605483e-05,
+      "loss": 1.0443,
+      "step": 6083
+    },
+    {
+      "epoch": 0.22126050114557952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8142051870426177e-05,
+      "loss": 1.0669,
+      "step": 6084
+    },
+    {
+      "epoch": 0.22129686874931812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814136797428745e-05,
+      "loss": 1.2744,
+      "step": 6085
+    },
+    {
+      "epoch": 0.22133323635305668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8140683965198794e-05,
+      "loss": 1.0546,
+      "step": 6086
+    },
+    {
+      "epoch": 0.22136960395679528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8139999843169697e-05,
+      "loss": 1.1763,
+      "step": 6087
+    },
+    {
+      "epoch": 0.22140597156053388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8139315608209655e-05,
+      "loss": 1.1249,
+      "step": 6088
+    },
+    {
+      "epoch": 0.22144233916427247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8138631260328154e-05,
+      "loss": 1.1615,
+      "step": 6089
+    },
+    {
+      "epoch": 0.22147870676801107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8137946799534694e-05,
+      "loss": 1.0076,
+      "step": 6090
+    },
+    {
+      "epoch": 0.22151507437174964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8137262225838766e-05,
+      "loss": 1.1002,
+      "step": 6091
+    },
+    {
+      "epoch": 0.22155144197548823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.813657753924987e-05,
+      "loss": 1.19,
+      "step": 6092
+    },
+    {
+      "epoch": 0.22158780957922683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8135892739777504e-05,
+      "loss": 1.1353,
+      "step": 6093
+    },
+    {
+      "epoch": 0.22162417718296543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8135207827431173e-05,
+      "loss": 1.0972,
+      "step": 6094
+    },
+    {
+      "epoch": 0.221660544786704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8134522802220372e-05,
+      "loss": 1.1069,
+      "step": 6095
+    },
+    {
+      "epoch": 0.2216969123904426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.813383766415461e-05,
+      "loss": 1.0999,
+      "step": 6096
+    },
+    {
+      "epoch": 0.2217332799941812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.813315241324339e-05,
+      "loss": 1.0851,
+      "step": 6097
+    },
+    {
+      "epoch": 0.22176964759791978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8132467049496224e-05,
+      "loss": 1.1172,
+      "step": 6098
+    },
+    {
+      "epoch": 0.22180601520165835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8131781572922612e-05,
+      "loss": 1.1778,
+      "step": 6099
+    },
+    {
+      "epoch": 0.22184238280539695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.813109598353207e-05,
+      "loss": 1.0605,
+      "step": 6100
+    },
+    {
+      "epoch": 0.22187875040913554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8130410281334105e-05,
+      "loss": 1.1011,
+      "step": 6101
+    },
+    {
+      "epoch": 0.22191511801287414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8129724466338237e-05,
+      "loss": 1.1324,
+      "step": 6102
+    },
+    {
+      "epoch": 0.2219514856166127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8129038538553972e-05,
+      "loss": 0.9998,
+      "step": 6103
+    },
+    {
+      "epoch": 0.2219878532203513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8128352497990835e-05,
+      "loss": 1.1169,
+      "step": 6104
+    },
+    {
+      "epoch": 0.2220242208240899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8127666344658336e-05,
+      "loss": 1.1457,
+      "step": 6105
+    },
+    {
+      "epoch": 0.2220605884278285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8126980078566002e-05,
+      "loss": 1.222,
+      "step": 6106
+    },
+    {
+      "epoch": 0.22209695603156707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.812629369972335e-05,
+      "loss": 1.2105,
+      "step": 6107
+    },
+    {
+      "epoch": 0.22213332363530566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81256072081399e-05,
+      "loss": 1.0127,
+      "step": 6108
+    },
+    {
+      "epoch": 0.22216969123904426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.812492060382518e-05,
+      "loss": 1.1282,
+      "step": 6109
+    },
+    {
+      "epoch": 0.22220605884278286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8124233886788715e-05,
+      "loss": 1.0894,
+      "step": 6110
+    },
+    {
+      "epoch": 0.22224242644652145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.812354705704003e-05,
+      "loss": 1.1418,
+      "step": 6111
+    },
+    {
+      "epoch": 0.22227879405026002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8122860114588653e-05,
+      "loss": 1.2759,
+      "step": 6112
+    },
+    {
+      "epoch": 0.22231516165399862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8122173059444122e-05,
+      "loss": 0.9609,
+      "step": 6113
+    },
+    {
+      "epoch": 0.2223515292577372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.812148589161596e-05,
+      "loss": 1.2074,
+      "step": 6114
+    },
+    {
+      "epoch": 0.2223878968614758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8120798611113704e-05,
+      "loss": 1.0332,
+      "step": 6115
+    },
+    {
+      "epoch": 0.22242426446521438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8120111217946893e-05,
+      "loss": 1.2101,
+      "step": 6116
+    },
+    {
+      "epoch": 0.22246063206895297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8119423712125055e-05,
+      "loss": 1.056,
+      "step": 6117
+    },
+    {
+      "epoch": 0.22249699967269157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8118736093657736e-05,
+      "loss": 1.1743,
+      "step": 6118
+    },
+    {
+      "epoch": 0.22253336727643017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8118048362554472e-05,
+      "loss": 1.1558,
+      "step": 6119
+    },
+    {
+      "epoch": 0.22256973488016873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8117360518824805e-05,
+      "loss": 1.0848,
+      "step": 6120
+    },
+    {
+      "epoch": 0.22260610248390733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.811667256247828e-05,
+      "loss": 1.266,
+      "step": 6121
+    },
+    {
+      "epoch": 0.22264247008764593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8115984493524435e-05,
+      "loss": 1.0025,
+      "step": 6122
+    },
+    {
+      "epoch": 0.22267883769138452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8115296311972824e-05,
+      "loss": 1.0235,
+      "step": 6123
+    },
+    {
+      "epoch": 0.2227152052951231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8114608017832987e-05,
+      "loss": 1.1169,
+      "step": 6124
+    },
+    {
+      "epoch": 0.2227515728988617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.811391961111448e-05,
+      "loss": 1.1777,
+      "step": 6125
+    },
+    {
+      "epoch": 0.22278794050260028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8113231091826848e-05,
+      "loss": 1.1264,
+      "step": 6126
+    },
+    {
+      "epoch": 0.22282430810633888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8112542459979648e-05,
+      "loss": 1.2545,
+      "step": 6127
+    },
+    {
+      "epoch": 0.22286067571007745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.811185371558243e-05,
+      "loss": 1.2172,
+      "step": 6128
+    },
+    {
+      "epoch": 0.22289704331381605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.811116485864475e-05,
+      "loss": 1.3401,
+      "step": 6129
+    },
+    {
+      "epoch": 0.22293341091755464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8110475889176167e-05,
+      "loss": 1.0549,
+      "step": 6130
+    },
+    {
+      "epoch": 0.22296977852129324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8109786807186243e-05,
+      "loss": 1.0698,
+      "step": 6131
+    },
+    {
+      "epoch": 0.22300614612503183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810909761268453e-05,
+      "loss": 1.2076,
+      "step": 6132
+    },
+    {
+      "epoch": 0.2230425137287704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810840830568059e-05,
+      "loss": 1.1855,
+      "step": 6133
+    },
+    {
+      "epoch": 0.223078881332509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810771888618399e-05,
+      "loss": 1.1555,
+      "step": 6134
+    },
+    {
+      "epoch": 0.2231152489362476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8107029354204297e-05,
+      "loss": 1.1016,
+      "step": 6135
+    },
+    {
+      "epoch": 0.2231516165399862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810633970975107e-05,
+      "loss": 1.2157,
+      "step": 6136
+    },
+    {
+      "epoch": 0.22318798414372476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8105649952833883e-05,
+      "loss": 1.0631,
+      "step": 6137
+    },
+    {
+      "epoch": 0.22322435174746336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8104960083462308e-05,
+      "loss": 1.0474,
+      "step": 6138
+    },
+    {
+      "epoch": 0.22326071935120195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8104270101645905e-05,
+      "loss": 1.1664,
+      "step": 6139
+    },
+    {
+      "epoch": 0.22329708695494055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810358000739425e-05,
+      "loss": 1.1067,
+      "step": 6140
+    },
+    {
+      "epoch": 0.22333345455867912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8102889800716927e-05,
+      "loss": 1.1091,
+      "step": 6141
+    },
+    {
+      "epoch": 0.2233698221624177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.81021994816235e-05,
+      "loss": 1.0947,
+      "step": 6142
+    },
+    {
+      "epoch": 0.2234061897661563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8101509050123555e-05,
+      "loss": 1.044,
+      "step": 6143
+    },
+    {
+      "epoch": 0.2234425573698949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8100818506226663e-05,
+      "loss": 1.1751,
+      "step": 6144
+    },
+    {
+      "epoch": 0.22347892497363347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8100127849942407e-05,
+      "loss": 1.0547,
+      "step": 6145
+    },
+    {
+      "epoch": 0.22351529257737207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8099437081280373e-05,
+      "loss": 1.0769,
+      "step": 6146
+    },
+    {
+      "epoch": 0.22355166018111067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8098746200250138e-05,
+      "loss": 1.1868,
+      "step": 6147
+    },
+    {
+      "epoch": 0.22358802778484926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8098055206861294e-05,
+      "loss": 1.0715,
+      "step": 6148
+    },
+    {
+      "epoch": 0.22362439538858783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.809736410112342e-05,
+      "loss": 1.1028,
+      "step": 6149
+    },
+    {
+      "epoch": 0.22366076299232643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.809667288304611e-05,
+      "loss": 1.224,
+      "step": 6150
+    },
+    {
+      "epoch": 0.22369713059606502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.809598155263895e-05,
+      "loss": 1.0632,
+      "step": 6151
+    },
+    {
+      "epoch": 0.22373349819980362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8095290109911533e-05,
+      "loss": 1.0793,
+      "step": 6152
+    },
+    {
+      "epoch": 0.22376986580354222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.809459855487345e-05,
+      "loss": 1.0971,
+      "step": 6153
+    },
+    {
+      "epoch": 0.22380623340728079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.80939068875343e-05,
+      "loss": 1.0593,
+      "step": 6154
+    },
+    {
+      "epoch": 0.22384260101101938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8093215107903672e-05,
+      "loss": 1.1439,
+      "step": 6155
+    },
+    {
+      "epoch": 0.22387896861475798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8092523215991172e-05,
+      "loss": 1.1179,
+      "step": 6156
+    },
+    {
+      "epoch": 0.22391533621849657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.809183121180639e-05,
+      "loss": 1.1233,
+      "step": 6157
+    },
+    {
+      "epoch": 0.22395170382223514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8091139095358932e-05,
+      "loss": 1.2173,
+      "step": 6158
+    },
+    {
+      "epoch": 0.22398807142597374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.80904468666584e-05,
+      "loss": 1.2311,
+      "step": 6159
+    },
+    {
+      "epoch": 0.22402443902971234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8089754525714392e-05,
+      "loss": 1.1379,
+      "step": 6160
+    },
+    {
+      "epoch": 0.22406080663345093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8089062072536526e-05,
+      "loss": 1.0544,
+      "step": 6161
+    },
+    {
+      "epoch": 0.2240971742371895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8088369507134394e-05,
+      "loss": 1.2143,
+      "step": 6162
+    },
+    {
+      "epoch": 0.2241335418409281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8087676829517614e-05,
+      "loss": 1.0223,
+      "step": 6163
+    },
+    {
+      "epoch": 0.2241699094446667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808698403969579e-05,
+      "loss": 1.0714,
+      "step": 6164
+    },
+    {
+      "epoch": 0.2242062770484053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8086291137678536e-05,
+      "loss": 1.2262,
+      "step": 6165
+    },
+    {
+      "epoch": 0.22424264465214386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808559812347547e-05,
+      "loss": 1.2311,
+      "step": 6166
+    },
+    {
+      "epoch": 0.22427901225588245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.80849049970962e-05,
+      "loss": 1.0387,
+      "step": 6167
+    },
+    {
+      "epoch": 0.22431537985962105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8084211758550343e-05,
+      "loss": 1.2056,
+      "step": 6168
+    },
+    {
+      "epoch": 0.22435174746335965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808351840784752e-05,
+      "loss": 1.1265,
+      "step": 6169
+    },
+    {
+      "epoch": 0.22438811506709824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8082824944997344e-05,
+      "loss": 1.116,
+      "step": 6170
+    },
+    {
+      "epoch": 0.2244244826708368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8082131370009442e-05,
+      "loss": 1.0776,
+      "step": 6171
+    },
+    {
+      "epoch": 0.2244608502745754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8081437682893436e-05,
+      "loss": 1.0696,
+      "step": 6172
+    },
+    {
+      "epoch": 0.224497217878314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8080743883658945e-05,
+      "loss": 1.0754,
+      "step": 6173
+    },
+    {
+      "epoch": 0.2245335854820526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8080049972315602e-05,
+      "loss": 1.2323,
+      "step": 6174
+    },
+    {
+      "epoch": 0.22456995308579117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.807935594887303e-05,
+      "loss": 1.0062,
+      "step": 6175
+    },
+    {
+      "epoch": 0.22460632068952976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8078661813340853e-05,
+      "loss": 1.1992,
+      "step": 6176
+    },
+    {
+      "epoch": 0.22464268829326836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.807796756572871e-05,
+      "loss": 1.1815,
+      "step": 6177
+    },
+    {
+      "epoch": 0.22467905589700696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8077273206046223e-05,
+      "loss": 0.9745,
+      "step": 6178
+    },
+    {
+      "epoch": 0.22471542350074553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8076578734303032e-05,
+      "loss": 1.0989,
+      "step": 6179
+    },
+    {
+      "epoch": 0.22475179110448412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8075884150508773e-05,
+      "loss": 1.0634,
+      "step": 6180
+    },
+    {
+      "epoch": 0.22478815870822272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8075189454673075e-05,
+      "loss": 0.9897,
+      "step": 6181
+    },
+    {
+      "epoch": 0.22482452631196131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8074494646805585e-05,
+      "loss": 1.0921,
+      "step": 6182
+    },
+    {
+      "epoch": 0.22486089391569988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8073799726915937e-05,
+      "loss": 1.049,
+      "step": 6183
+    },
+    {
+      "epoch": 0.22489726151943848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.807310469501377e-05,
+      "loss": 1.2279,
+      "step": 6184
+    },
+    {
+      "epoch": 0.22493362912317708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8072409551108734e-05,
+      "loss": 1.095,
+      "step": 6185
+    },
+    {
+      "epoch": 0.22496999672691567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8071714295210464e-05,
+      "loss": 1.0721,
+      "step": 6186
+    },
+    {
+      "epoch": 0.22500636433065424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8071018927328614e-05,
+      "loss": 1.1428,
+      "step": 6187
+    },
+    {
+      "epoch": 0.22504273193439284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8070323447472827e-05,
+      "loss": 0.9778,
+      "step": 6188
+    },
+    {
+      "epoch": 0.22507909953813143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8069627855652752e-05,
+      "loss": 1.1225,
+      "step": 6189
+    },
+    {
+      "epoch": 0.22511546714187003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806893215187804e-05,
+      "loss": 1.2201,
+      "step": 6190
+    },
+    {
+      "epoch": 0.22515183474560863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8068236336158343e-05,
+      "loss": 1.1995,
+      "step": 6191
+    },
+    {
+      "epoch": 0.2251882023493472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806754040850331e-05,
+      "loss": 1.0575,
+      "step": 6192
+    },
+    {
+      "epoch": 0.2252245699530858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8066844368922606e-05,
+      "loss": 1.0301,
+      "step": 6193
+    },
+    {
+      "epoch": 0.2252609375568244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806614821742588e-05,
+      "loss": 1.1609,
+      "step": 6194
+    },
+    {
+      "epoch": 0.22529730516056298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806545195402279e-05,
+      "loss": 1.0875,
+      "step": 6195
+    },
+    {
+      "epoch": 0.22533367276430155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8064755578723e-05,
+      "loss": 1.2429,
+      "step": 6196
+    },
+    {
+      "epoch": 0.22537004036804015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8064059091536167e-05,
+      "loss": 1.1855,
+      "step": 6197
+    },
+    {
+      "epoch": 0.22540640797177874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806336249247196e-05,
+      "loss": 1.1847,
+      "step": 6198
+    },
+    {
+      "epoch": 0.22544277557551734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8062665781540034e-05,
+      "loss": 1.081,
+      "step": 6199
+    },
+    {
+      "epoch": 0.2254791431792559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806196895875006e-05,
+      "loss": 1.1696,
+      "step": 6200
+    },
+    {
+      "epoch": 0.2255155107829945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8061272024111707e-05,
+      "loss": 1.1296,
+      "step": 6201
+    },
+    {
+      "epoch": 0.2255518783867331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.806057497763464e-05,
+      "loss": 1.1855,
+      "step": 6202
+    },
+    {
+      "epoch": 0.2255882459904717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8059877819328533e-05,
+      "loss": 1.0952,
+      "step": 6203
+    },
+    {
+      "epoch": 0.22562461359421027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8059180549203054e-05,
+      "loss": 1.2106,
+      "step": 6204
+    },
+    {
+      "epoch": 0.22566098119794886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8058483167267886e-05,
+      "loss": 1.1197,
+      "step": 6205
+    },
+    {
+      "epoch": 0.22569734880168746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805778567353269e-05,
+      "loss": 1.0558,
+      "step": 6206
+    },
+    {
+      "epoch": 0.22573371640542605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8057088068007155e-05,
+      "loss": 1.1517,
+      "step": 6207
+    },
+    {
+      "epoch": 0.22577008400916462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8056390350700952e-05,
+      "loss": 1.182,
+      "step": 6208
+    },
+    {
+      "epoch": 0.22580645161290322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8055692521623765e-05,
+      "loss": 0.9071,
+      "step": 6209
+    },
+    {
+      "epoch": 0.22584281921664182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8054994580785273e-05,
+      "loss": 1.0889,
+      "step": 6210
+    },
+    {
+      "epoch": 0.2258791868203804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805429652819516e-05,
+      "loss": 1.12,
+      "step": 6211
+    },
+    {
+      "epoch": 0.225915554424119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805359836386311e-05,
+      "loss": 1.0934,
+      "step": 6212
+    },
+    {
+      "epoch": 0.22595192202785758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8052900087798812e-05,
+      "loss": 1.0156,
+      "step": 6213
+    },
+    {
+      "epoch": 0.22598828963159617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805220170001195e-05,
+      "loss": 1.1029,
+      "step": 6214
+    },
+    {
+      "epoch": 0.22602465723533477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.805150320051221e-05,
+      "loss": 1.1747,
+      "step": 6215
+    },
+    {
+      "epoch": 0.22606102483907337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8050804589309286e-05,
+      "loss": 1.0249,
+      "step": 6216
+    },
+    {
+      "epoch": 0.22609739244281193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8050105866412877e-05,
+      "loss": 1.2495,
+      "step": 6217
+    },
+    {
+      "epoch": 0.22613376004655053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8049407031832667e-05,
+      "loss": 1.1877,
+      "step": 6218
+    },
+    {
+      "epoch": 0.22617012765028913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804870808557836e-05,
+      "loss": 1.0828,
+      "step": 6219
+    },
+    {
+      "epoch": 0.22620649525402772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804800902765964e-05,
+      "loss": 1.1935,
+      "step": 6220
+    },
+    {
+      "epoch": 0.2262428628577663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8047309858086216e-05,
+      "loss": 1.3195,
+      "step": 6221
+    },
+    {
+      "epoch": 0.2262792304615049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804661057686779e-05,
+      "loss": 1.1123,
+      "step": 6222
+    },
+    {
+      "epoch": 0.22631559806524348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804591118401405e-05,
+      "loss": 1.1058,
+      "step": 6223
+    },
+    {
+      "epoch": 0.22635196566898208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8045211679534714e-05,
+      "loss": 1.1325,
+      "step": 6224
+    },
+    {
+      "epoch": 0.22638833327272065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804451206343948e-05,
+      "loss": 1.2169,
+      "step": 6225
+    },
+    {
+      "epoch": 0.22642470087645925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8043812335738054e-05,
+      "loss": 1.0241,
+      "step": 6226
+    },
+    {
+      "epoch": 0.22646106848019784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804311249644015e-05,
+      "loss": 1.033,
+      "step": 6227
+    },
+    {
+      "epoch": 0.22649743608393644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8042412545555464e-05,
+      "loss": 1.1126,
+      "step": 6228
+    },
+    {
+      "epoch": 0.226533803687675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804171248309372e-05,
+      "loss": 1.2468,
+      "step": 6229
+    },
+    {
+      "epoch": 0.2265701712914136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.804101230906462e-05,
+      "loss": 1.0099,
+      "step": 6230
+    },
+    {
+      "epoch": 0.2266065388951522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8040312023477887e-05,
+      "loss": 1.1399,
+      "step": 6231
+    },
+    {
+      "epoch": 0.2266429064988908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8039611626343232e-05,
+      "loss": 1.0898,
+      "step": 6232
+    },
+    {
+      "epoch": 0.2266792741026294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803891111767037e-05,
+      "loss": 1.1666,
+      "step": 6233
+    },
+    {
+      "epoch": 0.22671564170636796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8038210497469023e-05,
+      "loss": 1.0715,
+      "step": 6234
+    },
+    {
+      "epoch": 0.22675200931010656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8037509765748914e-05,
+      "loss": 1.1157,
+      "step": 6235
+    },
+    {
+      "epoch": 0.22678837691384515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8036808922519756e-05,
+      "loss": 0.9596,
+      "step": 6236
+    },
+    {
+      "epoch": 0.22682474451758375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8036107967791278e-05,
+      "loss": 1.1012,
+      "step": 6237
+    },
+    {
+      "epoch": 0.22686111212132232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8035406901573202e-05,
+      "loss": 1.1905,
+      "step": 6238
+    },
+    {
+      "epoch": 0.2268974797250609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803470572387526e-05,
+      "loss": 1.0537,
+      "step": 6239
+    },
+    {
+      "epoch": 0.2269338473287995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803400443470717e-05,
+      "loss": 1.0996,
+      "step": 6240
+    },
+    {
+      "epoch": 0.2269702149325381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8033303034078674e-05,
+      "loss": 1.1862,
+      "step": 6241
+    },
+    {
+      "epoch": 0.22700658253627667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803260152199949e-05,
+      "loss": 1.0845,
+      "step": 6242
+    },
+    {
+      "epoch": 0.22704295014001527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803189989847936e-05,
+      "loss": 1.1617,
+      "step": 6243
+    },
+    {
+      "epoch": 0.22707931774375387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8031198163528017e-05,
+      "loss": 1.1002,
+      "step": 6244
+    },
+    {
+      "epoch": 0.22711568534749246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8030496317155192e-05,
+      "loss": 1.1185,
+      "step": 6245
+    },
+    {
+      "epoch": 0.22715205295123103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8029794359370623e-05,
+      "loss": 1.0958,
+      "step": 6246
+    },
+    {
+      "epoch": 0.22718842055496963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802909229018405e-05,
+      "loss": 1.125,
+      "step": 6247
+    },
+    {
+      "epoch": 0.22722478815870822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8028390109605217e-05,
+      "loss": 1.0664,
+      "step": 6248
+    },
+    {
+      "epoch": 0.22726115576244682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8027687817643856e-05,
+      "loss": 1.1085,
+      "step": 6249
+    },
+    {
+      "epoch": 0.2272975233661854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802698541430972e-05,
+      "loss": 1.1354,
+      "step": 6250
+    },
+    {
+      "epoch": 0.22733389096992399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802628289961255e-05,
+      "loss": 1.1782,
+      "step": 6251
+    },
+    {
+      "epoch": 0.22737025857366258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802558027356209e-05,
+      "loss": 1.09,
+      "step": 6252
+    },
+    {
+      "epoch": 0.22740662617740118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8024877536168096e-05,
+      "loss": 1.1009,
+      "step": 6253
+    },
+    {
+      "epoch": 0.22744299378113977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802417468744031e-05,
+      "loss": 1.0007,
+      "step": 6254
+    },
+    {
+      "epoch": 0.22747936138487834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8023471727388485e-05,
+      "loss": 1.2105,
+      "step": 6255
+    },
+    {
+      "epoch": 0.22751572898861694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.802276865602237e-05,
+      "loss": 1.1027,
+      "step": 6256
+    },
+    {
+      "epoch": 0.22755209659235553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8022065473351724e-05,
+      "loss": 1.0802,
+      "step": 6257
+    },
+    {
+      "epoch": 0.22758846419609413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8021362179386304e-05,
+      "loss": 1.1902,
+      "step": 6258
+    },
+    {
+      "epoch": 0.2276248317998327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8020658774135866e-05,
+      "loss": 1.1514,
+      "step": 6259
+    },
+    {
+      "epoch": 0.2276611994035713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8019955257610164e-05,
+      "loss": 1.1463,
+      "step": 6260
+    },
+    {
+      "epoch": 0.2276975670073099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801925162981896e-05,
+      "loss": 1.1573,
+      "step": 6261
+    },
+    {
+      "epoch": 0.2277339346110485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801854789077202e-05,
+      "loss": 1.2302,
+      "step": 6262
+    },
+    {
+      "epoch": 0.22777030221478706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8017844040479104e-05,
+      "loss": 0.9928,
+      "step": 6263
+    },
+    {
+      "epoch": 0.22780666981852565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8017140078949976e-05,
+      "loss": 1.1655,
+      "step": 6264
+    },
+    {
+      "epoch": 0.22784303742226425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8016436006194407e-05,
+      "loss": 1.021,
+      "step": 6265
+    },
+    {
+      "epoch": 0.22787940502600285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8015731822222165e-05,
+      "loss": 1.093,
+      "step": 6266
+    },
+    {
+      "epoch": 0.22791577262974141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8015027527043016e-05,
+      "loss": 1.0026,
+      "step": 6267
+    },
+    {
+      "epoch": 0.22795214023348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8014323120666726e-05,
+      "loss": 1.1702,
+      "step": 6268
+    },
+    {
+      "epoch": 0.2279885078372186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8013618603103077e-05,
+      "loss": 1.1514,
+      "step": 6269
+    },
+    {
+      "epoch": 0.2280248754409572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801291397436184e-05,
+      "loss": 1.1729,
+      "step": 6270
+    },
+    {
+      "epoch": 0.22806124304469577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.801220923445279e-05,
+      "loss": 1.1419,
+      "step": 6271
+    },
+    {
+      "epoch": 0.22809761064843437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8011504383385706e-05,
+      "loss": 1.0736,
+      "step": 6272
+    },
+    {
+      "epoch": 0.22813397825217296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8010799421170363e-05,
+      "loss": 1.2234,
+      "step": 6273
+    },
+    {
+      "epoch": 0.22817034585591156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8010094347816545e-05,
+      "loss": 1.0056,
+      "step": 6274
+    },
+    {
+      "epoch": 0.22820671345965016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8009389163334035e-05,
+      "loss": 1.0498,
+      "step": 6275
+    },
+    {
+      "epoch": 0.22824308106338873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.800868386773261e-05,
+      "loss": 1.0494,
+      "step": 6276
+    },
+    {
+      "epoch": 0.22827944866712732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8007978461022063e-05,
+      "loss": 1.2231,
+      "step": 6277
+    },
+    {
+      "epoch": 0.22831581627086592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8007272943212176e-05,
+      "loss": 1.1166,
+      "step": 6278
+    },
+    {
+      "epoch": 0.22835218387460451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8006567314312735e-05,
+      "loss": 1.1406,
+      "step": 6279
+    },
+    {
+      "epoch": 0.22838855147834308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8005861574333536e-05,
+      "loss": 1.13,
+      "step": 6280
+    },
+    {
+      "epoch": 0.22842491908208168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8005155723284363e-05,
+      "loss": 1.1977,
+      "step": 6281
+    },
+    {
+      "epoch": 0.22846128668582028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8004449761175015e-05,
+      "loss": 1.1337,
+      "step": 6282
+    },
+    {
+      "epoch": 0.22849765428955887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8003743688015284e-05,
+      "loss": 1.0205,
+      "step": 6283
+    },
+    {
+      "epoch": 0.22853402189329744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.800303750381496e-05,
+      "loss": 1.0262,
+      "step": 6284
+    },
+    {
+      "epoch": 0.22857038949703604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.800233120858385e-05,
+      "loss": 1.1643,
+      "step": 6285
+    },
+    {
+      "epoch": 0.22860675710077463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8001624802331745e-05,
+      "loss": 1.0834,
+      "step": 6286
+    },
+    {
+      "epoch": 0.22864312470451323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8000918285068452e-05,
+      "loss": 0.9994,
+      "step": 6287
+    },
+    {
+      "epoch": 0.2286794923082518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8000211656803768e-05,
+      "loss": 1.1914,
+      "step": 6288
+    },
+    {
+      "epoch": 0.2287158599119904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79995049175475e-05,
+      "loss": 0.97,
+      "step": 6289
+    },
+    {
+      "epoch": 0.228752227515729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.799879806730945e-05,
+      "loss": 1.1443,
+      "step": 6290
+    },
+    {
+      "epoch": 0.22878859511946759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7998091106099423e-05,
+      "loss": 1.0071,
+      "step": 6291
+    },
+    {
+      "epoch": 0.22882496272320615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7997384033927235e-05,
+      "loss": 1.1569,
+      "step": 6292
+    },
+    {
+      "epoch": 0.22886133032694475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7996676850802685e-05,
+      "loss": 1.0042,
+      "step": 6293
+    },
+    {
+      "epoch": 0.22889769793068335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7995969556735593e-05,
+      "loss": 1.0851,
+      "step": 6294
+    },
+    {
+      "epoch": 0.22893406553442194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7995262151735767e-05,
+      "loss": 1.0726,
+      "step": 6295
+    },
+    {
+      "epoch": 0.22897043313816054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7994554635813023e-05,
+      "loss": 0.988,
+      "step": 6296
+    },
+    {
+      "epoch": 0.2290068007418991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7993847008977176e-05,
+      "loss": 1.225,
+      "step": 6297
+    },
+    {
+      "epoch": 0.2290431683456377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.799313927123804e-05,
+      "loss": 1.0385,
+      "step": 6298
+    },
+    {
+      "epoch": 0.2290795359493763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.799243142260544e-05,
+      "loss": 0.9981,
+      "step": 6299
+    },
+    {
+      "epoch": 0.2291159035531149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7991723463089194e-05,
+      "loss": 1.0727,
+      "step": 6300
+    },
+    {
+      "epoch": 0.22915227115685347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7991015392699127e-05,
+      "loss": 1.0216,
+      "step": 6301
+    },
+    {
+      "epoch": 0.22918863876059206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7990307211445057e-05,
+      "loss": 1.1096,
+      "step": 6302
+    },
+    {
+      "epoch": 0.22922500636433066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7989598919336807e-05,
+      "loss": 1.1589,
+      "step": 6303
+    },
+    {
+      "epoch": 0.22926137396806925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7988890516384213e-05,
+      "loss": 1.0907,
+      "step": 6304
+    },
+    {
+      "epoch": 0.22929774157180782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7988182002597097e-05,
+      "loss": 1.1993,
+      "step": 6305
+    },
+    {
+      "epoch": 0.22933410917554642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7987473377985288e-05,
+      "loss": 1.1025,
+      "step": 6306
+    },
+    {
+      "epoch": 0.22937047677928502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7986764642558618e-05,
+      "loss": 1.0125,
+      "step": 6307
+    },
+    {
+      "epoch": 0.2294068443830236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7986055796326923e-05,
+      "loss": 1.0666,
+      "step": 6308
+    },
+    {
+      "epoch": 0.22944321198676218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7985346839300032e-05,
+      "loss": 1.187,
+      "step": 6309
+    },
+    {
+      "epoch": 0.22947957959050078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7984637771487785e-05,
+      "loss": 1.1634,
+      "step": 6310
+    },
+    {
+      "epoch": 0.22951594719423937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7983928592900016e-05,
+      "loss": 1.0949,
+      "step": 6311
+    },
+    {
+      "epoch": 0.22955231479797797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7983219303546565e-05,
+      "loss": 1.096,
+      "step": 6312
+    },
+    {
+      "epoch": 0.22958868240171654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7982509903437275e-05,
+      "loss": 1.1506,
+      "step": 6313
+    },
+    {
+      "epoch": 0.22962505000545513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7981800392581988e-05,
+      "loss": 1.1579,
+      "step": 6314
+    },
+    {
+      "epoch": 0.22966141760919373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.798109077099054e-05,
+      "loss": 0.9589,
+      "step": 6315
+    },
+    {
+      "epoch": 0.22969778521293233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7980381038672786e-05,
+      "loss": 1.097,
+      "step": 6316
+    },
+    {
+      "epoch": 0.22973415281667092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7979671195638564e-05,
+      "loss": 1.2476,
+      "step": 6317
+    },
+    {
+      "epoch": 0.2297705204204095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7978961241897727e-05,
+      "loss": 1.0703,
+      "step": 6318
+    },
+    {
+      "epoch": 0.2298068880241481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7978251177460124e-05,
+      "loss": 1.0503,
+      "step": 6319
+    },
+    {
+      "epoch": 0.22984325562788668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7977541002335606e-05,
+      "loss": 1.2743,
+      "step": 6320
+    },
+    {
+      "epoch": 0.22987962323162528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7976830716534023e-05,
+      "loss": 1.0027,
+      "step": 6321
+    },
+    {
+      "epoch": 0.22991599083536385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7976120320065233e-05,
+      "loss": 1.1315,
+      "step": 6322
+    },
+    {
+      "epoch": 0.22995235843910244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.797540981293909e-05,
+      "loss": 1.1785,
+      "step": 6323
+    },
+    {
+      "epoch": 0.22998872604284104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7974699195165454e-05,
+      "loss": 1.1004,
+      "step": 6324
+    },
+    {
+      "epoch": 0.23002509364657964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7973988466754177e-05,
+      "loss": 1.0733,
+      "step": 6325
+    },
+    {
+      "epoch": 0.2300614612503182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7973277627715128e-05,
+      "loss": 1.1397,
+      "step": 6326
+    },
+    {
+      "epoch": 0.2300978288540568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.797256667805816e-05,
+      "loss": 1.1306,
+      "step": 6327
+    },
+    {
+      "epoch": 0.2301341964577954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.797185561779314e-05,
+      "loss": 1.0389,
+      "step": 6328
+    },
+    {
+      "epoch": 0.230170564061534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.797114444692994e-05,
+      "loss": 1.1095,
+      "step": 6329
+    },
+    {
+      "epoch": 0.23020693166527256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7970433165478417e-05,
+      "loss": 1.0153,
+      "step": 6330
+    },
+    {
+      "epoch": 0.23024329926901116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7969721773448443e-05,
+      "loss": 1.2726,
+      "step": 6331
+    },
+    {
+      "epoch": 0.23027966687274976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7969010270849885e-05,
+      "loss": 1.1123,
+      "step": 6332
+    },
+    {
+      "epoch": 0.23031603447648835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.796829865769262e-05,
+      "loss": 1.1148,
+      "step": 6333
+    },
+    {
+      "epoch": 0.23035240208022692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7967586933986513e-05,
+      "loss": 1.0453,
+      "step": 6334
+    },
+    {
+      "epoch": 0.23038876968396552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7966875099741444e-05,
+      "loss": 1.1174,
+      "step": 6335
+    },
+    {
+      "epoch": 0.2304251372877041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7966163154967283e-05,
+      "loss": 1.1063,
+      "step": 6336
+    },
+    {
+      "epoch": 0.2304615048914427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7965451099673913e-05,
+      "loss": 1.1418,
+      "step": 6337
+    },
+    {
+      "epoch": 0.2304978724951813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7964738933871214e-05,
+      "loss": 1.0898,
+      "step": 6338
+    },
+    {
+      "epoch": 0.23053424009891987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.796402665756906e-05,
+      "loss": 1.0189,
+      "step": 6339
+    },
+    {
+      "epoch": 0.23057060770265847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7963314270777334e-05,
+      "loss": 1.1403,
+      "step": 6340
+    },
+    {
+      "epoch": 0.23060697530639707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7962601773505923e-05,
+      "loss": 1.001,
+      "step": 6341
+    },
+    {
+      "epoch": 0.23064334291013566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.796188916576471e-05,
+      "loss": 1.1962,
+      "step": 6342
+    },
+    {
+      "epoch": 0.23067971051387423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.796117644756358e-05,
+      "loss": 1.2068,
+      "step": 6343
+    },
+    {
+      "epoch": 0.23071607811761283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.796046361891242e-05,
+      "loss": 1.1135,
+      "step": 6344
+    },
+    {
+      "epoch": 0.23075244572135142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7959750679821125e-05,
+      "loss": 1.1873,
+      "step": 6345
+    },
+    {
+      "epoch": 0.23078881332509002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.795903763029958e-05,
+      "loss": 1.0204,
+      "step": 6346
+    },
+    {
+      "epoch": 0.2308251809288286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7958324470357683e-05,
+      "loss": 1.0769,
+      "step": 6347
+    },
+    {
+      "epoch": 0.23086154853256718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7957611200005325e-05,
+      "loss": 1.0265,
+      "step": 6348
+    },
+    {
+      "epoch": 0.23089791613630578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79568978192524e-05,
+      "loss": 1.1054,
+      "step": 6349
+    },
+    {
+      "epoch": 0.23093428374004438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7956184328108812e-05,
+      "loss": 1.0318,
+      "step": 6350
+    },
+    {
+      "epoch": 0.23097065134378295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.795547072658445e-05,
+      "loss": 1.1071,
+      "step": 6351
+    },
+    {
+      "epoch": 0.23100701894752154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.795475701468922e-05,
+      "loss": 1.1094,
+      "step": 6352
+    },
+    {
+      "epoch": 0.23104338655126014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7954043192433023e-05,
+      "loss": 1.085,
+      "step": 6353
+    },
+    {
+      "epoch": 0.23107975415499873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7953329259825763e-05,
+      "loss": 1.0512,
+      "step": 6354
+    },
+    {
+      "epoch": 0.23111612175873733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7952615216877344e-05,
+      "loss": 1.0712,
+      "step": 6355
+    },
+    {
+      "epoch": 0.2311524893624759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.795190106359767e-05,
+      "loss": 1.0511,
+      "step": 6356
+    },
+    {
+      "epoch": 0.2311888569662145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7951186799996652e-05,
+      "loss": 0.9733,
+      "step": 6357
+    },
+    {
+      "epoch": 0.2312252245699531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79504724260842e-05,
+      "loss": 1.0585,
+      "step": 6358
+    },
+    {
+      "epoch": 0.2312615921736917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794975794187022e-05,
+      "loss": 1.0387,
+      "step": 6359
+    },
+    {
+      "epoch": 0.23129795977743026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794904334736463e-05,
+      "loss": 1.0206,
+      "step": 6360
+    },
+    {
+      "epoch": 0.23133432738116885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7948328642577343e-05,
+      "loss": 1.0922,
+      "step": 6361
+    },
+    {
+      "epoch": 0.23137069498490745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794761382751827e-05,
+      "loss": 1.1527,
+      "step": 6362
+    },
+    {
+      "epoch": 0.23140706258864605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7946898902197337e-05,
+      "loss": 0.9729,
+      "step": 6363
+    },
+    {
+      "epoch": 0.2314434301923846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794618386662445e-05,
+      "loss": 1.0924,
+      "step": 6364
+    },
+    {
+      "epoch": 0.2314797977961232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794546872080954e-05,
+      "loss": 1.0171,
+      "step": 6365
+    },
+    {
+      "epoch": 0.2315161653998618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7944753464762522e-05,
+      "loss": 1.0786,
+      "step": 6366
+    },
+    {
+      "epoch": 0.2315525330036004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7944038098493327e-05,
+      "loss": 1.0628,
+      "step": 6367
+    },
+    {
+      "epoch": 0.23158890060733897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794332262201187e-05,
+      "loss": 1.051,
+      "step": 6368
+    },
+    {
+      "epoch": 0.23162526821107757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794260703532808e-05,
+      "loss": 1.1772,
+      "step": 6369
+    },
+    {
+      "epoch": 0.23166163581481616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7941891338451893e-05,
+      "loss": 1.1473,
+      "step": 6370
+    },
+    {
+      "epoch": 0.23169800341855476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7941175531393223e-05,
+      "loss": 0.9973,
+      "step": 6371
+    },
+    {
+      "epoch": 0.23173437102229333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7940459614162014e-05,
+      "loss": 1.1437,
+      "step": 6372
+    },
+    {
+      "epoch": 0.23177073862603192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7939743586768198e-05,
+      "loss": 1.081,
+      "step": 6373
+    },
+    {
+      "epoch": 0.23180710622977052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79390274492217e-05,
+      "loss": 1.1641,
+      "step": 6374
+    },
+    {
+      "epoch": 0.23184347383350912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7938311201532463e-05,
+      "loss": 1.201,
+      "step": 6375
+    },
+    {
+      "epoch": 0.2318798414372477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7937594843710416e-05,
+      "loss": 1.062,
+      "step": 6376
+    },
+    {
+      "epoch": 0.23191620904098628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7936878375765507e-05,
+      "loss": 0.995,
+      "step": 6377
+    },
+    {
+      "epoch": 0.23195257664472488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.793616179770767e-05,
+      "loss": 1.0854,
+      "step": 6378
+    },
+    {
+      "epoch": 0.23198894424846347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7935445109546853e-05,
+      "loss": 0.9593,
+      "step": 6379
+    },
+    {
+      "epoch": 0.23202531185220207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7934728311292986e-05,
+      "loss": 1.076,
+      "step": 6380
+    },
+    {
+      "epoch": 0.23206167945594064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7934011402956028e-05,
+      "loss": 1.0845,
+      "step": 6381
+    },
+    {
+      "epoch": 0.23209804705967924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7933294384545914e-05,
+      "loss": 1.1996,
+      "step": 6382
+    },
+    {
+      "epoch": 0.23213441466341783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7932577256072597e-05,
+      "loss": 1.1564,
+      "step": 6383
+    },
+    {
+      "epoch": 0.23217078226715643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7931860017546024e-05,
+      "loss": 1.0001,
+      "step": 6384
+    },
+    {
+      "epoch": 0.232207149870895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7931142668976152e-05,
+      "loss": 1.0814,
+      "step": 6385
+    },
+    {
+      "epoch": 0.2322435174746336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7930425210372927e-05,
+      "loss": 1.1709,
+      "step": 6386
+    },
+    {
+      "epoch": 0.2322798850783722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79297076417463e-05,
+      "loss": 1.0711,
+      "step": 6387
+    },
+    {
+      "epoch": 0.23231625268211079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7928989963106236e-05,
+      "loss": 1.1096,
+      "step": 6388
+    },
+    {
+      "epoch": 0.23235262028584935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.792827217446268e-05,
+      "loss": 1.0446,
+      "step": 6389
+    },
+    {
+      "epoch": 0.23238898788958795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79275542758256e-05,
+      "loss": 1.1074,
+      "step": 6390
+    },
+    {
+      "epoch": 0.23242535549332655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7926836267204954e-05,
+      "loss": 1.134,
+      "step": 6391
+    },
+    {
+      "epoch": 0.23246172309706514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7926118148610698e-05,
+      "loss": 1.1862,
+      "step": 6392
+    },
+    {
+      "epoch": 0.2324980907008037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7925399920052798e-05,
+      "loss": 1.096,
+      "step": 6393
+    },
+    {
+      "epoch": 0.2325344583045423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.792468158154122e-05,
+      "loss": 1.1641,
+      "step": 6394
+    },
+    {
+      "epoch": 0.2325708259082809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.792396313308593e-05,
+      "loss": 1.0369,
+      "step": 6395
+    },
+    {
+      "epoch": 0.2326071935120195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.792324457469689e-05,
+      "loss": 1.1384,
+      "step": 6396
+    },
+    {
+      "epoch": 0.2326435611157581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.792252590638408e-05,
+      "loss": 1.2132,
+      "step": 6397
+    },
+    {
+      "epoch": 0.23267992871949666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7921807128157458e-05,
+      "loss": 1.0809,
+      "step": 6398
+    },
+    {
+      "epoch": 0.23271629632323526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7921088240027003e-05,
+      "loss": 1.045,
+      "step": 6399
+    },
+    {
+      "epoch": 0.23275266392697386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7920369242002687e-05,
+      "loss": 1.1437,
+      "step": 6400
+    },
+    {
+      "epoch": 0.23278903153071245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7919650134094488e-05,
+      "loss": 0.9776,
+      "step": 6401
+    },
+    {
+      "epoch": 0.23282539913445102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.791893091631238e-05,
+      "loss": 1.1657,
+      "step": 6402
+    },
+    {
+      "epoch": 0.23286176673818962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7918211588666337e-05,
+      "loss": 1.2568,
+      "step": 6403
+    },
+    {
+      "epoch": 0.23289813434192821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7917492151166345e-05,
+      "loss": 1.1073,
+      "step": 6404
+    },
+    {
+      "epoch": 0.2329345019456668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7916772603822383e-05,
+      "loss": 1.1662,
+      "step": 6405
+    },
+    {
+      "epoch": 0.23297086954940538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7916052946644433e-05,
+      "loss": 1.134,
+      "step": 6406
+    },
+    {
+      "epoch": 0.23300723715314398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.791533317964248e-05,
+      "loss": 1.1075,
+      "step": 6407
+    },
+    {
+      "epoch": 0.23304360475688257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7914613302826508e-05,
+      "loss": 1.0407,
+      "step": 6408
+    },
+    {
+      "epoch": 0.23307997236062117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.791389331620651e-05,
+      "loss": 1.0996,
+      "step": 6409
+    },
+    {
+      "epoch": 0.23311633996435974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7913173219792466e-05,
+      "loss": 1.1329,
+      "step": 6410
+    },
+    {
+      "epoch": 0.23315270756809833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.791245301359437e-05,
+      "loss": 1.0516,
+      "step": 6411
+    },
+    {
+      "epoch": 0.23318907517183693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7911732697622218e-05,
+      "loss": 1.1826,
+      "step": 6412
+    },
+    {
+      "epoch": 0.23322544277557553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7911012271886e-05,
+      "loss": 1.0763,
+      "step": 6413
+    },
+    {
+      "epoch": 0.2332618103793141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7910291736395712e-05,
+      "loss": 1.084,
+      "step": 6414
+    },
+    {
+      "epoch": 0.2332981779830527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7909571091161348e-05,
+      "loss": 1.2211,
+      "step": 6415
+    },
+    {
+      "epoch": 0.2333345455867913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7908850336192907e-05,
+      "loss": 1.1752,
+      "step": 6416
+    },
+    {
+      "epoch": 0.23337091319052988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7908129471500387e-05,
+      "loss": 1.0009,
+      "step": 6417
+    },
+    {
+      "epoch": 0.23340728079426848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7907408497093793e-05,
+      "loss": 1.1225,
+      "step": 6418
+    },
+    {
+      "epoch": 0.23344364839800705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7906687412983125e-05,
+      "loss": 1.1343,
+      "step": 6419
+    },
+    {
+      "epoch": 0.23348001600174564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.790596621917839e-05,
+      "loss": 1.2266,
+      "step": 6420
+    },
+    {
+      "epoch": 0.23351638360548424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7905244915689585e-05,
+      "loss": 1.0364,
+      "step": 6421
+    },
+    {
+      "epoch": 0.23355275120922284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.790452350252673e-05,
+      "loss": 1.2156,
+      "step": 6422
+    },
+    {
+      "epoch": 0.2335891188129614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7903801979699817e-05,
+      "loss": 1.1728,
+      "step": 6423
+    },
+    {
+      "epoch": 0.2336254864167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7903080347218873e-05,
+      "loss": 1.0803,
+      "step": 6424
+    },
+    {
+      "epoch": 0.2336618540204386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.79023586050939e-05,
+      "loss": 1.0845,
+      "step": 6425
+    },
+    {
+      "epoch": 0.2336982216241772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7901636753334915e-05,
+      "loss": 1.0303,
+      "step": 6426
+    },
+    {
+      "epoch": 0.23373458922791576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7900914791951928e-05,
+      "loss": 1.1866,
+      "step": 6427
+    },
+    {
+      "epoch": 0.23377095683165436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.790019272095496e-05,
+      "loss": 1.087,
+      "step": 6428
+    },
+    {
+      "epoch": 0.23380732443539295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7899470540354028e-05,
+      "loss": 1.1378,
+      "step": 6429
+    },
+    {
+      "epoch": 0.23384369203913155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7898748250159146e-05,
+      "loss": 0.9465,
+      "step": 6430
+    },
+    {
+      "epoch": 0.23388005964287012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7898025850380342e-05,
+      "loss": 1.1292,
+      "step": 6431
+    },
+    {
+      "epoch": 0.23391642724660872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7897303341027635e-05,
+      "loss": 1.0316,
+      "step": 6432
+    },
+    {
+      "epoch": 0.2339527948503473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.789658072211105e-05,
+      "loss": 1.0977,
+      "step": 6433
+    },
+    {
+      "epoch": 0.2339891624540859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.789585799364061e-05,
+      "loss": 1.1008,
+      "step": 6434
+    },
+    {
+      "epoch": 0.23402553005782448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7895135155626342e-05,
+      "loss": 0.9695,
+      "step": 6435
+    },
+    {
+      "epoch": 0.23406189766156307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7894412208078277e-05,
+      "loss": 1.0198,
+      "step": 6436
+    },
+    {
+      "epoch": 0.23409826526530167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7893689151006443e-05,
+      "loss": 1.0971,
+      "step": 6437
+    },
+    {
+      "epoch": 0.23413463286904027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7892965984420872e-05,
+      "loss": 1.1445,
+      "step": 6438
+    },
+    {
+      "epoch": 0.23417100047277886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7892242708331596e-05,
+      "loss": 1.021,
+      "step": 6439
+    },
+    {
+      "epoch": 0.23420736807651743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.789151932274865e-05,
+      "loss": 1.0144,
+      "step": 6440
+    },
+    {
+      "epoch": 0.23424373568025603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7890795827682067e-05,
+      "loss": 0.9917,
+      "step": 6441
+    },
+    {
+      "epoch": 0.23428010328399462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7890072223141892e-05,
+      "loss": 1.037,
+      "step": 6442
+    },
+    {
+      "epoch": 0.23431647088773322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.788934850913816e-05,
+      "loss": 1.3046,
+      "step": 6443
+    },
+    {
+      "epoch": 0.2343528384914718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7888624685680905e-05,
+      "loss": 1.049,
+      "step": 6444
+    },
+    {
+      "epoch": 0.23438920609521038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.788790075278018e-05,
+      "loss": 1.0789,
+      "step": 6445
+    },
+    {
+      "epoch": 0.23442557369894898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7887176710446023e-05,
+      "loss": 1.2265,
+      "step": 6446
+    },
+    {
+      "epoch": 0.23446194130268758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.788645255868848e-05,
+      "loss": 1.1896,
+      "step": 6447
+    },
+    {
+      "epoch": 0.23449830890642615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7885728297517596e-05,
+      "loss": 1.2432,
+      "step": 6448
+    },
+    {
+      "epoch": 0.23453467651016474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7885003926943422e-05,
+      "loss": 1.049,
+      "step": 6449
+    },
+    {
+      "epoch": 0.23457104411390334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7884279446976003e-05,
+      "loss": 1.0705,
+      "step": 6450
+    },
+    {
+      "epoch": 0.23460741171764193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7883554857625393e-05,
+      "loss": 1.0726,
+      "step": 6451
+    },
+    {
+      "epoch": 0.2346437793213805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7882830158901646e-05,
+      "loss": 1.1123,
+      "step": 6452
+    },
+    {
+      "epoch": 0.2346801469251191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.788210535081482e-05,
+      "loss": 1.1409,
+      "step": 6453
+    },
+    {
+      "epoch": 0.2347165145288577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.788138043337496e-05,
+      "loss": 1.1163,
+      "step": 6454
+    },
+    {
+      "epoch": 0.2347528821325963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7880655406592125e-05,
+      "loss": 1.0477,
+      "step": 6455
+    },
+    {
+      "epoch": 0.23478924973633486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7879930270476384e-05,
+      "loss": 1.0619,
+      "step": 6456
+    },
+    {
+      "epoch": 0.23482561734007346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787920502503779e-05,
+      "loss": 1.0607,
+      "step": 6457
+    },
+    {
+      "epoch": 0.23486198494381205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.78784796702864e-05,
+      "loss": 1.1616,
+      "step": 6458
+    },
+    {
+      "epoch": 0.23489835254755065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7877754206232286e-05,
+      "loss": 1.1266,
+      "step": 6459
+    },
+    {
+      "epoch": 0.23493472015128924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787702863288551e-05,
+      "loss": 0.986,
+      "step": 6460
+    },
+    {
+      "epoch": 0.2349710877550278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787630295025614e-05,
+      "loss": 1.0356,
+      "step": 6461
+    },
+    {
+      "epoch": 0.2350074553587664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7875577158354235e-05,
+      "loss": 1.0536,
+      "step": 6462
+    },
+    {
+      "epoch": 0.235043822962505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7874851257189875e-05,
+      "loss": 1.1611,
+      "step": 6463
+    },
+    {
+      "epoch": 0.2350801905662436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787412524677313e-05,
+      "loss": 1.0253,
+      "step": 6464
+    },
+    {
+      "epoch": 0.23511655816998217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7873399127114063e-05,
+      "loss": 1.1783,
+      "step": 6465
+    },
+    {
+      "epoch": 0.23515292577372077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787267289822276e-05,
+      "loss": 1.0852,
+      "step": 6466
+    },
+    {
+      "epoch": 0.23518929337745936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7871946560109284e-05,
+      "loss": 1.166,
+      "step": 6467
+    },
+    {
+      "epoch": 0.23522566098119796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787122011278372e-05,
+      "loss": 1.2179,
+      "step": 6468
+    },
+    {
+      "epoch": 0.23526202858493653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.787049355625615e-05,
+      "loss": 1.1057,
+      "step": 6469
+    },
+    {
+      "epoch": 0.23529839618867512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7869766890536647e-05,
+      "loss": 1.1776,
+      "step": 6470
+    },
+    {
+      "epoch": 0.23533476379241372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.786904011563529e-05,
+      "loss": 1.1274,
+      "step": 6471
+    },
+    {
+      "epoch": 0.23537113139615232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7868313231562167e-05,
+      "loss": 1.1649,
+      "step": 6472
+    },
+    {
+      "epoch": 0.23540749899989089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7867586238327366e-05,
+      "loss": 1.0911,
+      "step": 6473
+    },
+    {
+      "epoch": 0.23544386660362948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7866859135940963e-05,
+      "loss": 1.0374,
+      "step": 6474
+    },
+    {
+      "epoch": 0.23548023420736808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7866131924413052e-05,
+      "loss": 1.1,
+      "step": 6475
+    },
+    {
+      "epoch": 0.23551660181110667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7865404603753725e-05,
+      "loss": 1.0987,
+      "step": 6476
+    },
+    {
+      "epoch": 0.23555296941484524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7864677173973065e-05,
+      "loss": 1.1421,
+      "step": 6477
+    },
+    {
+      "epoch": 0.23558933701858384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.786394963508117e-05,
+      "loss": 1.0864,
+      "step": 6478
+    },
+    {
+      "epoch": 0.23562570462232243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.786322198708813e-05,
+      "loss": 1.2505,
+      "step": 6479
+    },
+    {
+      "epoch": 0.23566207222606103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7862494230004042e-05,
+      "loss": 1.0454,
+      "step": 6480
+    },
+    {
+      "epoch": 0.23569843982979963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7861766363839e-05,
+      "loss": 1.1392,
+      "step": 6481
+    },
+    {
+      "epoch": 0.2357348074335382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7861038388603108e-05,
+      "loss": 1.1449,
+      "step": 6482
+    },
+    {
+      "epoch": 0.2357711750372768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.786031030430646e-05,
+      "loss": 1.1376,
+      "step": 6483
+    },
+    {
+      "epoch": 0.2358075426410154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7859582110959156e-05,
+      "loss": 1.1442,
+      "step": 6484
+    },
+    {
+      "epoch": 0.23584391024475398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7858853808571303e-05,
+      "loss": 1.025,
+      "step": 6485
+    },
+    {
+      "epoch": 0.23588027784849255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7858125397153006e-05,
+      "loss": 1.0964,
+      "step": 6486
+    },
+    {
+      "epoch": 0.23591664545223115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.785739687671437e-05,
+      "loss": 1.1348,
+      "step": 6487
+    },
+    {
+      "epoch": 0.23595301305596975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7856668247265492e-05,
+      "loss": 1.0482,
+      "step": 6488
+    },
+    {
+      "epoch": 0.23598938065970834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.78559395088165e-05,
+      "loss": 0.9382,
+      "step": 6489
+    },
+    {
+      "epoch": 0.2360257482634469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7855210661377483e-05,
+      "loss": 1.0995,
+      "step": 6490
+    },
+    {
+      "epoch": 0.2360621158671855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7854481704958567e-05,
+      "loss": 1.0781,
+      "step": 6491
+    },
+    {
+      "epoch": 0.2360984834709241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7853752639569866e-05,
+      "loss": 0.9795,
+      "step": 6492
+    },
+    {
+      "epoch": 0.2361348510746627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7853023465221485e-05,
+      "loss": 1.0648,
+      "step": 6493
+    },
+    {
+      "epoch": 0.23617121867840127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.785229418192355e-05,
+      "loss": 1.0863,
+      "step": 6494
+    },
+    {
+      "epoch": 0.23620758628213986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7851564789686172e-05,
+      "loss": 1.1743,
+      "step": 6495
+    },
+    {
+      "epoch": 0.23624395388587846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.785083528851947e-05,
+      "loss": 1.1438,
+      "step": 6496
+    },
+    {
+      "epoch": 0.23628032148961706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.785010567843357e-05,
+      "loss": 1.0851,
+      "step": 6497
+    },
+    {
+      "epoch": 0.23631668909335563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.784937595943859e-05,
+      "loss": 0.9419,
+      "step": 6498
+    },
+    {
+      "epoch": 0.23635305669709422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7848646131544653e-05,
+      "loss": 1.2214,
+      "step": 6499
+    },
+    {
+      "epoch": 0.23638942430083282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7847916194761893e-05,
+      "loss": 1.1653,
+      "step": 6500
+    },
+    {
+      "epoch": 0.23642579190457141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7847186149100427e-05,
+      "loss": 1.0665,
+      "step": 6501
+    },
+    {
+      "epoch": 0.23646215950831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7846455994570386e-05,
+      "loss": 1.1727,
+      "step": 6502
+    },
+    {
+      "epoch": 0.23649852711204858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7845725731181902e-05,
+      "loss": 1.0523,
+      "step": 6503
+    },
+    {
+      "epoch": 0.23653489471578718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7844995358945105e-05,
+      "loss": 1.1848,
+      "step": 6504
+    },
+    {
+      "epoch": 0.23657126231952577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7844264877870126e-05,
+      "loss": 1.1122,
+      "step": 6505
+    },
+    {
+      "epoch": 0.23660762992326437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7843534287967105e-05,
+      "loss": 1.1703,
+      "step": 6506
+    },
+    {
+      "epoch": 0.23664399752700294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.784280358924617e-05,
+      "loss": 1.126,
+      "step": 6507
+    },
+    {
+      "epoch": 0.23668036513074153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7842072781717464e-05,
+      "loss": 1.2121,
+      "step": 6508
+    },
+    {
+      "epoch": 0.23671673273448013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.784134186539113e-05,
+      "loss": 1.143,
+      "step": 6509
+    },
+    {
+      "epoch": 0.23675310033821872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7840610840277298e-05,
+      "loss": 1.1545,
+      "step": 6510
+    },
+    {
+      "epoch": 0.2367894679419573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7839879706386116e-05,
+      "loss": 1.0939,
+      "step": 6511
+    },
+    {
+      "epoch": 0.2368258355456959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7839148463727728e-05,
+      "loss": 1.1715,
+      "step": 6512
+    },
+    {
+      "epoch": 0.23686220314943449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7838417112312274e-05,
+      "loss": 1.0421,
+      "step": 6513
+    },
+    {
+      "epoch": 0.23689857075317308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7837685652149907e-05,
+      "loss": 1.0213,
+      "step": 6514
+    },
+    {
+      "epoch": 0.23693493835691165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.783695408325077e-05,
+      "loss": 1.1224,
+      "step": 6515
+    },
+    {
+      "epoch": 0.23697130596065025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.783622240562501e-05,
+      "loss": 1.1898,
+      "step": 6516
+    },
+    {
+      "epoch": 0.23700767356438884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7835490619282788e-05,
+      "loss": 1.1583,
+      "step": 6517
+    },
+    {
+      "epoch": 0.23704404116812744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.783475872423425e-05,
+      "loss": 1.1599,
+      "step": 6518
+    },
+    {
+      "epoch": 0.237080408771866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.783402672048955e-05,
+      "loss": 0.982,
+      "step": 6519
+    },
+    {
+      "epoch": 0.2371167763756046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7833294608058845e-05,
+      "loss": 1.0112,
+      "step": 6520
+    },
+    {
+      "epoch": 0.2371531439793432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7832562386952293e-05,
+      "loss": 1.1552,
+      "step": 6521
+    },
+    {
+      "epoch": 0.2371895115830818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7831830057180046e-05,
+      "loss": 1.1219,
+      "step": 6522
+    },
+    {
+      "epoch": 0.2372258791868204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7831097618752274e-05,
+      "loss": 1.0334,
+      "step": 6523
+    },
+    {
+      "epoch": 0.23726224679055896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.783036507167913e-05,
+      "loss": 1.0774,
+      "step": 6524
+    },
+    {
+      "epoch": 0.23729861439429756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7829632415970783e-05,
+      "loss": 1.1692,
+      "step": 6525
+    },
+    {
+      "epoch": 0.23733498199803615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7828899651637393e-05,
+      "loss": 1.0368,
+      "step": 6526
+    },
+    {
+      "epoch": 0.23737134960177475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7828166778689125e-05,
+      "loss": 1.0239,
+      "step": 6527
+    },
+    {
+      "epoch": 0.23740771720551332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7827433797136156e-05,
+      "loss": 1.0246,
+      "step": 6528
+    },
+    {
+      "epoch": 0.23744408480925192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7826700706988646e-05,
+      "loss": 1.1425,
+      "step": 6529
+    },
+    {
+      "epoch": 0.2374804524129905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7825967508256768e-05,
+      "loss": 1.1564,
+      "step": 6530
+    },
+    {
+      "epoch": 0.2375168200167291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7825234200950692e-05,
+      "loss": 1.1121,
+      "step": 6531
+    },
+    {
+      "epoch": 0.23755318762046768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7824500785080597e-05,
+      "loss": 1.1246,
+      "step": 6532
+    },
+    {
+      "epoch": 0.23758955522420627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7823767260656656e-05,
+      "loss": 1.1924,
+      "step": 6533
+    },
+    {
+      "epoch": 0.23762592282794487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7823033627689045e-05,
+      "loss": 1.0692,
+      "step": 6534
+    },
+    {
+      "epoch": 0.23766229043168347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.782229988618794e-05,
+      "loss": 1.0961,
+      "step": 6535
+    },
+    {
+      "epoch": 0.23769865803542203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7821566036163522e-05,
+      "loss": 1.0304,
+      "step": 6536
+    },
+    {
+      "epoch": 0.23773502563916063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7820832077625977e-05,
+      "loss": 1.2112,
+      "step": 6537
+    },
+    {
+      "epoch": 0.23777139324289923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7820098010585476e-05,
+      "loss": 1.1521,
+      "step": 6538
+    },
+    {
+      "epoch": 0.23780776084663782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7819363835052215e-05,
+      "loss": 1.1385,
+      "step": 6539
+    },
+    {
+      "epoch": 0.2378441284503764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7818629551036374e-05,
+      "loss": 1.0735,
+      "step": 6540
+    },
+    {
+      "epoch": 0.237880496054115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7817895158548143e-05,
+      "loss": 1.1362,
+      "step": 6541
+    },
+    {
+      "epoch": 0.23791686365785358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7817160657597705e-05,
+      "loss": 1.088,
+      "step": 6542
+    },
+    {
+      "epoch": 0.23795323126159218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7816426048195255e-05,
+      "loss": 1.0436,
+      "step": 6543
+    },
+    {
+      "epoch": 0.23798959886533078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7815691330350986e-05,
+      "loss": 1.1039,
+      "step": 6544
+    },
+    {
+      "epoch": 0.23802596646906934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7814956504075088e-05,
+      "loss": 1.1889,
+      "step": 6545
+    },
+    {
+      "epoch": 0.23806233407280794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7814221569377753e-05,
+      "loss": 1.0634,
+      "step": 6546
+    },
+    {
+      "epoch": 0.23809870167654654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7813486526269186e-05,
+      "loss": 1.0598,
+      "step": 6547
+    },
+    {
+      "epoch": 0.23813506928028513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.781275137475958e-05,
+      "loss": 1.1987,
+      "step": 6548
+    },
+    {
+      "epoch": 0.2381714368840237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7812016114859127e-05,
+      "loss": 1.1683,
+      "step": 6549
+    },
+    {
+      "epoch": 0.2382078044877623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.781128074657804e-05,
+      "loss": 1.1364,
+      "step": 6550
+    },
+    {
+      "epoch": 0.2382441720915009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.781054526992651e-05,
+      "loss": 1.0432,
+      "step": 6551
+    },
+    {
+      "epoch": 0.2382805396952395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7809809684914753e-05,
+      "loss": 1.1851,
+      "step": 6552
+    },
+    {
+      "epoch": 0.23831690729897806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.780907399155296e-05,
+      "loss": 1.0269,
+      "step": 6553
+    },
+    {
+      "epoch": 0.23835327490271666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.780833818985135e-05,
+      "loss": 1.1007,
+      "step": 6554
+    },
+    {
+      "epoch": 0.23838964250645525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7807602279820123e-05,
+      "loss": 1.1137,
+      "step": 6555
+    },
+    {
+      "epoch": 0.23842601011019385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7806866261469497e-05,
+      "loss": 1.1875,
+      "step": 6556
+    },
+    {
+      "epoch": 0.23846237771393242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7806130134809675e-05,
+      "loss": 1.1663,
+      "step": 6557
+    },
+    {
+      "epoch": 0.238498745317671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.780539389985087e-05,
+      "loss": 1.0918,
+      "step": 6558
+    },
+    {
+      "epoch": 0.2385351129214096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.78046575566033e-05,
+      "loss": 1.1262,
+      "step": 6559
+    },
+    {
+      "epoch": 0.2385714805251482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7803921105077183e-05,
+      "loss": 1.2014,
+      "step": 6560
+    },
+    {
+      "epoch": 0.2386078481288868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7803184545282734e-05,
+      "loss": 1.1245,
+      "step": 6561
+    },
+    {
+      "epoch": 0.23864421573262537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7802447877230166e-05,
+      "loss": 1.1493,
+      "step": 6562
+    },
+    {
+      "epoch": 0.23868058333636397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7801711100929708e-05,
+      "loss": 1.1074,
+      "step": 6563
+    },
+    {
+      "epoch": 0.23871695094010256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7800974216391575e-05,
+      "loss": 1.1171,
+      "step": 6564
+    },
+    {
+      "epoch": 0.23875331854384116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7800237223625992e-05,
+      "loss": 1.0907,
+      "step": 6565
+    },
+    {
+      "epoch": 0.23878968614757973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7799500122643186e-05,
+      "loss": 1.0319,
+      "step": 6566
+    },
+    {
+      "epoch": 0.23882605375131832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.779876291345338e-05,
+      "loss": 1.2588,
+      "step": 6567
+    },
+    {
+      "epoch": 0.23886242135505692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.779802559606681e-05,
+      "loss": 1.0883,
+      "step": 6568
+    },
+    {
+      "epoch": 0.23889878895879552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7797288170493692e-05,
+      "loss": 1.1608,
+      "step": 6569
+    },
+    {
+      "epoch": 0.23893515656253408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7796550636744267e-05,
+      "loss": 1.076,
+      "step": 6570
+    },
+    {
+      "epoch": 0.23897152416627268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.779581299482876e-05,
+      "loss": 1.0428,
+      "step": 6571
+    },
+    {
+      "epoch": 0.23900789177001128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.779507524475741e-05,
+      "loss": 0.9842,
+      "step": 6572
+    },
+    {
+      "epoch": 0.23904425937374987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7794337386540454e-05,
+      "loss": 1.056,
+      "step": 6573
+    },
+    {
+      "epoch": 0.23908062697748844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7793599420188125e-05,
+      "loss": 1.1494,
+      "step": 6574
+    },
+    {
+      "epoch": 0.23911699458122704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.779286134571066e-05,
+      "loss": 1.0092,
+      "step": 6575
+    },
+    {
+      "epoch": 0.23915336218496563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.77921231631183e-05,
+      "loss": 1.166,
+      "step": 6576
+    },
+    {
+      "epoch": 0.23918972978870423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7791384872421286e-05,
+      "loss": 1.1811,
+      "step": 6577
+    },
+    {
+      "epoch": 0.2392260973924428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.779064647362986e-05,
+      "loss": 1.0345,
+      "step": 6578
+    },
+    {
+      "epoch": 0.2392624649961814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7789907966754274e-05,
+      "loss": 0.9239,
+      "step": 6579
+    },
+    {
+      "epoch": 0.23929883259992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778916935180476e-05,
+      "loss": 1.0028,
+      "step": 6580
+    },
+    {
+      "epoch": 0.2393352002036586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778843062879158e-05,
+      "loss": 1.114,
+      "step": 6581
+    },
+    {
+      "epoch": 0.23937156780739718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7787691797724972e-05,
+      "loss": 1.0868,
+      "step": 6582
+    },
+    {
+      "epoch": 0.23940793541113575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778695285861519e-05,
+      "loss": 1.1461,
+      "step": 6583
+    },
+    {
+      "epoch": 0.23944430301487435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7786213811472483e-05,
+      "loss": 1.1005,
+      "step": 6584
+    },
+    {
+      "epoch": 0.23948067061861295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7785474656307108e-05,
+      "loss": 1.0231,
+      "step": 6585
+    },
+    {
+      "epoch": 0.23951703822235154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7784735393129317e-05,
+      "loss": 1.0628,
+      "step": 6586
+    },
+    {
+      "epoch": 0.2395534058260901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778399602194937e-05,
+      "loss": 1.1317,
+      "step": 6587
+    },
+    {
+      "epoch": 0.2395897734298287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7783256542777523e-05,
+      "loss": 1.1188,
+      "step": 6588
+    },
+    {
+      "epoch": 0.2396261410335673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7782516955624033e-05,
+      "loss": 1.1076,
+      "step": 6589
+    },
+    {
+      "epoch": 0.2396625086373059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7781777260499162e-05,
+      "loss": 1.1306,
+      "step": 6590
+    },
+    {
+      "epoch": 0.23969887624104447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778103745741317e-05,
+      "loss": 1.1321,
+      "step": 6591
+    },
+    {
+      "epoch": 0.23973524384478306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7780297546376327e-05,
+      "loss": 1.1719,
+      "step": 6592
+    },
+    {
+      "epoch": 0.23977161144852166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7779557527398892e-05,
+      "loss": 1.1154,
+      "step": 6593
+    },
+    {
+      "epoch": 0.23980797905226026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7778817400491136e-05,
+      "loss": 1.1783,
+      "step": 6594
+    },
+    {
+      "epoch": 0.23984434665599882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7778077165663323e-05,
+      "loss": 1.1212,
+      "step": 6595
+    },
+    {
+      "epoch": 0.23988071425973742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7777336822925727e-05,
+      "loss": 1.0359,
+      "step": 6596
+    },
+    {
+      "epoch": 0.23991708186347602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7776596372288616e-05,
+      "loss": 1.144,
+      "step": 6597
+    },
+    {
+      "epoch": 0.2399534494672146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7775855813762263e-05,
+      "loss": 1.1059,
+      "step": 6598
+    },
+    {
+      "epoch": 0.23998981707095318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7775115147356946e-05,
+      "loss": 1.0209,
+      "step": 6599
+    },
+    {
+      "epoch": 0.24002618467469178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7774374373082935e-05,
+      "loss": 1.1844,
+      "step": 6600
+    },
+    {
+      "epoch": 0.24006255227843037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777363349095051e-05,
+      "loss": 1.1642,
+      "step": 6601
+    },
+    {
+      "epoch": 0.24009891988216897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777289250096995e-05,
+      "loss": 0.9343,
+      "step": 6602
+    },
+    {
+      "epoch": 0.24013528748590757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7772151403151537e-05,
+      "loss": 1.1315,
+      "step": 6603
+    },
+    {
+      "epoch": 0.24017165508964614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777141019750555e-05,
+      "loss": 1.1077,
+      "step": 6604
+    },
+    {
+      "epoch": 0.24020802269338473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777066888404227e-05,
+      "loss": 1.1009,
+      "step": 6605
+    },
+    {
+      "epoch": 0.24024439029712333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7769927462771986e-05,
+      "loss": 1.1092,
+      "step": 6606
+    },
+    {
+      "epoch": 0.24028075790086192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7769185933704985e-05,
+      "loss": 1.1797,
+      "step": 6607
+    },
+    {
+      "epoch": 0.2403171255046005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7768444296851547e-05,
+      "loss": 1.0839,
+      "step": 6608
+    },
+    {
+      "epoch": 0.2403534931083391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.776770255222197e-05,
+      "loss": 1.0905,
+      "step": 6609
+    },
+    {
+      "epoch": 0.24038986071207769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7766960699826537e-05,
+      "loss": 1.0892,
+      "step": 6610
+    },
+    {
+      "epoch": 0.24042622831581628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.776621873967555e-05,
+      "loss": 1.1017,
+      "step": 6611
+    },
+    {
+      "epoch": 0.24046259591955485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7765476671779295e-05,
+      "loss": 1.1219,
+      "step": 6612
+    },
+    {
+      "epoch": 0.24049896352329345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7764734496148067e-05,
+      "loss": 1.1784,
+      "step": 6613
+    },
+    {
+      "epoch": 0.24053533112703204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7763992212792165e-05,
+      "loss": 1.0229,
+      "step": 6614
+    },
+    {
+      "epoch": 0.24057169873077064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.776324982172189e-05,
+      "loss": 1.053,
+      "step": 6615
+    },
+    {
+      "epoch": 0.2406080663345092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7762507322947534e-05,
+      "loss": 1.145,
+      "step": 6616
+    },
+    {
+      "epoch": 0.2406444339382478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7761764716479405e-05,
+      "loss": 1.233,
+      "step": 6617
+    },
+    {
+      "epoch": 0.2406808015419864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.77610220023278e-05,
+      "loss": 1.1203,
+      "step": 6618
+    },
+    {
+      "epoch": 0.240717169145725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7760279180503032e-05,
+      "loss": 1.1024,
+      "step": 6619
+    },
+    {
+      "epoch": 0.24075353674946356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7759536251015395e-05,
+      "loss": 1.0342,
+      "step": 6620
+    },
+    {
+      "epoch": 0.24078990435320216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7758793213875204e-05,
+      "loss": 1.0476,
+      "step": 6621
+    },
+    {
+      "epoch": 0.24082627195694076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7758050069092764e-05,
+      "loss": 1.0647,
+      "step": 6622
+    },
+    {
+      "epoch": 0.24086263956067935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.775730681667839e-05,
+      "loss": 1.1336,
+      "step": 6623
+    },
+    {
+      "epoch": 0.24089900716441795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7756563456642388e-05,
+      "loss": 1.1861,
+      "step": 6624
+    },
+    {
+      "epoch": 0.24093537476815652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7755819988995072e-05,
+      "loss": 1.102,
+      "step": 6625
+    },
+    {
+      "epoch": 0.24097174237189511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7755076413746758e-05,
+      "loss": 0.9841,
+      "step": 6626
+    },
+    {
+      "epoch": 0.2410081099756337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7754332730907764e-05,
+      "loss": 1.0731,
+      "step": 6627
+    },
+    {
+      "epoch": 0.2410444775793723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7753588940488406e-05,
+      "loss": 1.1085,
+      "step": 6628
+    },
+    {
+      "epoch": 0.24108084518311088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7752845042498997e-05,
+      "loss": 0.9699,
+      "step": 6629
+    },
+    {
+      "epoch": 0.24111721278684947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.775210103694987e-05,
+      "loss": 1.0686,
+      "step": 6630
+    },
+    {
+      "epoch": 0.24115358039058807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7751356923851335e-05,
+      "loss": 1.0733,
+      "step": 6631
+    },
+    {
+      "epoch": 0.24118994799432666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.775061270321372e-05,
+      "loss": 1.0572,
+      "step": 6632
+    },
+    {
+      "epoch": 0.24122631559806523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774986837504735e-05,
+      "loss": 1.0876,
+      "step": 6633
+    },
+    {
+      "epoch": 0.24126268320180383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774912393936256e-05,
+      "loss": 1.2046,
+      "step": 6634
+    },
+    {
+      "epoch": 0.24129905080554243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774837939616966e-05,
+      "loss": 1.1044,
+      "step": 6635
+    },
+    {
+      "epoch": 0.24133541840928102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7747634745478995e-05,
+      "loss": 1.0724,
+      "step": 6636
+    },
+    {
+      "epoch": 0.2413717860130196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774688998730089e-05,
+      "loss": 1.1044,
+      "step": 6637
+    },
+    {
+      "epoch": 0.2414081536167582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7746145121645674e-05,
+      "loss": 1.1931,
+      "step": 6638
+    },
+    {
+      "epoch": 0.24144452122049678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7745400148523688e-05,
+      "loss": 1.1237,
+      "step": 6639
+    },
+    {
+      "epoch": 0.24148088882423538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7744655067945263e-05,
+      "loss": 1.1005,
+      "step": 6640
+    },
+    {
+      "epoch": 0.24151725642797395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774390987992074e-05,
+      "loss": 1.0748,
+      "step": 6641
+    },
+    {
+      "epoch": 0.24155362403171254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774316458446045e-05,
+      "loss": 1.1349,
+      "step": 6642
+    },
+    {
+      "epoch": 0.24158999163545114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774241918157474e-05,
+      "loss": 1.099,
+      "step": 6643
+    },
+    {
+      "epoch": 0.24162635923918974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7741673671273945e-05,
+      "loss": 1.1125,
+      "step": 6644
+    },
+    {
+      "epoch": 0.24166272684292833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7740928053568418e-05,
+      "loss": 1.0939,
+      "step": 6645
+    },
+    {
+      "epoch": 0.2416990944466669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7740182328468495e-05,
+      "loss": 1.1879,
+      "step": 6646
+    },
+    {
+      "epoch": 0.2417354620504055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7739436495984527e-05,
+      "loss": 1.0427,
+      "step": 6647
+    },
+    {
+      "epoch": 0.2417718296541441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7738690556126853e-05,
+      "loss": 1.0362,
+      "step": 6648
+    },
+    {
+      "epoch": 0.2418081972578827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773794450890583e-05,
+      "loss": 1.1251,
+      "step": 6649
+    },
+    {
+      "epoch": 0.24184456486162126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7737198354331803e-05,
+      "loss": 1.1525,
+      "step": 6650
+    },
+    {
+      "epoch": 0.24188093246535985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7736452092415127e-05,
+      "loss": 1.0475,
+      "step": 6651
+    },
+    {
+      "epoch": 0.24191730006909845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773570572316616e-05,
+      "loss": 0.9489,
+      "step": 6652
+    },
+    {
+      "epoch": 0.24195366767283705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773495924659525e-05,
+      "loss": 1.2771,
+      "step": 6653
+    },
+    {
+      "epoch": 0.24199003527657562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773421266271275e-05,
+      "loss": 1.0453,
+      "step": 6654
+    },
+    {
+      "epoch": 0.2420264028803142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7733465971529026e-05,
+      "loss": 1.1347,
+      "step": 6655
+    },
+    {
+      "epoch": 0.2420627704840528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7732719173054434e-05,
+      "loss": 1.0293,
+      "step": 6656
+    },
+    {
+      "epoch": 0.2420991380877914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7731972267299335e-05,
+      "loss": 1.0141,
+      "step": 6657
+    },
+    {
+      "epoch": 0.24213550569152997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773122525427409e-05,
+      "loss": 1.0546,
+      "step": 6658
+    },
+    {
+      "epoch": 0.24217187329526857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773047813398906e-05,
+      "loss": 1.1198,
+      "step": 6659
+    },
+    {
+      "epoch": 0.24220824089900717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7729730906454622e-05,
+      "loss": 1.1276,
+      "step": 6660
+    },
+    {
+      "epoch": 0.24224460850274576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.772898357168113e-05,
+      "loss": 1.0747,
+      "step": 6661
+    },
+    {
+      "epoch": 0.24228097610648433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7728236129678958e-05,
+      "loss": 1.0634,
+      "step": 6662
+    },
+    {
+      "epoch": 0.24231734371022293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7727488580458475e-05,
+      "loss": 1.1947,
+      "step": 6663
+    },
+    {
+      "epoch": 0.24235371131396152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7726740924030052e-05,
+      "loss": 1.1011,
+      "step": 6664
+    },
+    {
+      "epoch": 0.24239007891770012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.772599316040406e-05,
+      "loss": 1.1506,
+      "step": 6665
+    },
+    {
+      "epoch": 0.24242644652143872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7725245289590875e-05,
+      "loss": 1.0795,
+      "step": 6666
+    },
+    {
+      "epoch": 0.24246281412517728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.772449731160087e-05,
+      "loss": 1.0695,
+      "step": 6667
+    },
+    {
+      "epoch": 0.24249918172891588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7723749226444428e-05,
+      "loss": 1.0479,
+      "step": 6668
+    },
+    {
+      "epoch": 0.24253554933265448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7723001034131927e-05,
+      "loss": 1.034,
+      "step": 6669
+    },
+    {
+      "epoch": 0.24257191693639307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.772225273467374e-05,
+      "loss": 1.195,
+      "step": 6670
+    },
+    {
+      "epoch": 0.24260828454013164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.772150432808025e-05,
+      "loss": 0.9684,
+      "step": 6671
+    },
+    {
+      "epoch": 0.24264465214387024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7720755814361848e-05,
+      "loss": 0.9317,
+      "step": 6672
+    },
+    {
+      "epoch": 0.24268101974760883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.772000719352891e-05,
+      "loss": 1.0504,
+      "step": 6673
+    },
+    {
+      "epoch": 0.24271738735134743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771925846559183e-05,
+      "loss": 1.1421,
+      "step": 6674
+    },
+    {
+      "epoch": 0.242753754955086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7718509630560987e-05,
+      "loss": 1.1427,
+      "step": 6675
+    },
+    {
+      "epoch": 0.2427901225588246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7717760688446776e-05,
+      "loss": 1.1312,
+      "step": 6676
+    },
+    {
+      "epoch": 0.2428264901625632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7717011639259584e-05,
+      "loss": 1.117,
+      "step": 6677
+    },
+    {
+      "epoch": 0.2428628577663018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7716262483009802e-05,
+      "loss": 1.1218,
+      "step": 6678
+    },
+    {
+      "epoch": 0.24289922537004036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771551321970783e-05,
+      "loss": 0.9218,
+      "step": 6679
+    },
+    {
+      "epoch": 0.24293559297377895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7714763849364057e-05,
+      "loss": 1.1772,
+      "step": 6680
+    },
+    {
+      "epoch": 0.24297196057751755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7714014371988883e-05,
+      "loss": 1.1616,
+      "step": 6681
+    },
+    {
+      "epoch": 0.24300832818125614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7713264787592702e-05,
+      "loss": 1.1406,
+      "step": 6682
+    },
+    {
+      "epoch": 0.2430446957849947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771251509618592e-05,
+      "loss": 1.0086,
+      "step": 6683
+    },
+    {
+      "epoch": 0.2430810633887333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771176529777893e-05,
+      "loss": 1.0512,
+      "step": 6684
+    },
+    {
+      "epoch": 0.2431174309924719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771101539238214e-05,
+      "loss": 0.9948,
+      "step": 6685
+    },
+    {
+      "epoch": 0.2431537985962105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.771026538000595e-05,
+      "loss": 1.126,
+      "step": 6686
+    },
+    {
+      "epoch": 0.2431901661999491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7709515260660767e-05,
+      "loss": 1.0943,
+      "step": 6687
+    },
+    {
+      "epoch": 0.24322653380368767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7708765034357e-05,
+      "loss": 1.1608,
+      "step": 6688
+    },
+    {
+      "epoch": 0.24326290140742626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7708014701105053e-05,
+      "loss": 1.0842,
+      "step": 6689
+    },
+    {
+      "epoch": 0.24329926901116486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7707264260915344e-05,
+      "loss": 1.1268,
+      "step": 6690
+    },
+    {
+      "epoch": 0.24333563661490346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7706513713798273e-05,
+      "loss": 1.0508,
+      "step": 6691
+    },
+    {
+      "epoch": 0.24337200421864202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.770576305976426e-05,
+      "loss": 1.0757,
+      "step": 6692
+    },
+    {
+      "epoch": 0.24340837182238062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7705012298823717e-05,
+      "loss": 1.0423,
+      "step": 6693
+    },
+    {
+      "epoch": 0.24344473942611922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7704261430987063e-05,
+      "loss": 1.1965,
+      "step": 6694
+    },
+    {
+      "epoch": 0.2434811070298578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.770351045626471e-05,
+      "loss": 1.0763,
+      "step": 6695
+    },
+    {
+      "epoch": 0.24351747463359638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7702759374667078e-05,
+      "loss": 1.019,
+      "step": 6696
+    },
+    {
+      "epoch": 0.24355384223733498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.770200818620459e-05,
+      "loss": 1.0333,
+      "step": 6697
+    },
+    {
+      "epoch": 0.24359020984107357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7701256890887667e-05,
+      "loss": 1.0811,
+      "step": 6698
+    },
+    {
+      "epoch": 0.24362657744481217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7700505488726733e-05,
+      "loss": 1.1622,
+      "step": 6699
+    },
+    {
+      "epoch": 0.24366294504855074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.769975397973221e-05,
+      "loss": 1.0536,
+      "step": 6700
+    },
+    {
+      "epoch": 0.24369931265228933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7699002363914522e-05,
+      "loss": 1.0803,
+      "step": 6701
+    },
+    {
+      "epoch": 0.24373568025602793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7698250641284097e-05,
+      "loss": 1.1196,
+      "step": 6702
+    },
+    {
+      "epoch": 0.24377204785976653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7697498811851375e-05,
+      "loss": 1.1797,
+      "step": 6703
+    },
+    {
+      "epoch": 0.2438084154635051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7696746875626772e-05,
+      "loss": 0.906,
+      "step": 6704
+    },
+    {
+      "epoch": 0.2438447830672437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.769599483262073e-05,
+      "loss": 1.0474,
+      "step": 6705
+    },
+    {
+      "epoch": 0.2438811506709823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.769524268284368e-05,
+      "loss": 1.0972,
+      "step": 6706
+    },
+    {
+      "epoch": 0.24391751827472088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.769449042630605e-05,
+      "loss": 1.1569,
+      "step": 6707
+    },
+    {
+      "epoch": 0.24395388587845948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7693738063018286e-05,
+      "loss": 1.077,
+      "step": 6708
+    },
+    {
+      "epoch": 0.24399025348219805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7692985592990823e-05,
+      "loss": 1.089,
+      "step": 6709
+    },
+    {
+      "epoch": 0.24402662108593665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7692233016234098e-05,
+      "loss": 1.0943,
+      "step": 6710
+    },
+    {
+      "epoch": 0.24406298868967524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7691480332758554e-05,
+      "loss": 1.0353,
+      "step": 6711
+    },
+    {
+      "epoch": 0.24409935629341384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7690727542574636e-05,
+      "loss": 1.0125,
+      "step": 6712
+    },
+    {
+      "epoch": 0.2441357238971524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7689974645692786e-05,
+      "loss": 1.1713,
+      "step": 6713
+    },
+    {
+      "epoch": 0.244172091500891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7689221642123444e-05,
+      "loss": 1.0954,
+      "step": 6714
+    },
+    {
+      "epoch": 0.2442084591046296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.768846853187706e-05,
+      "loss": 1.2283,
+      "step": 6715
+    },
+    {
+      "epoch": 0.2442448267083682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.768771531496409e-05,
+      "loss": 1.1425,
+      "step": 6716
+    },
+    {
+      "epoch": 0.24428119431210676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7686961991394973e-05,
+      "loss": 0.9631,
+      "step": 6717
+    },
+    {
+      "epoch": 0.24431756191584536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7686208561180163e-05,
+      "loss": 1.1166,
+      "step": 6718
+    },
+    {
+      "epoch": 0.24435392951958396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.768545502433012e-05,
+      "loss": 1.0016,
+      "step": 6719
+    },
+    {
+      "epoch": 0.24439029712332255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7684701380855286e-05,
+      "loss": 1.1419,
+      "step": 6720
+    },
+    {
+      "epoch": 0.24442666472706112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.768394763076613e-05,
+      "loss": 1.1844,
+      "step": 6721
+    },
+    {
+      "epoch": 0.24446303233079972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7683193774073098e-05,
+      "loss": 1.2083,
+      "step": 6722
+    },
+    {
+      "epoch": 0.24449939993453831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7682439810786654e-05,
+      "loss": 0.9955,
+      "step": 6723
+    },
+    {
+      "epoch": 0.2445357675382769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7681685740917262e-05,
+      "loss": 1.0565,
+      "step": 6724
+    },
+    {
+      "epoch": 0.24457213514201548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7680931564475377e-05,
+      "loss": 1.0666,
+      "step": 6725
+    },
+    {
+      "epoch": 0.24460850274575408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7680177281471462e-05,
+      "loss": 1.0609,
+      "step": 6726
+    },
+    {
+      "epoch": 0.24464487034949267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7679422891915985e-05,
+      "loss": 1.1379,
+      "step": 6727
+    },
+    {
+      "epoch": 0.24468123795323127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7678668395819414e-05,
+      "loss": 0.9566,
+      "step": 6728
+    },
+    {
+      "epoch": 0.24471760555696986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.767791379319221e-05,
+      "loss": 1.1298,
+      "step": 6729
+    },
+    {
+      "epoch": 0.24475397316070843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7677159084044847e-05,
+      "loss": 1.1453,
+      "step": 6730
+    },
+    {
+      "epoch": 0.24479034076444703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7676404268387793e-05,
+      "loss": 1.1943,
+      "step": 6731
+    },
+    {
+      "epoch": 0.24482670836818562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7675649346231525e-05,
+      "loss": 1.0425,
+      "step": 6732
+    },
+    {
+      "epoch": 0.24486307597192422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7674894317586512e-05,
+      "loss": 1.2364,
+      "step": 6733
+    },
+    {
+      "epoch": 0.2448994435756628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7674139182463227e-05,
+      "loss": 1.0364,
+      "step": 6734
+    },
+    {
+      "epoch": 0.24493581117940139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7673383940872146e-05,
+      "loss": 1.1964,
+      "step": 6735
+    },
+    {
+      "epoch": 0.24497217878313998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7672628592823754e-05,
+      "loss": 1.0743,
+      "step": 6736
+    },
+    {
+      "epoch": 0.24500854638687858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7671873138328526e-05,
+      "loss": 1.2236,
+      "step": 6737
+    },
+    {
+      "epoch": 0.24504491399061715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7671117577396943e-05,
+      "loss": 1.1494,
+      "step": 6738
+    },
+    {
+      "epoch": 0.24508128159435574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.767036191003949e-05,
+      "loss": 1.1992,
+      "step": 6739
+    },
+    {
+      "epoch": 0.24511764919809434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7669606136266643e-05,
+      "loss": 1.0944,
+      "step": 6740
+    },
+    {
+      "epoch": 0.24515401680183294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7668850256088895e-05,
+      "loss": 1.1216,
+      "step": 6741
+    },
+    {
+      "epoch": 0.2451903844055715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.766809426951673e-05,
+      "loss": 0.991,
+      "step": 6742
+    },
+    {
+      "epoch": 0.2452267520093101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7667338176560637e-05,
+      "loss": 1.2402,
+      "step": 6743
+    },
+    {
+      "epoch": 0.2452631196130487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.766658197723111e-05,
+      "loss": 1.0384,
+      "step": 6744
+    },
+    {
+      "epoch": 0.2452994872167873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7665825671538628e-05,
+      "loss": 1.0838,
+      "step": 6745
+    },
+    {
+      "epoch": 0.2453358548205259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7665069259493693e-05,
+      "loss": 1.1407,
+      "step": 6746
+    },
+    {
+      "epoch": 0.24537222242426446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.76643127411068e-05,
+      "loss": 1.1014,
+      "step": 6747
+    },
+    {
+      "epoch": 0.24540859002800305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7663556116388444e-05,
+      "loss": 1.0573,
+      "step": 6748
+    },
+    {
+      "epoch": 0.24544495763174165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7662799385349115e-05,
+      "loss": 1.155,
+      "step": 6749
+    },
+    {
+      "epoch": 0.24548132523548025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.766204254799932e-05,
+      "loss": 1.13,
+      "step": 6750
+    },
+    {
+      "epoch": 0.24551769283921882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7661285604349553e-05,
+      "loss": 1.1792,
+      "step": 6751
+    },
+    {
+      "epoch": 0.2455540604429574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.766052855441032e-05,
+      "loss": 1.06,
+      "step": 6752
+    },
+    {
+      "epoch": 0.245590428046696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765977139819212e-05,
+      "loss": 1.1339,
+      "step": 6753
+    },
+    {
+      "epoch": 0.2456267956504346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7659014135705462e-05,
+      "loss": 1.0367,
+      "step": 6754
+    },
+    {
+      "epoch": 0.24566316325417317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7658256766960845e-05,
+      "loss": 1.0518,
+      "step": 6755
+    },
+    {
+      "epoch": 0.24569953085791177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7657499291968786e-05,
+      "loss": 1.0861,
+      "step": 6756
+    },
+    {
+      "epoch": 0.24573589846165036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7656741710739788e-05,
+      "loss": 1.1757,
+      "step": 6757
+    },
+    {
+      "epoch": 0.24577226606538896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765598402328436e-05,
+      "loss": 1.0424,
+      "step": 6758
+    },
+    {
+      "epoch": 0.24580863366912753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765522622961302e-05,
+      "loss": 1.107,
+      "step": 6759
+    },
+    {
+      "epoch": 0.24584500127286613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7654468329736277e-05,
+      "loss": 1.1693,
+      "step": 6760
+    },
+    {
+      "epoch": 0.24588136887660472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7653710323664643e-05,
+      "loss": 1.1466,
+      "step": 6761
+    },
+    {
+      "epoch": 0.24591773648034332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765295221140864e-05,
+      "loss": 1.0993,
+      "step": 6762
+    },
+    {
+      "epoch": 0.2459541040840819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.765219399297878e-05,
+      "loss": 1.0271,
+      "step": 6763
+    },
+    {
+      "epoch": 0.24599047168782048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7651435668385592e-05,
+      "loss": 1.1488,
+      "step": 6764
+    },
+    {
+      "epoch": 0.24602683929155908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7650677237639583e-05,
+      "loss": 0.9697,
+      "step": 6765
+    },
+    {
+      "epoch": 0.24606320689529768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7649918700751287e-05,
+      "loss": 1.0658,
+      "step": 6766
+    },
+    {
+      "epoch": 0.24609957449903627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.764916005773122e-05,
+      "loss": 1.0387,
+      "step": 6767
+    },
+    {
+      "epoch": 0.24613594210277484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.764840130858991e-05,
+      "loss": 1.1306,
+      "step": 6768
+    },
+    {
+      "epoch": 0.24617230970651344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7647642453337883e-05,
+      "loss": 1.2213,
+      "step": 6769
+    },
+    {
+      "epoch": 0.24620867731025203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.764688349198567e-05,
+      "loss": 1.1142,
+      "step": 6770
+    },
+    {
+      "epoch": 0.24624504491399063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7646124424543795e-05,
+      "loss": 1.002,
+      "step": 6771
+    },
+    {
+      "epoch": 0.2462814125177292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7645365251022794e-05,
+      "loss": 1.2159,
+      "step": 6772
+    },
+    {
+      "epoch": 0.2463177801214678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7644605971433197e-05,
+      "loss": 1.0349,
+      "step": 6773
+    },
+    {
+      "epoch": 0.2463541477252064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7643846585785538e-05,
+      "loss": 1.0723,
+      "step": 6774
+    },
+    {
+      "epoch": 0.246390515328945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7643087094090353e-05,
+      "loss": 1.246,
+      "step": 6775
+    },
+    {
+      "epoch": 0.24642688293268356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7642327496358183e-05,
+      "loss": 0.8767,
+      "step": 6776
+    },
+    {
+      "epoch": 0.24646325053642215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7641567792599558e-05,
+      "loss": 1.0845,
+      "step": 6777
+    },
+    {
+      "epoch": 0.24649961814016075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.764080798282502e-05,
+      "loss": 1.0245,
+      "step": 6778
+    },
+    {
+      "epoch": 0.24653598574389934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7640048067045116e-05,
+      "loss": 1.1031,
+      "step": 6779
+    },
+    {
+      "epoch": 0.2465723533476379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7639288045270383e-05,
+      "loss": 1.0708,
+      "step": 6780
+    },
+    {
+      "epoch": 0.2466087209513765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7638527917511365e-05,
+      "loss": 1.1111,
+      "step": 6781
+    },
+    {
+      "epoch": 0.2466450885551151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763776768377861e-05,
+      "loss": 1.0421,
+      "step": 6782
+    },
+    {
+      "epoch": 0.2466814561588537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763700734408267e-05,
+      "loss": 1.1223,
+      "step": 6783
+    },
+    {
+      "epoch": 0.24671782376259227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7636246898434083e-05,
+      "loss": 1.0037,
+      "step": 6784
+    },
+    {
+      "epoch": 0.24675419136633087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763548634684341e-05,
+      "loss": 1.0581,
+      "step": 6785
+    },
+    {
+      "epoch": 0.24679055897006946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7634725689321193e-05,
+      "loss": 1.121,
+      "step": 6786
+    },
+    {
+      "epoch": 0.24682692657380806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7633964925877992e-05,
+      "loss": 0.9993,
+      "step": 6787
+    },
+    {
+      "epoch": 0.24686329417754665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763320405652436e-05,
+      "loss": 1.1494,
+      "step": 6788
+    },
+    {
+      "epoch": 0.24689966178128522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7632443081270848e-05,
+      "loss": 1.1192,
+      "step": 6789
+    },
+    {
+      "epoch": 0.24693602938502382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7631682000128017e-05,
+      "loss": 1.2076,
+      "step": 6790
+    },
+    {
+      "epoch": 0.24697239698876242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.763092081310643e-05,
+      "loss": 1.1157,
+      "step": 6791
+    },
+    {
+      "epoch": 0.247008764592501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7630159520216642e-05,
+      "loss": 1.1758,
+      "step": 6792
+    },
+    {
+      "epoch": 0.24704513219623958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7629398121469216e-05,
+      "loss": 1.1451,
+      "step": 6793
+    },
+    {
+      "epoch": 0.24708149979997818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7628636616874713e-05,
+      "loss": 1.0797,
+      "step": 6794
+    },
+    {
+      "epoch": 0.24711786740371677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7627875006443704e-05,
+      "loss": 1.0394,
+      "step": 6795
+    },
+    {
+      "epoch": 0.24715423500745537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762711329018675e-05,
+      "loss": 1.1871,
+      "step": 6796
+    },
+    {
+      "epoch": 0.24719060261119394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7626351468114422e-05,
+      "loss": 1.1487,
+      "step": 6797
+    },
+    {
+      "epoch": 0.24722697021493253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7625589540237288e-05,
+      "loss": 1.1183,
+      "step": 6798
+    },
+    {
+      "epoch": 0.24726333781867113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7624827506565918e-05,
+      "loss": 1.1726,
+      "step": 6799
+    },
+    {
+      "epoch": 0.24729970542240973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7624065367110887e-05,
+      "loss": 1.0754,
+      "step": 6800
+    },
+    {
+      "epoch": 0.2473360730261483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7623303121882762e-05,
+      "loss": 1.0477,
+      "step": 6801
+    },
+    {
+      "epoch": 0.2473724406298869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762254077089212e-05,
+      "loss": 1.2137,
+      "step": 6802
+    },
+    {
+      "epoch": 0.2474088082336255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7621778314149546e-05,
+      "loss": 1.1439,
+      "step": 6803
+    },
+    {
+      "epoch": 0.24744517583736408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762101575166561e-05,
+      "loss": 1.0763,
+      "step": 6804
+    },
+    {
+      "epoch": 0.24748154344110265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762025308345089e-05,
+      "loss": 1.0308,
+      "step": 6805
+    },
+    {
+      "epoch": 0.24751791104484125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7619490309515975e-05,
+      "loss": 1.0768,
+      "step": 6806
+    },
+    {
+      "epoch": 0.24755427864857985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7618727429871442e-05,
+      "loss": 1.143,
+      "step": 6807
+    },
+    {
+      "epoch": 0.24759064625231844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7617964444527873e-05,
+      "loss": 1.1222,
+      "step": 6808
+    },
+    {
+      "epoch": 0.24762701385605704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7617201353495855e-05,
+      "loss": 1.0909,
+      "step": 6809
+    },
+    {
+      "epoch": 0.2476633814597956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761643815678598e-05,
+      "loss": 1.1339,
+      "step": 6810
+    },
+    {
+      "epoch": 0.2476997490635342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761567485440883e-05,
+      "loss": 1.037,
+      "step": 6811
+    },
+    {
+      "epoch": 0.2477361166672728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7614911446374992e-05,
+      "loss": 1.0736,
+      "step": 6812
+    },
+    {
+      "epoch": 0.2477724842710114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7614147932695066e-05,
+      "loss": 1.1406,
+      "step": 6813
+    },
+    {
+      "epoch": 0.24780885187474996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761338431337964e-05,
+      "loss": 1.1835,
+      "step": 6814
+    },
+    {
+      "epoch": 0.24784521947848856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761262058843931e-05,
+      "loss": 1.1884,
+      "step": 6815
+    },
+    {
+      "epoch": 0.24788158708222716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761185675788467e-05,
+      "loss": 1.0147,
+      "step": 6816
+    },
+    {
+      "epoch": 0.24791795468596575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7611092821726313e-05,
+      "loss": 1.078,
+      "step": 6817
+    },
+    {
+      "epoch": 0.24795432228970432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761032877997485e-05,
+      "loss": 1.1054,
+      "step": 6818
+    },
+    {
+      "epoch": 0.24799068989344292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7609564632640865e-05,
+      "loss": 1.2327,
+      "step": 6819
+    },
+    {
+      "epoch": 0.2480270574971815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.760880037973497e-05,
+      "loss": 1.0868,
+      "step": 6820
+    },
+    {
+      "epoch": 0.2480634251009201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7608036021267767e-05,
+      "loss": 1.1599,
+      "step": 6821
+    },
+    {
+      "epoch": 0.24809979270465868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7607271557249858e-05,
+      "loss": 1.097,
+      "step": 6822
+    },
+    {
+      "epoch": 0.24813616030839727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7606506987691846e-05,
+      "loss": 1.0701,
+      "step": 6823
+    },
+    {
+      "epoch": 0.24817252791213587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7605742312604344e-05,
+      "loss": 1.0752,
+      "step": 6824
+    },
+    {
+      "epoch": 0.24820889551587447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.760497753199796e-05,
+      "loss": 1.0773,
+      "step": 6825
+    },
+    {
+      "epoch": 0.24824526311961304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.76042126458833e-05,
+      "loss": 1.11,
+      "step": 6826
+    },
+    {
+      "epoch": 0.24828163072335163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7603447654270983e-05,
+      "loss": 1.013,
+      "step": 6827
+    },
+    {
+      "epoch": 0.24831799832709023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7602682557171612e-05,
+      "loss": 1.0767,
+      "step": 6828
+    },
+    {
+      "epoch": 0.24835436593082882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7601917354595813e-05,
+      "loss": 1.0543,
+      "step": 6829
+    },
+    {
+      "epoch": 0.24839073353456742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7601152046554193e-05,
+      "loss": 1.0281,
+      "step": 6830
+    },
+    {
+      "epoch": 0.248427101138306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7600386633057375e-05,
+      "loss": 1.222,
+      "step": 6831
+    },
+    {
+      "epoch": 0.24846346874204459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7599621114115973e-05,
+      "loss": 1.1089,
+      "step": 6832
+    },
+    {
+      "epoch": 0.24849983634578318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7598855489740612e-05,
+      "loss": 1.0522,
+      "step": 6833
+    },
+    {
+      "epoch": 0.24853620394952178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759808975994191e-05,
+      "loss": 1.1392,
+      "step": 6834
+    },
+    {
+      "epoch": 0.24857257155326035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7597323924730498e-05,
+      "loss": 1.0884,
+      "step": 6835
+    },
+    {
+      "epoch": 0.24860893915699894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7596557984116992e-05,
+      "loss": 1.081,
+      "step": 6836
+    },
+    {
+      "epoch": 0.24864530676073754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7595791938112022e-05,
+      "loss": 1.0927,
+      "step": 6837
+    },
+    {
+      "epoch": 0.24868167436447614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759502578672622e-05,
+      "loss": 1.2342,
+      "step": 6838
+    },
+    {
+      "epoch": 0.2487180419682147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759425952997021e-05,
+      "loss": 1.0289,
+      "step": 6839
+    },
+    {
+      "epoch": 0.2487544095719533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759349316785462e-05,
+      "loss": 1.2068,
+      "step": 6840
+    },
+    {
+      "epoch": 0.2487907771756919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7592726700390087e-05,
+      "loss": 1.1432,
+      "step": 6841
+    },
+    {
+      "epoch": 0.2488271447794305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7591960127587242e-05,
+      "loss": 1.0345,
+      "step": 6842
+    },
+    {
+      "epoch": 0.24886351238316906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7591193449456728e-05,
+      "loss": 1.0785,
+      "step": 6843
+    },
+    {
+      "epoch": 0.24889987998690766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759042666600917e-05,
+      "loss": 1.11,
+      "step": 6844
+    },
+    {
+      "epoch": 0.24893624759064625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7589659777255213e-05,
+      "loss": 1.0367,
+      "step": 6845
+    },
+    {
+      "epoch": 0.24897261519438485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7588892783205496e-05,
+      "loss": 1.0753,
+      "step": 6846
+    },
+    {
+      "epoch": 0.24900898279812342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7588125683870658e-05,
+      "loss": 0.9778,
+      "step": 6847
+    },
+    {
+      "epoch": 0.24904535040186201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7587358479261344e-05,
+      "loss": 1.1261,
+      "step": 6848
+    },
+    {
+      "epoch": 0.2490817180056006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7586591169388193e-05,
+      "loss": 0.937,
+      "step": 6849
+    },
+    {
+      "epoch": 0.2491180856093392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7585823754261853e-05,
+      "loss": 1.077,
+      "step": 6850
+    },
+    {
+      "epoch": 0.2491544532130778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7585056233892978e-05,
+      "loss": 1.1464,
+      "step": 6851
+    },
+    {
+      "epoch": 0.24919082081681637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7584288608292202e-05,
+      "loss": 1.1034,
+      "step": 6852
+    },
+    {
+      "epoch": 0.24922718842055497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7583520877470185e-05,
+      "loss": 1.0027,
+      "step": 6853
+    },
+    {
+      "epoch": 0.24926355602429356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7582753041437575e-05,
+      "loss": 1.2211,
+      "step": 6854
+    },
+    {
+      "epoch": 0.24929992362803216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7581985100205025e-05,
+      "loss": 1.1361,
+      "step": 6855
+    },
+    {
+      "epoch": 0.24933629123177073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.758121705378319e-05,
+      "loss": 1.1213,
+      "step": 6856
+    },
+    {
+      "epoch": 0.24937265883550933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7580448902182724e-05,
+      "loss": 0.9498,
+      "step": 6857
+    },
+    {
+      "epoch": 0.24940902643924792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7579680645414287e-05,
+      "loss": 1.0831,
+      "step": 6858
+    },
+    {
+      "epoch": 0.24944539404298652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7578912283488532e-05,
+      "loss": 1.0506,
+      "step": 6859
+    },
+    {
+      "epoch": 0.2494817616467251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7578143816416126e-05,
+      "loss": 1.0742,
+      "step": 6860
+    },
+    {
+      "epoch": 0.24951812925046368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7577375244207724e-05,
+      "loss": 1.1112,
+      "step": 6861
+    },
+    {
+      "epoch": 0.24955449685420228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.757660656687399e-05,
+      "loss": 1.1989,
+      "step": 6862
+    },
+    {
+      "epoch": 0.24959086445794088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7575837784425594e-05,
+      "loss": 0.9463,
+      "step": 6863
+    },
+    {
+      "epoch": 0.24962723206167944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7575068896873192e-05,
+      "loss": 1.1684,
+      "step": 6864
+    },
+    {
+      "epoch": 0.24966359966541804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7574299904227457e-05,
+      "loss": 1.1042,
+      "step": 6865
+    },
+    {
+      "epoch": 0.24969996726915664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7573530806499062e-05,
+      "loss": 1.2301,
+      "step": 6866
+    },
+    {
+      "epoch": 0.24973633487289523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7572761603698673e-05,
+      "loss": 1.1978,
+      "step": 6867
+    },
+    {
+      "epoch": 0.2497727024766338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.757199229583696e-05,
+      "loss": 1.0818,
+      "step": 6868
+    },
+    {
+      "epoch": 0.2498090700803724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7571222882924592e-05,
+      "loss": 1.1428,
+      "step": 6869
+    },
+    {
+      "epoch": 0.249845437684111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7570453364972252e-05,
+      "loss": 1.1672,
+      "step": 6870
+    },
+    {
+      "epoch": 0.2498818052878496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7569683741990612e-05,
+      "loss": 1.191,
+      "step": 6871
+    },
+    {
+      "epoch": 0.2499181728915882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756891401399035e-05,
+      "loss": 1.0982,
+      "step": 6872
+    },
+    {
+      "epoch": 0.24995454049532675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7568144180982148e-05,
+      "loss": 1.0216,
+      "step": 6873
+    },
+    {
+      "epoch": 0.24999090809906535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756737424297668e-05,
+      "loss": 1.0381,
+      "step": 6874
+    },
+    {
+      "epoch": 0.2500272757028039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756660419998463e-05,
+      "loss": 1.0357,
+      "step": 6875
+    },
+    {
+      "epoch": 0.2500636433065425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7565834052016686e-05,
+      "loss": 1.1495,
+      "step": 6876
+    },
+    {
+      "epoch": 0.2501000109102811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7565063799083528e-05,
+      "loss": 1.0817,
+      "step": 6877
+    },
+    {
+      "epoch": 0.2501363785140197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756429344119584e-05,
+      "loss": 1.107,
+      "step": 6878
+    },
+    {
+      "epoch": 0.2501727461177583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7563522978364317e-05,
+      "loss": 1.0899,
+      "step": 6879
+    },
+    {
+      "epoch": 0.2502091137214969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7562752410599644e-05,
+      "loss": 1.0798,
+      "step": 6880
+    },
+    {
+      "epoch": 0.2502454813252355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.756198173791251e-05,
+      "loss": 1.1301,
+      "step": 6881
+    },
+    {
+      "epoch": 0.2502818489289741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7561210960313607e-05,
+      "loss": 1.0808,
+      "step": 6882
+    },
+    {
+      "epoch": 0.25031821653271263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7560440077813632e-05,
+      "loss": 1.0481,
+      "step": 6883
+    },
+    {
+      "epoch": 0.25035458413645123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7559669090423277e-05,
+      "loss": 1.1945,
+      "step": 6884
+    },
+    {
+      "epoch": 0.2503909517401898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7558897998153238e-05,
+      "loss": 1.1659,
+      "step": 6885
+    },
+    {
+      "epoch": 0.2504273193439284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7558126801014218e-05,
+      "loss": 1.1926,
+      "step": 6886
+    },
+    {
+      "epoch": 0.250463686947667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7557355499016907e-05,
+      "loss": 1.0116,
+      "step": 6887
+    },
+    {
+      "epoch": 0.2505000545514056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7556584092172015e-05,
+      "loss": 1.067,
+      "step": 6888
+    },
+    {
+      "epoch": 0.2505364221551442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7555812580490234e-05,
+      "loss": 1.109,
+      "step": 6889
+    },
+    {
+      "epoch": 0.2505727897588828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7555040963982278e-05,
+      "loss": 1.0585,
+      "step": 6890
+    },
+    {
+      "epoch": 0.2506091573626214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.755426924265885e-05,
+      "loss": 1.0077,
+      "step": 6891
+    },
+    {
+      "epoch": 0.25064552496635994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7553497416530653e-05,
+      "loss": 1.0667,
+      "step": 6892
+    },
+    {
+      "epoch": 0.25068189257009854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7552725485608394e-05,
+      "loss": 1.0797,
+      "step": 6893
+    },
+    {
+      "epoch": 0.25071826017383714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.755195344990279e-05,
+      "loss": 1.0309,
+      "step": 6894
+    },
+    {
+      "epoch": 0.25075462777757573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.755118130942454e-05,
+      "loss": 1.0712,
+      "step": 6895
+    },
+    {
+      "epoch": 0.25079099538131433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7550409064184367e-05,
+      "loss": 0.9909,
+      "step": 6896
+    },
+    {
+      "epoch": 0.2508273629850529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7549636714192977e-05,
+      "loss": 1.234,
+      "step": 6897
+    },
+    {
+      "epoch": 0.2508637305887915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7548864259461095e-05,
+      "loss": 1.0742,
+      "step": 6898
+    },
+    {
+      "epoch": 0.2509000981925301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.754809169999943e-05,
+      "loss": 1.0383,
+      "step": 6899
+    },
+    {
+      "epoch": 0.25093646579626866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7547319035818698e-05,
+      "loss": 1.1389,
+      "step": 6900
+    },
+    {
+      "epoch": 0.25097283340000726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7546546266929628e-05,
+      "loss": 1.0642,
+      "step": 6901
+    },
+    {
+      "epoch": 0.25100920100374585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.754577339334293e-05,
+      "loss": 1.1215,
+      "step": 6902
+    },
+    {
+      "epoch": 0.25104556860748445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7545000415069338e-05,
+      "loss": 0.9973,
+      "step": 6903
+    },
+    {
+      "epoch": 0.25108193621122304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7544227332119566e-05,
+      "loss": 0.9651,
+      "step": 6904
+    },
+    {
+      "epoch": 0.25111830381496164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.754345414450435e-05,
+      "loss": 1.0334,
+      "step": 6905
+    },
+    {
+      "epoch": 0.25115467141870024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7542680852234403e-05,
+      "loss": 1.0574,
+      "step": 6906
+    },
+    {
+      "epoch": 0.25119103902243883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7541907455320465e-05,
+      "loss": 1.1182,
+      "step": 6907
+    },
+    {
+      "epoch": 0.2512274066261774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.754113395377326e-05,
+      "loss": 1.1462,
+      "step": 6908
+    },
+    {
+      "epoch": 0.25126377422991597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7540360347603517e-05,
+      "loss": 1.151,
+      "step": 6909
+    },
+    {
+      "epoch": 0.25130014183365457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7539586636821978e-05,
+      "loss": 1.0934,
+      "step": 6910
+    },
+    {
+      "epoch": 0.25133650943739316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.753881282143937e-05,
+      "loss": 1.0539,
+      "step": 6911
+    },
+    {
+      "epoch": 0.25137287704113176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7538038901466433e-05,
+      "loss": 1.0928,
+      "step": 6912
+    },
+    {
+      "epoch": 0.25140924464487036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.75372648769139e-05,
+      "loss": 1.1969,
+      "step": 6913
+    },
+    {
+      "epoch": 0.25144561224860895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7536490747792507e-05,
+      "loss": 1.1904,
+      "step": 6914
+    },
+    {
+      "epoch": 0.25148197985234755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7535716514113003e-05,
+      "loss": 0.98,
+      "step": 6915
+    },
+    {
+      "epoch": 0.25151834745608614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.753494217588612e-05,
+      "loss": 1.1672,
+      "step": 6916
+    },
+    {
+      "epoch": 0.2515547150598247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7534167733122607e-05,
+      "loss": 1.1404,
+      "step": 6917
+    },
+    {
+      "epoch": 0.2515910826635633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7533393185833206e-05,
+      "loss": 1.0707,
+      "step": 6918
+    },
+    {
+      "epoch": 0.2516274502673019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7532618534028664e-05,
+      "loss": 1.2111,
+      "step": 6919
+    },
+    {
+      "epoch": 0.2516638178710405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7531843777719728e-05,
+      "loss": 1.1486,
+      "step": 6920
+    },
+    {
+      "epoch": 0.25170018547477907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.753106891691714e-05,
+      "loss": 1.0868,
+      "step": 6921
+    },
+    {
+      "epoch": 0.25173655307851767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.753029395163166e-05,
+      "loss": 1.1056,
+      "step": 6922
+    },
+    {
+      "epoch": 0.25177292068225626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7529518881874036e-05,
+      "loss": 1.1633,
+      "step": 6923
+    },
+    {
+      "epoch": 0.25180928828599486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7528743707655022e-05,
+      "loss": 1.0884,
+      "step": 6924
+    },
+    {
+      "epoch": 0.2518456558897334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.752796842898537e-05,
+      "loss": 1.2104,
+      "step": 6925
+    },
+    {
+      "epoch": 0.251882023493472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7527193045875834e-05,
+      "loss": 1.1076,
+      "step": 6926
+    },
+    {
+      "epoch": 0.2519183910972106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7526417558337175e-05,
+      "loss": 1.1142,
+      "step": 6927
+    },
+    {
+      "epoch": 0.2519547587009492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7525641966380153e-05,
+      "loss": 1.1028,
+      "step": 6928
+    },
+    {
+      "epoch": 0.2519911263046878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7524866270015522e-05,
+      "loss": 1.0239,
+      "step": 6929
+    },
+    {
+      "epoch": 0.2520274939084264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7524090469254057e-05,
+      "loss": 1.1369,
+      "step": 6930
+    },
+    {
+      "epoch": 0.252063861512165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7523314564106505e-05,
+      "loss": 0.9684,
+      "step": 6931
+    },
+    {
+      "epoch": 0.2521002291159036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7522538554583636e-05,
+      "loss": 1.048,
+      "step": 6932
+    },
+    {
+      "epoch": 0.25213659671964217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.752176244069622e-05,
+      "loss": 1.157,
+      "step": 6933
+    },
+    {
+      "epoch": 0.2521729643233807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7520986222455025e-05,
+      "loss": 1.11,
+      "step": 6934
+    },
+    {
+      "epoch": 0.2522093319271193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7520209899870815e-05,
+      "loss": 1.0899,
+      "step": 6935
+    },
+    {
+      "epoch": 0.2522456995308579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7519433472954362e-05,
+      "loss": 0.9934,
+      "step": 6936
+    },
+    {
+      "epoch": 0.2522820671345965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.751865694171644e-05,
+      "loss": 1.0024,
+      "step": 6937
+    },
+    {
+      "epoch": 0.2523184347383351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7517880306167817e-05,
+      "loss": 1.0841,
+      "step": 6938
+    },
+    {
+      "epoch": 0.2523548023420737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7517103566319273e-05,
+      "loss": 1.0672,
+      "step": 6939
+    },
+    {
+      "epoch": 0.2523911699458123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.751632672218158e-05,
+      "loss": 1.0407,
+      "step": 6940
+    },
+    {
+      "epoch": 0.2524275375495509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.751554977376552e-05,
+      "loss": 1.0591,
+      "step": 6941
+    },
+    {
+      "epoch": 0.2524639051532894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7514772721081873e-05,
+      "loss": 1.2043,
+      "step": 6942
+    },
+    {
+      "epoch": 0.252500272757028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7513995564141414e-05,
+      "loss": 1.1533,
+      "step": 6943
+    },
+    {
+      "epoch": 0.2525366403607666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.751321830295493e-05,
+      "loss": 1.2201,
+      "step": 6944
+    },
+    {
+      "epoch": 0.2525730079645052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7512440937533198e-05,
+      "loss": 1.1862,
+      "step": 6945
+    },
+    {
+      "epoch": 0.2526093755682438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7511663467887006e-05,
+      "loss": 1.178,
+      "step": 6946
+    },
+    {
+      "epoch": 0.2526457431719824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7510885894027143e-05,
+      "loss": 0.9724,
+      "step": 6947
+    },
+    {
+      "epoch": 0.252682110775721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7510108215964395e-05,
+      "loss": 1.1703,
+      "step": 6948
+    },
+    {
+      "epoch": 0.2527184783794596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7509330433709554e-05,
+      "loss": 1.0707,
+      "step": 6949
+    },
+    {
+      "epoch": 0.25275484598319814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7508552547273404e-05,
+      "loss": 1.0825,
+      "step": 6950
+    },
+    {
+      "epoch": 0.25279121358693674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.750777455666674e-05,
+      "loss": 1.0283,
+      "step": 6951
+    },
+    {
+      "epoch": 0.25282758119067533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7506996461900357e-05,
+      "loss": 1.0679,
+      "step": 6952
+    },
+    {
+      "epoch": 0.25286394879441393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.750621826298505e-05,
+      "loss": 1.2,
+      "step": 6953
+    },
+    {
+      "epoch": 0.2529003163981525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7505439959931614e-05,
+      "loss": 1.0631,
+      "step": 6954
+    },
+    {
+      "epoch": 0.2529366840018911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7504661552750847e-05,
+      "loss": 1.0081,
+      "step": 6955
+    },
+    {
+      "epoch": 0.2529730516056297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.750388304145355e-05,
+      "loss": 1.0045,
+      "step": 6956
+    },
+    {
+      "epoch": 0.2530094192093683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7503104426050518e-05,
+      "loss": 1.0471,
+      "step": 6957
+    },
+    {
+      "epoch": 0.2530457868131069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7502325706552562e-05,
+      "loss": 1.1034,
+      "step": 6958
+    },
+    {
+      "epoch": 0.25308215441684545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7501546882970483e-05,
+      "loss": 1.0029,
+      "step": 6959
+    },
+    {
+      "epoch": 0.25311852202058405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7500767955315078e-05,
+      "loss": 1.0874,
+      "step": 6960
+    },
+    {
+      "epoch": 0.25315488962432264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7499988923597166e-05,
+      "loss": 1.1267,
+      "step": 6961
+    },
+    {
+      "epoch": 0.25319125722806124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7499209787827545e-05,
+      "loss": 1.1318,
+      "step": 6962
+    },
+    {
+      "epoch": 0.25322762483179984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7498430548017032e-05,
+      "loss": 1.1227,
+      "step": 6963
+    },
+    {
+      "epoch": 0.25326399243553843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7497651204176427e-05,
+      "loss": 1.1825,
+      "step": 6964
+    },
+    {
+      "epoch": 0.25330036003927703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7496871756316555e-05,
+      "loss": 1.0515,
+      "step": 6965
+    },
+    {
+      "epoch": 0.2533367276430156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7496092204448225e-05,
+      "loss": 0.9731,
+      "step": 6966
+    },
+    {
+      "epoch": 0.25337309524675417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7495312548582245e-05,
+      "loss": 1.0703,
+      "step": 6967
+    },
+    {
+      "epoch": 0.25340946285049276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7494532788729444e-05,
+      "loss": 1.0599,
+      "step": 6968
+    },
+    {
+      "epoch": 0.25344583045423136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7493752924900632e-05,
+      "loss": 1.0088,
+      "step": 6969
+    },
+    {
+      "epoch": 0.25348219805796995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.749297295710663e-05,
+      "loss": 1.0913,
+      "step": 6970
+    },
+    {
+      "epoch": 0.25351856566170855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.749219288535826e-05,
+      "loss": 1.017,
+      "step": 6971
+    },
+    {
+      "epoch": 0.25355493326544715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7491412709666344e-05,
+      "loss": 1.0657,
+      "step": 6972
+    },
+    {
+      "epoch": 0.25359130086918574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7490632430041703e-05,
+      "loss": 1.1905,
+      "step": 6973
+    },
+    {
+      "epoch": 0.25362766847292434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748985204649517e-05,
+      "loss": 1.09,
+      "step": 6974
+    },
+    {
+      "epoch": 0.25366403607666294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7489071559037563e-05,
+      "loss": 1.1389,
+      "step": 6975
+    },
+    {
+      "epoch": 0.2537004036804015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748829096767971e-05,
+      "loss": 1.1503,
+      "step": 6976
+    },
+    {
+      "epoch": 0.2537367712841401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748751027243245e-05,
+      "loss": 1.134,
+      "step": 6977
+    },
+    {
+      "epoch": 0.25377313888787867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7486729473306608e-05,
+      "loss": 1.2355,
+      "step": 6978
+    },
+    {
+      "epoch": 0.25380950649161726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7485948570313015e-05,
+      "loss": 1.1635,
+      "step": 6979
+    },
+    {
+      "epoch": 0.25384587409535586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7485167563462505e-05,
+      "loss": 0.964,
+      "step": 6980
+    },
+    {
+      "epoch": 0.25388224169909446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748438645276592e-05,
+      "loss": 1.0431,
+      "step": 6981
+    },
+    {
+      "epoch": 0.25391860930283305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7483605238234088e-05,
+      "loss": 1.1924,
+      "step": 6982
+    },
+    {
+      "epoch": 0.25395497690657165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7482823919877853e-05,
+      "loss": 1.0333,
+      "step": 6983
+    },
+    {
+      "epoch": 0.2539913445103102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7482042497708053e-05,
+      "loss": 1.0922,
+      "step": 6984
+    },
+    {
+      "epoch": 0.2540277121140488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748126097173553e-05,
+      "loss": 1.0357,
+      "step": 6985
+    },
+    {
+      "epoch": 0.2540640797177874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748047934197112e-05,
+      "loss": 1.1647,
+      "step": 6986
+    },
+    {
+      "epoch": 0.254100447321526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7479697608425677e-05,
+      "loss": 1.1878,
+      "step": 6987
+    },
+    {
+      "epoch": 0.2541368149252646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.747891577111004e-05,
+      "loss": 1.1051,
+      "step": 6988
+    },
+    {
+      "epoch": 0.25417318252900317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.747813383003506e-05,
+      "loss": 0.9921,
+      "step": 6989
+    },
+    {
+      "epoch": 0.25420955013274177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7477351785211586e-05,
+      "loss": 1.1617,
+      "step": 6990
+    },
+    {
+      "epoch": 0.25424591773648036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7476569636650458e-05,
+      "loss": 1.1212,
+      "step": 6991
+    },
+    {
+      "epoch": 0.2542822853402189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.747578738436254e-05,
+      "loss": 1.1465,
+      "step": 6992
+    },
+    {
+      "epoch": 0.2543186529439575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7475005028358675e-05,
+      "loss": 1.2018,
+      "step": 6993
+    },
+    {
+      "epoch": 0.2543550205476961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7474222568649724e-05,
+      "loss": 1.0782,
+      "step": 6994
+    },
+    {
+      "epoch": 0.2543913881514347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7473440005246536e-05,
+      "loss": 1.0609,
+      "step": 6995
+    },
+    {
+      "epoch": 0.2544277557551733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7472657338159973e-05,
+      "loss": 0.9405,
+      "step": 6996
+    },
+    {
+      "epoch": 0.2544641233589119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.747187456740089e-05,
+      "loss": 1.1118,
+      "step": 6997
+    },
+    {
+      "epoch": 0.2545004909626505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7471091692980153e-05,
+      "loss": 1.0621,
+      "step": 6998
+    },
+    {
+      "epoch": 0.2545368585663891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7470308714908614e-05,
+      "loss": 1.0591,
+      "step": 6999
+    },
+    {
+      "epoch": 0.2545732261701277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7469525633197143e-05,
+      "loss": 1.1105,
+      "step": 7000
+    },
+    {
+      "epoch": 0.2546095937738662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.74687424478566e-05,
+      "loss": 1.1805,
+      "step": 7001
+    },
+    {
+      "epoch": 0.2546459613776048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7467959158897856e-05,
+      "loss": 1.2494,
+      "step": 7002
+    },
+    {
+      "epoch": 0.2546823289813434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746717576633177e-05,
+      "loss": 1.0372,
+      "step": 7003
+    },
+    {
+      "epoch": 0.254718696585082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746639227016922e-05,
+      "loss": 1.0603,
+      "step": 7004
+    },
+    {
+      "epoch": 0.2547550641888206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7465608670421065e-05,
+      "loss": 1.1497,
+      "step": 7005
+    },
+    {
+      "epoch": 0.2547914317925592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7464824967098185e-05,
+      "loss": 0.9373,
+      "step": 7006
+    },
+    {
+      "epoch": 0.2548277993962978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746404116021145e-05,
+      "loss": 1.1259,
+      "step": 7007
+    },
+    {
+      "epoch": 0.2548641670000364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7463257249771732e-05,
+      "loss": 1.1149,
+      "step": 7008
+    },
+    {
+      "epoch": 0.25490053460377493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746247323578991e-05,
+      "loss": 1.0422,
+      "step": 7009
+    },
+    {
+      "epoch": 0.2549369022075135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746168911827686e-05,
+      "loss": 1.1996,
+      "step": 7010
+    },
+    {
+      "epoch": 0.2549732698112521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746090489724346e-05,
+      "loss": 1.1435,
+      "step": 7011
+    },
+    {
+      "epoch": 0.2550096374149907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746012057270059e-05,
+      "loss": 1.0471,
+      "step": 7012
+    },
+    {
+      "epoch": 0.2550460050187293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7459336144659132e-05,
+      "loss": 1.1286,
+      "step": 7013
+    },
+    {
+      "epoch": 0.2550823726224679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7458551613129968e-05,
+      "loss": 1.1672,
+      "step": 7014
+    },
+    {
+      "epoch": 0.2551187402262065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7457766978123982e-05,
+      "loss": 1.0959,
+      "step": 7015
+    },
+    {
+      "epoch": 0.2551551078299451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.745698223965206e-05,
+      "loss": 1.1926,
+      "step": 7016
+    },
+    {
+      "epoch": 0.2551914754336837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7456197397725092e-05,
+      "loss": 1.2061,
+      "step": 7017
+    },
+    {
+      "epoch": 0.25522784303742224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7455412452353963e-05,
+      "loss": 1.1234,
+      "step": 7018
+    },
+    {
+      "epoch": 0.25526421064116084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7454627403549564e-05,
+      "loss": 1.1634,
+      "step": 7019
+    },
+    {
+      "epoch": 0.25530057824489943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7453842251322787e-05,
+      "loss": 1.1195,
+      "step": 7020
+    },
+    {
+      "epoch": 0.25533694584863803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7453056995684523e-05,
+      "loss": 1.2047,
+      "step": 7021
+    },
+    {
+      "epoch": 0.2553733134523766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7452271636645667e-05,
+      "loss": 0.9867,
+      "step": 7022
+    },
+    {
+      "epoch": 0.2554096810561152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7451486174217115e-05,
+      "loss": 1.1195,
+      "step": 7023
+    },
+    {
+      "epoch": 0.2554460486598538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.745070060840977e-05,
+      "loss": 1.0853,
+      "step": 7024
+    },
+    {
+      "epoch": 0.2554824162635924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.744991493923452e-05,
+      "loss": 1.1089,
+      "step": 7025
+    },
+    {
+      "epoch": 0.25551878386733096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7449129166702272e-05,
+      "loss": 1.0278,
+      "step": 7026
+    },
+    {
+      "epoch": 0.25555515147106955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7448343290823925e-05,
+      "loss": 1.0482,
+      "step": 7027
+    },
+    {
+      "epoch": 0.25559151907480815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7447557311610385e-05,
+      "loss": 1.1496,
+      "step": 7028
+    },
+    {
+      "epoch": 0.25562788667854675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.744677122907255e-05,
+      "loss": 1.0852,
+      "step": 7029
+    },
+    {
+      "epoch": 0.25566425428228534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7445985043221334e-05,
+      "loss": 1.1332,
+      "step": 7030
+    },
+    {
+      "epoch": 0.25570062188602394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7445198754067637e-05,
+      "loss": 1.0065,
+      "step": 7031
+    },
+    {
+      "epoch": 0.25573698948976253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.744441236162237e-05,
+      "loss": 1.0499,
+      "step": 7032
+    },
+    {
+      "epoch": 0.25577335709350113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7443625865896447e-05,
+      "loss": 0.9797,
+      "step": 7033
+    },
+    {
+      "epoch": 0.2558097246972397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7442839266900775e-05,
+      "loss": 0.9971,
+      "step": 7034
+    },
+    {
+      "epoch": 0.25584609230097827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7442052564646266e-05,
+      "loss": 0.942,
+      "step": 7035
+    },
+    {
+      "epoch": 0.25588245990471686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.744126575914384e-05,
+      "loss": 1.0841,
+      "step": 7036
+    },
+    {
+      "epoch": 0.25591882750845546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7440478850404406e-05,
+      "loss": 1.0604,
+      "step": 7037
+    },
+    {
+      "epoch": 0.25595519511219406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7439691838438887e-05,
+      "loss": 1.0954,
+      "step": 7038
+    },
+    {
+      "epoch": 0.25599156271593265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.74389047232582e-05,
+      "loss": 1.075,
+      "step": 7039
+    },
+    {
+      "epoch": 0.25602793031967125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743811750487326e-05,
+      "loss": 1.1695,
+      "step": 7040
+    },
+    {
+      "epoch": 0.25606429792340984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7437330183294998e-05,
+      "loss": 1.0483,
+      "step": 7041
+    },
+    {
+      "epoch": 0.25610066552714844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7436542758534328e-05,
+      "loss": 1.1602,
+      "step": 7042
+    },
+    {
+      "epoch": 0.256137033130887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7435755230602182e-05,
+      "loss": 0.9896,
+      "step": 7043
+    },
+    {
+      "epoch": 0.2561734007346256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743496759950948e-05,
+      "loss": 1.0909,
+      "step": 7044
+    },
+    {
+      "epoch": 0.2562097683383642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743417986526715e-05,
+      "loss": 0.9758,
+      "step": 7045
+    },
+    {
+      "epoch": 0.25624613594210277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7433392027886123e-05,
+      "loss": 1.1029,
+      "step": 7046
+    },
+    {
+      "epoch": 0.25628250354584137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.743260408737733e-05,
+      "loss": 1.2351,
+      "step": 7047
+    },
+    {
+      "epoch": 0.25631887114957996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.74318160437517e-05,
+      "loss": 1.0753,
+      "step": 7048
+    },
+    {
+      "epoch": 0.25635523875331856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7431027897020167e-05,
+      "loss": 0.981,
+      "step": 7049
+    },
+    {
+      "epoch": 0.25639160635705716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7430239647193668e-05,
+      "loss": 1.1066,
+      "step": 7050
+    },
+    {
+      "epoch": 0.2564279739607957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742945129428313e-05,
+      "loss": 1.1283,
+      "step": 7051
+    },
+    {
+      "epoch": 0.2564643415645343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.74286628382995e-05,
+      "loss": 1.1383,
+      "step": 7052
+    },
+    {
+      "epoch": 0.2565007091682729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742787427925371e-05,
+      "loss": 1.254,
+      "step": 7053
+    },
+    {
+      "epoch": 0.2565370767720115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7427085617156706e-05,
+      "loss": 1.0026,
+      "step": 7054
+    },
+    {
+      "epoch": 0.2565734443757501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7426296852019425e-05,
+      "loss": 1.1436,
+      "step": 7055
+    },
+    {
+      "epoch": 0.2566098119794887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742550798385281e-05,
+      "loss": 1.0768,
+      "step": 7056
+    },
+    {
+      "epoch": 0.2566461795832273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7424719012667816e-05,
+      "loss": 1.2274,
+      "step": 7057
+    },
+    {
+      "epoch": 0.25668254718696587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742392993847537e-05,
+      "loss": 1.0179,
+      "step": 7058
+    },
+    {
+      "epoch": 0.25671891479070447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7423140761286435e-05,
+      "loss": 0.9712,
+      "step": 7059
+    },
+    {
+      "epoch": 0.256755282394443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742235148111195e-05,
+      "loss": 1.0978,
+      "step": 7060
+    },
+    {
+      "epoch": 0.2567916499981816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7421562097962873e-05,
+      "loss": 1.1187,
+      "step": 7061
+    },
+    {
+      "epoch": 0.2568280176019202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742077261185015e-05,
+      "loss": 1.1703,
+      "step": 7062
+    },
+    {
+      "epoch": 0.2568643852056588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7419983022784735e-05,
+      "loss": 1.1499,
+      "step": 7063
+    },
+    {
+      "epoch": 0.2569007528093974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7419193330777582e-05,
+      "loss": 1.0769,
+      "step": 7064
+    },
+    {
+      "epoch": 0.256937120413136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7418403535839653e-05,
+      "loss": 1.0409,
+      "step": 7065
+    },
+    {
+      "epoch": 0.2569734880168746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7417613637981895e-05,
+      "loss": 1.1394,
+      "step": 7066
+    },
+    {
+      "epoch": 0.2570098556206132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7416823637215275e-05,
+      "loss": 0.9973,
+      "step": 7067
+    },
+    {
+      "epoch": 0.2570462232243517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7416033533550747e-05,
+      "loss": 1.1183,
+      "step": 7068
+    },
+    {
+      "epoch": 0.2570825908280903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7415243326999278e-05,
+      "loss": 1.0552,
+      "step": 7069
+    },
+    {
+      "epoch": 0.2571189584318289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.741445301757183e-05,
+      "loss": 1.1075,
+      "step": 7070
+    },
+    {
+      "epoch": 0.2571553260355675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7413662605279365e-05,
+      "loss": 1.0832,
+      "step": 7071
+    },
+    {
+      "epoch": 0.2571916936393061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7412872090132848e-05,
+      "loss": 1.1497,
+      "step": 7072
+    },
+    {
+      "epoch": 0.2572280612430447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.741208147214325e-05,
+      "loss": 1.037,
+      "step": 7073
+    },
+    {
+      "epoch": 0.2572644288467833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7411290751321538e-05,
+      "loss": 1.0804,
+      "step": 7074
+    },
+    {
+      "epoch": 0.2573007964505219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7410499927678682e-05,
+      "loss": 1.1027,
+      "step": 7075
+    },
+    {
+      "epoch": 0.2573371640542605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7409709001225655e-05,
+      "loss": 1.0495,
+      "step": 7076
+    },
+    {
+      "epoch": 0.25737353165799903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7408917971973426e-05,
+      "loss": 1.1023,
+      "step": 7077
+    },
+    {
+      "epoch": 0.25740989926173763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7408126839932973e-05,
+      "loss": 1.0884,
+      "step": 7078
+    },
+    {
+      "epoch": 0.2574462668654762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7407335605115273e-05,
+      "loss": 1.0344,
+      "step": 7079
+    },
+    {
+      "epoch": 0.2574826344692148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7406544267531295e-05,
+      "loss": 1.034,
+      "step": 7080
+    },
+    {
+      "epoch": 0.2575190020729534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7405752827192028e-05,
+      "loss": 1.1041,
+      "step": 7081
+    },
+    {
+      "epoch": 0.257555369676692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.740496128410845e-05,
+      "loss": 1.109,
+      "step": 7082
+    },
+    {
+      "epoch": 0.2575917372804306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7404169638291537e-05,
+      "loss": 1.0783,
+      "step": 7083
+    },
+    {
+      "epoch": 0.2576281048841692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.740337788975228e-05,
+      "loss": 1.0699,
+      "step": 7084
+    },
+    {
+      "epoch": 0.25766447248790775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7402586038501652e-05,
+      "loss": 0.9666,
+      "step": 7085
+    },
+    {
+      "epoch": 0.25770084009164634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.740179408455065e-05,
+      "loss": 1.2473,
+      "step": 7086
+    },
+    {
+      "epoch": 0.25773720769538494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7401002027910254e-05,
+      "loss": 1.2524,
+      "step": 7087
+    },
+    {
+      "epoch": 0.25777357529912354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.740020986859146e-05,
+      "loss": 1.0638,
+      "step": 7088
+    },
+    {
+      "epoch": 0.25780994290286213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739941760660525e-05,
+      "loss": 1.0846,
+      "step": 7089
+    },
+    {
+      "epoch": 0.25784631050660073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739862524196262e-05,
+      "loss": 1.1627,
+      "step": 7090
+    },
+    {
+      "epoch": 0.2578826781103393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739783277467456e-05,
+      "loss": 1.0749,
+      "step": 7091
+    },
+    {
+      "epoch": 0.2579190457140779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739704020475207e-05,
+      "loss": 1.2113,
+      "step": 7092
+    },
+    {
+      "epoch": 0.25795541331781646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7396247532206142e-05,
+      "loss": 1.1589,
+      "step": 7093
+    },
+    {
+      "epoch": 0.25799178092155506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7395454757047772e-05,
+      "loss": 1.1284,
+      "step": 7094
+    },
+    {
+      "epoch": 0.25802814852529365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739466187928796e-05,
+      "loss": 1.2198,
+      "step": 7095
+    },
+    {
+      "epoch": 0.25806451612903225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7393868898937705e-05,
+      "loss": 1.0482,
+      "step": 7096
+    },
+    {
+      "epoch": 0.25810088373277085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739307581600801e-05,
+      "loss": 1.0749,
+      "step": 7097
+    },
+    {
+      "epoch": 0.25813725133650944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739228263050988e-05,
+      "loss": 1.0978,
+      "step": 7098
+    },
+    {
+      "epoch": 0.25817361894024804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7391489342454314e-05,
+      "loss": 1.005,
+      "step": 7099
+    },
+    {
+      "epoch": 0.25820998654398664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739069595185232e-05,
+      "loss": 0.9839,
+      "step": 7100
+    },
+    {
+      "epoch": 0.25824635414772523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738990245871491e-05,
+      "loss": 1.0224,
+      "step": 7101
+    },
+    {
+      "epoch": 0.2582827217514638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7389108863053086e-05,
+      "loss": 1.0388,
+      "step": 7102
+    },
+    {
+      "epoch": 0.25831908935520237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738831516487786e-05,
+      "loss": 1.0749,
+      "step": 7103
+    },
+    {
+      "epoch": 0.25835545695894097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738752136420024e-05,
+      "loss": 1.0199,
+      "step": 7104
+    },
+    {
+      "epoch": 0.25839182456267956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738672746103125e-05,
+      "loss": 1.2164,
+      "step": 7105
+    },
+    {
+      "epoch": 0.25842819216641816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7385933455381893e-05,
+      "loss": 1.08,
+      "step": 7106
+    },
+    {
+      "epoch": 0.25846455977015675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7385139347263187e-05,
+      "loss": 1.101,
+      "step": 7107
+    },
+    {
+      "epoch": 0.25850092737389535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7384345136686153e-05,
+      "loss": 1.1133,
+      "step": 7108
+    },
+    {
+      "epoch": 0.25853729497763395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7383550823661805e-05,
+      "loss": 1.1375,
+      "step": 7109
+    },
+    {
+      "epoch": 0.2585736625813725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7382756408201168e-05,
+      "loss": 1.1541,
+      "step": 7110
+    },
+    {
+      "epoch": 0.2586100301851111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.738196189031526e-05,
+      "loss": 1.0511,
+      "step": 7111
+    },
+    {
+      "epoch": 0.2586463977888497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.73811672700151e-05,
+      "loss": 1.0253,
+      "step": 7112
+    },
+    {
+      "epoch": 0.2586827653925883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7380372547311722e-05,
+      "loss": 1.0313,
+      "step": 7113
+    },
+    {
+      "epoch": 0.2587191329963269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7379577722216147e-05,
+      "loss": 1.056,
+      "step": 7114
+    },
+    {
+      "epoch": 0.25875550060006547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7378782794739398e-05,
+      "loss": 1.0534,
+      "step": 7115
+    },
+    {
+      "epoch": 0.25879186820380407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7377987764892506e-05,
+      "loss": 1.1083,
+      "step": 7116
+    },
+    {
+      "epoch": 0.25882823580754266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7377192632686505e-05,
+      "loss": 1.103,
+      "step": 7117
+    },
+    {
+      "epoch": 0.25886460341128126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.737639739813242e-05,
+      "loss": 1.1609,
+      "step": 7118
+    },
+    {
+      "epoch": 0.2589009710150198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.737560206124129e-05,
+      "loss": 1.1034,
+      "step": 7119
+    },
+    {
+      "epoch": 0.2589373386187584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7374806622024145e-05,
+      "loss": 1.0182,
+      "step": 7120
+    },
+    {
+      "epoch": 0.258973706222497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.737401108049202e-05,
+      "loss": 1.0403,
+      "step": 7121
+    },
+    {
+      "epoch": 0.2590100738262356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7373215436655956e-05,
+      "loss": 1.1605,
+      "step": 7122
+    },
+    {
+      "epoch": 0.2590464414299742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7372419690526987e-05,
+      "loss": 1.0465,
+      "step": 7123
+    },
+    {
+      "epoch": 0.2590828090337128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7371623842116156e-05,
+      "loss": 1.1367,
+      "step": 7124
+    },
+    {
+      "epoch": 0.2591191766374514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7370827891434503e-05,
+      "loss": 1.0603,
+      "step": 7125
+    },
+    {
+      "epoch": 0.25915554424119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7370031838493072e-05,
+      "loss": 1.164,
+      "step": 7126
+    },
+    {
+      "epoch": 0.2591919118449285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7369235683302904e-05,
+      "loss": 1.1908,
+      "step": 7127
+    },
+    {
+      "epoch": 0.2592282794486671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7368439425875045e-05,
+      "loss": 1.1697,
+      "step": 7128
+    },
+    {
+      "epoch": 0.2592646470524057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.736764306622055e-05,
+      "loss": 1.0862,
+      "step": 7129
+    },
+    {
+      "epoch": 0.2593010146561443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7366846604350453e-05,
+      "loss": 1.1864,
+      "step": 7130
+    },
+    {
+      "epoch": 0.2593373822598829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7366050040275813e-05,
+      "loss": 1.063,
+      "step": 7131
+    },
+    {
+      "epoch": 0.2593737498636215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7365253374007682e-05,
+      "loss": 1.0181,
+      "step": 7132
+    },
+    {
+      "epoch": 0.2594101174673601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.736445660555711e-05,
+      "loss": 1.1449,
+      "step": 7133
+    },
+    {
+      "epoch": 0.2594464850710987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7363659734935156e-05,
+      "loss": 1.0889,
+      "step": 7134
+    },
+    {
+      "epoch": 0.2594828526748372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7362862762152862e-05,
+      "loss": 1.1607,
+      "step": 7135
+    },
+    {
+      "epoch": 0.2595192202785758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7362065687221298e-05,
+      "loss": 1.1076,
+      "step": 7136
+    },
+    {
+      "epoch": 0.2595555878823144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.736126851015152e-05,
+      "loss": 1.0621,
+      "step": 7137
+    },
+    {
+      "epoch": 0.259591955486053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7360471230954583e-05,
+      "loss": 1.1754,
+      "step": 7138
+    },
+    {
+      "epoch": 0.2596283230897916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735967384964155e-05,
+      "loss": 1.2139,
+      "step": 7139
+    },
+    {
+      "epoch": 0.2596646906935302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735887636622349e-05,
+      "loss": 1.1191,
+      "step": 7140
+    },
+    {
+      "epoch": 0.2597010582972688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7358078780711452e-05,
+      "loss": 1.2838,
+      "step": 7141
+    },
+    {
+      "epoch": 0.2597374259010074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735728109311652e-05,
+      "loss": 1.1035,
+      "step": 7142
+    },
+    {
+      "epoch": 0.259773793504746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7356483303449746e-05,
+      "loss": 1.1089,
+      "step": 7143
+    },
+    {
+      "epoch": 0.25981016110848454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7355685411722203e-05,
+      "loss": 1.1431,
+      "step": 7144
+    },
+    {
+      "epoch": 0.25984652871222313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7354887417944962e-05,
+      "loss": 1.1123,
+      "step": 7145
+    },
+    {
+      "epoch": 0.25988289631596173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7354089322129097e-05,
+      "loss": 0.9891,
+      "step": 7146
+    },
+    {
+      "epoch": 0.2599192639197003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7353291124285674e-05,
+      "loss": 1.1161,
+      "step": 7147
+    },
+    {
+      "epoch": 0.2599556315234389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735249282442577e-05,
+      "loss": 1.133,
+      "step": 7148
+    },
+    {
+      "epoch": 0.2599919991271775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.735169442256046e-05,
+      "loss": 1.1392,
+      "step": 7149
+    },
+    {
+      "epoch": 0.2600283667309161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7350895918700818e-05,
+      "loss": 1.1225,
+      "step": 7150
+    },
+    {
+      "epoch": 0.2600647343346547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7350097312857928e-05,
+      "loss": 1.075,
+      "step": 7151
+    },
+    {
+      "epoch": 0.26010110193839325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7349298605042862e-05,
+      "loss": 1.1401,
+      "step": 7152
+    },
+    {
+      "epoch": 0.26013746954213185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734849979526671e-05,
+      "loss": 1.0282,
+      "step": 7153
+    },
+    {
+      "epoch": 0.26017383714587045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7347700883540552e-05,
+      "loss": 1.0591,
+      "step": 7154
+    },
+    {
+      "epoch": 0.26021020474960904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734690186987546e-05,
+      "loss": 0.9891,
+      "step": 7155
+    },
+    {
+      "epoch": 0.26024657235334764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7346102754282537e-05,
+      "loss": 1.1792,
+      "step": 7156
+    },
+    {
+      "epoch": 0.26028293995708623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7345303536772856e-05,
+      "loss": 1.1118,
+      "step": 7157
+    },
+    {
+      "epoch": 0.26031930756082483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734450421735751e-05,
+      "loss": 1.0706,
+      "step": 7158
+    },
+    {
+      "epoch": 0.2603556751645634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734370479604759e-05,
+      "loss": 1.0831,
+      "step": 7159
+    },
+    {
+      "epoch": 0.260392042768302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7342905272854188e-05,
+      "loss": 1.1007,
+      "step": 7160
+    },
+    {
+      "epoch": 0.26042841037204056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734210564778839e-05,
+      "loss": 1.0422,
+      "step": 7161
+    },
+    {
+      "epoch": 0.26046477797577916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7341305920861294e-05,
+      "loss": 1.2513,
+      "step": 7162
+    },
+    {
+      "epoch": 0.26050114557951776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734050609208399e-05,
+      "loss": 1.0901,
+      "step": 7163
+    },
+    {
+      "epoch": 0.26053751318325635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.733970616146759e-05,
+      "loss": 1.0165,
+      "step": 7164
+    },
+    {
+      "epoch": 0.26057388078699495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7338906129023173e-05,
+      "loss": 1.0466,
+      "step": 7165
+    },
+    {
+      "epoch": 0.26061024839073355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7338105994761846e-05,
+      "loss": 0.9789,
+      "step": 7166
+    },
+    {
+      "epoch": 0.26064661599447214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.733730575869471e-05,
+      "loss": 0.9081,
+      "step": 7167
+    },
+    {
+      "epoch": 0.26068298359821074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7336505420832866e-05,
+      "loss": 1.0223,
+      "step": 7168
+    },
+    {
+      "epoch": 0.2607193512019493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7335704981187423e-05,
+      "loss": 1.1578,
+      "step": 7169
+    },
+    {
+      "epoch": 0.2607557188056879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7334904439769474e-05,
+      "loss": 1.0771,
+      "step": 7170
+    },
+    {
+      "epoch": 0.26079208640942647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.733410379659014e-05,
+      "loss": 1.0058,
+      "step": 7171
+    },
+    {
+      "epoch": 0.26082845401316507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7333303051660516e-05,
+      "loss": 1.0935,
+      "step": 7172
+    },
+    {
+      "epoch": 0.26086482161690366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.733250220499172e-05,
+      "loss": 1.0779,
+      "step": 7173
+    },
+    {
+      "epoch": 0.26090118922064226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.733170125659486e-05,
+      "loss": 1.0657,
+      "step": 7174
+    },
+    {
+      "epoch": 0.26093755682438086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7330900206481046e-05,
+      "loss": 1.1206,
+      "step": 7175
+    },
+    {
+      "epoch": 0.26097392442811945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7330099054661393e-05,
+      "loss": 1.0182,
+      "step": 7176
+    },
+    {
+      "epoch": 0.261010292031858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732929780114702e-05,
+      "loss": 1.092,
+      "step": 7177
+    },
+    {
+      "epoch": 0.2610466596355966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732849644594903e-05,
+      "loss": 1.1464,
+      "step": 7178
+    },
+    {
+      "epoch": 0.2610830272393352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7327694989078553e-05,
+      "loss": 0.996,
+      "step": 7179
+    },
+    {
+      "epoch": 0.2611193948430738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7326893430546707e-05,
+      "loss": 1.218,
+      "step": 7180
+    },
+    {
+      "epoch": 0.2611557624468124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732609177036461e-05,
+      "loss": 1.1261,
+      "step": 7181
+    },
+    {
+      "epoch": 0.261192130050551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7325290008543382e-05,
+      "loss": 1.1931,
+      "step": 7182
+    },
+    {
+      "epoch": 0.26122849765428957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732448814509415e-05,
+      "loss": 1.1423,
+      "step": 7183
+    },
+    {
+      "epoch": 0.26126486525802817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7323686180028037e-05,
+      "loss": 1.1634,
+      "step": 7184
+    },
+    {
+      "epoch": 0.26130123286176676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732288411335617e-05,
+      "loss": 1.1053,
+      "step": 7185
+    },
+    {
+      "epoch": 0.2613376004655053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7322081945089672e-05,
+      "loss": 0.9571,
+      "step": 7186
+    },
+    {
+      "epoch": 0.2613739680692439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732127967523968e-05,
+      "loss": 1.2297,
+      "step": 7187
+    },
+    {
+      "epoch": 0.2614103356729825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7320477303817316e-05,
+      "loss": 1.1695,
+      "step": 7188
+    },
+    {
+      "epoch": 0.2614467032767211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731967483083372e-05,
+      "loss": 1.0526,
+      "step": 7189
+    },
+    {
+      "epoch": 0.2614830708804597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7318872256300018e-05,
+      "loss": 1.1044,
+      "step": 7190
+    },
+    {
+      "epoch": 0.2615194384841983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731806958022735e-05,
+      "loss": 1.0822,
+      "step": 7191
+    },
+    {
+      "epoch": 0.2615558060879369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7317266802626847e-05,
+      "loss": 1.1717,
+      "step": 7192
+    },
+    {
+      "epoch": 0.2615921736916755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731646392350965e-05,
+      "loss": 1.0694,
+      "step": 7193
+    },
+    {
+      "epoch": 0.261628541295414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7315660942886897e-05,
+      "loss": 1.0652,
+      "step": 7194
+    },
+    {
+      "epoch": 0.2616649088991526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7314857860769726e-05,
+      "loss": 1.0322,
+      "step": 7195
+    },
+    {
+      "epoch": 0.2617012765028912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7314054677169284e-05,
+      "loss": 1.0188,
+      "step": 7196
+    },
+    {
+      "epoch": 0.2617376441066298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.731325139209671e-05,
+      "loss": 1.0818,
+      "step": 7197
+    },
+    {
+      "epoch": 0.2617740117103684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7312448005563145e-05,
+      "loss": 1.1837,
+      "step": 7198
+    },
+    {
+      "epoch": 0.261810379314107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7311644517579746e-05,
+      "loss": 1.1725,
+      "step": 7199
+    },
+    {
+      "epoch": 0.2618467469178456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7310840928157646e-05,
+      "loss": 1.0615,
+      "step": 7200
+    },
+    {
+      "epoch": 0.2618831145215842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7310037237308006e-05,
+      "loss": 1.1102,
+      "step": 7201
+    },
+    {
+      "epoch": 0.2619194821253228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7309233445041967e-05,
+      "loss": 1.0057,
+      "step": 7202
+    },
+    {
+      "epoch": 0.26195584972906133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7308429551370687e-05,
+      "loss": 1.0738,
+      "step": 7203
+    },
+    {
+      "epoch": 0.2619922173327999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7307625556305316e-05,
+      "loss": 1.1821,
+      "step": 7204
+    },
+    {
+      "epoch": 0.2620285849365385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730682145985701e-05,
+      "loss": 0.9937,
+      "step": 7205
+    },
+    {
+      "epoch": 0.2620649525402771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730601726203692e-05,
+      "loss": 1.1707,
+      "step": 7206
+    },
+    {
+      "epoch": 0.2621013201440157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7305212962856206e-05,
+      "loss": 1.0407,
+      "step": 7207
+    },
+    {
+      "epoch": 0.2621376877477543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730440856232603e-05,
+      "loss": 1.0739,
+      "step": 7208
+    },
+    {
+      "epoch": 0.2621740553514929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7303604060457545e-05,
+      "loss": 1.02,
+      "step": 7209
+    },
+    {
+      "epoch": 0.2622104229552315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730279945726192e-05,
+      "loss": 1.0212,
+      "step": 7210
+    },
+    {
+      "epoch": 0.26224679055897004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7301994752750315e-05,
+      "loss": 1.0403,
+      "step": 7211
+    },
+    {
+      "epoch": 0.26228315816270864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730118994693389e-05,
+      "loss": 1.0876,
+      "step": 7212
+    },
+    {
+      "epoch": 0.26231952576644724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7300385039823813e-05,
+      "loss": 1.0332,
+      "step": 7213
+    },
+    {
+      "epoch": 0.26235589337018583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7299580031431253e-05,
+      "loss": 1.0915,
+      "step": 7214
+    },
+    {
+      "epoch": 0.26239226097392443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7298774921767376e-05,
+      "loss": 1.0542,
+      "step": 7215
+    },
+    {
+      "epoch": 0.262428628577663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7297969710843352e-05,
+      "loss": 1.1626,
+      "step": 7216
+    },
+    {
+      "epoch": 0.2624649961814016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7297164398670354e-05,
+      "loss": 1.1475,
+      "step": 7217
+    },
+    {
+      "epoch": 0.2625013637851402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729635898525955e-05,
+      "loss": 1.142,
+      "step": 7218
+    },
+    {
+      "epoch": 0.2625377313888788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7295553470622123e-05,
+      "loss": 1.0815,
+      "step": 7219
+    },
+    {
+      "epoch": 0.26257409899261736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729474785476924e-05,
+      "loss": 1.0297,
+      "step": 7220
+    },
+    {
+      "epoch": 0.26261046659635595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7293942137712082e-05,
+      "loss": 1.2098,
+      "step": 7221
+    },
+    {
+      "epoch": 0.26264683420009455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729313631946182e-05,
+      "loss": 1.1525,
+      "step": 7222
+    },
+    {
+      "epoch": 0.26268320180383314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7292330400029645e-05,
+      "loss": 0.9917,
+      "step": 7223
+    },
+    {
+      "epoch": 0.26271956940757174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.729152437942673e-05,
+      "loss": 1.0061,
+      "step": 7224
+    },
+    {
+      "epoch": 0.26275593701131034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7290718257664258e-05,
+      "loss": 1.1267,
+      "step": 7225
+    },
+    {
+      "epoch": 0.26279230461504893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7289912034753418e-05,
+      "loss": 1.188,
+      "step": 7226
+    },
+    {
+      "epoch": 0.26282867221878753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.728910571070539e-05,
+      "loss": 1.0231,
+      "step": 7227
+    },
+    {
+      "epoch": 0.26286503982252607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.728829928553136e-05,
+      "loss": 1.1532,
+      "step": 7228
+    },
+    {
+      "epoch": 0.26290140742626467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.728749275924252e-05,
+      "loss": 0.9783,
+      "step": 7229
+    },
+    {
+      "epoch": 0.26293777503000326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.728668613185006e-05,
+      "loss": 1.1038,
+      "step": 7230
+    },
+    {
+      "epoch": 0.26297414263374186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7285879403365166e-05,
+      "loss": 1.0483,
+      "step": 7231
+    },
+    {
+      "epoch": 0.26301051023748045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.728507257379903e-05,
+      "loss": 1.1357,
+      "step": 7232
+    },
+    {
+      "epoch": 0.26304687784121905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.728426564316285e-05,
+      "loss": 1.0886,
+      "step": 7233
+    },
+    {
+      "epoch": 0.26308324544495765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.728345861146782e-05,
+      "loss": 1.0,
+      "step": 7234
+    },
+    {
+      "epoch": 0.26311961304869624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7282651478725135e-05,
+      "loss": 1.0033,
+      "step": 7235
+    },
+    {
+      "epoch": 0.2631559806524348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7281844244945992e-05,
+      "loss": 1.1104,
+      "step": 7236
+    },
+    {
+      "epoch": 0.2631923482561734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.72810369101416e-05,
+      "loss": 1.1476,
+      "step": 7237
+    },
+    {
+      "epoch": 0.263228715859912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7280229474323143e-05,
+      "loss": 1.1894,
+      "step": 7238
+    },
+    {
+      "epoch": 0.2632650834636506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7279421937501833e-05,
+      "loss": 1.0662,
+      "step": 7239
+    },
+    {
+      "epoch": 0.26330145106738917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7278614299688872e-05,
+      "loss": 1.0052,
+      "step": 7240
+    },
+    {
+      "epoch": 0.26333781867112777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7277806560895466e-05,
+      "loss": 1.0593,
+      "step": 7241
+    },
+    {
+      "epoch": 0.26337418627486636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7276998721132815e-05,
+      "loss": 1.0752,
+      "step": 7242
+    },
+    {
+      "epoch": 0.26341055387860496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7276190780412134e-05,
+      "loss": 1.1104,
+      "step": 7243
+    },
+    {
+      "epoch": 0.26344692148234355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.727538273874463e-05,
+      "loss": 1.1591,
+      "step": 7244
+    },
+    {
+      "epoch": 0.2634832890860821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7274574596141515e-05,
+      "loss": 1.0997,
+      "step": 7245
+    },
+    {
+      "epoch": 0.2635196566898207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7273766352613996e-05,
+      "loss": 1.1638,
+      "step": 7246
+    },
+    {
+      "epoch": 0.2635560242935593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7272958008173293e-05,
+      "loss": 1.2278,
+      "step": 7247
+    },
+    {
+      "epoch": 0.2635923918972979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7272149562830613e-05,
+      "loss": 1.1332,
+      "step": 7248
+    },
+    {
+      "epoch": 0.2636287595010365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7271341016597178e-05,
+      "loss": 1.1228,
+      "step": 7249
+    },
+    {
+      "epoch": 0.2636651271047751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.72705323694842e-05,
+      "loss": 1.035,
+      "step": 7250
+    },
+    {
+      "epoch": 0.2637014947085137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.72697236215029e-05,
+      "loss": 0.9945,
+      "step": 7251
+    },
+    {
+      "epoch": 0.26373786231225227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7268914772664503e-05,
+      "loss": 1.175,
+      "step": 7252
+    },
+    {
+      "epoch": 0.2637742299159908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7268105822980222e-05,
+      "loss": 1.0343,
+      "step": 7253
+    },
+    {
+      "epoch": 0.2638105975197294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.726729677246129e-05,
+      "loss": 1.0055,
+      "step": 7254
+    },
+    {
+      "epoch": 0.263846965123468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7266487621118922e-05,
+      "loss": 1.1403,
+      "step": 7255
+    },
+    {
+      "epoch": 0.2638833327272066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7265678368964353e-05,
+      "loss": 1.087,
+      "step": 7256
+    },
+    {
+      "epoch": 0.2639197003309452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7264869016008804e-05,
+      "loss": 0.9433,
+      "step": 7257
+    },
+    {
+      "epoch": 0.2639560679346838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7264059562263504e-05,
+      "loss": 1.1128,
+      "step": 7258
+    },
+    {
+      "epoch": 0.2639924355384224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7263250007739683e-05,
+      "loss": 1.0498,
+      "step": 7259
+    },
+    {
+      "epoch": 0.264028803142161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.726244035244857e-05,
+      "loss": 1.1409,
+      "step": 7260
+    },
+    {
+      "epoch": 0.2640651707458996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7261630596401407e-05,
+      "loss": 1.0727,
+      "step": 7261
+    },
+    {
+      "epoch": 0.2641015383496381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.726082073960942e-05,
+      "loss": 1.1845,
+      "step": 7262
+    },
+    {
+      "epoch": 0.2641379059533767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7260010782083848e-05,
+      "loss": 1.1562,
+      "step": 7263
+    },
+    {
+      "epoch": 0.2641742735571153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7259200723835924e-05,
+      "loss": 1.2012,
+      "step": 7264
+    },
+    {
+      "epoch": 0.2642106411608539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7258390564876894e-05,
+      "loss": 1.2323,
+      "step": 7265
+    },
+    {
+      "epoch": 0.2642470087645925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.725758030521799e-05,
+      "loss": 0.9582,
+      "step": 7266
+    },
+    {
+      "epoch": 0.2642833763683311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7256769944870456e-05,
+      "loss": 1.0741,
+      "step": 7267
+    },
+    {
+      "epoch": 0.2643197439720697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7255959483845538e-05,
+      "loss": 1.132,
+      "step": 7268
+    },
+    {
+      "epoch": 0.2643561115758083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7255148922154478e-05,
+      "loss": 1.0763,
+      "step": 7269
+    },
+    {
+      "epoch": 0.26439247917954684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7254338259808518e-05,
+      "loss": 1.1819,
+      "step": 7270
+    },
+    {
+      "epoch": 0.26442884678328543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7253527496818905e-05,
+      "loss": 0.9968,
+      "step": 7271
+    },
+    {
+      "epoch": 0.26446521438702403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.725271663319689e-05,
+      "loss": 1.0519,
+      "step": 7272
+    },
+    {
+      "epoch": 0.2645015819907626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7251905668953727e-05,
+      "loss": 1.1081,
+      "step": 7273
+    },
+    {
+      "epoch": 0.2645379495945012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7251094604100654e-05,
+      "loss": 1.1349,
+      "step": 7274
+    },
+    {
+      "epoch": 0.2645743171982398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7250283438648937e-05,
+      "loss": 0.9795,
+      "step": 7275
+    },
+    {
+      "epoch": 0.2646106848019784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7249472172609826e-05,
+      "loss": 0.9954,
+      "step": 7276
+    },
+    {
+      "epoch": 0.264647052405717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7248660805994567e-05,
+      "loss": 1.2323,
+      "step": 7277
+    },
+    {
+      "epoch": 0.26468342000945555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724784933881443e-05,
+      "loss": 1.1498,
+      "step": 7278
+    },
+    {
+      "epoch": 0.26471978761319415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7247037771080662e-05,
+      "loss": 1.2023,
+      "step": 7279
+    },
+    {
+      "epoch": 0.26475615521693274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724622610280453e-05,
+      "loss": 1.0613,
+      "step": 7280
+    },
+    {
+      "epoch": 0.26479252282067134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7245414333997296e-05,
+      "loss": 1.1754,
+      "step": 7281
+    },
+    {
+      "epoch": 0.26482889042440994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724460246467021e-05,
+      "loss": 1.0415,
+      "step": 7282
+    },
+    {
+      "epoch": 0.26486525802814853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7243790494834546e-05,
+      "loss": 1.1609,
+      "step": 7283
+    },
+    {
+      "epoch": 0.2649016256318871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724297842450157e-05,
+      "loss": 1.0887,
+      "step": 7284
+    },
+    {
+      "epoch": 0.2649379932356257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7242166253682542e-05,
+      "loss": 1.0598,
+      "step": 7285
+    },
+    {
+      "epoch": 0.2649743608393643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724135398238873e-05,
+      "loss": 1.0147,
+      "step": 7286
+    },
+    {
+      "epoch": 0.26501072844310286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724054161063141e-05,
+      "loss": 1.0334,
+      "step": 7287
+    },
+    {
+      "epoch": 0.26504709604684146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.723972913842185e-05,
+      "loss": 1.0332,
+      "step": 7288
+    },
+    {
+      "epoch": 0.26508346365058005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7238916565771316e-05,
+      "loss": 0.9995,
+      "step": 7289
+    },
+    {
+      "epoch": 0.26511983125431865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7238103892691084e-05,
+      "loss": 1.0649,
+      "step": 7290
+    },
+    {
+      "epoch": 0.26515619885805725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7237291119192434e-05,
+      "loss": 1.1473,
+      "step": 7291
+    },
+    {
+      "epoch": 0.26519256646179584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7236478245286633e-05,
+      "loss": 1.1862,
+      "step": 7292
+    },
+    {
+      "epoch": 0.26522893406553444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.723566527098497e-05,
+      "loss": 1.1117,
+      "step": 7293
+    },
+    {
+      "epoch": 0.26526530166927303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7234852196298715e-05,
+      "loss": 0.985,
+      "step": 7294
+    },
+    {
+      "epoch": 0.2653016692730116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7234039021239146e-05,
+      "loss": 1.0306,
+      "step": 7295
+    },
+    {
+      "epoch": 0.26533803687675017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7233225745817556e-05,
+      "loss": 1.113,
+      "step": 7296
+    },
+    {
+      "epoch": 0.26537440448048877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7232412370045216e-05,
+      "loss": 1.0969,
+      "step": 7297
+    },
+    {
+      "epoch": 0.26541077208422736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.723159889393342e-05,
+      "loss": 1.1721,
+      "step": 7298
+    },
+    {
+      "epoch": 0.26544713968796596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7230785317493448e-05,
+      "loss": 1.1012,
+      "step": 7299
+    },
+    {
+      "epoch": 0.26548350729170456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722997164073659e-05,
+      "loss": 1.0211,
+      "step": 7300
+    },
+    {
+      "epoch": 0.26551987489544315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722915786367413e-05,
+      "loss": 1.0574,
+      "step": 7301
+    },
+    {
+      "epoch": 0.26555624249918175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7228343986317365e-05,
+      "loss": 1.2336,
+      "step": 7302
+    },
+    {
+      "epoch": 0.26559261010292035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722753000867758e-05,
+      "loss": 1.122,
+      "step": 7303
+    },
+    {
+      "epoch": 0.2656289777066589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7226715930766077e-05,
+      "loss": 1.0524,
+      "step": 7304
+    },
+    {
+      "epoch": 0.2656653453103975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7225901752594137e-05,
+      "loss": 1.065,
+      "step": 7305
+    },
+    {
+      "epoch": 0.2657017129141361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7225087474173062e-05,
+      "loss": 1.1168,
+      "step": 7306
+    },
+    {
+      "epoch": 0.2657380805178747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7224273095514156e-05,
+      "loss": 1.095,
+      "step": 7307
+    },
+    {
+      "epoch": 0.26577444812161327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7223458616628707e-05,
+      "loss": 1.1841,
+      "step": 7308
+    },
+    {
+      "epoch": 0.26581081572535187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7222644037528015e-05,
+      "loss": 1.0523,
+      "step": 7309
+    },
+    {
+      "epoch": 0.26584718332909046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722182935822339e-05,
+      "loss": 1.1083,
+      "step": 7310
+    },
+    {
+      "epoch": 0.26588355093282906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722101457872613e-05,
+      "loss": 1.1461,
+      "step": 7311
+    },
+    {
+      "epoch": 0.2659199185365676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7220199699047532e-05,
+      "loss": 1.1896,
+      "step": 7312
+    },
+    {
+      "epoch": 0.2659562861403062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.721938471919891e-05,
+      "loss": 1.0549,
+      "step": 7313
+    },
+    {
+      "epoch": 0.2659926537440448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7218569639191567e-05,
+      "loss": 1.0913,
+      "step": 7314
+    },
+    {
+      "epoch": 0.2660290213477834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7217754459036814e-05,
+      "loss": 1.1043,
+      "step": 7315
+    },
+    {
+      "epoch": 0.266065388951522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7216939178745956e-05,
+      "loss": 1.0115,
+      "step": 7316
+    },
+    {
+      "epoch": 0.2661017565552606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.721612379833031e-05,
+      "loss": 1.0032,
+      "step": 7317
+    },
+    {
+      "epoch": 0.2661381241589992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.721530831780118e-05,
+      "loss": 1.0508,
+      "step": 7318
+    },
+    {
+      "epoch": 0.2661744917627378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.721449273716989e-05,
+      "loss": 1.1647,
+      "step": 7319
+    },
+    {
+      "epoch": 0.2662108593664763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7213677056447747e-05,
+      "loss": 1.1957,
+      "step": 7320
+    },
+    {
+      "epoch": 0.2662472269702149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7212861275646065e-05,
+      "loss": 0.9963,
+      "step": 7321
+    },
+    {
+      "epoch": 0.2662835945739535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7212045394776172e-05,
+      "loss": 1.143,
+      "step": 7322
+    },
+    {
+      "epoch": 0.2663199621776921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.721122941384938e-05,
+      "loss": 1.0699,
+      "step": 7323
+    },
+    {
+      "epoch": 0.2663563297814307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7210413332877012e-05,
+      "loss": 0.949,
+      "step": 7324
+    },
+    {
+      "epoch": 0.2663926973851693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720959715187039e-05,
+      "loss": 1.1903,
+      "step": 7325
+    },
+    {
+      "epoch": 0.2664290649889079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7208780870840837e-05,
+      "loss": 1.1249,
+      "step": 7326
+    },
+    {
+      "epoch": 0.2664654325926465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7207964489799675e-05,
+      "loss": 0.999,
+      "step": 7327
+    },
+    {
+      "epoch": 0.2665018001963851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720714800875823e-05,
+      "loss": 1.1301,
+      "step": 7328
+    },
+    {
+      "epoch": 0.2665381678001236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720633142772784e-05,
+      "loss": 1.0768,
+      "step": 7329
+    },
+    {
+      "epoch": 0.2665745354038622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720551474671982e-05,
+      "loss": 1.0724,
+      "step": 7330
+    },
+    {
+      "epoch": 0.2666109030076008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7204697965745506e-05,
+      "loss": 1.0701,
+      "step": 7331
+    },
+    {
+      "epoch": 0.2666472706113394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720388108481623e-05,
+      "loss": 1.0638,
+      "step": 7332
+    },
+    {
+      "epoch": 0.266683638215078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7203064103943326e-05,
+      "loss": 1.051,
+      "step": 7333
+    },
+    {
+      "epoch": 0.2667200058188166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7202247023138126e-05,
+      "loss": 1.0934,
+      "step": 7334
+    },
+    {
+      "epoch": 0.2667563734225552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720142984241197e-05,
+      "loss": 1.1897,
+      "step": 7335
+    },
+    {
+      "epoch": 0.2667927410262938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720061256177619e-05,
+      "loss": 1.1656,
+      "step": 7336
+    },
+    {
+      "epoch": 0.26682910863003234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.719979518124213e-05,
+      "loss": 1.0654,
+      "step": 7337
+    },
+    {
+      "epoch": 0.26686547623377094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7198977700821125e-05,
+      "loss": 1.1673,
+      "step": 7338
+    },
+    {
+      "epoch": 0.26690184383750953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.719816012052452e-05,
+      "loss": 1.0334,
+      "step": 7339
+    },
+    {
+      "epoch": 0.26693821144124813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7197342440363654e-05,
+      "loss": 0.9493,
+      "step": 7340
+    },
+    {
+      "epoch": 0.2669745790449867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.719652466034987e-05,
+      "loss": 1.1718,
+      "step": 7341
+    },
+    {
+      "epoch": 0.2670109466487253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7195706780494523e-05,
+      "loss": 1.0313,
+      "step": 7342
+    },
+    {
+      "epoch": 0.2670473142524639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.719488880080895e-05,
+      "loss": 0.9955,
+      "step": 7343
+    },
+    {
+      "epoch": 0.2670836818562025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7194070721304506e-05,
+      "loss": 1.1321,
+      "step": 7344
+    },
+    {
+      "epoch": 0.2671200494599411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7193252541992536e-05,
+      "loss": 1.1781,
+      "step": 7345
+    },
+    {
+      "epoch": 0.26715641706367965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7192434262884395e-05,
+      "loss": 1.0896,
+      "step": 7346
+    },
+    {
+      "epoch": 0.26719278466741825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7191615883991432e-05,
+      "loss": 1.0572,
+      "step": 7347
+    },
+    {
+      "epoch": 0.26722915227115684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7190797405325e-05,
+      "loss": 1.1259,
+      "step": 7348
+    },
+    {
+      "epoch": 0.26726551987489544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718997882689646e-05,
+      "loss": 1.116,
+      "step": 7349
+    },
+    {
+      "epoch": 0.26730188747863404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718916014871716e-05,
+      "loss": 1.1622,
+      "step": 7350
+    },
+    {
+      "epoch": 0.26733825508237263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718834137079847e-05,
+      "loss": 1.222,
+      "step": 7351
+    },
+    {
+      "epoch": 0.26737462268611123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7187522493151738e-05,
+      "loss": 1.0289,
+      "step": 7352
+    },
+    {
+      "epoch": 0.2674109902898498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7186703515788327e-05,
+      "loss": 1.0819,
+      "step": 7353
+    },
+    {
+      "epoch": 0.26744735789358837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7185884438719608e-05,
+      "loss": 1.1612,
+      "step": 7354
+    },
+    {
+      "epoch": 0.26748372549732696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7185065261956933e-05,
+      "loss": 1.1012,
+      "step": 7355
+    },
+    {
+      "epoch": 0.26752009310106556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718424598551167e-05,
+      "loss": 1.1938,
+      "step": 7356
+    },
+    {
+      "epoch": 0.26755646070480416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7183426609395194e-05,
+      "loss": 1.134,
+      "step": 7357
+    },
+    {
+      "epoch": 0.26759282830854275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718260713361886e-05,
+      "loss": 1.0343,
+      "step": 7358
+    },
+    {
+      "epoch": 0.26762919591228135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7181787558194045e-05,
+      "loss": 1.0663,
+      "step": 7359
+    },
+    {
+      "epoch": 0.26766556351601994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718096788313212e-05,
+      "loss": 1.1914,
+      "step": 7360
+    },
+    {
+      "epoch": 0.26770193111975854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.718014810844445e-05,
+      "loss": 1.1221,
+      "step": 7361
+    },
+    {
+      "epoch": 0.2677382987234971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.717932823414241e-05,
+      "loss": 1.0806,
+      "step": 7362
+    },
+    {
+      "epoch": 0.2677746663272357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7178508260237378e-05,
+      "loss": 0.9754,
+      "step": 7363
+    },
+    {
+      "epoch": 0.2678110339309743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7177688186740733e-05,
+      "loss": 1.1432,
+      "step": 7364
+    },
+    {
+      "epoch": 0.26784740153471287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7176868013663845e-05,
+      "loss": 1.1934,
+      "step": 7365
+    },
+    {
+      "epoch": 0.26788376913845147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.71760477410181e-05,
+      "loss": 1.035,
+      "step": 7366
+    },
+    {
+      "epoch": 0.26792013674219006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7175227368814867e-05,
+      "loss": 1.0316,
+      "step": 7367
+    },
+    {
+      "epoch": 0.26795650434592866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.717440689706554e-05,
+      "loss": 1.0468,
+      "step": 7368
+    },
+    {
+      "epoch": 0.26799287194966726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7173586325781495e-05,
+      "loss": 1.1284,
+      "step": 7369
+    },
+    {
+      "epoch": 0.26802923955340585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.717276565497412e-05,
+      "loss": 1.2201,
+      "step": 7370
+    },
+    {
+      "epoch": 0.2680656071571444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7171944884654794e-05,
+      "loss": 1.1227,
+      "step": 7371
+    },
+    {
+      "epoch": 0.268101974760883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7171124014834913e-05,
+      "loss": 1.0709,
+      "step": 7372
+    },
+    {
+      "epoch": 0.2681383423646216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7170303045525858e-05,
+      "loss": 1.0006,
+      "step": 7373
+    },
+    {
+      "epoch": 0.2681747099683602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.716948197673902e-05,
+      "loss": 1.0471,
+      "step": 7374
+    },
+    {
+      "epoch": 0.2682110775720988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7168660808485797e-05,
+      "loss": 1.0163,
+      "step": 7375
+    },
+    {
+      "epoch": 0.2682474451758374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7167839540777572e-05,
+      "loss": 1.102,
+      "step": 7376
+    },
+    {
+      "epoch": 0.26828381277957597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.716701817362574e-05,
+      "loss": 0.9731,
+      "step": 7377
+    },
+    {
+      "epoch": 0.26832018038331457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7166196707041705e-05,
+      "loss": 0.9652,
+      "step": 7378
+    },
+    {
+      "epoch": 0.2683565479870531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.716537514103686e-05,
+      "loss": 1.1339,
+      "step": 7379
+    },
+    {
+      "epoch": 0.2683929155907917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.71645534756226e-05,
+      "loss": 1.0995,
+      "step": 7380
+    },
+    {
+      "epoch": 0.2684292831945303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.716373171081032e-05,
+      "loss": 0.9504,
+      "step": 7381
+    },
+    {
+      "epoch": 0.2684656507982689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7162909846611432e-05,
+      "loss": 1.0952,
+      "step": 7382
+    },
+    {
+      "epoch": 0.2685020184020075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.716208788303733e-05,
+      "loss": 1.0453,
+      "step": 7383
+    },
+    {
+      "epoch": 0.2685383860057461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7161265820099423e-05,
+      "loss": 1.1259,
+      "step": 7384
+    },
+    {
+      "epoch": 0.2685747536094847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.716044365780911e-05,
+      "loss": 1.0547,
+      "step": 7385
+    },
+    {
+      "epoch": 0.2686111212132233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7159621396177802e-05,
+      "loss": 1.087,
+      "step": 7386
+    },
+    {
+      "epoch": 0.2686474888169619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7158799035216906e-05,
+      "loss": 1.1339,
+      "step": 7387
+    },
+    {
+      "epoch": 0.2686838564207004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7157976574937828e-05,
+      "loss": 0.9668,
+      "step": 7388
+    },
+    {
+      "epoch": 0.268720224024439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7157154015351984e-05,
+      "loss": 1.0544,
+      "step": 7389
+    },
+    {
+      "epoch": 0.2687565916281776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7156331356470782e-05,
+      "loss": 1.0416,
+      "step": 7390
+    },
+    {
+      "epoch": 0.2687929592319162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7155508598305636e-05,
+      "loss": 1.2097,
+      "step": 7391
+    },
+    {
+      "epoch": 0.2688293268356548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.715468574086796e-05,
+      "loss": 1.0605,
+      "step": 7392
+    },
+    {
+      "epoch": 0.2688656944393934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7153862784169172e-05,
+      "loss": 1.2072,
+      "step": 7393
+    },
+    {
+      "epoch": 0.268902062043132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7153039728220683e-05,
+      "loss": 1.164,
+      "step": 7394
+    },
+    {
+      "epoch": 0.2689384296468706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.715221657303392e-05,
+      "loss": 1.1181,
+      "step": 7395
+    },
+    {
+      "epoch": 0.26897479725060913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.71513933186203e-05,
+      "loss": 1.1616,
+      "step": 7396
+    },
+    {
+      "epoch": 0.26901116485434773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7150569964991243e-05,
+      "loss": 1.0159,
+      "step": 7397
+    },
+    {
+      "epoch": 0.2690475324580863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7149746512158175e-05,
+      "loss": 0.9949,
+      "step": 7398
+    },
+    {
+      "epoch": 0.2690839000618249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7148922960132518e-05,
+      "loss": 1.0455,
+      "step": 7399
+    },
+    {
+      "epoch": 0.2691202676655635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7148099308925696e-05,
+      "loss": 1.0373,
+      "step": 7400
+    },
+    {
+      "epoch": 0.2691566352693021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.714727555854914e-05,
+      "loss": 1.0525,
+      "step": 7401
+    },
+    {
+      "epoch": 0.2691930028730407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7146451709014277e-05,
+      "loss": 1.0198,
+      "step": 7402
+    },
+    {
+      "epoch": 0.2692293704767793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7145627760332533e-05,
+      "loss": 1.1303,
+      "step": 7403
+    },
+    {
+      "epoch": 0.2692657380805179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7144803712515344e-05,
+      "loss": 1.0646,
+      "step": 7404
+    },
+    {
+      "epoch": 0.26930210568425644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.714397956557414e-05,
+      "loss": 1.0516,
+      "step": 7405
+    },
+    {
+      "epoch": 0.26933847328799504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7143155319520356e-05,
+      "loss": 1.0953,
+      "step": 7406
+    },
+    {
+      "epoch": 0.26937484089173364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7142330974365427e-05,
+      "loss": 1.0746,
+      "step": 7407
+    },
+    {
+      "epoch": 0.26941120849547223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.714150653012079e-05,
+      "loss": 1.1503,
+      "step": 7408
+    },
+    {
+      "epoch": 0.26944757609921083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.714068198679788e-05,
+      "loss": 1.155,
+      "step": 7409
+    },
+    {
+      "epoch": 0.2694839437029494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7139857344408137e-05,
+      "loss": 1.0511,
+      "step": 7410
+    },
+    {
+      "epoch": 0.269520311306688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713903260296301e-05,
+      "loss": 1.1573,
+      "step": 7411
+    },
+    {
+      "epoch": 0.2695566789104266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713820776247393e-05,
+      "loss": 1.0732,
+      "step": 7412
+    },
+    {
+      "epoch": 0.26959304651416516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7137382822952346e-05,
+      "loss": 1.1801,
+      "step": 7413
+    },
+    {
+      "epoch": 0.26962941411790375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.71365577844097e-05,
+      "loss": 1.0944,
+      "step": 7414
+    },
+    {
+      "epoch": 0.26966578172164235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7135732646857442e-05,
+      "loss": 1.1562,
+      "step": 7415
+    },
+    {
+      "epoch": 0.26970214932538095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713490741030702e-05,
+      "loss": 1.1668,
+      "step": 7416
+    },
+    {
+      "epoch": 0.26973851692911954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713408207476988e-05,
+      "loss": 1.1289,
+      "step": 7417
+    },
+    {
+      "epoch": 0.26977488453285814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713325664025747e-05,
+      "loss": 1.0138,
+      "step": 7418
+    },
+    {
+      "epoch": 0.26981125213659674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7132431106781246e-05,
+      "loss": 1.0763,
+      "step": 7419
+    },
+    {
+      "epoch": 0.26984761974033533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713160547435266e-05,
+      "loss": 1.0977,
+      "step": 7420
+    },
+    {
+      "epoch": 0.2698839873440739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713077974298317e-05,
+      "loss": 0.9908,
+      "step": 7421
+    },
+    {
+      "epoch": 0.26992035494781247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7129953912684226e-05,
+      "loss": 1.0877,
+      "step": 7422
+    },
+    {
+      "epoch": 0.26995672255155106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7129127983467287e-05,
+      "loss": 1.0991,
+      "step": 7423
+    },
+    {
+      "epoch": 0.26999309015528966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7128301955343812e-05,
+      "loss": 1.0984,
+      "step": 7424
+    },
+    {
+      "epoch": 0.27002945775902826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7127475828325262e-05,
+      "loss": 1.0269,
+      "step": 7425
+    },
+    {
+      "epoch": 0.27006582536276685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7126649602423097e-05,
+      "loss": 1.0568,
+      "step": 7426
+    },
+    {
+      "epoch": 0.27010219296650545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7125823277648778e-05,
+      "loss": 1.1755,
+      "step": 7427
+    },
+    {
+      "epoch": 0.27013856057024405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7124996854013775e-05,
+      "loss": 1.1264,
+      "step": 7428
+    },
+    {
+      "epoch": 0.27017492817398264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7124170331529545e-05,
+      "loss": 1.0844,
+      "step": 7429
+    },
+    {
+      "epoch": 0.2702112957777212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.712334371020756e-05,
+      "loss": 1.1653,
+      "step": 7430
+    },
+    {
+      "epoch": 0.2702476633814598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7122516990059293e-05,
+      "loss": 1.0357,
+      "step": 7431
+    },
+    {
+      "epoch": 0.2702840309851984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.712169017109621e-05,
+      "loss": 1.1403,
+      "step": 7432
+    },
+    {
+      "epoch": 0.27032039858893697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7120863253329773e-05,
+      "loss": 1.1498,
+      "step": 7433
+    },
+    {
+      "epoch": 0.27035676619267557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7120036236771466e-05,
+      "loss": 1.1403,
+      "step": 7434
+    },
+    {
+      "epoch": 0.27039313379641416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7119209121432752e-05,
+      "loss": 1.1436,
+      "step": 7435
+    },
+    {
+      "epoch": 0.27042950140015276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711838190732512e-05,
+      "loss": 1.1234,
+      "step": 7436
+    },
+    {
+      "epoch": 0.27046586900389136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7117554594460036e-05,
+      "loss": 0.9675,
+      "step": 7437
+    },
+    {
+      "epoch": 0.2705022366076299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711672718284898e-05,
+      "loss": 1.0251,
+      "step": 7438
+    },
+    {
+      "epoch": 0.2705386042113685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7115899672503432e-05,
+      "loss": 1.0059,
+      "step": 7439
+    },
+    {
+      "epoch": 0.2705749718151071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711507206343487e-05,
+      "loss": 1.1237,
+      "step": 7440
+    },
+    {
+      "epoch": 0.2706113394188457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7114244355654782e-05,
+      "loss": 1.0526,
+      "step": 7441
+    },
+    {
+      "epoch": 0.2706477070225843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7113416549174643e-05,
+      "loss": 1.0896,
+      "step": 7442
+    },
+    {
+      "epoch": 0.2706840746263229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7112588644005946e-05,
+      "loss": 1.1069,
+      "step": 7443
+    },
+    {
+      "epoch": 0.2707204422300615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711176064016017e-05,
+      "loss": 1.0473,
+      "step": 7444
+    },
+    {
+      "epoch": 0.27075680983380007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7110932537648807e-05,
+      "loss": 1.0606,
+      "step": 7445
+    },
+    {
+      "epoch": 0.27079317743753867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7110104336483342e-05,
+      "loss": 0.9932,
+      "step": 7446
+    },
+    {
+      "epoch": 0.2708295450412772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.710927603667527e-05,
+      "loss": 1.0287,
+      "step": 7447
+    },
+    {
+      "epoch": 0.2708659126450158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.710844763823608e-05,
+      "loss": 1.0945,
+      "step": 7448
+    },
+    {
+      "epoch": 0.2709022802487544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7107619141177262e-05,
+      "loss": 1.0837,
+      "step": 7449
+    },
+    {
+      "epoch": 0.270938647852493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7106790545510315e-05,
+      "loss": 0.8914,
+      "step": 7450
+    },
+    {
+      "epoch": 0.2709750154562316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.710596185124673e-05,
+      "loss": 1.0352,
+      "step": 7451
+    },
+    {
+      "epoch": 0.2710113830599702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.710513305839801e-05,
+      "loss": 1.1985,
+      "step": 7452
+    },
+    {
+      "epoch": 0.2710477506637088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7104304166975648e-05,
+      "loss": 1.1249,
+      "step": 7453
+    },
+    {
+      "epoch": 0.2710841182674474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7103475176991144e-05,
+      "loss": 1.071,
+      "step": 7454
+    },
+    {
+      "epoch": 0.2711204858711859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7102646088456e-05,
+      "loss": 0.9923,
+      "step": 7455
+    },
+    {
+      "epoch": 0.2711568534749245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7101816901381722e-05,
+      "loss": 1.1136,
+      "step": 7456
+    },
+    {
+      "epoch": 0.2711932210786631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7100987615779806e-05,
+      "loss": 1.1127,
+      "step": 7457
+    },
+    {
+      "epoch": 0.2712295886824017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7100158231661767e-05,
+      "loss": 1.0936,
+      "step": 7458
+    },
+    {
+      "epoch": 0.2712659562861403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7099328749039107e-05,
+      "loss": 1.0955,
+      "step": 7459
+    },
+    {
+      "epoch": 0.2713023238898789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.709849916792333e-05,
+      "loss": 1.0308,
+      "step": 7460
+    },
+    {
+      "epoch": 0.2713386914936175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.709766948832595e-05,
+      "loss": 1.0651,
+      "step": 7461
+    },
+    {
+      "epoch": 0.2713750590973561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.709683971025847e-05,
+      "loss": 1.0972,
+      "step": 7462
+    },
+    {
+      "epoch": 0.27141142670109464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7096009833732414e-05,
+      "loss": 1.118,
+      "step": 7463
+    },
+    {
+      "epoch": 0.27144779430483323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.709517985875929e-05,
+      "loss": 1.0524,
+      "step": 7464
+    },
+    {
+      "epoch": 0.27148416190857183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.709434978535061e-05,
+      "loss": 1.0882,
+      "step": 7465
+    },
+    {
+      "epoch": 0.2715205295123104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7093519613517894e-05,
+      "loss": 1.1326,
+      "step": 7466
+    },
+    {
+      "epoch": 0.271556897116049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7092689343272658e-05,
+      "loss": 1.0536,
+      "step": 7467
+    },
+    {
+      "epoch": 0.2715932647197876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.709185897462642e-05,
+      "loss": 1.1396,
+      "step": 7468
+    },
+    {
+      "epoch": 0.2716296323235262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7091028507590697e-05,
+      "loss": 1.1226,
+      "step": 7469
+    },
+    {
+      "epoch": 0.2716659999272648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7090197942177017e-05,
+      "loss": 1.1376,
+      "step": 7470
+    },
+    {
+      "epoch": 0.2717023675310034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.70893672783969e-05,
+      "loss": 1.1241,
+      "step": 7471
+    },
+    {
+      "epoch": 0.27173873513474195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7088536516261874e-05,
+      "loss": 1.074,
+      "step": 7472
+    },
+    {
+      "epoch": 0.27177510273848055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7087705655783458e-05,
+      "loss": 1.1447,
+      "step": 7473
+    },
+    {
+      "epoch": 0.27181147034221914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7086874696973183e-05,
+      "loss": 0.9799,
+      "step": 7474
+    },
+    {
+      "epoch": 0.27184783794595774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7086043639842572e-05,
+      "loss": 1.1566,
+      "step": 7475
+    },
+    {
+      "epoch": 0.27188420554969633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7085212484403167e-05,
+      "loss": 1.1723,
+      "step": 7476
+    },
+    {
+      "epoch": 0.27192057315343493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7084381230666485e-05,
+      "loss": 1.1562,
+      "step": 7477
+    },
+    {
+      "epoch": 0.2719569407571735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.708354987864407e-05,
+      "loss": 1.1204,
+      "step": 7478
+    },
+    {
+      "epoch": 0.2719933083609121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7082718428347446e-05,
+      "loss": 0.9264,
+      "step": 7479
+    },
+    {
+      "epoch": 0.27202967596465066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7081886879788157e-05,
+      "loss": 1.1303,
+      "step": 7480
+    },
+    {
+      "epoch": 0.27206604356838926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7081055232977735e-05,
+      "loss": 1.1672,
+      "step": 7481
+    },
+    {
+      "epoch": 0.27210241117212786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7080223487927714e-05,
+      "loss": 1.1644,
+      "step": 7482
+    },
+    {
+      "epoch": 0.27213877877586645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.707939164464964e-05,
+      "loss": 1.1309,
+      "step": 7483
+    },
+    {
+      "epoch": 0.27217514637960505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7078559703155054e-05,
+      "loss": 1.0292,
+      "step": 7484
+    },
+    {
+      "epoch": 0.27221151398334364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.707772766345549e-05,
+      "loss": 1.0708,
+      "step": 7485
+    },
+    {
+      "epoch": 0.27224788158708224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.70768955255625e-05,
+      "loss": 1.0391,
+      "step": 7486
+    },
+    {
+      "epoch": 0.27228424919082084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7076063289487628e-05,
+      "loss": 1.1164,
+      "step": 7487
+    },
+    {
+      "epoch": 0.27232061679455943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7075230955242413e-05,
+      "loss": 1.0486,
+      "step": 7488
+    },
+    {
+      "epoch": 0.272356984398298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.707439852283841e-05,
+      "loss": 1.0961,
+      "step": 7489
+    },
+    {
+      "epoch": 0.27239335200203657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.707356599228716e-05,
+      "loss": 1.1557,
+      "step": 7490
+    },
+    {
+      "epoch": 0.27242971960577517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7072733363600226e-05,
+      "loss": 1.157,
+      "step": 7491
+    },
+    {
+      "epoch": 0.27246608720951376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7071900636789146e-05,
+      "loss": 0.9877,
+      "step": 7492
+    },
+    {
+      "epoch": 0.27250245481325236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 1.0764,
+      "step": 7493
+    },
+    {
+      "epoch": 0.27253882241699096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7070234888840777e-05,
+      "loss": 1.0541,
+      "step": 7494
+    },
+    {
+      "epoch": 0.27257519002072955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7069401867726598e-05,
+      "loss": 1.1884,
+      "step": 7495
+    },
+    {
+      "epoch": 0.27261155762446815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7068568748534498e-05,
+      "loss": 1.0444,
+      "step": 7496
+    },
+    {
+      "epoch": 0.2726479252282067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.706773553127604e-05,
+      "loss": 0.9751,
+      "step": 7497
+    },
+    {
+      "epoch": 0.2726842928319453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7066902215962776e-05,
+      "loss": 1.0877,
+      "step": 7498
+    },
+    {
+      "epoch": 0.2727206604356839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.706606880260627e-05,
+      "loss": 1.114,
+      "step": 7499
+    },
+    {
+      "epoch": 0.2727570280394225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7065235291218085e-05,
+      "loss": 1.0843,
+      "step": 7500
+    },
+    {
+      "epoch": 0.2727933956431611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7064401681809784e-05,
+      "loss": 1.0393,
+      "step": 7501
+    },
+    {
+      "epoch": 0.27282976324689967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7063567974392933e-05,
+      "loss": 1.2511,
+      "step": 7502
+    },
+    {
+      "epoch": 0.27286613085063827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.70627341689791e-05,
+      "loss": 1.0598,
+      "step": 7503
+    },
+    {
+      "epoch": 0.27290249845437686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7061900265579852e-05,
+      "loss": 1.1335,
+      "step": 7504
+    },
+    {
+      "epoch": 0.2729388660581154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7061066264206753e-05,
+      "loss": 1.1398,
+      "step": 7505
+    },
+    {
+      "epoch": 0.272975233661854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7060232164871383e-05,
+      "loss": 1.1286,
+      "step": 7506
+    },
+    {
+      "epoch": 0.2730116012655926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7059397967585302e-05,
+      "loss": 1.0995,
+      "step": 7507
+    },
+    {
+      "epoch": 0.2730479688693312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7058563672360095e-05,
+      "loss": 1.0975,
+      "step": 7508
+    },
+    {
+      "epoch": 0.2730843364730698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7057729279207332e-05,
+      "loss": 1.1792,
+      "step": 7509
+    },
+    {
+      "epoch": 0.2731207040768084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705689478813859e-05,
+      "loss": 1.2974,
+      "step": 7510
+    },
+    {
+      "epoch": 0.273157071680547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705606019916544e-05,
+      "loss": 1.1153,
+      "step": 7511
+    },
+    {
+      "epoch": 0.2731934392842856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705522551229947e-05,
+      "loss": 1.0491,
+      "step": 7512
+    },
+    {
+      "epoch": 0.2732298068880242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705439072755226e-05,
+      "loss": 1.0813,
+      "step": 7513
+    },
+    {
+      "epoch": 0.2732661744917627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7053555844935378e-05,
+      "loss": 1.1213,
+      "step": 7514
+    },
+    {
+      "epoch": 0.2733025420955013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705272086446042e-05,
+      "loss": 0.9736,
+      "step": 7515
+    },
+    {
+      "epoch": 0.2733389096992399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705188578613897e-05,
+      "loss": 1.0846,
+      "step": 7516
+    },
+    {
+      "epoch": 0.2733752773029785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7051050609982608e-05,
+      "loss": 1.03,
+      "step": 7517
+    },
+    {
+      "epoch": 0.2734116449067171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.705021533600292e-05,
+      "loss": 1.1504,
+      "step": 7518
+    },
+    {
+      "epoch": 0.2734480125104557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.70493799642115e-05,
+      "loss": 0.959,
+      "step": 7519
+    },
+    {
+      "epoch": 0.2734843801141943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7048544494619938e-05,
+      "loss": 1.2357,
+      "step": 7520
+    },
+    {
+      "epoch": 0.2735207477179329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7047708927239813e-05,
+      "loss": 1.1283,
+      "step": 7521
+    },
+    {
+      "epoch": 0.27355711532167143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7046873262082735e-05,
+      "loss": 1.059,
+      "step": 7522
+    },
+    {
+      "epoch": 0.27359348292541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7046037499160284e-05,
+      "loss": 1.0799,
+      "step": 7523
+    },
+    {
+      "epoch": 0.2736298505291486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704520163848406e-05,
+      "loss": 1.1005,
+      "step": 7524
+    },
+    {
+      "epoch": 0.2736662181328872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7044365680065664e-05,
+      "loss": 0.9859,
+      "step": 7525
+    },
+    {
+      "epoch": 0.2737025857366258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7043529623916683e-05,
+      "loss": 1.0587,
+      "step": 7526
+    },
+    {
+      "epoch": 0.2737389533403644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7042693470048723e-05,
+      "loss": 1.1143,
+      "step": 7527
+    },
+    {
+      "epoch": 0.273775320944103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7041857218473385e-05,
+      "loss": 0.9813,
+      "step": 7528
+    },
+    {
+      "epoch": 0.2738116885478416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7041020869202266e-05,
+      "loss": 1.1698,
+      "step": 7529
+    },
+    {
+      "epoch": 0.2738480561515802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7040184422246974e-05,
+      "loss": 1.0913,
+      "step": 7530
+    },
+    {
+      "epoch": 0.27388442375531874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7039347877619114e-05,
+      "loss": 1.0166,
+      "step": 7531
+    },
+    {
+      "epoch": 0.27392079135905734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7038511235330286e-05,
+      "loss": 1.1758,
+      "step": 7532
+    },
+    {
+      "epoch": 0.27395715896279593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7037674495392105e-05,
+      "loss": 1.042,
+      "step": 7533
+    },
+    {
+      "epoch": 0.27399352656653453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7036837657816172e-05,
+      "loss": 1.245,
+      "step": 7534
+    },
+    {
+      "epoch": 0.2740298941702731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.70360007226141e-05,
+      "loss": 1.116,
+      "step": 7535
+    },
+    {
+      "epoch": 0.2740662617740117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7035163689797504e-05,
+      "loss": 1.1924,
+      "step": 7536
+    },
+    {
+      "epoch": 0.2741026293777503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7034326559377992e-05,
+      "loss": 0.9535,
+      "step": 7537
+    },
+    {
+      "epoch": 0.2741389969814889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.703348933136718e-05,
+      "loss": 1.1079,
+      "step": 7538
+    },
+    {
+      "epoch": 0.27417536458522745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7032652005776682e-05,
+      "loss": 1.0937,
+      "step": 7539
+    },
+    {
+      "epoch": 0.27421173218896605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7031814582618115e-05,
+      "loss": 1.1882,
+      "step": 7540
+    },
+    {
+      "epoch": 0.27424809979270465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.70309770619031e-05,
+      "loss": 1.0093,
+      "step": 7541
+    },
+    {
+      "epoch": 0.27428446739644324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7030139443643252e-05,
+      "loss": 1.0083,
+      "step": 7542
+    },
+    {
+      "epoch": 0.27432083500018184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7029301727850193e-05,
+      "loss": 1.1641,
+      "step": 7543
+    },
+    {
+      "epoch": 0.27435720260392044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7028463914535547e-05,
+      "loss": 1.1501,
+      "step": 7544
+    },
+    {
+      "epoch": 0.27439357020765903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7027626003710937e-05,
+      "loss": 1.0496,
+      "step": 7545
+    },
+    {
+      "epoch": 0.27442993781139763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7026787995387985e-05,
+      "loss": 0.9938,
+      "step": 7546
+    },
+    {
+      "epoch": 0.27446630541513617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.702594988957832e-05,
+      "loss": 1.1083,
+      "step": 7547
+    },
+    {
+      "epoch": 0.27450267301887477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.702511168629357e-05,
+      "loss": 1.0524,
+      "step": 7548
+    },
+    {
+      "epoch": 0.27453904062261336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7024273385545364e-05,
+      "loss": 1.0297,
+      "step": 7549
+    },
+    {
+      "epoch": 0.27457540822635196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7023434987345324e-05,
+      "loss": 1.1624,
+      "step": 7550
+    },
+    {
+      "epoch": 0.27461177583009055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7022596491705097e-05,
+      "loss": 1.1521,
+      "step": 7551
+    },
+    {
+      "epoch": 0.27464814343382915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.70217578986363e-05,
+      "loss": 1.1804,
+      "step": 7552
+    },
+    {
+      "epoch": 0.27468451103756775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7020919208150578e-05,
+      "loss": 1.0614,
+      "step": 7553
+    },
+    {
+      "epoch": 0.27472087864130634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7020080420259563e-05,
+      "loss": 1.1205,
+      "step": 7554
+    },
+    {
+      "epoch": 0.27475724624504494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7019241534974897e-05,
+      "loss": 1.0687,
+      "step": 7555
+    },
+    {
+      "epoch": 0.2747936138487835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7018402552308208e-05,
+      "loss": 0.987,
+      "step": 7556
+    },
+    {
+      "epoch": 0.2748299814525221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7017563472271145e-05,
+      "loss": 1.1567,
+      "step": 7557
+    },
+    {
+      "epoch": 0.2748663490562607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7016724294875342e-05,
+      "loss": 0.913,
+      "step": 7558
+    },
+    {
+      "epoch": 0.27490271665999927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7015885020132445e-05,
+      "loss": 1.0804,
+      "step": 7559
+    },
+    {
+      "epoch": 0.27493908426373787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7015045648054102e-05,
+      "loss": 1.0115,
+      "step": 7560
+    },
+    {
+      "epoch": 0.27497545186747646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7014206178651948e-05,
+      "loss": 1.1348,
+      "step": 7561
+    },
+    {
+      "epoch": 0.27501181947121506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7013366611937637e-05,
+      "loss": 1.0939,
+      "step": 7562
+    },
+    {
+      "epoch": 0.27504818707495365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.701252694792282e-05,
+      "loss": 1.2098,
+      "step": 7563
+    },
+    {
+      "epoch": 0.2750845546786922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7011687186619135e-05,
+      "loss": 1.0553,
+      "step": 7564
+    },
+    {
+      "epoch": 0.2751209222824308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.701084732803824e-05,
+      "loss": 1.0507,
+      "step": 7565
+    },
+    {
+      "epoch": 0.2751572898861694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7010007372191787e-05,
+      "loss": 0.8985,
+      "step": 7566
+    },
+    {
+      "epoch": 0.275193657489908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7009167319091426e-05,
+      "loss": 1.0689,
+      "step": 7567
+    },
+    {
+      "epoch": 0.2752300250936466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7008327168748812e-05,
+      "loss": 1.0215,
+      "step": 7568
+    },
+    {
+      "epoch": 0.2752663926973852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7007486921175606e-05,
+      "loss": 1.0294,
+      "step": 7569
+    },
+    {
+      "epoch": 0.2753027603011238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7006646576383457e-05,
+      "loss": 1.0372,
+      "step": 7570
+    },
+    {
+      "epoch": 0.27533912790486237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7005806134384032e-05,
+      "loss": 1.1887,
+      "step": 7571
+    },
+    {
+      "epoch": 0.27537549550860096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7004965595188983e-05,
+      "loss": 1.1071,
+      "step": 7572
+    },
+    {
+      "epoch": 0.2754118631123395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7004124958809977e-05,
+      "loss": 0.9392,
+      "step": 7573
+    },
+    {
+      "epoch": 0.2754482307160781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7003284225258675e-05,
+      "loss": 1.1774,
+      "step": 7574
+    },
+    {
+      "epoch": 0.2754845983198167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.700244339454674e-05,
+      "loss": 1.1223,
+      "step": 7575
+    },
+    {
+      "epoch": 0.2755209659235553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.700160246668584e-05,
+      "loss": 0.9803,
+      "step": 7576
+    },
+    {
+      "epoch": 0.2755573335272939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7000761441687636e-05,
+      "loss": 1.0256,
+      "step": 7577
+    },
+    {
+      "epoch": 0.2755937011310325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6999920319563804e-05,
+      "loss": 1.1458,
+      "step": 7578
+    },
+    {
+      "epoch": 0.2756300687347711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6999079100326005e-05,
+      "loss": 1.2206,
+      "step": 7579
+    },
+    {
+      "epoch": 0.2756664363385097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6998237783985917e-05,
+      "loss": 1.0284,
+      "step": 7580
+    },
+    {
+      "epoch": 0.2757028039422482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6997396370555207e-05,
+      "loss": 1.196,
+      "step": 7581
+    },
+    {
+      "epoch": 0.2757391715459868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6996554860045552e-05,
+      "loss": 1.0512,
+      "step": 7582
+    },
+    {
+      "epoch": 0.2757755391497254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6995713252468622e-05,
+      "loss": 1.187,
+      "step": 7583
+    },
+    {
+      "epoch": 0.275811906753464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6994871547836102e-05,
+      "loss": 1.1047,
+      "step": 7584
+    },
+    {
+      "epoch": 0.2758482743572026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6994029746159658e-05,
+      "loss": 1.1717,
+      "step": 7585
+    },
+    {
+      "epoch": 0.2758846419609412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6993187847450975e-05,
+      "loss": 1.0545,
+      "step": 7586
+    },
+    {
+      "epoch": 0.2759210095646798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6992345851721735e-05,
+      "loss": 1.1304,
+      "step": 7587
+    },
+    {
+      "epoch": 0.2759573771684184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6991503758983618e-05,
+      "loss": 1.0759,
+      "step": 7588
+    },
+    {
+      "epoch": 0.275993744772157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6990661569248302e-05,
+      "loss": 1.0129,
+      "step": 7589
+    },
+    {
+      "epoch": 0.27603011237589553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.698981928252748e-05,
+      "loss": 1.1781,
+      "step": 7590
+    },
+    {
+      "epoch": 0.2760664799796341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.698897689883283e-05,
+      "loss": 1.0951,
+      "step": 7591
+    },
+    {
+      "epoch": 0.2761028475833727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6988134418176043e-05,
+      "loss": 1.1094,
+      "step": 7592
+    },
+    {
+      "epoch": 0.2761392151871113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6987291840568803e-05,
+      "loss": 1.0089,
+      "step": 7593
+    },
+    {
+      "epoch": 0.2761755827908499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6986449166022806e-05,
+      "loss": 1.0375,
+      "step": 7594
+    },
+    {
+      "epoch": 0.2762119503945885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6985606394549738e-05,
+      "loss": 0.9768,
+      "step": 7595
+    },
+    {
+      "epoch": 0.2762483179983271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.698476352616129e-05,
+      "loss": 1.0714,
+      "step": 7596
+    },
+    {
+      "epoch": 0.2762846856020657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6983920560869158e-05,
+      "loss": 1.2945,
+      "step": 7597
+    },
+    {
+      "epoch": 0.27632105320580425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.698307749868504e-05,
+      "loss": 1.0047,
+      "step": 7598
+    },
+    {
+      "epoch": 0.27635742080954284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6982234339620633e-05,
+      "loss": 1.081,
+      "step": 7599
+    },
+    {
+      "epoch": 0.27639378841328144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6981391083687623e-05,
+      "loss": 1.0348,
+      "step": 7600
+    },
+    {
+      "epoch": 0.27643015601702003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6980547730897723e-05,
+      "loss": 1.0867,
+      "step": 7601
+    },
+    {
+      "epoch": 0.27646652362075863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6979704281262623e-05,
+      "loss": 1.1449,
+      "step": 7602
+    },
+    {
+      "epoch": 0.2765028912244972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6978860734794033e-05,
+      "loss": 1.0454,
+      "step": 7603
+    },
+    {
+      "epoch": 0.2765392588282358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.697801709150365e-05,
+      "loss": 1.1353,
+      "step": 7604
+    },
+    {
+      "epoch": 0.2765756264319744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.697717335140318e-05,
+      "loss": 1.1086,
+      "step": 7605
+    },
+    {
+      "epoch": 0.27661199403571296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6976329514504327e-05,
+      "loss": 0.9954,
+      "step": 7606
+    },
+    {
+      "epoch": 0.27664836163945156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6975485580818806e-05,
+      "loss": 1.1221,
+      "step": 7607
+    },
+    {
+      "epoch": 0.27668472924319015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6974641550358316e-05,
+      "loss": 0.955,
+      "step": 7608
+    },
+    {
+      "epoch": 0.27672109684692875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.697379742313457e-05,
+      "loss": 1.0811,
+      "step": 7609
+    },
+    {
+      "epoch": 0.27675746445066735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.697295319915928e-05,
+      "loss": 1.0772,
+      "step": 7610
+    },
+    {
+      "epoch": 0.27679383205440594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6972108878444156e-05,
+      "loss": 0.9996,
+      "step": 7611
+    },
+    {
+      "epoch": 0.27683019965814454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6971264461000917e-05,
+      "loss": 1.1182,
+      "step": 7612
+    },
+    {
+      "epoch": 0.27686656726188313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.697041994684127e-05,
+      "loss": 1.1477,
+      "step": 7613
+    },
+    {
+      "epoch": 0.27690293486562173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696957533597694e-05,
+      "loss": 1.1346,
+      "step": 7614
+    },
+    {
+      "epoch": 0.27693930246936027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696873062841964e-05,
+      "loss": 1.1223,
+      "step": 7615
+    },
+    {
+      "epoch": 0.27697567007309887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696788582418109e-05,
+      "loss": 1.0666,
+      "step": 7616
+    },
+    {
+      "epoch": 0.27701203767683746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6967040923273005e-05,
+      "loss": 1.0573,
+      "step": 7617
+    },
+    {
+      "epoch": 0.27704840528057606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696619592570712e-05,
+      "loss": 1.0388,
+      "step": 7618
+    },
+    {
+      "epoch": 0.27708477288431466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6965350831495143e-05,
+      "loss": 1.151,
+      "step": 7619
+    },
+    {
+      "epoch": 0.27712114048805325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6964505640648813e-05,
+      "loss": 1.0445,
+      "step": 7620
+    },
+    {
+      "epoch": 0.27715750809179185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6963660353179842e-05,
+      "loss": 1.0789,
+      "step": 7621
+    },
+    {
+      "epoch": 0.27719387569553045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6962814969099968e-05,
+      "loss": 1.1014,
+      "step": 7622
+    },
+    {
+      "epoch": 0.277230243299269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.696196948842091e-05,
+      "loss": 0.9593,
+      "step": 7623
+    },
+    {
+      "epoch": 0.2772666109030076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6961123911154407e-05,
+      "loss": 0.9854,
+      "step": 7624
+    },
+    {
+      "epoch": 0.2773029785067462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6960278237312188e-05,
+      "loss": 1.1333,
+      "step": 7625
+    },
+    {
+      "epoch": 0.2773393461104848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6959432466905984e-05,
+      "loss": 1.2028,
+      "step": 7626
+    },
+    {
+      "epoch": 0.27737571371422337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6958586599947527e-05,
+      "loss": 0.9864,
+      "step": 7627
+    },
+    {
+      "epoch": 0.27741208131796197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.695774063644855e-05,
+      "loss": 1.119,
+      "step": 7628
+    },
+    {
+      "epoch": 0.27744844892170056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.69568945764208e-05,
+      "loss": 1.107,
+      "step": 7629
+    },
+    {
+      "epoch": 0.27748481652543916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6956048419876007e-05,
+      "loss": 1.0712,
+      "step": 7630
+    },
+    {
+      "epoch": 0.27752118412917776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.695520216682591e-05,
+      "loss": 1.0807,
+      "step": 7631
+    },
+    {
+      "epoch": 0.2775575517329163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6954355817282245e-05,
+      "loss": 1.1153,
+      "step": 7632
+    },
+    {
+      "epoch": 0.2775939193366549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6953509371256768e-05,
+      "loss": 1.0688,
+      "step": 7633
+    },
+    {
+      "epoch": 0.2776302869403935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6952662828761214e-05,
+      "loss": 1.2136,
+      "step": 7634
+    },
+    {
+      "epoch": 0.2776666545441321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6951816189807323e-05,
+      "loss": 0.931,
+      "step": 7635
+    },
+    {
+      "epoch": 0.2777030221478707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.695096945440685e-05,
+      "loss": 1.0479,
+      "step": 7636
+    },
+    {
+      "epoch": 0.2777393897516093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6950122622571536e-05,
+      "loss": 1.1128,
+      "step": 7637
+    },
+    {
+      "epoch": 0.2777757573553479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6949275694313133e-05,
+      "loss": 1.14,
+      "step": 7638
+    },
+    {
+      "epoch": 0.27781212495908647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.694842866964339e-05,
+      "loss": 1.0057,
+      "step": 7639
+    },
+    {
+      "epoch": 0.277848492562825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6947581548574056e-05,
+      "loss": 1.0448,
+      "step": 7640
+    },
+    {
+      "epoch": 0.2778848601665636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6946734331116886e-05,
+      "loss": 1.1814,
+      "step": 7641
+    },
+    {
+      "epoch": 0.2779212277703022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.694588701728363e-05,
+      "loss": 1.1644,
+      "step": 7642
+    },
+    {
+      "epoch": 0.2779575953740408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6945039607086055e-05,
+      "loss": 1.0396,
+      "step": 7643
+    },
+    {
+      "epoch": 0.2779939629777794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.69441921005359e-05,
+      "loss": 0.9827,
+      "step": 7644
+    },
+    {
+      "epoch": 0.278030330581518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6943344497644936e-05,
+      "loss": 1.0959,
+      "step": 7645
+    },
+    {
+      "epoch": 0.2780666981852566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.694249679842492e-05,
+      "loss": 1.1392,
+      "step": 7646
+    },
+    {
+      "epoch": 0.2781030657889952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6941649002887608e-05,
+      "loss": 0.9214,
+      "step": 7647
+    },
+    {
+      "epoch": 0.2781394333927337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6940801111044764e-05,
+      "loss": 1.0529,
+      "step": 7648
+    },
+    {
+      "epoch": 0.2781758009964723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.693995312290816e-05,
+      "loss": 1.1432,
+      "step": 7649
+    },
+    {
+      "epoch": 0.2782121686002109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6939105038489545e-05,
+      "loss": 1.1129,
+      "step": 7650
+    },
+    {
+      "epoch": 0.2782485362039495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.693825685780069e-05,
+      "loss": 1.0645,
+      "step": 7651
+    },
+    {
+      "epoch": 0.2782849038076881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6937408580853376e-05,
+      "loss": 1.0541,
+      "step": 7652
+    },
+    {
+      "epoch": 0.2783212714114267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6936560207659353e-05,
+      "loss": 1.0945,
+      "step": 7653
+    },
+    {
+      "epoch": 0.2783576390151653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.69357117382304e-05,
+      "loss": 1.1697,
+      "step": 7654
+    },
+    {
+      "epoch": 0.2783940066189039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6934863172578292e-05,
+      "loss": 1.0072,
+      "step": 7655
+    },
+    {
+      "epoch": 0.2784303742226425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.693401451071479e-05,
+      "loss": 1.1085,
+      "step": 7656
+    },
+    {
+      "epoch": 0.27846674182638104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6933165752651678e-05,
+      "loss": 1.1558,
+      "step": 7657
+    },
+    {
+      "epoch": 0.27850310943011963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.693231689840073e-05,
+      "loss": 1.1412,
+      "step": 7658
+    },
+    {
+      "epoch": 0.27853947703385823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.693146794797372e-05,
+      "loss": 1.0956,
+      "step": 7659
+    },
+    {
+      "epoch": 0.2785758446375968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6930618901382423e-05,
+      "loss": 1.096,
+      "step": 7660
+    },
+    {
+      "epoch": 0.2786122122413354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6929769758638625e-05,
+      "loss": 1.1178,
+      "step": 7661
+    },
+    {
+      "epoch": 0.278648579845074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6928920519754102e-05,
+      "loss": 1.1057,
+      "step": 7662
+    },
+    {
+      "epoch": 0.2786849474488126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692807118474064e-05,
+      "loss": 1.0377,
+      "step": 7663
+    },
+    {
+      "epoch": 0.2787213150525512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6927221753610018e-05,
+      "loss": 1.0836,
+      "step": 7664
+    },
+    {
+      "epoch": 0.27875768265628975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6926372226374025e-05,
+      "loss": 1.0614,
+      "step": 7665
+    },
+    {
+      "epoch": 0.27879405026002835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692552260304444e-05,
+      "loss": 0.9866,
+      "step": 7666
+    },
+    {
+      "epoch": 0.27883041786376694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.692467288363306e-05,
+      "loss": 1.151,
+      "step": 7667
+    },
+    {
+      "epoch": 0.27886678546750554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6923823068151663e-05,
+      "loss": 1.1241,
+      "step": 7668
+    },
+    {
+      "epoch": 0.27890315307124414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6922973156612048e-05,
+      "loss": 1.1279,
+      "step": 7669
+    },
+    {
+      "epoch": 0.27893952067498273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6922123149026002e-05,
+      "loss": 1.0667,
+      "step": 7670
+    },
+    {
+      "epoch": 0.27897588827872133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6921273045405323e-05,
+      "loss": 1.0698,
+      "step": 7671
+    },
+    {
+      "epoch": 0.2790122558824599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6920422845761793e-05,
+      "loss": 1.1198,
+      "step": 7672
+    },
+    {
+      "epoch": 0.2790486234861985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6919572550107218e-05,
+      "loss": 1.0786,
+      "step": 7673
+    },
+    {
+      "epoch": 0.27908499108993706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6918722158453394e-05,
+      "loss": 1.0072,
+      "step": 7674
+    },
+    {
+      "epoch": 0.27912135869367566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6917871670812113e-05,
+      "loss": 1.0869,
+      "step": 7675
+    },
+    {
+      "epoch": 0.27915772629741425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6917021087195183e-05,
+      "loss": 1.1546,
+      "step": 7676
+    },
+    {
+      "epoch": 0.27919409390115285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6916170407614393e-05,
+      "loss": 1.0464,
+      "step": 7677
+    },
+    {
+      "epoch": 0.27923046150489145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6915319632081555e-05,
+      "loss": 1.1032,
+      "step": 7678
+    },
+    {
+      "epoch": 0.27926682910863004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6914468760608466e-05,
+      "loss": 1.1382,
+      "step": 7679
+    },
+    {
+      "epoch": 0.27930319671236864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6913617793206934e-05,
+      "loss": 0.9925,
+      "step": 7680
+    },
+    {
+      "epoch": 0.27933956431610724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6912766729888767e-05,
+      "loss": 1.1002,
+      "step": 7681
+    },
+    {
+      "epoch": 0.2793759319198458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6911915570665766e-05,
+      "loss": 1.0518,
+      "step": 7682
+    },
+    {
+      "epoch": 0.2794122995235844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6911064315549747e-05,
+      "loss": 1.1281,
+      "step": 7683
+    },
+    {
+      "epoch": 0.27944866712732297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6910212964552513e-05,
+      "loss": 1.0615,
+      "step": 7684
+    },
+    {
+      "epoch": 0.27948503473106157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6909361517685878e-05,
+      "loss": 1.0754,
+      "step": 7685
+    },
+    {
+      "epoch": 0.27952140233480016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6908509974961655e-05,
+      "loss": 1.0313,
+      "step": 7686
+    },
+    {
+      "epoch": 0.27955776993853876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690765833639166e-05,
+      "loss": 1.0438,
+      "step": 7687
+    },
+    {
+      "epoch": 0.27959413754227735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6906806601987707e-05,
+      "loss": 1.0392,
+      "step": 7688
+    },
+    {
+      "epoch": 0.27963050514601595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690595477176161e-05,
+      "loss": 1.0914,
+      "step": 7689
+    },
+    {
+      "epoch": 0.2796668727497545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690510284572519e-05,
+      "loss": 1.1228,
+      "step": 7690
+    },
+    {
+      "epoch": 0.2797032403534931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6904250823890262e-05,
+      "loss": 0.9598,
+      "step": 7691
+    },
+    {
+      "epoch": 0.2797396079572317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690339870626865e-05,
+      "loss": 1.1987,
+      "step": 7692
+    },
+    {
+      "epoch": 0.2797759755609703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.690254649287218e-05,
+      "loss": 0.9962,
+      "step": 7693
+    },
+    {
+      "epoch": 0.2798123431647089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6901694183712672e-05,
+      "loss": 1.1248,
+      "step": 7694
+    },
+    {
+      "epoch": 0.2798487107684475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6900841778801944e-05,
+      "loss": 1.2037,
+      "step": 7695
+    },
+    {
+      "epoch": 0.27988507837218607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.689998927815183e-05,
+      "loss": 1.0875,
+      "step": 7696
+    },
+    {
+      "epoch": 0.27992144597592467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6899136681774157e-05,
+      "loss": 1.001,
+      "step": 7697
+    },
+    {
+      "epoch": 0.27995781357966326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6898283989680747e-05,
+      "loss": 1.1224,
+      "step": 7698
+    },
+    {
+      "epoch": 0.2799941811834018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6897431201883435e-05,
+      "loss": 1.0242,
+      "step": 7699
+    },
+    {
+      "epoch": 0.2800305487871404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.689657831839405e-05,
+      "loss": 1.0341,
+      "step": 7700
+    },
+    {
+      "epoch": 0.280066916390879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6895725339224423e-05,
+      "loss": 1.022,
+      "step": 7701
+    },
+    {
+      "epoch": 0.2801032839946176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6894872264386396e-05,
+      "loss": 1.0742,
+      "step": 7702
+    },
+    {
+      "epoch": 0.2801396515983562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.68940190938918e-05,
+      "loss": 1.0165,
+      "step": 7703
+    },
+    {
+      "epoch": 0.2801760192020948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6893165827752465e-05,
+      "loss": 1.1484,
+      "step": 7704
+    },
+    {
+      "epoch": 0.2802123868058334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6892312465980238e-05,
+      "loss": 1.1006,
+      "step": 7705
+    },
+    {
+      "epoch": 0.280248754409572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6891459008586956e-05,
+      "loss": 0.9926,
+      "step": 7706
+    },
+    {
+      "epoch": 0.2802851220133105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6890605455584454e-05,
+      "loss": 0.9844,
+      "step": 7707
+    },
+    {
+      "epoch": 0.2803214896170491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688975180698458e-05,
+      "loss": 1.1042,
+      "step": 7708
+    },
+    {
+      "epoch": 0.2803578572207877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6888898062799173e-05,
+      "loss": 1.1802,
+      "step": 7709
+    },
+    {
+      "epoch": 0.2803942248245263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688804422304008e-05,
+      "loss": 1.0972,
+      "step": 7710
+    },
+    {
+      "epoch": 0.2804305924282649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688719028771915e-05,
+      "loss": 1.1824,
+      "step": 7711
+    },
+    {
+      "epoch": 0.2804669600320035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6886336256848223e-05,
+      "loss": 1.1297,
+      "step": 7712
+    },
+    {
+      "epoch": 0.2805033276357421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688548213043915e-05,
+      "loss": 1.0278,
+      "step": 7713
+    },
+    {
+      "epoch": 0.2805396952394807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6884627908503787e-05,
+      "loss": 1.0369,
+      "step": 7714
+    },
+    {
+      "epoch": 0.2805760628432193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6883773591053975e-05,
+      "loss": 1.0647,
+      "step": 7715
+    },
+    {
+      "epoch": 0.28061243044695783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688291917810157e-05,
+      "loss": 1.0442,
+      "step": 7716
+    },
+    {
+      "epoch": 0.2806487980506964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.688206466965843e-05,
+      "loss": 1.0635,
+      "step": 7717
+    },
+    {
+      "epoch": 0.280685165654435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6881210065736405e-05,
+      "loss": 1.2421,
+      "step": 7718
+    },
+    {
+      "epoch": 0.2807215332581736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6880355366347356e-05,
+      "loss": 1.1296,
+      "step": 7719
+    },
+    {
+      "epoch": 0.2807579008619122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6879500571503138e-05,
+      "loss": 1.0826,
+      "step": 7720
+    },
+    {
+      "epoch": 0.2807942684656508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.687864568121561e-05,
+      "loss": 0.9654,
+      "step": 7721
+    },
+    {
+      "epoch": 0.2808306360693894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.687779069549663e-05,
+      "loss": 1.1897,
+      "step": 7722
+    },
+    {
+      "epoch": 0.280867003673128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6876935614358062e-05,
+      "loss": 1.1647,
+      "step": 7723
+    },
+    {
+      "epoch": 0.28090337127686654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6876080437811775e-05,
+      "loss": 0.9835,
+      "step": 7724
+    },
+    {
+      "epoch": 0.28093973888060514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6875225165869623e-05,
+      "loss": 1.068,
+      "step": 7725
+    },
+    {
+      "epoch": 0.28097610648434374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.687436979854348e-05,
+      "loss": 1.0898,
+      "step": 7726
+    },
+    {
+      "epoch": 0.28101247408808233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6873514335845205e-05,
+      "loss": 0.9502,
+      "step": 7727
+    },
+    {
+      "epoch": 0.2810488416918209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6872658777786673e-05,
+      "loss": 1.0849,
+      "step": 7728
+    },
+    {
+      "epoch": 0.2810852092955595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6871803124379753e-05,
+      "loss": 1.0487,
+      "step": 7729
+    },
+    {
+      "epoch": 0.2811215768992981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6870947375636313e-05,
+      "loss": 1.0348,
+      "step": 7730
+    },
+    {
+      "epoch": 0.2811579445030367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6870091531568226e-05,
+      "loss": 1.1359,
+      "step": 7731
+    },
+    {
+      "epoch": 0.28119431210677526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686923559218737e-05,
+      "loss": 1.1625,
+      "step": 7732
+    },
+    {
+      "epoch": 0.28123067971051385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6868379557505612e-05,
+      "loss": 1.048,
+      "step": 7733
+    },
+    {
+      "epoch": 0.28126704731425245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6867523427534835e-05,
+      "loss": 1.097,
+      "step": 7734
+    },
+    {
+      "epoch": 0.28130341491799105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6866667202286916e-05,
+      "loss": 1.1973,
+      "step": 7735
+    },
+    {
+      "epoch": 0.28133978252172964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6865810881773727e-05,
+      "loss": 0.9319,
+      "step": 7736
+    },
+    {
+      "epoch": 0.28137615012546824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6864954466007156e-05,
+      "loss": 1.1551,
+      "step": 7737
+    },
+    {
+      "epoch": 0.28141251772920683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6864097954999082e-05,
+      "loss": 1.1442,
+      "step": 7738
+    },
+    {
+      "epoch": 0.28144888533294543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6863241348761388e-05,
+      "loss": 1.0876,
+      "step": 7739
+    },
+    {
+      "epoch": 0.281485252936684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686238464730596e-05,
+      "loss": 1.1248,
+      "step": 7740
+    },
+    {
+      "epoch": 0.28152162054042257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6861527850644677e-05,
+      "loss": 1.054,
+      "step": 7741
+    },
+    {
+      "epoch": 0.28155798814416116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.686067095878943e-05,
+      "loss": 1.0549,
+      "step": 7742
+    },
+    {
+      "epoch": 0.28159435574789976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685981397175211e-05,
+      "loss": 1.0927,
+      "step": 7743
+    },
+    {
+      "epoch": 0.28163072335163836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6858956889544608e-05,
+      "loss": 1.1228,
+      "step": 7744
+    },
+    {
+      "epoch": 0.28166709095537695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6858099712178805e-05,
+      "loss": 1.0371,
+      "step": 7745
+    },
+    {
+      "epoch": 0.28170345855911555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.68572424396666e-05,
+      "loss": 1.0174,
+      "step": 7746
+    },
+    {
+      "epoch": 0.28173982616285415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6856385072019887e-05,
+      "loss": 1.2437,
+      "step": 7747
+    },
+    {
+      "epoch": 0.28177619376659274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6855527609250554e-05,
+      "loss": 1.1228,
+      "step": 7748
+    },
+    {
+      "epoch": 0.2818125613703313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6854670051370506e-05,
+      "loss": 1.0712,
+      "step": 7749
+    },
+    {
+      "epoch": 0.2818489289740699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6853812398391637e-05,
+      "loss": 1.0609,
+      "step": 7750
+    },
+    {
+      "epoch": 0.2818852965778085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6852954650325846e-05,
+      "loss": 1.0811,
+      "step": 7751
+    },
+    {
+      "epoch": 0.28192166418154707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685209680718503e-05,
+      "loss": 1.091,
+      "step": 7752
+    },
+    {
+      "epoch": 0.28195803178528567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6851238868981095e-05,
+      "loss": 1.1245,
+      "step": 7753
+    },
+    {
+      "epoch": 0.28199439938902426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6850380835725937e-05,
+      "loss": 1.0202,
+      "step": 7754
+    },
+    {
+      "epoch": 0.28203076699276286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.684952270743147e-05,
+      "loss": 1.0391,
+      "step": 7755
+    },
+    {
+      "epoch": 0.28206713459650146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.684866448410959e-05,
+      "loss": 1.1296,
+      "step": 7756
+    },
+    {
+      "epoch": 0.28210350220024005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.684780616577221e-05,
+      "loss": 1.1326,
+      "step": 7757
+    },
+    {
+      "epoch": 0.2821398698039786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6846947752431234e-05,
+      "loss": 1.0308,
+      "step": 7758
+    },
+    {
+      "epoch": 0.2821762374077172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6846089244098573e-05,
+      "loss": 0.978,
+      "step": 7759
+    },
+    {
+      "epoch": 0.2822126050114558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.684523064078613e-05,
+      "loss": 1.0107,
+      "step": 7760
+    },
+    {
+      "epoch": 0.2822489726151944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6844371942505834e-05,
+      "loss": 1.0618,
+      "step": 7761
+    },
+    {
+      "epoch": 0.282285340218933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6843513149269585e-05,
+      "loss": 1.0694,
+      "step": 7762
+    },
+    {
+      "epoch": 0.2823217078226716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6842654261089298e-05,
+      "loss": 1.0902,
+      "step": 7763
+    },
+    {
+      "epoch": 0.28235807542641017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6841795277976892e-05,
+      "loss": 1.1464,
+      "step": 7764
+    },
+    {
+      "epoch": 0.28239444303014877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6840936199944287e-05,
+      "loss": 1.0646,
+      "step": 7765
+    },
+    {
+      "epoch": 0.2824308106338873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6840077027003395e-05,
+      "loss": 1.0914,
+      "step": 7766
+    },
+    {
+      "epoch": 0.2824671782376259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.683921775916614e-05,
+      "loss": 1.1693,
+      "step": 7767
+    },
+    {
+      "epoch": 0.2825035458413645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6838358396444443e-05,
+      "loss": 1.1157,
+      "step": 7768
+    },
+    {
+      "epoch": 0.2825399134451031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6837498938850228e-05,
+      "loss": 1.0812,
+      "step": 7769
+    },
+    {
+      "epoch": 0.2825762810488417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.683663938639541e-05,
+      "loss": 1.0767,
+      "step": 7770
+    },
+    {
+      "epoch": 0.2826126486525803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6835779739091925e-05,
+      "loss": 1.0663,
+      "step": 7771
+    },
+    {
+      "epoch": 0.2826490162563189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6834919996951696e-05,
+      "loss": 1.0784,
+      "step": 7772
+    },
+    {
+      "epoch": 0.2826853838600575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6834060159986647e-05,
+      "loss": 1.1525,
+      "step": 7773
+    },
+    {
+      "epoch": 0.282721751463796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6833200228208708e-05,
+      "loss": 1.041,
+      "step": 7774
+    },
+    {
+      "epoch": 0.2827581190675346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6832340201629814e-05,
+      "loss": 1.1465,
+      "step": 7775
+    },
+    {
+      "epoch": 0.2827944866712732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6831480080261893e-05,
+      "loss": 1.0103,
+      "step": 7776
+    },
+    {
+      "epoch": 0.2828308542750118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6830619864116876e-05,
+      "loss": 1.1746,
+      "step": 7777
+    },
+    {
+      "epoch": 0.2828672218787504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6829759553206705e-05,
+      "loss": 1.0557,
+      "step": 7778
+    },
+    {
+      "epoch": 0.282903589482489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6828899147543305e-05,
+      "loss": 1.1107,
+      "step": 7779
+    },
+    {
+      "epoch": 0.2829399570862276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.682803864713862e-05,
+      "loss": 1.1613,
+      "step": 7780
+    },
+    {
+      "epoch": 0.2829763246899662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6827178052004588e-05,
+      "loss": 1.1346,
+      "step": 7781
+    },
+    {
+      "epoch": 0.2830126922937048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6826317362153143e-05,
+      "loss": 1.101,
+      "step": 7782
+    },
+    {
+      "epoch": 0.28304905989744333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6825456577596235e-05,
+      "loss": 1.0558,
+      "step": 7783
+    },
+    {
+      "epoch": 0.28308542750118193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6824595698345798e-05,
+      "loss": 1.1027,
+      "step": 7784
+    },
+    {
+      "epoch": 0.2831217951049205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6823734724413777e-05,
+      "loss": 1.0214,
+      "step": 7785
+    },
+    {
+      "epoch": 0.2831581627086591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6822873655812123e-05,
+      "loss": 1.1217,
+      "step": 7786
+    },
+    {
+      "epoch": 0.2831945303123977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6822012492552775e-05,
+      "loss": 1.1162,
+      "step": 7787
+    },
+    {
+      "epoch": 0.2832308979161363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6821151234647683e-05,
+      "loss": 1.0434,
+      "step": 7788
+    },
+    {
+      "epoch": 0.2832672655198749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6820289882108794e-05,
+      "loss": 1.0707,
+      "step": 7789
+    },
+    {
+      "epoch": 0.2833036331236135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6819428434948064e-05,
+      "loss": 1.0774,
+      "step": 7790
+    },
+    {
+      "epoch": 0.28334000072735205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6818566893177438e-05,
+      "loss": 1.1751,
+      "step": 7791
+    },
+    {
+      "epoch": 0.28337636833109064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6817705256808867e-05,
+      "loss": 1.1692,
+      "step": 7792
+    },
+    {
+      "epoch": 0.28341273593482924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.681684352585431e-05,
+      "loss": 1.1929,
+      "step": 7793
+    },
+    {
+      "epoch": 0.28344910353856784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6815981700325724e-05,
+      "loss": 1.1236,
+      "step": 7794
+    },
+    {
+      "epoch": 0.28348547114230643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6815119780235058e-05,
+      "loss": 0.9952,
+      "step": 7795
+    },
+    {
+      "epoch": 0.28352183874604503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.681425776559428e-05,
+      "loss": 1.0465,
+      "step": 7796
+    },
+    {
+      "epoch": 0.2835582063497836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.681339565641534e-05,
+      "loss": 1.2195,
+      "step": 7797
+    },
+    {
+      "epoch": 0.2835945739535222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6812533452710202e-05,
+      "loss": 1.0803,
+      "step": 7798
+    },
+    {
+      "epoch": 0.2836309415572608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6811671154490828e-05,
+      "loss": 1.2049,
+      "step": 7799
+    },
+    {
+      "epoch": 0.28366730916099936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6810808761769185e-05,
+      "loss": 1.0375,
+      "step": 7800
+    },
+    {
+      "epoch": 0.28370367676473796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6809946274557228e-05,
+      "loss": 1.1158,
+      "step": 7801
+    },
+    {
+      "epoch": 0.28374004436847655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6809083692866935e-05,
+      "loss": 1.1084,
+      "step": 7802
+    },
+    {
+      "epoch": 0.28377641197221515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680822101671026e-05,
+      "loss": 1.1275,
+      "step": 7803
+    },
+    {
+      "epoch": 0.28381277957595374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680735824609918e-05,
+      "loss": 1.0564,
+      "step": 7804
+    },
+    {
+      "epoch": 0.28384914717969234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6806495381045668e-05,
+      "loss": 1.0756,
+      "step": 7805
+    },
+    {
+      "epoch": 0.28388551478343094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6805632421561686e-05,
+      "loss": 1.0435,
+      "step": 7806
+    },
+    {
+      "epoch": 0.28392188238716953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6804769367659205e-05,
+      "loss": 1.0717,
+      "step": 7807
+    },
+    {
+      "epoch": 0.2839582499909081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680390621935021e-05,
+      "loss": 1.1693,
+      "step": 7808
+    },
+    {
+      "epoch": 0.28399461759464667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6803042976646667e-05,
+      "loss": 1.0118,
+      "step": 7809
+    },
+    {
+      "epoch": 0.28403098519838527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6802179639560557e-05,
+      "loss": 1.2147,
+      "step": 7810
+    },
+    {
+      "epoch": 0.28406735280212386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6801316208103854e-05,
+      "loss": 1.2036,
+      "step": 7811
+    },
+    {
+      "epoch": 0.28410372040586246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680045268228854e-05,
+      "loss": 1.0102,
+      "step": 7812
+    },
+    {
+      "epoch": 0.28414008800960106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6799589062126595e-05,
+      "loss": 1.1181,
+      "step": 7813
+    },
+    {
+      "epoch": 0.28417645561333965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6798725347629993e-05,
+      "loss": 1.152,
+      "step": 7814
+    },
+    {
+      "epoch": 0.28421282321707825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6797861538810726e-05,
+      "loss": 1.0725,
+      "step": 7815
+    },
+    {
+      "epoch": 0.28424919082081684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6796997635680775e-05,
+      "loss": 1.0664,
+      "step": 7816
+    },
+    {
+      "epoch": 0.2842855584245554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6796133638252126e-05,
+      "loss": 1.0762,
+      "step": 7817
+    },
+    {
+      "epoch": 0.284321926028294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6795269546536768e-05,
+      "loss": 0.9235,
+      "step": 7818
+    },
+    {
+      "epoch": 0.2843582936320326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6794405360546682e-05,
+      "loss": 1.0575,
+      "step": 7819
+    },
+    {
+      "epoch": 0.2843946612357712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6793541080293865e-05,
+      "loss": 1.019,
+      "step": 7820
+    },
+    {
+      "epoch": 0.28443102883950977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6792676705790302e-05,
+      "loss": 1.0926,
+      "step": 7821
+    },
+    {
+      "epoch": 0.28446739644324837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6791812237047988e-05,
+      "loss": 1.07,
+      "step": 7822
+    },
+    {
+      "epoch": 0.28450376404698696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6790947674078913e-05,
+      "loss": 1.2498,
+      "step": 7823
+    },
+    {
+      "epoch": 0.28454013165072556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.679008301689508e-05,
+      "loss": 0.9676,
+      "step": 7824
+    },
+    {
+      "epoch": 0.2845764992544641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6789218265508472e-05,
+      "loss": 1.0455,
+      "step": 7825
+    },
+    {
+      "epoch": 0.2846128668582027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67883534199311e-05,
+      "loss": 1.0678,
+      "step": 7826
+    },
+    {
+      "epoch": 0.2846492344619413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678748848017495e-05,
+      "loss": 1.0991,
+      "step": 7827
+    },
+    {
+      "epoch": 0.2846856020656799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678662344625203e-05,
+      "loss": 1.0874,
+      "step": 7828
+    },
+    {
+      "epoch": 0.2847219696694185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678575831817434e-05,
+      "loss": 1.0359,
+      "step": 7829
+    },
+    {
+      "epoch": 0.2847583372731571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6784893095953878e-05,
+      "loss": 1.2518,
+      "step": 7830
+    },
+    {
+      "epoch": 0.2847947048768957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6784027779602654e-05,
+      "loss": 1.1459,
+      "step": 7831
+    },
+    {
+      "epoch": 0.2848310724806343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6783162369132667e-05,
+      "loss": 1.094,
+      "step": 7832
+    },
+    {
+      "epoch": 0.2848674400843728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6782296864555924e-05,
+      "loss": 1.0553,
+      "step": 7833
+    },
+    {
+      "epoch": 0.2849038076881114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678143126588444e-05,
+      "loss": 1.1025,
+      "step": 7834
+    },
+    {
+      "epoch": 0.28494017529185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6780565573130218e-05,
+      "loss": 0.9737,
+      "step": 7835
+    },
+    {
+      "epoch": 0.2849765428955886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6779699786305268e-05,
+      "loss": 1.1443,
+      "step": 7836
+    },
+    {
+      "epoch": 0.2850129104993272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67788339054216e-05,
+      "loss": 1.0758,
+      "step": 7837
+    },
+    {
+      "epoch": 0.2850492781030658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6777967930491233e-05,
+      "loss": 1.2001,
+      "step": 7838
+    },
+    {
+      "epoch": 0.2850856457068044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6777101861526177e-05,
+      "loss": 1.2479,
+      "step": 7839
+    },
+    {
+      "epoch": 0.285122013310543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.677623569853845e-05,
+      "loss": 0.9838,
+      "step": 7840
+    },
+    {
+      "epoch": 0.2851583809142816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6775369441540065e-05,
+      "loss": 1.0593,
+      "step": 7841
+    },
+    {
+      "epoch": 0.2851947485180201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.677450309054304e-05,
+      "loss": 1.1553,
+      "step": 7842
+    },
+    {
+      "epoch": 0.2852311161217587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6773636645559398e-05,
+      "loss": 0.9399,
+      "step": 7843
+    },
+    {
+      "epoch": 0.2852674837254973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6772770106601157e-05,
+      "loss": 0.9781,
+      "step": 7844
+    },
+    {
+      "epoch": 0.2853038513292359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6771903473680342e-05,
+      "loss": 1.1577,
+      "step": 7845
+    },
+    {
+      "epoch": 0.2853402189329745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6771036746808974e-05,
+      "loss": 1.0132,
+      "step": 7846
+    },
+    {
+      "epoch": 0.2853765865367131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6770169925999077e-05,
+      "loss": 1.1957,
+      "step": 7847
+    },
+    {
+      "epoch": 0.2854129541404517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6769303011262682e-05,
+      "loss": 1.1441,
+      "step": 7848
+    },
+    {
+      "epoch": 0.2854493217441903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6768436002611808e-05,
+      "loss": 1.1347,
+      "step": 7849
+    },
+    {
+      "epoch": 0.28548568934792884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6767568900058488e-05,
+      "loss": 1.1433,
+      "step": 7850
+    },
+    {
+      "epoch": 0.28552205695166744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.676670170361475e-05,
+      "loss": 1.1445,
+      "step": 7851
+    },
+    {
+      "epoch": 0.28555842455540603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.676583441329263e-05,
+      "loss": 1.0574,
+      "step": 7852
+    },
+    {
+      "epoch": 0.28559479215914463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6764967029104153e-05,
+      "loss": 1.11,
+      "step": 7853
+    },
+    {
+      "epoch": 0.2856311597628832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6764099551061358e-05,
+      "loss": 1.1518,
+      "step": 7854
+    },
+    {
+      "epoch": 0.2856675273666218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.676323197917628e-05,
+      "loss": 1.1138,
+      "step": 7855
+    },
+    {
+      "epoch": 0.2857038949703604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6762364313460956e-05,
+      "loss": 1.0013,
+      "step": 7856
+    },
+    {
+      "epoch": 0.285740262574099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6761496553927416e-05,
+      "loss": 1.0249,
+      "step": 7857
+    },
+    {
+      "epoch": 0.2857766301778376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.676062870058771e-05,
+      "loss": 1.2062,
+      "step": 7858
+    },
+    {
+      "epoch": 0.28581299778157615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.675976075345387e-05,
+      "loss": 1.0294,
+      "step": 7859
+    },
+    {
+      "epoch": 0.28584936538531475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6758892712537943e-05,
+      "loss": 1.0761,
+      "step": 7860
+    },
+    {
+      "epoch": 0.28588573298905334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6758024577851965e-05,
+      "loss": 1.0508,
+      "step": 7861
+    },
+    {
+      "epoch": 0.28592210059279194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6757156349407988e-05,
+      "loss": 1.2611,
+      "step": 7862
+    },
+    {
+      "epoch": 0.28595846819653054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6756288027218053e-05,
+      "loss": 1.0608,
+      "step": 7863
+    },
+    {
+      "epoch": 0.28599483580026913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6755419611294206e-05,
+      "loss": 1.1475,
+      "step": 7864
+    },
+    {
+      "epoch": 0.28603120340400773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6754551101648497e-05,
+      "loss": 1.2181,
+      "step": 7865
+    },
+    {
+      "epoch": 0.2860675710077463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6753682498292975e-05,
+      "loss": 1.0629,
+      "step": 7866
+    },
+    {
+      "epoch": 0.28610393861148486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6752813801239692e-05,
+      "loss": 1.0871,
+      "step": 7867
+    },
+    {
+      "epoch": 0.28614030621522346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67519450105007e-05,
+      "loss": 1.2493,
+      "step": 7868
+    },
+    {
+      "epoch": 0.28617667381896206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6751076126088048e-05,
+      "loss": 0.9201,
+      "step": 7869
+    },
+    {
+      "epoch": 0.28621304142270065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6750207148013797e-05,
+      "loss": 1.1788,
+      "step": 7870
+    },
+    {
+      "epoch": 0.28624940902643925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674933807628999e-05,
+      "loss": 1.1136,
+      "step": 7871
+    },
+    {
+      "epoch": 0.28628577663017785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67484689109287e-05,
+      "loss": 1.0975,
+      "step": 7872
+    },
+    {
+      "epoch": 0.28632214423391644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674759965194198e-05,
+      "loss": 1.0517,
+      "step": 7873
+    },
+    {
+      "epoch": 0.28635851183765504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6746730299341886e-05,
+      "loss": 0.9681,
+      "step": 7874
+    },
+    {
+      "epoch": 0.2863948794413936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674586085314048e-05,
+      "loss": 1.129,
+      "step": 7875
+    },
+    {
+      "epoch": 0.2864312470451322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674499131334983e-05,
+      "loss": 1.1731,
+      "step": 7876
+    },
+    {
+      "epoch": 0.28646761464887077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6744121679981992e-05,
+      "loss": 1.1508,
+      "step": 7877
+    },
+    {
+      "epoch": 0.28650398225260937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6743251953049036e-05,
+      "loss": 1.1382,
+      "step": 7878
+    },
+    {
+      "epoch": 0.28654034985634796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6742382132563023e-05,
+      "loss": 1.1821,
+      "step": 7879
+    },
+    {
+      "epoch": 0.28657671746008656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674151221853603e-05,
+      "loss": 1.0344,
+      "step": 7880
+    },
+    {
+      "epoch": 0.28661308506382516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6740642210980117e-05,
+      "loss": 1.0984,
+      "step": 7881
+    },
+    {
+      "epoch": 0.28664945266756375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6739772109907355e-05,
+      "loss": 0.9339,
+      "step": 7882
+    },
+    {
+      "epoch": 0.28668582027130235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.673890191532982e-05,
+      "loss": 1.0843,
+      "step": 7883
+    },
+    {
+      "epoch": 0.2867221878750409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6738031627259583e-05,
+      "loss": 1.18,
+      "step": 7884
+    },
+    {
+      "epoch": 0.2867585554787795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6737161245708713e-05,
+      "loss": 1.0271,
+      "step": 7885
+    },
+    {
+      "epoch": 0.2867949230825181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6736290770689295e-05,
+      "loss": 1.0743,
+      "step": 7886
+    },
+    {
+      "epoch": 0.2868312906862567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6735420202213396e-05,
+      "loss": 1.1283,
+      "step": 7887
+    },
+    {
+      "epoch": 0.2868676582899953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67345495402931e-05,
+      "loss": 1.0487,
+      "step": 7888
+    },
+    {
+      "epoch": 0.28690402589373387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.673367878494048e-05,
+      "loss": 1.113,
+      "step": 7889
+    },
+    {
+      "epoch": 0.28694039349747247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6732807936167626e-05,
+      "loss": 1.0416,
+      "step": 7890
+    },
+    {
+      "epoch": 0.28697676110121106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6731936993986614e-05,
+      "loss": 1.0146,
+      "step": 7891
+    },
+    {
+      "epoch": 0.2870131287049496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6731065958409524e-05,
+      "loss": 0.9344,
+      "step": 7892
+    },
+    {
+      "epoch": 0.2870494963086882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6730194829448446e-05,
+      "loss": 1.0959,
+      "step": 7893
+    },
+    {
+      "epoch": 0.2870858639124268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6729323607115465e-05,
+      "loss": 1.003,
+      "step": 7894
+    },
+    {
+      "epoch": 0.2871222315161654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6728452291422664e-05,
+      "loss": 1.1003,
+      "step": 7895
+    },
+    {
+      "epoch": 0.287158599119904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6727580882382137e-05,
+      "loss": 0.9537,
+      "step": 7896
+    },
+    {
+      "epoch": 0.2871949667236426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6726709380005968e-05,
+      "loss": 1.1246,
+      "step": 7897
+    },
+    {
+      "epoch": 0.2872313343273812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6725837784306254e-05,
+      "loss": 0.973,
+      "step": 7898
+    },
+    {
+      "epoch": 0.2872677019311198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672496609529508e-05,
+      "loss": 1.1701,
+      "step": 7899
+    },
+    {
+      "epoch": 0.2873040695348584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672409431298454e-05,
+      "loss": 1.0819,
+      "step": 7900
+    },
+    {
+      "epoch": 0.2873404371385969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6723222437386736e-05,
+      "loss": 1.0366,
+      "step": 7901
+    },
+    {
+      "epoch": 0.2873768047423355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6722350468513758e-05,
+      "loss": 1.103,
+      "step": 7902
+    },
+    {
+      "epoch": 0.2874131723460741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6721478406377706e-05,
+      "loss": 1.0987,
+      "step": 7903
+    },
+    {
+      "epoch": 0.2874495399498127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672060625099068e-05,
+      "loss": 1.0373,
+      "step": 7904
+    },
+    {
+      "epoch": 0.2874859075535513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6719734002364775e-05,
+      "loss": 1.108,
+      "step": 7905
+    },
+    {
+      "epoch": 0.2875222751572899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6718861660512094e-05,
+      "loss": 1.0588,
+      "step": 7906
+    },
+    {
+      "epoch": 0.2875586427610285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6717989225444742e-05,
+      "loss": 1.0165,
+      "step": 7907
+    },
+    {
+      "epoch": 0.2875950103647671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.671711669717482e-05,
+      "loss": 1.1912,
+      "step": 7908
+    },
+    {
+      "epoch": 0.28763137796850563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6716244075714433e-05,
+      "loss": 1.0284,
+      "step": 7909
+    },
+    {
+      "epoch": 0.2876677455722442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.671537136107569e-05,
+      "loss": 1.0258,
+      "step": 7910
+    },
+    {
+      "epoch": 0.2877041131759828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6714498553270698e-05,
+      "loss": 1.0167,
+      "step": 7911
+    },
+    {
+      "epoch": 0.2877404807797214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6713625652311566e-05,
+      "loss": 1.0754,
+      "step": 7912
+    },
+    {
+      "epoch": 0.28777684838346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.67127526582104e-05,
+      "loss": 1.0536,
+      "step": 7913
+    },
+    {
+      "epoch": 0.2878132159871986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6711879570979317e-05,
+      "loss": 1.0061,
+      "step": 7914
+    },
+    {
+      "epoch": 0.2878495835909372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6711006390630426e-05,
+      "loss": 1.152,
+      "step": 7915
+    },
+    {
+      "epoch": 0.2878859511946758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6710133117175846e-05,
+      "loss": 1.0618,
+      "step": 7916
+    },
+    {
+      "epoch": 0.28792231879841435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670925975062769e-05,
+      "loss": 1.0688,
+      "step": 7917
+    },
+    {
+      "epoch": 0.28795868640215294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670838629099807e-05,
+      "loss": 0.998,
+      "step": 7918
+    },
+    {
+      "epoch": 0.28799505400589154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670751273829911e-05,
+      "loss": 0.9883,
+      "step": 7919
+    },
+    {
+      "epoch": 0.28803142160963013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6706639092542927e-05,
+      "loss": 1.1026,
+      "step": 7920
+    },
+    {
+      "epoch": 0.28806778921336873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6705765353741643e-05,
+      "loss": 1.0392,
+      "step": 7921
+    },
+    {
+      "epoch": 0.2881041568171073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670489152190738e-05,
+      "loss": 1.0767,
+      "step": 7922
+    },
+    {
+      "epoch": 0.2881405244208459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670401759705226e-05,
+      "loss": 1.0894,
+      "step": 7923
+    },
+    {
+      "epoch": 0.2881768920245845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6703143579188407e-05,
+      "loss": 1.0795,
+      "step": 7924
+    },
+    {
+      "epoch": 0.2882132596283231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6702269468327948e-05,
+      "loss": 1.2001,
+      "step": 7925
+    },
+    {
+      "epoch": 0.28824962723206166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6701395264483007e-05,
+      "loss": 1.1749,
+      "step": 7926
+    },
+    {
+      "epoch": 0.28828599483580025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6700520967665718e-05,
+      "loss": 1.1111,
+      "step": 7927
+    },
+    {
+      "epoch": 0.28832236243953885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6699646577888202e-05,
+      "loss": 0.9526,
+      "step": 7928
+    },
+    {
+      "epoch": 0.28835873004327744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.66987720951626e-05,
+      "loss": 1.1257,
+      "step": 7929
+    },
+    {
+      "epoch": 0.28839509764701604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6697897519501038e-05,
+      "loss": 1.1358,
+      "step": 7930
+    },
+    {
+      "epoch": 0.28843146525075464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.669702285091565e-05,
+      "loss": 1.0738,
+      "step": 7931
+    },
+    {
+      "epoch": 0.28846783285449323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6696148089418574e-05,
+      "loss": 1.0401,
+      "step": 7932
+    },
+    {
+      "epoch": 0.28850420045823183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6695273235021945e-05,
+      "loss": 1.0806,
+      "step": 7933
+    },
+    {
+      "epoch": 0.28854056806197037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6694398287737897e-05,
+      "loss": 0.9738,
+      "step": 7934
+    },
+    {
+      "epoch": 0.28857693566570897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.669352324757857e-05,
+      "loss": 1.1046,
+      "step": 7935
+    },
+    {
+      "epoch": 0.28861330326944756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6692648114556105e-05,
+      "loss": 1.0182,
+      "step": 7936
+    },
+    {
+      "epoch": 0.28864967087318616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6691772888682643e-05,
+      "loss": 1.0859,
+      "step": 7937
+    },
+    {
+      "epoch": 0.28868603847692476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.669089756997033e-05,
+      "loss": 1.1028,
+      "step": 7938
+    },
+    {
+      "epoch": 0.28872240608066335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6690022158431304e-05,
+      "loss": 1.0615,
+      "step": 7939
+    },
+    {
+      "epoch": 0.28875877368440195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6689146654077712e-05,
+      "loss": 1.0887,
+      "step": 7940
+    },
+    {
+      "epoch": 0.28879514128814054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.66882710569217e-05,
+      "loss": 1.1195,
+      "step": 7941
+    },
+    {
+      "epoch": 0.28883150889187914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6687395366975417e-05,
+      "loss": 1.1529,
+      "step": 7942
+    },
+    {
+      "epoch": 0.2888678764956177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6686519584251013e-05,
+      "loss": 1.0922,
+      "step": 7943
+    },
+    {
+      "epoch": 0.2889042440993563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668564370876064e-05,
+      "loss": 1.045,
+      "step": 7944
+    },
+    {
+      "epoch": 0.2889406117030949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668476774051644e-05,
+      "loss": 1.128,
+      "step": 7945
+    },
+    {
+      "epoch": 0.28897697930683347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6683891679530573e-05,
+      "loss": 1.0659,
+      "step": 7946
+    },
+    {
+      "epoch": 0.28901334691057207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668301552581519e-05,
+      "loss": 1.1817,
+      "step": 7947
+    },
+    {
+      "epoch": 0.28904971451431066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668213927938245e-05,
+      "loss": 1.0549,
+      "step": 7948
+    },
+    {
+      "epoch": 0.28908608211804926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6681262940244514e-05,
+      "loss": 1.1232,
+      "step": 7949
+    },
+    {
+      "epoch": 0.28912244972178786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.668038650841353e-05,
+      "loss": 1.2034,
+      "step": 7950
+    },
+    {
+      "epoch": 0.2891588173255264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.667950998390166e-05,
+      "loss": 1.0067,
+      "step": 7951
+    },
+    {
+      "epoch": 0.289195184929265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6678633366721068e-05,
+      "loss": 1.0106,
+      "step": 7952
+    },
+    {
+      "epoch": 0.2892315525330036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6677756656883915e-05,
+      "loss": 1.0462,
+      "step": 7953
+    },
+    {
+      "epoch": 0.2892679201367422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.667687985440236e-05,
+      "loss": 1.0508,
+      "step": 7954
+    },
+    {
+      "epoch": 0.2893042877404808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.667600295928857e-05,
+      "loss": 1.1984,
+      "step": 7955
+    },
+    {
+      "epoch": 0.2893406553442194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6675125971554713e-05,
+      "loss": 0.8975,
+      "step": 7956
+    },
+    {
+      "epoch": 0.289377022947958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6674248891212954e-05,
+      "loss": 1.0883,
+      "step": 7957
+    },
+    {
+      "epoch": 0.28941339055169657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6673371718275458e-05,
+      "loss": 0.897,
+      "step": 7958
+    },
+    {
+      "epoch": 0.2894497581554351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.66724944527544e-05,
+      "loss": 1.1867,
+      "step": 7959
+    },
+    {
+      "epoch": 0.2894861257591737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.667161709466195e-05,
+      "loss": 1.0992,
+      "step": 7960
+    },
+    {
+      "epoch": 0.2895224933629123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6670739644010276e-05,
+      "loss": 1.1427,
+      "step": 7961
+    },
+    {
+      "epoch": 0.2895588609666509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6669862100811555e-05,
+      "loss": 0.9895,
+      "step": 7962
+    },
+    {
+      "epoch": 0.2895952285703895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6668984465077963e-05,
+      "loss": 1.0489,
+      "step": 7963
+    },
+    {
+      "epoch": 0.2896315961741281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666810673682167e-05,
+      "loss": 1.0313,
+      "step": 7964
+    },
+    {
+      "epoch": 0.2896679637778667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666722891605486e-05,
+      "loss": 1.2294,
+      "step": 7965
+    },
+    {
+      "epoch": 0.2897043313816053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6666351002789706e-05,
+      "loss": 1.1087,
+      "step": 7966
+    },
+    {
+      "epoch": 0.2897406989853439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666547299703839e-05,
+      "loss": 0.9663,
+      "step": 7967
+    },
+    {
+      "epoch": 0.2897770665890824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6664594898813094e-05,
+      "loss": 1.1176,
+      "step": 7968
+    },
+    {
+      "epoch": 0.289813434192821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6663716708126e-05,
+      "loss": 1.0632,
+      "step": 7969
+    },
+    {
+      "epoch": 0.2898498017965596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666283842498929e-05,
+      "loss": 1.1184,
+      "step": 7970
+    },
+    {
+      "epoch": 0.2898861694002982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666196004941515e-05,
+      "loss": 1.1024,
+      "step": 7971
+    },
+    {
+      "epoch": 0.2899225370040368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6661081581415768e-05,
+      "loss": 0.9823,
+      "step": 7972
+    },
+    {
+      "epoch": 0.2899589046077754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.666020302100333e-05,
+      "loss": 1.2016,
+      "step": 7973
+    },
+    {
+      "epoch": 0.289995272211514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6659324368190024e-05,
+      "loss": 1.0485,
+      "step": 7974
+    },
+    {
+      "epoch": 0.2900316398152526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.665844562298804e-05,
+      "loss": 1.0375,
+      "step": 7975
+    },
+    {
+      "epoch": 0.29006800741899114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.665756678540957e-05,
+      "loss": 1.0016,
+      "step": 7976
+    },
+    {
+      "epoch": 0.29010437502272973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6656687855466807e-05,
+      "loss": 1.1135,
+      "step": 7977
+    },
+    {
+      "epoch": 0.29014074262646833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6655808833171947e-05,
+      "loss": 1.0745,
+      "step": 7978
+    },
+    {
+      "epoch": 0.2901771102302069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.665492971853718e-05,
+      "loss": 1.1035,
+      "step": 7979
+    },
+    {
+      "epoch": 0.2902134778339455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6654050511574707e-05,
+      "loss": 1.1318,
+      "step": 7980
+    },
+    {
+      "epoch": 0.2902498454376841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6653171212296722e-05,
+      "loss": 1.0219,
+      "step": 7981
+    },
+    {
+      "epoch": 0.2902862130414227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.665229182071543e-05,
+      "loss": 0.9584,
+      "step": 7982
+    },
+    {
+      "epoch": 0.2903225806451613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6651412336843025e-05,
+      "loss": 0.9476,
+      "step": 7983
+    },
+    {
+      "epoch": 0.2903589482488999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6650532760691708e-05,
+      "loss": 1.1724,
+      "step": 7984
+    },
+    {
+      "epoch": 0.29039531585263845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6649653092273688e-05,
+      "loss": 1.0196,
+      "step": 7985
+    },
+    {
+      "epoch": 0.29043168345637704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6648773331601166e-05,
+      "loss": 0.9991,
+      "step": 7986
+    },
+    {
+      "epoch": 0.29046805106011564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6647893478686346e-05,
+      "loss": 1.1962,
+      "step": 7987
+    },
+    {
+      "epoch": 0.29050441866385424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6647013533541436e-05,
+      "loss": 1.0453,
+      "step": 7988
+    },
+    {
+      "epoch": 0.29054078626759283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6646133496178646e-05,
+      "loss": 1.156,
+      "step": 7989
+    },
+    {
+      "epoch": 0.29057715387133143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.664525336661018e-05,
+      "loss": 1.1666,
+      "step": 7990
+    },
+    {
+      "epoch": 0.29061352147507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6644373144848257e-05,
+      "loss": 1.0133,
+      "step": 7991
+    },
+    {
+      "epoch": 0.2906498890788086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.664349283090508e-05,
+      "loss": 1.0313,
+      "step": 7992
+    },
+    {
+      "epoch": 0.29068625668254716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.664261242479287e-05,
+      "loss": 1.0793,
+      "step": 7993
+    },
+    {
+      "epoch": 0.29072262428628576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.664173192652383e-05,
+      "loss": 1.0514,
+      "step": 7994
+    },
+    {
+      "epoch": 0.29075899189002435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.664085133611019e-05,
+      "loss": 1.0342,
+      "step": 7995
+    },
+    {
+      "epoch": 0.29079535949376295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6639970653564155e-05,
+      "loss": 1.0104,
+      "step": 7996
+    },
+    {
+      "epoch": 0.29083172709750155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663908987889795e-05,
+      "loss": 0.9663,
+      "step": 7997
+    },
+    {
+      "epoch": 0.29086809470124014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663820901212379e-05,
+      "loss": 1.1565,
+      "step": 7998
+    },
+    {
+      "epoch": 0.29090446230497874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.66373280532539e-05,
+      "loss": 1.1659,
+      "step": 7999
+    },
+    {
+      "epoch": 0.29094082990871734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6636447002300505e-05,
+      "loss": 1.0865,
+      "step": 8000
+    },
+    {
+      "epoch": 0.29097719751245593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663556585927582e-05,
+      "loss": 1.0828,
+      "step": 8001
+    },
+    {
+      "epoch": 0.2910135651161945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6634684624192075e-05,
+      "loss": 1.1784,
+      "step": 8002
+    },
+    {
+      "epoch": 0.29104993271993307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.663380329706149e-05,
+      "loss": 1.0763,
+      "step": 8003
+    },
+    {
+      "epoch": 0.29108630032367167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.66329218778963e-05,
+      "loss": 1.1135,
+      "step": 8004
+    },
+    {
+      "epoch": 0.29112266792741026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6632040366708727e-05,
+      "loss": 0.9597,
+      "step": 8005
+    },
+    {
+      "epoch": 0.29115903553114886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6631158763511005e-05,
+      "loss": 1.0793,
+      "step": 8006
+    },
+    {
+      "epoch": 0.29119540313488745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6630277068315365e-05,
+      "loss": 1.0353,
+      "step": 8007
+    },
+    {
+      "epoch": 0.29123177073862605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6629395281134037e-05,
+      "loss": 1.0426,
+      "step": 8008
+    },
+    {
+      "epoch": 0.29126813834236465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6628513401979252e-05,
+      "loss": 1.1445,
+      "step": 8009
+    },
+    {
+      "epoch": 0.2913045059461032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6627631430863252e-05,
+      "loss": 1.0792,
+      "step": 8010
+    },
+    {
+      "epoch": 0.2913408735498418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.662674936779827e-05,
+      "loss": 1.1199,
+      "step": 8011
+    },
+    {
+      "epoch": 0.2913772411535804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.662586721279654e-05,
+      "loss": 1.1892,
+      "step": 8012
+    },
+    {
+      "epoch": 0.291413608757319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6624984965870305e-05,
+      "loss": 1.1904,
+      "step": 8013
+    },
+    {
+      "epoch": 0.29144997636105757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.66241026270318e-05,
+      "loss": 1.1639,
+      "step": 8014
+    },
+    {
+      "epoch": 0.29148634396479617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6623220196293274e-05,
+      "loss": 1.0694,
+      "step": 8015
+    },
+    {
+      "epoch": 0.29152271156853476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6622337673666963e-05,
+      "loss": 1.1187,
+      "step": 8016
+    },
+    {
+      "epoch": 0.29155907917227336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6621455059165113e-05,
+      "loss": 0.9488,
+      "step": 8017
+    },
+    {
+      "epoch": 0.2915954467760119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6620572352799968e-05,
+      "loss": 1.0789,
+      "step": 8018
+    },
+    {
+      "epoch": 0.2916318143797505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6619689554583773e-05,
+      "loss": 1.0765,
+      "step": 8019
+    },
+    {
+      "epoch": 0.2916681819834891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6618806664528783e-05,
+      "loss": 1.1422,
+      "step": 8020
+    },
+    {
+      "epoch": 0.2917045495872277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.661792368264724e-05,
+      "loss": 1.0545,
+      "step": 8021
+    },
+    {
+      "epoch": 0.2917409171909663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6617040608951393e-05,
+      "loss": 1.1305,
+      "step": 8022
+    },
+    {
+      "epoch": 0.2917772847947049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6616157443453492e-05,
+      "loss": 0.9204,
+      "step": 8023
+    },
+    {
+      "epoch": 0.2918136523984435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6615274186165803e-05,
+      "loss": 0.9365,
+      "step": 8024
+    },
+    {
+      "epoch": 0.2918500200021821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6614390837100565e-05,
+      "loss": 0.9246,
+      "step": 8025
+    },
+    {
+      "epoch": 0.29188638760592067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6613507396270038e-05,
+      "loss": 1.1606,
+      "step": 8026
+    },
+    {
+      "epoch": 0.2919227552096592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6612623863686483e-05,
+      "loss": 1.1152,
+      "step": 8027
+    },
+    {
+      "epoch": 0.2919591228133978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6611740239362148e-05,
+      "loss": 1.1056,
+      "step": 8028
+    },
+    {
+      "epoch": 0.2919954904171364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.66108565233093e-05,
+      "loss": 1.1585,
+      "step": 8029
+    },
+    {
+      "epoch": 0.292031858020875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6609972715540198e-05,
+      "loss": 1.2048,
+      "step": 8030
+    },
+    {
+      "epoch": 0.2920682256246136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6609088816067104e-05,
+      "loss": 1.0538,
+      "step": 8031
+    },
+    {
+      "epoch": 0.2921045932283522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6608204824902283e-05,
+      "loss": 1.0264,
+      "step": 8032
+    },
+    {
+      "epoch": 0.2921409608320908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.660732074205799e-05,
+      "loss": 1.1061,
+      "step": 8033
+    },
+    {
+      "epoch": 0.2921773284358294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6606436567546497e-05,
+      "loss": 0.9982,
+      "step": 8034
+    },
+    {
+      "epoch": 0.2922136960395679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6605552301380073e-05,
+      "loss": 1.1086,
+      "step": 8035
+    },
+    {
+      "epoch": 0.2922500636433065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6604667943570983e-05,
+      "loss": 1.0524,
+      "step": 8036
+    },
+    {
+      "epoch": 0.2922864312470451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6603783494131495e-05,
+      "loss": 1.0161,
+      "step": 8037
+    },
+    {
+      "epoch": 0.2923227988507837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6602898953073877e-05,
+      "loss": 0.9596,
+      "step": 8038
+    },
+    {
+      "epoch": 0.2923591664545223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.660201432041041e-05,
+      "loss": 1.0997,
+      "step": 8039
+    },
+    {
+      "epoch": 0.2923955340582609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.660112959615336e-05,
+      "loss": 1.1138,
+      "step": 8040
+    },
+    {
+      "epoch": 0.2924319016619995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6600244780315e-05,
+      "loss": 1.0145,
+      "step": 8041
+    },
+    {
+      "epoch": 0.2924682692657381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659935987290761e-05,
+      "loss": 1.0286,
+      "step": 8042
+    },
+    {
+      "epoch": 0.2925046368694767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6598474873943464e-05,
+      "loss": 1.1545,
+      "step": 8043
+    },
+    {
+      "epoch": 0.29254100447321524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6597589783434842e-05,
+      "loss": 1.0623,
+      "step": 8044
+    },
+    {
+      "epoch": 0.29257737207695383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6596704601394024e-05,
+      "loss": 1.171,
+      "step": 8045
+    },
+    {
+      "epoch": 0.29261373968069243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6595819327833287e-05,
+      "loss": 1.0673,
+      "step": 8046
+    },
+    {
+      "epoch": 0.292650107284431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6594933962764915e-05,
+      "loss": 1.1235,
+      "step": 8047
+    },
+    {
+      "epoch": 0.2926864748881696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659404850620119e-05,
+      "loss": 1.0243,
+      "step": 8048
+    },
+    {
+      "epoch": 0.2927228424919082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65931629581544e-05,
+      "loss": 1.1378,
+      "step": 8049
+    },
+    {
+      "epoch": 0.2927592100956468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6592277318636828e-05,
+      "loss": 0.9479,
+      "step": 8050
+    },
+    {
+      "epoch": 0.2927955776993854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659139158766076e-05,
+      "loss": 1.001,
+      "step": 8051
+    },
+    {
+      "epoch": 0.29283194530312395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659050576523849e-05,
+      "loss": 1.0546,
+      "step": 8052
+    },
+    {
+      "epoch": 0.29286831290686255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65896198513823e-05,
+      "loss": 1.0431,
+      "step": 8053
+    },
+    {
+      "epoch": 0.29290468051060115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6588733846104483e-05,
+      "loss": 1.0798,
+      "step": 8054
+    },
+    {
+      "epoch": 0.29294104811433974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6587847749417332e-05,
+      "loss": 1.0964,
+      "step": 8055
+    },
+    {
+      "epoch": 0.29297741571807834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6586961561333144e-05,
+      "loss": 1.1125,
+      "step": 8056
+    },
+    {
+      "epoch": 0.29301378332181693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6586075281864205e-05,
+      "loss": 1.0256,
+      "step": 8057
+    },
+    {
+      "epoch": 0.29305015092555553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.658518891102282e-05,
+      "loss": 1.09,
+      "step": 8058
+    },
+    {
+      "epoch": 0.2930865185292941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.658430244882128e-05,
+      "loss": 1.1044,
+      "step": 8059
+    },
+    {
+      "epoch": 0.29312288613303267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6583415895271886e-05,
+      "loss": 1.0732,
+      "step": 8060
+    },
+    {
+      "epoch": 0.29315925373677126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.658252925038694e-05,
+      "loss": 1.1396,
+      "step": 8061
+    },
+    {
+      "epoch": 0.29319562134050986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6581642514178735e-05,
+      "loss": 1.0562,
+      "step": 8062
+    },
+    {
+      "epoch": 0.29323198894424846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.658075568665958e-05,
+      "loss": 1.1065,
+      "step": 8063
+    },
+    {
+      "epoch": 0.29326835654798705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6579868767841776e-05,
+      "loss": 1.1467,
+      "step": 8064
+    },
+    {
+      "epoch": 0.29330472415172565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6578981757737632e-05,
+      "loss": 1.1243,
+      "step": 8065
+    },
+    {
+      "epoch": 0.29334109175546424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6578094656359448e-05,
+      "loss": 1.1218,
+      "step": 8066
+    },
+    {
+      "epoch": 0.29337745935920284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.657720746371953e-05,
+      "loss": 1.1328,
+      "step": 8067
+    },
+    {
+      "epoch": 0.29341382696294144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6576320179830198e-05,
+      "loss": 1.0193,
+      "step": 8068
+    },
+    {
+      "epoch": 0.29345019456668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6575432804703748e-05,
+      "loss": 1.0993,
+      "step": 8069
+    },
+    {
+      "epoch": 0.2934865621704186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65745453383525e-05,
+      "loss": 1.0954,
+      "step": 8070
+    },
+    {
+      "epoch": 0.29352292977415717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.657365778078876e-05,
+      "loss": 1.0607,
+      "step": 8071
+    },
+    {
+      "epoch": 0.29355929737789577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.657277013202485e-05,
+      "loss": 1.0236,
+      "step": 8072
+    },
+    {
+      "epoch": 0.29359566498163436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6571882392073076e-05,
+      "loss": 1.0435,
+      "step": 8073
+    },
+    {
+      "epoch": 0.29363203258537296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.657099456094576e-05,
+      "loss": 1.1559,
+      "step": 8074
+    },
+    {
+      "epoch": 0.29366840018911156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6570106638655218e-05,
+      "loss": 1.1019,
+      "step": 8075
+    },
+    {
+      "epoch": 0.29370476779285015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6569218625213768e-05,
+      "loss": 0.9836,
+      "step": 8076
+    },
+    {
+      "epoch": 0.2937411353965887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.656833052063373e-05,
+      "loss": 1.1842,
+      "step": 8077
+    },
+    {
+      "epoch": 0.2937775030003273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6567442324927427e-05,
+      "loss": 1.0789,
+      "step": 8078
+    },
+    {
+      "epoch": 0.2938138706040659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6566554038107178e-05,
+      "loss": 1.1682,
+      "step": 8079
+    },
+    {
+      "epoch": 0.2938502382078045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6565665660185306e-05,
+      "loss": 1.0275,
+      "step": 8080
+    },
+    {
+      "epoch": 0.2938866058115431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.656477719117414e-05,
+      "loss": 1.048,
+      "step": 8081
+    },
+    {
+      "epoch": 0.2939229734152817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6563888631086003e-05,
+      "loss": 1.016,
+      "step": 8082
+    },
+    {
+      "epoch": 0.29395934101902027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.656299997993323e-05,
+      "loss": 1.1212,
+      "step": 8083
+    },
+    {
+      "epoch": 0.29399570862275887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6562111237728137e-05,
+      "loss": 1.0458,
+      "step": 8084
+    },
+    {
+      "epoch": 0.29403207622649746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6561222404483067e-05,
+      "loss": 1.1046,
+      "step": 8085
+    },
+    {
+      "epoch": 0.294068443830236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6560333480210342e-05,
+      "loss": 1.2161,
+      "step": 8086
+    },
+    {
+      "epoch": 0.2941048114339746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65594444649223e-05,
+      "loss": 1.1325,
+      "step": 8087
+    },
+    {
+      "epoch": 0.2941411790377132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.655855535863127e-05,
+      "loss": 1.2483,
+      "step": 8088
+    },
+    {
+      "epoch": 0.2941775466414518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.655766616134959e-05,
+      "loss": 1.1384,
+      "step": 8089
+    },
+    {
+      "epoch": 0.2942139142451904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6556776873089596e-05,
+      "loss": 1.1506,
+      "step": 8090
+    },
+    {
+      "epoch": 0.294250281848929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6555887493863627e-05,
+      "loss": 0.9931,
+      "step": 8091
+    },
+    {
+      "epoch": 0.2942866494526676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6554998023684025e-05,
+      "loss": 1.0895,
+      "step": 8092
+    },
+    {
+      "epoch": 0.2943230170564062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6554108462563117e-05,
+      "loss": 1.0914,
+      "step": 8093
+    },
+    {
+      "epoch": 0.2943593846601447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6553218810513256e-05,
+      "loss": 1.1453,
+      "step": 8094
+    },
+    {
+      "epoch": 0.2943957522638833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6552329067546785e-05,
+      "loss": 0.9855,
+      "step": 8095
+    },
+    {
+      "epoch": 0.2944321198676219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6551439233676043e-05,
+      "loss": 1.0801,
+      "step": 8096
+    },
+    {
+      "epoch": 0.2944684874713605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6550549308913378e-05,
+      "loss": 1.0573,
+      "step": 8097
+    },
+    {
+      "epoch": 0.2945048550750991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6549659293271134e-05,
+      "loss": 1.1658,
+      "step": 8098
+    },
+    {
+      "epoch": 0.2945412226788377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.654876918676166e-05,
+      "loss": 1.085,
+      "step": 8099
+    },
+    {
+      "epoch": 0.2945775902825763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6547878989397306e-05,
+      "loss": 0.9945,
+      "step": 8100
+    },
+    {
+      "epoch": 0.2946139578863149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.654698870119042e-05,
+      "loss": 1.0939,
+      "step": 8101
+    },
+    {
+      "epoch": 0.29465032549005343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6546098322153354e-05,
+      "loss": 1.1424,
+      "step": 8102
+    },
+    {
+      "epoch": 0.29468669309379203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6545207852298464e-05,
+      "loss": 1.1086,
+      "step": 8103
+    },
+    {
+      "epoch": 0.2947230606975306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65443172916381e-05,
+      "loss": 1.2238,
+      "step": 8104
+    },
+    {
+      "epoch": 0.2947594283012692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6543426640184618e-05,
+      "loss": 1.0507,
+      "step": 8105
+    },
+    {
+      "epoch": 0.2947957959050078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6542535897950377e-05,
+      "loss": 1.0462,
+      "step": 8106
+    },
+    {
+      "epoch": 0.2948321635087464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6541645064947733e-05,
+      "loss": 1.0309,
+      "step": 8107
+    },
+    {
+      "epoch": 0.294868531112485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6540754141189042e-05,
+      "loss": 0.9576,
+      "step": 8108
+    },
+    {
+      "epoch": 0.2949048987162236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.653986312668667e-05,
+      "loss": 1.0148,
+      "step": 8109
+    },
+    {
+      "epoch": 0.2949412663199622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6538972021452975e-05,
+      "loss": 1.1301,
+      "step": 8110
+    },
+    {
+      "epoch": 0.29497763392370074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.653808082550032e-05,
+      "loss": 1.1061,
+      "step": 8111
+    },
+    {
+      "epoch": 0.29501400152743934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6537189538841075e-05,
+      "loss": 1.042,
+      "step": 8112
+    },
+    {
+      "epoch": 0.29505036913117794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6536298161487594e-05,
+      "loss": 1.0582,
+      "step": 8113
+    },
+    {
+      "epoch": 0.29508673673491653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.653540669345225e-05,
+      "loss": 1.1229,
+      "step": 8114
+    },
+    {
+      "epoch": 0.29512310433865513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6534515134747414e-05,
+      "loss": 1.0494,
+      "step": 8115
+    },
+    {
+      "epoch": 0.2951594719423937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6533623485385454e-05,
+      "loss": 1.1171,
+      "step": 8116
+    },
+    {
+      "epoch": 0.2951958395461323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6532731745378738e-05,
+      "loss": 1.1754,
+      "step": 8117
+    },
+    {
+      "epoch": 0.2952322071498709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.653183991473963e-05,
+      "loss": 1.1509,
+      "step": 8118
+    },
+    {
+      "epoch": 0.29526857475360946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.653094799348052e-05,
+      "loss": 0.9618,
+      "step": 8119
+    },
+    {
+      "epoch": 0.29530494235734805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6530055981613763e-05,
+      "loss": 1.2193,
+      "step": 8120
+    },
+    {
+      "epoch": 0.29534130996108665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6529163879151752e-05,
+      "loss": 1.0753,
+      "step": 8121
+    },
+    {
+      "epoch": 0.29537767756482525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.652827168610685e-05,
+      "loss": 1.0607,
+      "step": 8122
+    },
+    {
+      "epoch": 0.29541404516856384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6527379402491445e-05,
+      "loss": 1.0212,
+      "step": 8123
+    },
+    {
+      "epoch": 0.29545041277230244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6526487028317914e-05,
+      "loss": 1.1429,
+      "step": 8124
+    },
+    {
+      "epoch": 0.29548678037604104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.652559456359863e-05,
+      "loss": 1.2027,
+      "step": 8125
+    },
+    {
+      "epoch": 0.29552314797977963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.652470200834598e-05,
+      "loss": 1.19,
+      "step": 8126
+    },
+    {
+      "epoch": 0.29555951558351823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.652380936257235e-05,
+      "loss": 1.0781,
+      "step": 8127
+    },
+    {
+      "epoch": 0.29559588318725677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.652291662629012e-05,
+      "loss": 0.9821,
+      "step": 8128
+    },
+    {
+      "epoch": 0.29563225079099537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.652202379951168e-05,
+      "loss": 1.0822,
+      "step": 8129
+    },
+    {
+      "epoch": 0.29566861839473396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6521130882249408e-05,
+      "loss": 1.176,
+      "step": 8130
+    },
+    {
+      "epoch": 0.29570498599847256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65202378745157e-05,
+      "loss": 1.1268,
+      "step": 8131
+    },
+    {
+      "epoch": 0.29574135360221115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.651934477632294e-05,
+      "loss": 1.0412,
+      "step": 8132
+    },
+    {
+      "epoch": 0.29577772120594975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6518451587683524e-05,
+      "loss": 1.0132,
+      "step": 8133
+    },
+    {
+      "epoch": 0.29581408880968835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6517558308609842e-05,
+      "loss": 1.177,
+      "step": 8134
+    },
+    {
+      "epoch": 0.29585045641342694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6516664939114284e-05,
+      "loss": 0.9718,
+      "step": 8135
+    },
+    {
+      "epoch": 0.2958868240171655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6515771479209245e-05,
+      "loss": 1.0596,
+      "step": 8136
+    },
+    {
+      "epoch": 0.2959231916209041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6514877928907122e-05,
+      "loss": 0.9569,
+      "step": 8137
+    },
+    {
+      "epoch": 0.2959595592246427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.651398428822031e-05,
+      "loss": 1.1623,
+      "step": 8138
+    },
+    {
+      "epoch": 0.2959959268283813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.651309055716121e-05,
+      "loss": 1.0557,
+      "step": 8139
+    },
+    {
+      "epoch": 0.29603229443211987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6512196735742217e-05,
+      "loss": 0.9846,
+      "step": 8140
+    },
+    {
+      "epoch": 0.29606866203585847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6511302823975736e-05,
+      "loss": 1.0659,
+      "step": 8141
+    },
+    {
+      "epoch": 0.29610502963959706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6510408821874167e-05,
+      "loss": 1.2373,
+      "step": 8142
+    },
+    {
+      "epoch": 0.29614139724333566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650951472944991e-05,
+      "loss": 1.0928,
+      "step": 8143
+    },
+    {
+      "epoch": 0.2961777648470742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6508620546715374e-05,
+      "loss": 1.1904,
+      "step": 8144
+    },
+    {
+      "epoch": 0.2962141324508128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650772627368296e-05,
+      "loss": 1.1251,
+      "step": 8145
+    },
+    {
+      "epoch": 0.2962505000545514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650683191036508e-05,
+      "loss": 1.135,
+      "step": 8146
+    },
+    {
+      "epoch": 0.29628686765829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650593745677414e-05,
+      "loss": 1.0987,
+      "step": 8147
+    },
+    {
+      "epoch": 0.2963232352620286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6505042912922546e-05,
+      "loss": 1.1923,
+      "step": 8148
+    },
+    {
+      "epoch": 0.2963596028657672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6504148278822712e-05,
+      "loss": 1.2004,
+      "step": 8149
+    },
+    {
+      "epoch": 0.2963959704695058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650325355448705e-05,
+      "loss": 1.1176,
+      "step": 8150
+    },
+    {
+      "epoch": 0.2964323380732444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6502358739927967e-05,
+      "loss": 1.1164,
+      "step": 8151
+    },
+    {
+      "epoch": 0.29646870567698297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6501463835157887e-05,
+      "loss": 1.0364,
+      "step": 8152
+    },
+    {
+      "epoch": 0.2965050732807215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650056884018922e-05,
+      "loss": 1.1298,
+      "step": 8153
+    },
+    {
+      "epoch": 0.2965414408844601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.649967375503438e-05,
+      "loss": 0.9547,
+      "step": 8154
+    },
+    {
+      "epoch": 0.2965778084881987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6498778579705793e-05,
+      "loss": 1.1536,
+      "step": 8155
+    },
+    {
+      "epoch": 0.2966141760919373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.649788331421587e-05,
+      "loss": 1.1461,
+      "step": 8156
+    },
+    {
+      "epoch": 0.2966505436956759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.649698795857704e-05,
+      "loss": 1.0134,
+      "step": 8157
+    },
+    {
+      "epoch": 0.2966869112994145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.649609251280172e-05,
+      "loss": 1.1395,
+      "step": 8158
+    },
+    {
+      "epoch": 0.2967232789031531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.649519697690233e-05,
+      "loss": 1.0988,
+      "step": 8159
+    },
+    {
+      "epoch": 0.2967596465068917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6494301350891298e-05,
+      "loss": 1.0168,
+      "step": 8160
+    },
+    {
+      "epoch": 0.2967960141106302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6493405634781047e-05,
+      "loss": 0.928,
+      "step": 8161
+    },
+    {
+      "epoch": 0.2968323817143688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.649250982858401e-05,
+      "loss": 1.0643,
+      "step": 8162
+    },
+    {
+      "epoch": 0.2968687493181074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6491613932312606e-05,
+      "loss": 1.1835,
+      "step": 8163
+    },
+    {
+      "epoch": 0.296905116921846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6490717945979275e-05,
+      "loss": 1.0958,
+      "step": 8164
+    },
+    {
+      "epoch": 0.2969414845255846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6489821869596438e-05,
+      "loss": 1.0154,
+      "step": 8165
+    },
+    {
+      "epoch": 0.2969778521293232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.648892570317653e-05,
+      "loss": 1.0797,
+      "step": 8166
+    },
+    {
+      "epoch": 0.2970142197330618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6488029446731985e-05,
+      "loss": 1.1165,
+      "step": 8167
+    },
+    {
+      "epoch": 0.2970505873368004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6487133100275237e-05,
+      "loss": 1.1072,
+      "step": 8168
+    },
+    {
+      "epoch": 0.297086954940539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6486236663818718e-05,
+      "loss": 1.1354,
+      "step": 8169
+    },
+    {
+      "epoch": 0.29712332254427753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6485340137374872e-05,
+      "loss": 1.0332,
+      "step": 8170
+    },
+    {
+      "epoch": 0.29715969014801613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6484443520956134e-05,
+      "loss": 1.0574,
+      "step": 8171
+    },
+    {
+      "epoch": 0.2971960577517547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.648354681457494e-05,
+      "loss": 1.0631,
+      "step": 8172
+    },
+    {
+      "epoch": 0.2972324253554933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.648265001824373e-05,
+      "loss": 0.9635,
+      "step": 8173
+    },
+    {
+      "epoch": 0.2972687929592319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6481753131974957e-05,
+      "loss": 1.0062,
+      "step": 8174
+    },
+    {
+      "epoch": 0.2973051605629705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.648085615578105e-05,
+      "loss": 1.0888,
+      "step": 8175
+    },
+    {
+      "epoch": 0.2973415281667091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.647995908967446e-05,
+      "loss": 1.1175,
+      "step": 8176
+    },
+    {
+      "epoch": 0.2973778957704477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.647906193366763e-05,
+      "loss": 1.0583,
+      "step": 8177
+    },
+    {
+      "epoch": 0.29741426337418625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6478164687773012e-05,
+      "loss": 1.0754,
+      "step": 8178
+    },
+    {
+      "epoch": 0.29745063097792485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6477267352003046e-05,
+      "loss": 1.0563,
+      "step": 8179
+    },
+    {
+      "epoch": 0.29748699858166344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.647636992637019e-05,
+      "loss": 1.1446,
+      "step": 8180
+    },
+    {
+      "epoch": 0.29752336618540204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6475472410886886e-05,
+      "loss": 1.0494,
+      "step": 8181
+    },
+    {
+      "epoch": 0.29755973378914063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.647457480556559e-05,
+      "loss": 1.1435,
+      "step": 8182
+    },
+    {
+      "epoch": 0.29759610139287923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.647367711041876e-05,
+      "loss": 1.2042,
+      "step": 8183
+    },
+    {
+      "epoch": 0.2976324689966178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6472779325458836e-05,
+      "loss": 1.0876,
+      "step": 8184
+    },
+    {
+      "epoch": 0.2976688366003564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.647188145069829e-05,
+      "loss": 1.0751,
+      "step": 8185
+    },
+    {
+      "epoch": 0.297705204204095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6470983486149568e-05,
+      "loss": 1.0678,
+      "step": 8186
+    },
+    {
+      "epoch": 0.29774157180783356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6470085431825134e-05,
+      "loss": 1.0788,
+      "step": 8187
+    },
+    {
+      "epoch": 0.29777793941157216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646918728773744e-05,
+      "loss": 1.0013,
+      "step": 8188
+    },
+    {
+      "epoch": 0.29781430701531075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6468289053898952e-05,
+      "loss": 1.049,
+      "step": 8189
+    },
+    {
+      "epoch": 0.29785067461904935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6467390730322134e-05,
+      "loss": 1.1248,
+      "step": 8190
+    },
+    {
+      "epoch": 0.29788704222278795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6466492317019445e-05,
+      "loss": 1.031,
+      "step": 8191
+    },
+    {
+      "epoch": 0.29792340982652654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646559381400335e-05,
+      "loss": 1.1252,
+      "step": 8192
+    },
+    {
+      "epoch": 0.29795977743026514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6464695221286312e-05,
+      "loss": 1.0978,
+      "step": 8193
+    },
+    {
+      "epoch": 0.29799614503400373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.64637965388808e-05,
+      "loss": 1.1004,
+      "step": 8194
+    },
+    {
+      "epoch": 0.2980325126377423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6462897766799282e-05,
+      "loss": 0.9868,
+      "step": 8195
+    },
+    {
+      "epoch": 0.29806888024148087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646199890505423e-05,
+      "loss": 1.0569,
+      "step": 8196
+    },
+    {
+      "epoch": 0.29810524784521947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.646109995365811e-05,
+      "loss": 1.0337,
+      "step": 8197
+    },
+    {
+      "epoch": 0.29814161544895806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6460200912623393e-05,
+      "loss": 1.2222,
+      "step": 8198
+    },
+    {
+      "epoch": 0.29817798305269666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6459301781962558e-05,
+      "loss": 1.0868,
+      "step": 8199
+    },
+    {
+      "epoch": 0.29821435065643526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6458402561688073e-05,
+      "loss": 1.1707,
+      "step": 8200
+    },
+    {
+      "epoch": 0.29825071826017385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6457503251812417e-05,
+      "loss": 1.0572,
+      "step": 8201
+    },
+    {
+      "epoch": 0.29828708586391245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6456603852348065e-05,
+      "loss": 0.9701,
+      "step": 8202
+    },
+    {
+      "epoch": 0.298323453467651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.645570436330749e-05,
+      "loss": 1.1868,
+      "step": 8203
+    },
+    {
+      "epoch": 0.2983598210713896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6454804784703186e-05,
+      "loss": 1.1262,
+      "step": 8204
+    },
+    {
+      "epoch": 0.2983961886751282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6453905116547616e-05,
+      "loss": 1.178,
+      "step": 8205
+    },
+    {
+      "epoch": 0.2984325562788668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6453005358853274e-05,
+      "loss": 1.0429,
+      "step": 8206
+    },
+    {
+      "epoch": 0.2984689238826054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6452105511632635e-05,
+      "loss": 0.9671,
+      "step": 8207
+    },
+    {
+      "epoch": 0.29850529148634397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.645120557489819e-05,
+      "loss": 1.1196,
+      "step": 8208
+    },
+    {
+      "epoch": 0.29854165909008257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6450305548662418e-05,
+      "loss": 1.103,
+      "step": 8209
+    },
+    {
+      "epoch": 0.29857802669382116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6449405432937807e-05,
+      "loss": 1.197,
+      "step": 8210
+    },
+    {
+      "epoch": 0.29861439429755976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.644850522773685e-05,
+      "loss": 1.0521,
+      "step": 8211
+    },
+    {
+      "epoch": 0.2986507619012983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6447604933072032e-05,
+      "loss": 1.1612,
+      "step": 8212
+    },
+    {
+      "epoch": 0.2986871295050369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.644670454895584e-05,
+      "loss": 1.0784,
+      "step": 8213
+    },
+    {
+      "epoch": 0.2987234971087755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6445804075400774e-05,
+      "loss": 1.2014,
+      "step": 8214
+    },
+    {
+      "epoch": 0.2987598647125141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6444903512419318e-05,
+      "loss": 1.1871,
+      "step": 8215
+    },
+    {
+      "epoch": 0.2987962323162527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6444002860023972e-05,
+      "loss": 1.0355,
+      "step": 8216
+    },
+    {
+      "epoch": 0.2988325999199913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6443102118227226e-05,
+      "loss": 1.215,
+      "step": 8217
+    },
+    {
+      "epoch": 0.2988689675237299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6442201287041583e-05,
+      "loss": 1.0137,
+      "step": 8218
+    },
+    {
+      "epoch": 0.2989053351274685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6441300366479533e-05,
+      "loss": 1.1804,
+      "step": 8219
+    },
+    {
+      "epoch": 0.298941702731207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6440399356553586e-05,
+      "loss": 1.0661,
+      "step": 8220
+    },
+    {
+      "epoch": 0.2989780703349456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.643949825727623e-05,
+      "loss": 1.0642,
+      "step": 8221
+    },
+    {
+      "epoch": 0.2990144379386842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6438597068659973e-05,
+      "loss": 1.1054,
+      "step": 8222
+    },
+    {
+      "epoch": 0.2990508055424228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6437695790717318e-05,
+      "loss": 1.019,
+      "step": 8223
+    },
+    {
+      "epoch": 0.2990871731461614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6436794423460768e-05,
+      "loss": 1.1713,
+      "step": 8224
+    },
+    {
+      "epoch": 0.2991235407499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.643589296690283e-05,
+      "loss": 1.0608,
+      "step": 8225
+    },
+    {
+      "epoch": 0.2991599083536386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6434991421056005e-05,
+      "loss": 1.0986,
+      "step": 8226
+    },
+    {
+      "epoch": 0.2991962759573772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6434089785932804e-05,
+      "loss": 1.0429,
+      "step": 8227
+    },
+    {
+      "epoch": 0.2992326435611158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6433188061545737e-05,
+      "loss": 1.199,
+      "step": 8228
+    },
+    {
+      "epoch": 0.2992690111648543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6432286247907313e-05,
+      "loss": 1.0568,
+      "step": 8229
+    },
+    {
+      "epoch": 0.2993053787685929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6431384345030044e-05,
+      "loss": 1.1989,
+      "step": 8230
+    },
+    {
+      "epoch": 0.2993417463723315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6430482352926444e-05,
+      "loss": 1.1772,
+      "step": 8231
+    },
+    {
+      "epoch": 0.2993781139760701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6429580271609024e-05,
+      "loss": 0.9856,
+      "step": 8232
+    },
+    {
+      "epoch": 0.2994144815798087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.64286781010903e-05,
+      "loss": 1.1073,
+      "step": 8233
+    },
+    {
+      "epoch": 0.2994508491835473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6427775841382787e-05,
+      "loss": 1.0339,
+      "step": 8234
+    },
+    {
+      "epoch": 0.2994872167872859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6426873492499006e-05,
+      "loss": 1.0512,
+      "step": 8235
+    },
+    {
+      "epoch": 0.2995235843910245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.642597105445147e-05,
+      "loss": 1.0638,
+      "step": 8236
+    },
+    {
+      "epoch": 0.29955995199476304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.642506852725271e-05,
+      "loss": 1.1171,
+      "step": 8237
+    },
+    {
+      "epoch": 0.29959631959850164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6424165910915232e-05,
+      "loss": 0.9412,
+      "step": 8238
+    },
+    {
+      "epoch": 0.29963268720224023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.642326320545157e-05,
+      "loss": 1.0812,
+      "step": 8239
+    },
+    {
+      "epoch": 0.29966905480597883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6422360410874247e-05,
+      "loss": 1.0809,
+      "step": 8240
+    },
+    {
+      "epoch": 0.2997054224097174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.642145752719578e-05,
+      "loss": 1.033,
+      "step": 8241
+    },
+    {
+      "epoch": 0.299741790013456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6420554554428706e-05,
+      "loss": 1.0356,
+      "step": 8242
+    },
+    {
+      "epoch": 0.2997781576171946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6419651492585543e-05,
+      "loss": 0.9455,
+      "step": 8243
+    },
+    {
+      "epoch": 0.2998145252209332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6418748341678827e-05,
+      "loss": 1.1469,
+      "step": 8244
+    },
+    {
+      "epoch": 0.29985089282467176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6417845101721085e-05,
+      "loss": 0.9556,
+      "step": 8245
+    },
+    {
+      "epoch": 0.29988726042841035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6416941772724845e-05,
+      "loss": 1.113,
+      "step": 8246
+    },
+    {
+      "epoch": 0.29992362803214895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6416038354702647e-05,
+      "loss": 0.9831,
+      "step": 8247
+    },
+    {
+      "epoch": 0.29995999563588754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6415134847667014e-05,
+      "loss": 1.0225,
+      "step": 8248
+    },
+    {
+      "epoch": 0.29999636323962614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.641423125163049e-05,
+      "loss": 0.9715,
+      "step": 8249
+    },
+    {
+      "epoch": 0.30003273084336474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.641332756660561e-05,
+      "loss": 1.115,
+      "step": 8250
+    },
+    {
+      "epoch": 0.30006909844710333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6412423792604906e-05,
+      "loss": 0.941,
+      "step": 8251
+    },
+    {
+      "epoch": 0.30010546605084193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6411519929640922e-05,
+      "loss": 1.09,
+      "step": 8252
+    },
+    {
+      "epoch": 0.3001418336545805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6410615977726193e-05,
+      "loss": 0.9757,
+      "step": 8253
+    },
+    {
+      "epoch": 0.30017820125831907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6409711936873266e-05,
+      "loss": 1.0326,
+      "step": 8254
+    },
+    {
+      "epoch": 0.30021456886205766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6408807807094677e-05,
+      "loss": 1.0231,
+      "step": 8255
+    },
+    {
+      "epoch": 0.30025093646579626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6407903588402974e-05,
+      "loss": 1.1181,
+      "step": 8256
+    },
+    {
+      "epoch": 0.30028730406953485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6406999280810702e-05,
+      "loss": 1.0684,
+      "step": 8257
+    },
+    {
+      "epoch": 0.30032367167327345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6406094884330402e-05,
+      "loss": 1.0971,
+      "step": 8258
+    },
+    {
+      "epoch": 0.30036003927701205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6405190398974625e-05,
+      "loss": 1.0093,
+      "step": 8259
+    },
+    {
+      "epoch": 0.30039640688075064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6404285824755924e-05,
+      "loss": 1.1749,
+      "step": 8260
+    },
+    {
+      "epoch": 0.30043277448448924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.640338116168684e-05,
+      "loss": 1.0679,
+      "step": 8261
+    },
+    {
+      "epoch": 0.3004691420882278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6402476409779925e-05,
+      "loss": 1.0052,
+      "step": 8262
+    },
+    {
+      "epoch": 0.3005055096919664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6401571569047738e-05,
+      "loss": 1.1472,
+      "step": 8263
+    },
+    {
+      "epoch": 0.300541877295705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.640066663950283e-05,
+      "loss": 1.0284,
+      "step": 8264
+    },
+    {
+      "epoch": 0.30057824489944357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6399761621157746e-05,
+      "loss": 1.0464,
+      "step": 8265
+    },
+    {
+      "epoch": 0.30061461250318217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6398856514025054e-05,
+      "loss": 0.9953,
+      "step": 8266
+    },
+    {
+      "epoch": 0.30065098010692076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639795131811731e-05,
+      "loss": 1.1951,
+      "step": 8267
+    },
+    {
+      "epoch": 0.30068734771065936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6397046033447063e-05,
+      "loss": 1.0681,
+      "step": 8268
+    },
+    {
+      "epoch": 0.30072371531439795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639614066002688e-05,
+      "loss": 1.064,
+      "step": 8269
+    },
+    {
+      "epoch": 0.30076008291813655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6395235197869323e-05,
+      "loss": 1.1332,
+      "step": 8270
+    },
+    {
+      "epoch": 0.3007964505218751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639432964698695e-05,
+      "loss": 1.1426,
+      "step": 8271
+    },
+    {
+      "epoch": 0.3008328181256137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6393424007392324e-05,
+      "loss": 1.0254,
+      "step": 8272
+    },
+    {
+      "epoch": 0.3008691857293523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6392518279098012e-05,
+      "loss": 1.1797,
+      "step": 8273
+    },
+    {
+      "epoch": 0.3009055533330909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639161246211658e-05,
+      "loss": 1.0424,
+      "step": 8274
+    },
+    {
+      "epoch": 0.3009419209368295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639070655646059e-05,
+      "loss": 0.9456,
+      "step": 8275
+    },
+    {
+      "epoch": 0.3009782885405681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6389800562142618e-05,
+      "loss": 0.9908,
+      "step": 8276
+    },
+    {
+      "epoch": 0.30101465614430667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6388894479175227e-05,
+      "loss": 0.9729,
+      "step": 8277
+    },
+    {
+      "epoch": 0.30105102374804527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.638798830757099e-05,
+      "loss": 1.0048,
+      "step": 8278
+    },
+    {
+      "epoch": 0.3010873913517838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.638708204734248e-05,
+      "loss": 0.9458,
+      "step": 8279
+    },
+    {
+      "epoch": 0.3011237589555224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6386175698502265e-05,
+      "loss": 0.9303,
+      "step": 8280
+    },
+    {
+      "epoch": 0.301160126559261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6385269261062923e-05,
+      "loss": 1.1379,
+      "step": 8281
+    },
+    {
+      "epoch": 0.3011964941629996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6384362735037033e-05,
+      "loss": 0.9056,
+      "step": 8282
+    },
+    {
+      "epoch": 0.3012328617667382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6383456120437168e-05,
+      "loss": 1.1525,
+      "step": 8283
+    },
+    {
+      "epoch": 0.3012692293704768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6382549417275906e-05,
+      "loss": 0.9535,
+      "step": 8284
+    },
+    {
+      "epoch": 0.3013055969742154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6381642625565826e-05,
+      "loss": 1.1296,
+      "step": 8285
+    },
+    {
+      "epoch": 0.301341964577954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6380735745319506e-05,
+      "loss": 1.0853,
+      "step": 8286
+    },
+    {
+      "epoch": 0.3013783321816925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6379828776549532e-05,
+      "loss": 1.0866,
+      "step": 8287
+    },
+    {
+      "epoch": 0.3014146997854311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6378921719268487e-05,
+      "loss": 1.0274,
+      "step": 8288
+    },
+    {
+      "epoch": 0.3014510673891697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.637801457348896e-05,
+      "loss": 1.1619,
+      "step": 8289
+    },
+    {
+      "epoch": 0.3014874349929083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.637710733922352e-05,
+      "loss": 1.15,
+      "step": 8290
+    },
+    {
+      "epoch": 0.3015238025966469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6376200016484766e-05,
+      "loss": 1.119,
+      "step": 8291
+    },
+    {
+      "epoch": 0.3015601702003855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6375292605285286e-05,
+      "loss": 1.1107,
+      "step": 8292
+    },
+    {
+      "epoch": 0.3015965378041241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6374385105637666e-05,
+      "loss": 0.9584,
+      "step": 8293
+    },
+    {
+      "epoch": 0.3016329054078627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6373477517554494e-05,
+      "loss": 1.1607,
+      "step": 8294
+    },
+    {
+      "epoch": 0.3016692730116013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6372569841048366e-05,
+      "loss": 1.1247,
+      "step": 8295
+    },
+    {
+      "epoch": 0.30170564061533983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6371662076131874e-05,
+      "loss": 1.0482,
+      "step": 8296
+    },
+    {
+      "epoch": 0.30174200821907843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6370754222817607e-05,
+      "loss": 1.2118,
+      "step": 8297
+    },
+    {
+      "epoch": 0.301778375822817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.636984628111817e-05,
+      "loss": 0.9634,
+      "step": 8298
+    },
+    {
+      "epoch": 0.3018147434265556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6368938251046148e-05,
+      "loss": 1.0563,
+      "step": 8299
+    },
+    {
+      "epoch": 0.3018511110302942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6368030132614147e-05,
+      "loss": 1.1109,
+      "step": 8300
+    },
+    {
+      "epoch": 0.3018874786340328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6367121925834762e-05,
+      "loss": 0.9373,
+      "step": 8301
+    },
+    {
+      "epoch": 0.3019238462377714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6366213630720595e-05,
+      "loss": 1.0695,
+      "step": 8302
+    },
+    {
+      "epoch": 0.30196021384151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6365305247284243e-05,
+      "loss": 1.0511,
+      "step": 8303
+    },
+    {
+      "epoch": 0.30199658144524855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6364396775538316e-05,
+      "loss": 1.1265,
+      "step": 8304
+    },
+    {
+      "epoch": 0.30203294904898714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6363488215495415e-05,
+      "loss": 1.136,
+      "step": 8305
+    },
+    {
+      "epoch": 0.30206931665272574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.636257956716814e-05,
+      "loss": 0.9784,
+      "step": 8306
+    },
+    {
+      "epoch": 0.30210568425646434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.63616708305691e-05,
+      "loss": 1.0306,
+      "step": 8307
+    },
+    {
+      "epoch": 0.30214205186020293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6360762005710907e-05,
+      "loss": 1.1105,
+      "step": 8308
+    },
+    {
+      "epoch": 0.3021784194639415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6359853092606163e-05,
+      "loss": 1.0519,
+      "step": 8309
+    },
+    {
+      "epoch": 0.3022147870676801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6358944091267482e-05,
+      "loss": 1.2115,
+      "step": 8310
+    },
+    {
+      "epoch": 0.3022511546714187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6358035001707473e-05,
+      "loss": 1.0717,
+      "step": 8311
+    },
+    {
+      "epoch": 0.3022875222751573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.635712582393875e-05,
+      "loss": 1.1528,
+      "step": 8312
+    },
+    {
+      "epoch": 0.30232388987889586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6356216557973926e-05,
+      "loss": 1.0231,
+      "step": 8313
+    },
+    {
+      "epoch": 0.30236025748263445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6355307203825616e-05,
+      "loss": 1.1218,
+      "step": 8314
+    },
+    {
+      "epoch": 0.30239662508637305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6354397761506434e-05,
+      "loss": 1.1439,
+      "step": 8315
+    },
+    {
+      "epoch": 0.30243299269011165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6353488231029e-05,
+      "loss": 1.0298,
+      "step": 8316
+    },
+    {
+      "epoch": 0.30246936029385024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.635257861240593e-05,
+      "loss": 1.1228,
+      "step": 8317
+    },
+    {
+      "epoch": 0.30250572789758884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6351668905649844e-05,
+      "loss": 1.0415,
+      "step": 8318
+    },
+    {
+      "epoch": 0.30254209550132743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.635075911077337e-05,
+      "loss": 1.0128,
+      "step": 8319
+    },
+    {
+      "epoch": 0.30257846310506603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6349849227789116e-05,
+      "loss": 1.253,
+      "step": 8320
+    },
+    {
+      "epoch": 0.30261483070880457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6348939256709715e-05,
+      "loss": 1.1563,
+      "step": 8321
+    },
+    {
+      "epoch": 0.30265119831254317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.634802919754779e-05,
+      "loss": 1.099,
+      "step": 8322
+    },
+    {
+      "epoch": 0.30268756591628176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6347119050315973e-05,
+      "loss": 1.1865,
+      "step": 8323
+    },
+    {
+      "epoch": 0.30272393352002036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6346208815026877e-05,
+      "loss": 1.0477,
+      "step": 8324
+    },
+    {
+      "epoch": 0.30276030112375896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.634529849169314e-05,
+      "loss": 1.0787,
+      "step": 8325
+    },
+    {
+      "epoch": 0.30279666872749755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.634438808032739e-05,
+      "loss": 1.0609,
+      "step": 8326
+    },
+    {
+      "epoch": 0.30283303633123615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6343477580942257e-05,
+      "loss": 1.0761,
+      "step": 8327
+    },
+    {
+      "epoch": 0.30286940393497475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6342566993550374e-05,
+      "loss": 1.0143,
+      "step": 8328
+    },
+    {
+      "epoch": 0.3029057715387133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.634165631816437e-05,
+      "loss": 1.0469,
+      "step": 8329
+    },
+    {
+      "epoch": 0.3029421391424519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.634074555479688e-05,
+      "loss": 0.8894,
+      "step": 8330
+    },
+    {
+      "epoch": 0.3029785067461905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.633983470346055e-05,
+      "loss": 1.1328,
+      "step": 8331
+    },
+    {
+      "epoch": 0.3030148743499291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6338923764168e-05,
+      "loss": 1.1588,
+      "step": 8332
+    },
+    {
+      "epoch": 0.30305124195366767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.633801273693188e-05,
+      "loss": 1.0139,
+      "step": 8333
+    },
+    {
+      "epoch": 0.30308760955740627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6337101621764826e-05,
+      "loss": 0.9882,
+      "step": 8334
+    },
+    {
+      "epoch": 0.30312397716114486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.633619041867948e-05,
+      "loss": 1.1425,
+      "step": 8335
+    },
+    {
+      "epoch": 0.30316034476488346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6335279127688477e-05,
+      "loss": 1.054,
+      "step": 8336
+    },
+    {
+      "epoch": 0.30319671236862206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6334367748804468e-05,
+      "loss": 1.0707,
+      "step": 8337
+    },
+    {
+      "epoch": 0.3032330799723606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6333456282040093e-05,
+      "loss": 1.0139,
+      "step": 8338
+    },
+    {
+      "epoch": 0.3032694475760992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6332544727408e-05,
+      "loss": 1.0391,
+      "step": 8339
+    },
+    {
+      "epoch": 0.3033058151798378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6331633084920828e-05,
+      "loss": 1.2162,
+      "step": 8340
+    },
+    {
+      "epoch": 0.3033421827835764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6330721354591234e-05,
+      "loss": 0.9071,
+      "step": 8341
+    },
+    {
+      "epoch": 0.303378550387315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6329809536431866e-05,
+      "loss": 1.0597,
+      "step": 8342
+    },
+    {
+      "epoch": 0.3034149179910536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6328897630455368e-05,
+      "loss": 1.106,
+      "step": 8343
+    },
+    {
+      "epoch": 0.3034512855947922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6327985636674398e-05,
+      "loss": 1.0909,
+      "step": 8344
+    },
+    {
+      "epoch": 0.30348765319853077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.63270735551016e-05,
+      "loss": 1.0532,
+      "step": 8345
+    },
+    {
+      "epoch": 0.3035240208022693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6326161385749642e-05,
+      "loss": 1.0705,
+      "step": 8346
+    },
+    {
+      "epoch": 0.3035603884060079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6325249128631164e-05,
+      "loss": 1.0451,
+      "step": 8347
+    },
+    {
+      "epoch": 0.3035967560097465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.632433678375883e-05,
+      "loss": 1.0798,
+      "step": 8348
+    },
+    {
+      "epoch": 0.3036331236134851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6323424351145293e-05,
+      "loss": 1.0888,
+      "step": 8349
+    },
+    {
+      "epoch": 0.3036694912172237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6322511830803222e-05,
+      "loss": 1.241,
+      "step": 8350
+    },
+    {
+      "epoch": 0.3037058588209623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6321599222745264e-05,
+      "loss": 1.1041,
+      "step": 8351
+    },
+    {
+      "epoch": 0.3037422264247009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.632068652698409e-05,
+      "loss": 1.1773,
+      "step": 8352
+    },
+    {
+      "epoch": 0.3037785940284395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6319773743532354e-05,
+      "loss": 1.1105,
+      "step": 8353
+    },
+    {
+      "epoch": 0.3038149616321781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6318860872402727e-05,
+      "loss": 0.9936,
+      "step": 8354
+    },
+    {
+      "epoch": 0.3038513292359166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.631794791360787e-05,
+      "loss": 1.1542,
+      "step": 8355
+    },
+    {
+      "epoch": 0.3038876968396552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.631703486716045e-05,
+      "loss": 1.1417,
+      "step": 8356
+    },
+    {
+      "epoch": 0.3039240644433938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.631612173307313e-05,
+      "loss": 0.9948,
+      "step": 8357
+    },
+    {
+      "epoch": 0.3039604320471324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6315208511358586e-05,
+      "loss": 1.1079,
+      "step": 8358
+    },
+    {
+      "epoch": 0.303996799650871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6314295202029488e-05,
+      "loss": 1.0856,
+      "step": 8359
+    },
+    {
+      "epoch": 0.3040331672546096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6313381805098494e-05,
+      "loss": 0.9901,
+      "step": 8360
+    },
+    {
+      "epoch": 0.3040695348583482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.631246832057829e-05,
+      "loss": 1.1369,
+      "step": 8361
+    },
+    {
+      "epoch": 0.3041059024620868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6311554748481544e-05,
+      "loss": 1.0285,
+      "step": 8362
+    },
+    {
+      "epoch": 0.30414227006582534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.631064108882093e-05,
+      "loss": 1.1426,
+      "step": 8363
+    },
+    {
+      "epoch": 0.30417863766956393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6309727341609124e-05,
+      "loss": 1.0259,
+      "step": 8364
+    },
+    {
+      "epoch": 0.30421500527330253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.630881350685881e-05,
+      "loss": 1.1875,
+      "step": 8365
+    },
+    {
+      "epoch": 0.3042513728770411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6307899584582652e-05,
+      "loss": 0.9677,
+      "step": 8366
+    },
+    {
+      "epoch": 0.3042877404807797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.630698557479334e-05,
+      "loss": 1.1386,
+      "step": 8367
+    },
+    {
+      "epoch": 0.3043241080845183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6306071477503552e-05,
+      "loss": 1.0703,
+      "step": 8368
+    },
+    {
+      "epoch": 0.3043604756882569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6305157292725967e-05,
+      "loss": 1.0531,
+      "step": 8369
+    },
+    {
+      "epoch": 0.3043968432919955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.630424302047327e-05,
+      "loss": 1.1577,
+      "step": 8370
+    },
+    {
+      "epoch": 0.3044332108957341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6303328660758147e-05,
+      "loss": 1.0956,
+      "step": 8371
+    },
+    {
+      "epoch": 0.30446957849947265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6302414213593283e-05,
+      "loss": 1.0454,
+      "step": 8372
+    },
+    {
+      "epoch": 0.30450594610321124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6301499678991365e-05,
+      "loss": 1.0855,
+      "step": 8373
+    },
+    {
+      "epoch": 0.30454231370694984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6300585056965077e-05,
+      "loss": 1.1508,
+      "step": 8374
+    },
+    {
+      "epoch": 0.30457868131068844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6299670347527112e-05,
+      "loss": 1.1837,
+      "step": 8375
+    },
+    {
+      "epoch": 0.30461504891442703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6298755550690157e-05,
+      "loss": 1.0753,
+      "step": 8376
+    },
+    {
+      "epoch": 0.30465141651816563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.629784066646691e-05,
+      "loss": 1.1695,
+      "step": 8377
+    },
+    {
+      "epoch": 0.3046877841219042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6296925694870053e-05,
+      "loss": 1.1044,
+      "step": 8378
+    },
+    {
+      "epoch": 0.3047241517256428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.629601063591229e-05,
+      "loss": 1.1114,
+      "step": 8379
+    },
+    {
+      "epoch": 0.30476051932938136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6295095489606313e-05,
+      "loss": 1.179,
+      "step": 8380
+    },
+    {
+      "epoch": 0.30479688693311996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6294180255964814e-05,
+      "loss": 0.9888,
+      "step": 8381
+    },
+    {
+      "epoch": 0.30483325453685856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6293264935000497e-05,
+      "loss": 1.0971,
+      "step": 8382
+    },
+    {
+      "epoch": 0.30486962214059715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6292349526726056e-05,
+      "loss": 1.0401,
+      "step": 8383
+    },
+    {
+      "epoch": 0.30490598974433575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6291434031154198e-05,
+      "loss": 1.165,
+      "step": 8384
+    },
+    {
+      "epoch": 0.30494235734807434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6290518448297614e-05,
+      "loss": 1.0626,
+      "step": 8385
+    },
+    {
+      "epoch": 0.30497872495181294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6289602778169012e-05,
+      "loss": 1.0347,
+      "step": 8386
+    },
+    {
+      "epoch": 0.30501509255555154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6288687020781096e-05,
+      "loss": 0.9789,
+      "step": 8387
+    },
+    {
+      "epoch": 0.3050514601592901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.628777117614657e-05,
+      "loss": 0.9751,
+      "step": 8388
+    },
+    {
+      "epoch": 0.3050878277630287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6286855244278144e-05,
+      "loss": 1.0947,
+      "step": 8389
+    },
+    {
+      "epoch": 0.30512419536676727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6285939225188516e-05,
+      "loss": 1.0762,
+      "step": 8390
+    },
+    {
+      "epoch": 0.30516056297050587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6285023118890405e-05,
+      "loss": 1.0271,
+      "step": 8391
+    },
+    {
+      "epoch": 0.30519693057424446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.628410692539651e-05,
+      "loss": 1.0535,
+      "step": 8392
+    },
+    {
+      "epoch": 0.30523329817798306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.628319064471955e-05,
+      "loss": 1.1238,
+      "step": 8393
+    },
+    {
+      "epoch": 0.30526966578172166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6282274276872235e-05,
+      "loss": 1.094,
+      "step": 8394
+    },
+    {
+      "epoch": 0.30530603338546025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.628135782186728e-05,
+      "loss": 1.1426,
+      "step": 8395
+    },
+    {
+      "epoch": 0.30534240098919885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.628044127971739e-05,
+      "loss": 1.1384,
+      "step": 8396
+    },
+    {
+      "epoch": 0.3053787685929374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6279524650435294e-05,
+      "loss": 1.0331,
+      "step": 8397
+    },
+    {
+      "epoch": 0.305415136196676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.62786079340337e-05,
+      "loss": 1.1686,
+      "step": 8398
+    },
+    {
+      "epoch": 0.3054515038004146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.627769113052533e-05,
+      "loss": 1.1021,
+      "step": 8399
+    },
+    {
+      "epoch": 0.3054878714041532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6276774239922904e-05,
+      "loss": 0.9952,
+      "step": 8400
+    },
+    {
+      "epoch": 0.3055242390078918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6275857262239137e-05,
+      "loss": 1.1229,
+      "step": 8401
+    },
+    {
+      "epoch": 0.30556060661163037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.627494019748676e-05,
+      "loss": 1.1931,
+      "step": 8402
+    },
+    {
+      "epoch": 0.30559697421536897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6274023045678488e-05,
+      "loss": 1.1312,
+      "step": 8403
+    },
+    {
+      "epoch": 0.30563334181910756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6273105806827053e-05,
+      "loss": 1.025,
+      "step": 8404
+    },
+    {
+      "epoch": 0.3056697094228461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.627218848094517e-05,
+      "loss": 1.0088,
+      "step": 8405
+    },
+    {
+      "epoch": 0.3057060770265847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.627127106804557e-05,
+      "loss": 1.0418,
+      "step": 8406
+    },
+    {
+      "epoch": 0.3057424446303233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.627035356814098e-05,
+      "loss": 1.1249,
+      "step": 8407
+    },
+    {
+      "epoch": 0.3057788122340619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6269435981244138e-05,
+      "loss": 1.143,
+      "step": 8408
+    },
+    {
+      "epoch": 0.3058151798378005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6268518307367763e-05,
+      "loss": 1.0388,
+      "step": 8409
+    },
+    {
+      "epoch": 0.3058515474415391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6267600546524587e-05,
+      "loss": 1.107,
+      "step": 8410
+    },
+    {
+      "epoch": 0.3058879150452777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.626668269872735e-05,
+      "loss": 1.1049,
+      "step": 8411
+    },
+    {
+      "epoch": 0.3059242826490163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6265764763988784e-05,
+      "loss": 1.207,
+      "step": 8412
+    },
+    {
+      "epoch": 0.3059606502527549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6264846742321616e-05,
+      "loss": 1.1521,
+      "step": 8413
+    },
+    {
+      "epoch": 0.3059970178564934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.626392863373859e-05,
+      "loss": 1.0953,
+      "step": 8414
+    },
+    {
+      "epoch": 0.306033385460232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6263010438252442e-05,
+      "loss": 1.0674,
+      "step": 8415
+    },
+    {
+      "epoch": 0.3060697530639706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6262092155875912e-05,
+      "loss": 1.0951,
+      "step": 8416
+    },
+    {
+      "epoch": 0.3061061206677092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6261173786621736e-05,
+      "loss": 0.9916,
+      "step": 8417
+    },
+    {
+      "epoch": 0.3061424882714478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.626025533050266e-05,
+      "loss": 1.002,
+      "step": 8418
+    },
+    {
+      "epoch": 0.3061788558751864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.625933678753142e-05,
+      "loss": 1.1082,
+      "step": 8419
+    },
+    {
+      "epoch": 0.306215223478925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6258418157720766e-05,
+      "loss": 0.951,
+      "step": 8420
+    },
+    {
+      "epoch": 0.3062515910826636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6257499441083434e-05,
+      "loss": 0.9921,
+      "step": 8421
+    },
+    {
+      "epoch": 0.30628795868640213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.625658063763218e-05,
+      "loss": 1.1727,
+      "step": 8422
+    },
+    {
+      "epoch": 0.3063243262901407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.625566174737975e-05,
+      "loss": 1.1506,
+      "step": 8423
+    },
+    {
+      "epoch": 0.3063606938938793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6254742770338885e-05,
+      "loss": 1.0807,
+      "step": 8424
+    },
+    {
+      "epoch": 0.3063970614976179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6253823706522334e-05,
+      "loss": 1.1511,
+      "step": 8425
+    },
+    {
+      "epoch": 0.3064334291013565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6252904555942857e-05,
+      "loss": 0.9342,
+      "step": 8426
+    },
+    {
+      "epoch": 0.3064697967050951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.62519853186132e-05,
+      "loss": 1.0948,
+      "step": 8427
+    },
+    {
+      "epoch": 0.3065061643088337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.625106599454612e-05,
+      "loss": 1.1638,
+      "step": 8428
+    },
+    {
+      "epoch": 0.3065425319125723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6250146583754366e-05,
+      "loss": 1.1076,
+      "step": 8429
+    },
+    {
+      "epoch": 0.30657889951631084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6249227086250696e-05,
+      "loss": 0.938,
+      "step": 8430
+    },
+    {
+      "epoch": 0.30661526712004944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6248307502047867e-05,
+      "loss": 1.1332,
+      "step": 8431
+    },
+    {
+      "epoch": 0.30665163472378804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6247387831158636e-05,
+      "loss": 1.2289,
+      "step": 8432
+    },
+    {
+      "epoch": 0.30668800232752663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.624646807359576e-05,
+      "loss": 1.0745,
+      "step": 8433
+    },
+    {
+      "epoch": 0.30672436993126523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6245548229372006e-05,
+      "loss": 1.0928,
+      "step": 8434
+    },
+    {
+      "epoch": 0.3067607375350038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.624462829850013e-05,
+      "loss": 0.9959,
+      "step": 8435
+    },
+    {
+      "epoch": 0.3067971051387424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6243708280992898e-05,
+      "loss": 1.0428,
+      "step": 8436
+    },
+    {
+      "epoch": 0.306833472742481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.624278817686307e-05,
+      "loss": 1.0311,
+      "step": 8437
+    },
+    {
+      "epoch": 0.3068698403462196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6241867986123415e-05,
+      "loss": 1.0952,
+      "step": 8438
+    },
+    {
+      "epoch": 0.30690620794995815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.62409477087867e-05,
+      "loss": 1.0561,
+      "step": 8439
+    },
+    {
+      "epoch": 0.30694257555369675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6240027344865684e-05,
+      "loss": 1.0876,
+      "step": 8440
+    },
+    {
+      "epoch": 0.30697894315743535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6239106894373146e-05,
+      "loss": 1.1455,
+      "step": 8441
+    },
+    {
+      "epoch": 0.30701531076117394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6238186357321853e-05,
+      "loss": 1.2011,
+      "step": 8442
+    },
+    {
+      "epoch": 0.30705167836491254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6237265733724573e-05,
+      "loss": 0.9855,
+      "step": 8443
+    },
+    {
+      "epoch": 0.30708804596865114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623634502359408e-05,
+      "loss": 0.9939,
+      "step": 8444
+    },
+    {
+      "epoch": 0.30712441357238973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6235424226943152e-05,
+      "loss": 1.0235,
+      "step": 8445
+    },
+    {
+      "epoch": 0.30716078117612833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6234503343784554e-05,
+      "loss": 1.2558,
+      "step": 8446
+    },
+    {
+      "epoch": 0.30719714877986687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6233582374131072e-05,
+      "loss": 1.0616,
+      "step": 8447
+    },
+    {
+      "epoch": 0.30723351638360547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623266131799548e-05,
+      "loss": 1.1035,
+      "step": 8448
+    },
+    {
+      "epoch": 0.30726988398734406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623174017539055e-05,
+      "loss": 1.0764,
+      "step": 8449
+    },
+    {
+      "epoch": 0.30730625159108266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623081894632907e-05,
+      "loss": 1.2328,
+      "step": 8450
+    },
+    {
+      "epoch": 0.30734261919482125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6229897630823814e-05,
+      "loss": 1.1553,
+      "step": 8451
+    },
+    {
+      "epoch": 0.30737898679855985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622897622888757e-05,
+      "loss": 1.0981,
+      "step": 8452
+    },
+    {
+      "epoch": 0.30741535440229845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622805474053312e-05,
+      "loss": 1.1392,
+      "step": 8453
+    },
+    {
+      "epoch": 0.30745172200603704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6227133165773244e-05,
+      "loss": 1.0745,
+      "step": 8454
+    },
+    {
+      "epoch": 0.30748808960977564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6226211504620736e-05,
+      "loss": 0.9849,
+      "step": 8455
+    },
+    {
+      "epoch": 0.3075244572135142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622528975708837e-05,
+      "loss": 1.0687,
+      "step": 8456
+    },
+    {
+      "epoch": 0.3075608248172528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6224367923188947e-05,
+      "loss": 0.9625,
+      "step": 8457
+    },
+    {
+      "epoch": 0.30759719242099137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6223446002935247e-05,
+      "loss": 1.0879,
+      "step": 8458
+    },
+    {
+      "epoch": 0.30763356002472997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6222523996340064e-05,
+      "loss": 1.0585,
+      "step": 8459
+    },
+    {
+      "epoch": 0.30766992762846856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.622160190341619e-05,
+      "loss": 1.0958,
+      "step": 8460
+    },
+    {
+      "epoch": 0.30770629523220716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6220679724176417e-05,
+      "loss": 1.0729,
+      "step": 8461
+    },
+    {
+      "epoch": 0.30774266283594576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6219757458633537e-05,
+      "loss": 1.1247,
+      "step": 8462
+    },
+    {
+      "epoch": 0.30777903043968435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6218835106800352e-05,
+      "loss": 1.0704,
+      "step": 8463
+    },
+    {
+      "epoch": 0.3078153980434229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6217912668689647e-05,
+      "loss": 1.1048,
+      "step": 8464
+    },
+    {
+      "epoch": 0.3078517656471615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6216990144314228e-05,
+      "loss": 0.9393,
+      "step": 8465
+    },
+    {
+      "epoch": 0.3078881332509001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6216067533686893e-05,
+      "loss": 0.9755,
+      "step": 8466
+    },
+    {
+      "epoch": 0.3079245008546387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.621514483682044e-05,
+      "loss": 1.0388,
+      "step": 8467
+    },
+    {
+      "epoch": 0.3079608684583773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6214222053727667e-05,
+      "loss": 1.1952,
+      "step": 8468
+    },
+    {
+      "epoch": 0.3079972360621159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6213299184421383e-05,
+      "loss": 1.0328,
+      "step": 8469
+    },
+    {
+      "epoch": 0.30803360366585447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6212376228914393e-05,
+      "loss": 1.0096,
+      "step": 8470
+    },
+    {
+      "epoch": 0.30806997126959307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.621145318721949e-05,
+      "loss": 0.9832,
+      "step": 8471
+    },
+    {
+      "epoch": 0.3081063388733316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.621053005934949e-05,
+      "loss": 1.1527,
+      "step": 8472
+    },
+    {
+      "epoch": 0.3081427064770702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6209606845317195e-05,
+      "loss": 1.0934,
+      "step": 8473
+    },
+    {
+      "epoch": 0.3081790740808088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6208683545135415e-05,
+      "loss": 1.2017,
+      "step": 8474
+    },
+    {
+      "epoch": 0.3082154416845474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6207760158816965e-05,
+      "loss": 1.0936,
+      "step": 8475
+    },
+    {
+      "epoch": 0.308251809288286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6206836686374643e-05,
+      "loss": 1.051,
+      "step": 8476
+    },
+    {
+      "epoch": 0.3082881768920246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6205913127821272e-05,
+      "loss": 1.0544,
+      "step": 8477
+    },
+    {
+      "epoch": 0.3083245444957632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.620498948316966e-05,
+      "loss": 1.055,
+      "step": 8478
+    },
+    {
+      "epoch": 0.3083609120995018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6204065752432625e-05,
+      "loss": 1.0953,
+      "step": 8479
+    },
+    {
+      "epoch": 0.3083972797032404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6203141935622978e-05,
+      "loss": 1.1505,
+      "step": 8480
+    },
+    {
+      "epoch": 0.3084336473069789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.620221803275354e-05,
+      "loss": 1.0982,
+      "step": 8481
+    },
+    {
+      "epoch": 0.3084700149107175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6201294043837127e-05,
+      "loss": 1.0403,
+      "step": 8482
+    },
+    {
+      "epoch": 0.3085063825144561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6200369968886555e-05,
+      "loss": 1.0258,
+      "step": 8483
+    },
+    {
+      "epoch": 0.3085427501181947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6199445807914647e-05,
+      "loss": 1.097,
+      "step": 8484
+    },
+    {
+      "epoch": 0.3085791177219333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6198521560934225e-05,
+      "loss": 1.1062,
+      "step": 8485
+    },
+    {
+      "epoch": 0.3086154853256719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6197597227958112e-05,
+      "loss": 1.158,
+      "step": 8486
+    },
+    {
+      "epoch": 0.3086518529294105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619667280899913e-05,
+      "loss": 1.0225,
+      "step": 8487
+    },
+    {
+      "epoch": 0.3086882205331491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6195748304070106e-05,
+      "loss": 0.9855,
+      "step": 8488
+    },
+    {
+      "epoch": 0.30872458813688763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619482371318386e-05,
+      "loss": 1.0236,
+      "step": 8489
+    },
+    {
+      "epoch": 0.30876095574062623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619389903635323e-05,
+      "loss": 1.137,
+      "step": 8490
+    },
+    {
+      "epoch": 0.3087973233443648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6192974273591037e-05,
+      "loss": 1.1248,
+      "step": 8491
+    },
+    {
+      "epoch": 0.3088336909481034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.619204942491011e-05,
+      "loss": 1.0125,
+      "step": 8492
+    },
+    {
+      "epoch": 0.308870058551842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6191124490323284e-05,
+      "loss": 1.0386,
+      "step": 8493
+    },
+    {
+      "epoch": 0.3089064261555806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6190199469843392e-05,
+      "loss": 1.2042,
+      "step": 8494
+    },
+    {
+      "epoch": 0.3089427937593192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6189274363483264e-05,
+      "loss": 1.0559,
+      "step": 8495
+    },
+    {
+      "epoch": 0.3089791613630578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6188349171255736e-05,
+      "loss": 1.0812,
+      "step": 8496
+    },
+    {
+      "epoch": 0.3090155289667964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.618742389317364e-05,
+      "loss": 1.1246,
+      "step": 8497
+    },
+    {
+      "epoch": 0.30905189657053495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.618649852924982e-05,
+      "loss": 1.0987,
+      "step": 8498
+    },
+    {
+      "epoch": 0.30908826417427354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.618557307949711e-05,
+      "loss": 1.0167,
+      "step": 8499
+    },
+    {
+      "epoch": 0.30912463177801214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.618464754392835e-05,
+      "loss": 1.0768,
+      "step": 8500
+    },
+    {
+      "epoch": 0.30916099938175073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6183721922556376e-05,
+      "loss": 1.0597,
+      "step": 8501
+    },
+    {
+      "epoch": 0.30919736698548933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.618279621539404e-05,
+      "loss": 1.0934,
+      "step": 8502
+    },
+    {
+      "epoch": 0.3092337345892279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6181870422454172e-05,
+      "loss": 1.0289,
+      "step": 8503
+    },
+    {
+      "epoch": 0.3092701021929665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.618094454374963e-05,
+      "loss": 1.0162,
+      "step": 8504
+    },
+    {
+      "epoch": 0.3093064697967051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6180018579293246e-05,
+      "loss": 1.0735,
+      "step": 8505
+    },
+    {
+      "epoch": 0.30934283740044366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6179092529097877e-05,
+      "loss": 1.035,
+      "step": 8506
+    },
+    {
+      "epoch": 0.30937920500418226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6178166393176364e-05,
+      "loss": 1.1058,
+      "step": 8507
+    },
+    {
+      "epoch": 0.30941557260792085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.617724017154156e-05,
+      "loss": 0.958,
+      "step": 8508
+    },
+    {
+      "epoch": 0.30945194021165945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.617631386420631e-05,
+      "loss": 1.0526,
+      "step": 8509
+    },
+    {
+      "epoch": 0.30948830781539804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6175387471183472e-05,
+      "loss": 0.9932,
+      "step": 8510
+    },
+    {
+      "epoch": 0.30952467541913664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.617446099248589e-05,
+      "loss": 1.1188,
+      "step": 8511
+    },
+    {
+      "epoch": 0.30956104302287524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6173534428126423e-05,
+      "loss": 0.974,
+      "step": 8512
+    },
+    {
+      "epoch": 0.30959741062661383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6172607778117927e-05,
+      "loss": 1.0421,
+      "step": 8513
+    },
+    {
+      "epoch": 0.3096337782303524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6171681042473255e-05,
+      "loss": 1.0687,
+      "step": 8514
+    },
+    {
+      "epoch": 0.30967014583409097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6170754221205265e-05,
+      "loss": 0.9983,
+      "step": 8515
+    },
+    {
+      "epoch": 0.30970651343782957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6169827314326814e-05,
+      "loss": 0.9996,
+      "step": 8516
+    },
+    {
+      "epoch": 0.30974288104156816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6168900321850765e-05,
+      "loss": 0.9542,
+      "step": 8517
+    },
+    {
+      "epoch": 0.30977924864530676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6167973243789975e-05,
+      "loss": 1.1193,
+      "step": 8518
+    },
+    {
+      "epoch": 0.30981561624904536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6167046080157308e-05,
+      "loss": 1.0233,
+      "step": 8519
+    },
+    {
+      "epoch": 0.30985198385278395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.616611883096563e-05,
+      "loss": 1.1085,
+      "step": 8520
+    },
+    {
+      "epoch": 0.30988835145652255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6165191496227795e-05,
+      "loss": 1.1364,
+      "step": 8521
+    },
+    {
+      "epoch": 0.30992471906026114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6164264075956682e-05,
+      "loss": 1.1048,
+      "step": 8522
+    },
+    {
+      "epoch": 0.3099610866639997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6163336570165148e-05,
+      "loss": 1.0805,
+      "step": 8523
+    },
+    {
+      "epoch": 0.3099974542677383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.616240897886606e-05,
+      "loss": 1.113,
+      "step": 8524
+    },
+    {
+      "epoch": 0.3100338218714769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6161481302072294e-05,
+      "loss": 0.9832,
+      "step": 8525
+    },
+    {
+      "epoch": 0.3100701894752155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6160553539796717e-05,
+      "loss": 1.1045,
+      "step": 8526
+    },
+    {
+      "epoch": 0.31010655707895407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.61596256920522e-05,
+      "loss": 1.0619,
+      "step": 8527
+    },
+    {
+      "epoch": 0.31014292468269267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6158697758851617e-05,
+      "loss": 1.1116,
+      "step": 8528
+    },
+    {
+      "epoch": 0.31017929228643126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.615776974020784e-05,
+      "loss": 1.0262,
+      "step": 8529
+    },
+    {
+      "epoch": 0.31021565989016986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6156841636133744e-05,
+      "loss": 1.0664,
+      "step": 8530
+    },
+    {
+      "epoch": 0.3102520274939084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6155913446642206e-05,
+      "loss": 1.1118,
+      "step": 8531
+    },
+    {
+      "epoch": 0.310288395097647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6154985171746104e-05,
+      "loss": 1.1653,
+      "step": 8532
+    },
+    {
+      "epoch": 0.3103247627013856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6154056811458314e-05,
+      "loss": 1.1978,
+      "step": 8533
+    },
+    {
+      "epoch": 0.3103611303051242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6153128365791716e-05,
+      "loss": 1.1277,
+      "step": 8534
+    },
+    {
+      "epoch": 0.3103974979088628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6152199834759197e-05,
+      "loss": 1.0684,
+      "step": 8535
+    },
+    {
+      "epoch": 0.3104338655126014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6151271218373628e-05,
+      "loss": 1.1424,
+      "step": 8536
+    },
+    {
+      "epoch": 0.31047023311634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6150342516647904e-05,
+      "loss": 1.0482,
+      "step": 8537
+    },
+    {
+      "epoch": 0.3105066007200786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.61494137295949e-05,
+      "loss": 1.0832,
+      "step": 8538
+    },
+    {
+      "epoch": 0.31054296832381717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614848485722751e-05,
+      "loss": 0.9765,
+      "step": 8539
+    },
+    {
+      "epoch": 0.3105793359275557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614755589955861e-05,
+      "loss": 1.0859,
+      "step": 8540
+    },
+    {
+      "epoch": 0.3106157035312943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.61466268566011e-05,
+      "loss": 0.9751,
+      "step": 8541
+    },
+    {
+      "epoch": 0.3106520711350329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614569772836786e-05,
+      "loss": 1.0557,
+      "step": 8542
+    },
+    {
+      "epoch": 0.3106884387387715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614476851487178e-05,
+      "loss": 1.1114,
+      "step": 8543
+    },
+    {
+      "epoch": 0.3107248063425101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6143839216125762e-05,
+      "loss": 1.0426,
+      "step": 8544
+    },
+    {
+      "epoch": 0.3107611739462487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614290983214269e-05,
+      "loss": 0.9342,
+      "step": 8545
+    },
+    {
+      "epoch": 0.3107975415499873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6141980362935455e-05,
+      "loss": 1.1688,
+      "step": 8546
+    },
+    {
+      "epoch": 0.3108339091537259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614105080851696e-05,
+      "loss": 1.1013,
+      "step": 8547
+    },
+    {
+      "epoch": 0.3108702767574644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.61401211689001e-05,
+      "loss": 1.0826,
+      "step": 8548
+    },
+    {
+      "epoch": 0.310906644361203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6139191444097767e-05,
+      "loss": 1.0326,
+      "step": 8549
+    },
+    {
+      "epoch": 0.3109430119649416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6138261634122865e-05,
+      "loss": 1.065,
+      "step": 8550
+    },
+    {
+      "epoch": 0.3109793795686802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.613733173898829e-05,
+      "loss": 1.0951,
+      "step": 8551
+    },
+    {
+      "epoch": 0.3110157471724188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6136401758706946e-05,
+      "loss": 1.0561,
+      "step": 8552
+    },
+    {
+      "epoch": 0.3110521147761574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6135471693291736e-05,
+      "loss": 1.0108,
+      "step": 8553
+    },
+    {
+      "epoch": 0.311088482379896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6134541542755563e-05,
+      "loss": 1.2384,
+      "step": 8554
+    },
+    {
+      "epoch": 0.3111248499836346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6133611307111325e-05,
+      "loss": 1.1776,
+      "step": 8555
+    },
+    {
+      "epoch": 0.3111612175873732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6132680986371935e-05,
+      "loss": 1.0012,
+      "step": 8556
+    },
+    {
+      "epoch": 0.31119758519111174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6131750580550297e-05,
+      "loss": 1.1377,
+      "step": 8557
+    },
+    {
+      "epoch": 0.31123395279485033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.613082008965932e-05,
+      "loss": 1.1529,
+      "step": 8558
+    },
+    {
+      "epoch": 0.31127032039858893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6129889513711912e-05,
+      "loss": 1.0794,
+      "step": 8559
+    },
+    {
+      "epoch": 0.3113066880023275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6128958852720983e-05,
+      "loss": 1.1115,
+      "step": 8560
+    },
+    {
+      "epoch": 0.3113430556060661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6128028106699448e-05,
+      "loss": 1.1362,
+      "step": 8561
+    },
+    {
+      "epoch": 0.3113794232098047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.612709727566022e-05,
+      "loss": 1.0612,
+      "step": 8562
+    },
+    {
+      "epoch": 0.3114157908135433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6126166359616205e-05,
+      "loss": 0.9526,
+      "step": 8563
+    },
+    {
+      "epoch": 0.3114521584172819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6125235358580328e-05,
+      "loss": 1.0284,
+      "step": 8564
+    },
+    {
+      "epoch": 0.31148852602102045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6124304272565497e-05,
+      "loss": 1.0954,
+      "step": 8565
+    },
+    {
+      "epoch": 0.31152489362475905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6123373101584636e-05,
+      "loss": 0.9997,
+      "step": 8566
+    },
+    {
+      "epoch": 0.31156126122849764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.612244184565066e-05,
+      "loss": 1.0263,
+      "step": 8567
+    },
+    {
+      "epoch": 0.31159762883223624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.612151050477649e-05,
+      "loss": 1.0578,
+      "step": 8568
+    },
+    {
+      "epoch": 0.31163399643597484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6120579078975047e-05,
+      "loss": 1.0812,
+      "step": 8569
+    },
+    {
+      "epoch": 0.31167036403971343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6119647568259255e-05,
+      "loss": 1.1482,
+      "step": 8570
+    },
+    {
+      "epoch": 0.31170673164345203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6118715972642034e-05,
+      "loss": 1.0297,
+      "step": 8571
+    },
+    {
+      "epoch": 0.3117430992471906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.611778429213631e-05,
+      "loss": 1.0442,
+      "step": 8572
+    },
+    {
+      "epoch": 0.31177946685092917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6116852526755008e-05,
+      "loss": 1.0142,
+      "step": 8573
+    },
+    {
+      "epoch": 0.31181583445466776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6115920676511057e-05,
+      "loss": 1.103,
+      "step": 8574
+    },
+    {
+      "epoch": 0.31185220205840636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6114988741417386e-05,
+      "loss": 1.0985,
+      "step": 8575
+    },
+    {
+      "epoch": 0.31188856966214495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.611405672148692e-05,
+      "loss": 0.947,
+      "step": 8576
+    },
+    {
+      "epoch": 0.31192493726588355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6113124616732592e-05,
+      "loss": 1.0681,
+      "step": 8577
+    },
+    {
+      "epoch": 0.31196130486962215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6112192427167338e-05,
+      "loss": 0.9835,
+      "step": 8578
+    },
+    {
+      "epoch": 0.31199767247336074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.611126015280408e-05,
+      "loss": 0.9714,
+      "step": 8579
+    },
+    {
+      "epoch": 0.31203404007709934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.611032779365576e-05,
+      "loss": 1.0067,
+      "step": 8580
+    },
+    {
+      "epoch": 0.31207040768083794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610939534973531e-05,
+      "loss": 1.1401,
+      "step": 8581
+    },
+    {
+      "epoch": 0.3121067752845765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610846282105567e-05,
+      "loss": 1.0187,
+      "step": 8582
+    },
+    {
+      "epoch": 0.3121431428883151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610753020762977e-05,
+      "loss": 1.0917,
+      "step": 8583
+    },
+    {
+      "epoch": 0.31217951049205367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610659750947056e-05,
+      "loss": 1.0113,
+      "step": 8584
+    },
+    {
+      "epoch": 0.31221587809579227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610566472659097e-05,
+      "loss": 1.1008,
+      "step": 8585
+    },
+    {
+      "epoch": 0.31225224569953086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610473185900395e-05,
+      "loss": 0.9776,
+      "step": 8586
+    },
+    {
+      "epoch": 0.31228861330326946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6103798906722432e-05,
+      "loss": 1.0653,
+      "step": 8587
+    },
+    {
+      "epoch": 0.31232498090700805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6102865869759362e-05,
+      "loss": 1.0141,
+      "step": 8588
+    },
+    {
+      "epoch": 0.31236134851074665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610193274812769e-05,
+      "loss": 1.106,
+      "step": 8589
+    },
+    {
+      "epoch": 0.3123977161144852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610099954184036e-05,
+      "loss": 0.9904,
+      "step": 8590
+    },
+    {
+      "epoch": 0.3124340837182238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6100066250910318e-05,
+      "loss": 1.1306,
+      "step": 8591
+    },
+    {
+      "epoch": 0.3124704513219624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.609913287535051e-05,
+      "loss": 1.0946,
+      "step": 8592
+    },
+    {
+      "epoch": 0.312506818925701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6098199415173885e-05,
+      "loss": 1.0988,
+      "step": 8593
+    },
+    {
+      "epoch": 0.3125431865294396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6097265870393398e-05,
+      "loss": 1.0962,
+      "step": 8594
+    },
+    {
+      "epoch": 0.3125795541331782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6096332241022e-05,
+      "loss": 1.0781,
+      "step": 8595
+    },
+    {
+      "epoch": 0.31261592173691677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6095398527072638e-05,
+      "loss": 1.1021,
+      "step": 8596
+    },
+    {
+      "epoch": 0.31265228934065536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.609446472855827e-05,
+      "loss": 1.1497,
+      "step": 8597
+    },
+    {
+      "epoch": 0.31268865694439396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6093530845491852e-05,
+      "loss": 1.0503,
+      "step": 8598
+    },
+    {
+      "epoch": 0.3127250245481325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.609259687788634e-05,
+      "loss": 1.028,
+      "step": 8599
+    },
+    {
+      "epoch": 0.3127613921518711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6091662825754687e-05,
+      "loss": 1.0137,
+      "step": 8600
+    },
+    {
+      "epoch": 0.3127977597556097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.609072868910986e-05,
+      "loss": 1.1323,
+      "step": 8601
+    },
+    {
+      "epoch": 0.3128341273593483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608979446796481e-05,
+      "loss": 0.9997,
+      "step": 8602
+    },
+    {
+      "epoch": 0.3128704949630869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6088860162332504e-05,
+      "loss": 0.9716,
+      "step": 8603
+    },
+    {
+      "epoch": 0.3129068625668255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60879257722259e-05,
+      "loss": 0.9866,
+      "step": 8604
+    },
+    {
+      "epoch": 0.3129432301705641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6086991297657965e-05,
+      "loss": 0.9911,
+      "step": 8605
+    },
+    {
+      "epoch": 0.3129795977743027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6086056738641663e-05,
+      "loss": 0.9818,
+      "step": 8606
+    },
+    {
+      "epoch": 0.3130159653780412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608512209518996e-05,
+      "loss": 1.0831,
+      "step": 8607
+    },
+    {
+      "epoch": 0.3130523329817798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608418736731582e-05,
+      "loss": 1.0873,
+      "step": 8608
+    },
+    {
+      "epoch": 0.3130887005855184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608325255503221e-05,
+      "loss": 1.1419,
+      "step": 8609
+    },
+    {
+      "epoch": 0.313125068189257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6082317658352106e-05,
+      "loss": 1.1785,
+      "step": 8610
+    },
+    {
+      "epoch": 0.3131614357929956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6081382677288472e-05,
+      "loss": 1.1393,
+      "step": 8611
+    },
+    {
+      "epoch": 0.3131978033967342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608044761185428e-05,
+      "loss": 1.1579,
+      "step": 8612
+    },
+    {
+      "epoch": 0.3132341710004728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6079512462062504e-05,
+      "loss": 1.1653,
+      "step": 8613
+    },
+    {
+      "epoch": 0.3132705386042114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6078577227926123e-05,
+      "loss": 1.0951,
+      "step": 8614
+    },
+    {
+      "epoch": 0.31330690620794993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6077641909458105e-05,
+      "loss": 1.067,
+      "step": 8615
+    },
+    {
+      "epoch": 0.3133432738116885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607670650667143e-05,
+      "loss": 1.0197,
+      "step": 8616
+    },
+    {
+      "epoch": 0.3133796414154271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6075771019579072e-05,
+      "loss": 1.2058,
+      "step": 8617
+    },
+    {
+      "epoch": 0.3134160090191657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607483544819401e-05,
+      "loss": 1.0396,
+      "step": 8618
+    },
+    {
+      "epoch": 0.3134523766229043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607389979252923e-05,
+      "loss": 1.0682,
+      "step": 8619
+    },
+    {
+      "epoch": 0.3134887442266429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6072964052597705e-05,
+      "loss": 1.0121,
+      "step": 8620
+    },
+    {
+      "epoch": 0.3135251118303815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.607202822841242e-05,
+      "loss": 1.0637,
+      "step": 8621
+    },
+    {
+      "epoch": 0.3135614794341201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6071092319986358e-05,
+      "loss": 1.0407,
+      "step": 8622
+    },
+    {
+      "epoch": 0.3135978470378587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6070156327332505e-05,
+      "loss": 1.016,
+      "step": 8623
+    },
+    {
+      "epoch": 0.31363421464159724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6069220250463848e-05,
+      "loss": 1.0579,
+      "step": 8624
+    },
+    {
+      "epoch": 0.31367058224533584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6068284089393363e-05,
+      "loss": 1.1035,
+      "step": 8625
+    },
+    {
+      "epoch": 0.31370694984907443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6067347844134055e-05,
+      "loss": 0.9371,
+      "step": 8626
+    },
+    {
+      "epoch": 0.31374331745281303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60664115146989e-05,
+      "loss": 1.0556,
+      "step": 8627
+    },
+    {
+      "epoch": 0.3137796850565516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606547510110089e-05,
+      "loss": 1.0558,
+      "step": 8628
+    },
+    {
+      "epoch": 0.3138160526602902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6064538603353023e-05,
+      "loss": 1.1445,
+      "step": 8629
+    },
+    {
+      "epoch": 0.3138524202640288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6063602021468286e-05,
+      "loss": 1.0137,
+      "step": 8630
+    },
+    {
+      "epoch": 0.3138887878677674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6062665355459672e-05,
+      "loss": 1.1793,
+      "step": 8631
+    },
+    {
+      "epoch": 0.31392515547150596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606172860534018e-05,
+      "loss": 1.1148,
+      "step": 8632
+    },
+    {
+      "epoch": 0.31396152307524455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6060791771122802e-05,
+      "loss": 1.1021,
+      "step": 8633
+    },
+    {
+      "epoch": 0.31399789067898315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6059854852820543e-05,
+      "loss": 0.9403,
+      "step": 8634
+    },
+    {
+      "epoch": 0.31403425828272175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.605891785044639e-05,
+      "loss": 1.1028,
+      "step": 8635
+    },
+    {
+      "epoch": 0.31407062588646034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6057980764013354e-05,
+      "loss": 1.0024,
+      "step": 8636
+    },
+    {
+      "epoch": 0.31410699349019894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6057043593534424e-05,
+      "loss": 1.0022,
+      "step": 8637
+    },
+    {
+      "epoch": 0.31414336109393753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.605610633902261e-05,
+      "loss": 1.0145,
+      "step": 8638
+    },
+    {
+      "epoch": 0.31417972869767613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6055169000490917e-05,
+      "loss": 0.9884,
+      "step": 8639
+    },
+    {
+      "epoch": 0.3142160963014147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.605423157795234e-05,
+      "loss": 1.1494,
+      "step": 8640
+    },
+    {
+      "epoch": 0.31425246390515327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6053294071419888e-05,
+      "loss": 1.0415,
+      "step": 8641
+    },
+    {
+      "epoch": 0.31428883150889186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6052356480906573e-05,
+      "loss": 1.0121,
+      "step": 8642
+    },
+    {
+      "epoch": 0.31432519911263046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60514188064254e-05,
+      "loss": 1.0382,
+      "step": 8643
+    },
+    {
+      "epoch": 0.31436156671636906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6050481047989376e-05,
+      "loss": 1.0127,
+      "step": 8644
+    },
+    {
+      "epoch": 0.31439793432010765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6049543205611512e-05,
+      "loss": 0.9381,
+      "step": 8645
+    },
+    {
+      "epoch": 0.31443430192384625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6048605279304816e-05,
+      "loss": 1.0278,
+      "step": 8646
+    },
+    {
+      "epoch": 0.31447066952758485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6047667269082308e-05,
+      "loss": 1.1166,
+      "step": 8647
+    },
+    {
+      "epoch": 0.31450703713132344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6046729174956996e-05,
+      "loss": 1.1273,
+      "step": 8648
+    },
+    {
+      "epoch": 0.314543404735062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6045790996941893e-05,
+      "loss": 1.112,
+      "step": 8649
+    },
+    {
+      "epoch": 0.3145797723388006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.604485273505002e-05,
+      "loss": 0.9504,
+      "step": 8650
+    },
+    {
+      "epoch": 0.3146161399425392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6043914389294393e-05,
+      "loss": 1.0763,
+      "step": 8651
+    },
+    {
+      "epoch": 0.31465250754627777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6042975959688028e-05,
+      "loss": 1.0901,
+      "step": 8652
+    },
+    {
+      "epoch": 0.31468887515001637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6042037446243944e-05,
+      "loss": 1.0539,
+      "step": 8653
+    },
+    {
+      "epoch": 0.31472524275375496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6041098848975166e-05,
+      "loss": 1.0479,
+      "step": 8654
+    },
+    {
+      "epoch": 0.31476161035749356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.604016016789471e-05,
+      "loss": 1.2226,
+      "step": 8655
+    },
+    {
+      "epoch": 0.31479797796123216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60392214030156e-05,
+      "loss": 1.0427,
+      "step": 8656
+    },
+    {
+      "epoch": 0.3148343455649707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6038282554350866e-05,
+      "loss": 1.0155,
+      "step": 8657
+    },
+    {
+      "epoch": 0.3148707131687093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.603734362191353e-05,
+      "loss": 1.0257,
+      "step": 8658
+    },
+    {
+      "epoch": 0.3149070807724479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.603640460571661e-05,
+      "loss": 0.9684,
+      "step": 8659
+    },
+    {
+      "epoch": 0.3149434483761865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6035465505773148e-05,
+      "loss": 0.9052,
+      "step": 8660
+    },
+    {
+      "epoch": 0.3149798159799251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.603452632209616e-05,
+      "loss": 1.1365,
+      "step": 8661
+    },
+    {
+      "epoch": 0.3150161835836637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6033587054698684e-05,
+      "loss": 1.0493,
+      "step": 8662
+    },
+    {
+      "epoch": 0.3150525511874023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6032647703593743e-05,
+      "loss": 1.067,
+      "step": 8663
+    },
+    {
+      "epoch": 0.31508891879114087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.603170826879438e-05,
+      "loss": 1.1231,
+      "step": 8664
+    },
+    {
+      "epoch": 0.31512528639487947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6030768750313624e-05,
+      "loss": 1.154,
+      "step": 8665
+    },
+    {
+      "epoch": 0.315161653998618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6029829148164504e-05,
+      "loss": 1.1188,
+      "step": 8666
+    },
+    {
+      "epoch": 0.3151980216023566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602888946236006e-05,
+      "loss": 1.1477,
+      "step": 8667
+    },
+    {
+      "epoch": 0.3152343892060952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602794969291333e-05,
+      "loss": 1.0244,
+      "step": 8668
+    },
+    {
+      "epoch": 0.3152707568098338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602700983983735e-05,
+      "loss": 1.2006,
+      "step": 8669
+    },
+    {
+      "epoch": 0.3153071244135724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602606990314516e-05,
+      "loss": 1.1084,
+      "step": 8670
+    },
+    {
+      "epoch": 0.315343492017311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60251298828498e-05,
+      "loss": 1.2125,
+      "step": 8671
+    },
+    {
+      "epoch": 0.3153798596210496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602418977896431e-05,
+      "loss": 0.9706,
+      "step": 8672
+    },
+    {
+      "epoch": 0.3154162272247882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602324959150174e-05,
+      "loss": 1.1683,
+      "step": 8673
+    },
+    {
+      "epoch": 0.3154525948285267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602230932047512e-05,
+      "loss": 1.1613,
+      "step": 8674
+    },
+    {
+      "epoch": 0.3154889624322653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6021368965897506e-05,
+      "loss": 1.1007,
+      "step": 8675
+    },
+    {
+      "epoch": 0.3155253300360039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602042852778194e-05,
+      "loss": 1.1299,
+      "step": 8676
+    },
+    {
+      "epoch": 0.3155616976397425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6019488006141472e-05,
+      "loss": 1.1138,
+      "step": 8677
+    },
+    {
+      "epoch": 0.3155980652434811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6018547400989146e-05,
+      "loss": 1.071,
+      "step": 8678
+    },
+    {
+      "epoch": 0.3156344328472197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6017606712338016e-05,
+      "loss": 0.978,
+      "step": 8679
+    },
+    {
+      "epoch": 0.3156708004509583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.601666594020113e-05,
+      "loss": 1.045,
+      "step": 8680
+    },
+    {
+      "epoch": 0.3157071680546969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.601572508459154e-05,
+      "loss": 0.9983,
+      "step": 8681
+    },
+    {
+      "epoch": 0.3157435356584355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60147841455223e-05,
+      "loss": 1.041,
+      "step": 8682
+    },
+    {
+      "epoch": 0.31577990326217403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6013843123006463e-05,
+      "loss": 1.129,
+      "step": 8683
+    },
+    {
+      "epoch": 0.31581627086591263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6012902017057085e-05,
+      "loss": 1.1795,
+      "step": 8684
+    },
+    {
+      "epoch": 0.3158526384696512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6011960827687225e-05,
+      "loss": 1.0817,
+      "step": 8685
+    },
+    {
+      "epoch": 0.3158890060733898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.601101955490994e-05,
+      "loss": 1.1251,
+      "step": 8686
+    },
+    {
+      "epoch": 0.3159253736771284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.601007819873828e-05,
+      "loss": 1.1227,
+      "step": 8687
+    },
+    {
+      "epoch": 0.315961741280867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6009136759185318e-05,
+      "loss": 1.1497,
+      "step": 8688
+    },
+    {
+      "epoch": 0.3159981088846056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.600819523626411e-05,
+      "loss": 1.1243,
+      "step": 8689
+    },
+    {
+      "epoch": 0.3160344764883442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6007253629987714e-05,
+      "loss": 1.119,
+      "step": 8690
+    },
+    {
+      "epoch": 0.31607084409208275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6006311940369198e-05,
+      "loss": 1.1309,
+      "step": 8691
+    },
+    {
+      "epoch": 0.31610721169582134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6005370167421628e-05,
+      "loss": 0.9725,
+      "step": 8692
+    },
+    {
+      "epoch": 0.31614357929955994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6004428311158063e-05,
+      "loss": 1.0919,
+      "step": 8693
+    },
+    {
+      "epoch": 0.31617994690329854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.600348637159158e-05,
+      "loss": 1.0537,
+      "step": 8694
+    },
+    {
+      "epoch": 0.31621631450703713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6002544348735236e-05,
+      "loss": 1.0538,
+      "step": 8695
+    },
+    {
+      "epoch": 0.31625268211077573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6001602242602106e-05,
+      "loss": 1.0165,
+      "step": 8696
+    },
+    {
+      "epoch": 0.3162890497145143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6000660053205265e-05,
+      "loss": 0.9781,
+      "step": 8697
+    },
+    {
+      "epoch": 0.3163254173182529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5999717780557776e-05,
+      "loss": 0.9926,
+      "step": 8698
+    },
+    {
+      "epoch": 0.31636178492199146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5998775424672716e-05,
+      "loss": 1.2135,
+      "step": 8699
+    },
+    {
+      "epoch": 0.31639815252573006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.599783298556316e-05,
+      "loss": 1.0522,
+      "step": 8700
+    },
+    {
+      "epoch": 0.31643452012946865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.599689046324218e-05,
+      "loss": 0.9508,
+      "step": 8701
+    },
+    {
+      "epoch": 0.31647088773320725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.599594785772285e-05,
+      "loss": 0.9908,
+      "step": 8702
+    },
+    {
+      "epoch": 0.31650725533694585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5995005169018254e-05,
+      "loss": 1.0421,
+      "step": 8703
+    },
+    {
+      "epoch": 0.31654362294068444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5994062397141466e-05,
+      "loss": 1.042,
+      "step": 8704
+    },
+    {
+      "epoch": 0.31657999054442304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5993119542105565e-05,
+      "loss": 1.1505,
+      "step": 8705
+    },
+    {
+      "epoch": 0.31661635814816164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.599217660392364e-05,
+      "loss": 1.0412,
+      "step": 8706
+    },
+    {
+      "epoch": 0.31665272575190023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5991233582608757e-05,
+      "loss": 0.9766,
+      "step": 8707
+    },
+    {
+      "epoch": 0.3166890933556388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5990290478174017e-05,
+      "loss": 1.1157,
+      "step": 8708
+    },
+    {
+      "epoch": 0.31672546095937737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598934729063249e-05,
+      "loss": 1.0527,
+      "step": 8709
+    },
+    {
+      "epoch": 0.31676182856311597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598840401999727e-05,
+      "loss": 0.8853,
+      "step": 8710
+    },
+    {
+      "epoch": 0.31679819616685456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5987460666281442e-05,
+      "loss": 0.9658,
+      "step": 8711
+    },
+    {
+      "epoch": 0.31683456377059316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5986517229498088e-05,
+      "loss": 1.0479,
+      "step": 8712
+    },
+    {
+      "epoch": 0.31687093137433175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5985573709660305e-05,
+      "loss": 1.0716,
+      "step": 8713
+    },
+    {
+      "epoch": 0.31690729897807035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598463010678118e-05,
+      "loss": 0.9927,
+      "step": 8714
+    },
+    {
+      "epoch": 0.31694366658180895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5983686420873798e-05,
+      "loss": 1.0403,
+      "step": 8715
+    },
+    {
+      "epoch": 0.3169800341855475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5982742651951258e-05,
+      "loss": 1.0873,
+      "step": 8716
+    },
+    {
+      "epoch": 0.3170164017892861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5981798800026655e-05,
+      "loss": 1.1326,
+      "step": 8717
+    },
+    {
+      "epoch": 0.3170527693930247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598085486511308e-05,
+      "loss": 0.9788,
+      "step": 8718
+    },
+    {
+      "epoch": 0.3170891369967633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597991084722363e-05,
+      "loss": 1.0818,
+      "step": 8719
+    },
+    {
+      "epoch": 0.3171255046005019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5978966746371397e-05,
+      "loss": 1.0574,
+      "step": 8720
+    },
+    {
+      "epoch": 0.31716187220424047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5978022562569487e-05,
+      "loss": 0.9828,
+      "step": 8721
+    },
+    {
+      "epoch": 0.31719823980797907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5977078295830997e-05,
+      "loss": 1.0829,
+      "step": 8722
+    },
+    {
+      "epoch": 0.31723460741171766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5976133946169024e-05,
+      "loss": 1.1111,
+      "step": 8723
+    },
+    {
+      "epoch": 0.31727097501545626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597518951359667e-05,
+      "loss": 1.048,
+      "step": 8724
+    },
+    {
+      "epoch": 0.3173073426191948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5974244998127044e-05,
+      "loss": 1.178,
+      "step": 8725
+    },
+    {
+      "epoch": 0.3173437102229334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597330039977324e-05,
+      "loss": 0.9551,
+      "step": 8726
+    },
+    {
+      "epoch": 0.317380077826672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597235571854837e-05,
+      "loss": 0.9488,
+      "step": 8727
+    },
+    {
+      "epoch": 0.3174164454304106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597141095446554e-05,
+      "loss": 1.1107,
+      "step": 8728
+    },
+    {
+      "epoch": 0.3174528130341492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5970466107537852e-05,
+      "loss": 1.0068,
+      "step": 8729
+    },
+    {
+      "epoch": 0.3174891806378878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5969521177778423e-05,
+      "loss": 0.9697,
+      "step": 8730
+    },
+    {
+      "epoch": 0.3175255482416264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5968576165200352e-05,
+      "loss": 1.1032,
+      "step": 8731
+    },
+    {
+      "epoch": 0.317561915845365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5967631069816753e-05,
+      "loss": 1.1801,
+      "step": 8732
+    },
+    {
+      "epoch": 0.3175982834491035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5966685891640748e-05,
+      "loss": 1.1034,
+      "step": 8733
+    },
+    {
+      "epoch": 0.3176346510528421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5965740630685433e-05,
+      "loss": 1.157,
+      "step": 8734
+    },
+    {
+      "epoch": 0.3176710186565807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5964795286963937e-05,
+      "loss": 0.9694,
+      "step": 8735
+    },
+    {
+      "epoch": 0.3177073862603193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.596384986048937e-05,
+      "loss": 1.0515,
+      "step": 8736
+    },
+    {
+      "epoch": 0.3177437538640579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5962904351274843e-05,
+      "loss": 1.1805,
+      "step": 8737
+    },
+    {
+      "epoch": 0.3177801214677965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.596195875933348e-05,
+      "loss": 1.0759,
+      "step": 8738
+    },
+    {
+      "epoch": 0.3178164890715351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.59610130846784e-05,
+      "loss": 1.0397,
+      "step": 8739
+    },
+    {
+      "epoch": 0.3178528566752737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.596006732732272e-05,
+      "loss": 1.0698,
+      "step": 8740
+    },
+    {
+      "epoch": 0.3178892242790123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.595912148727956e-05,
+      "loss": 0.9814,
+      "step": 8741
+    },
+    {
+      "epoch": 0.3179255918827508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5958175564562046e-05,
+      "loss": 1.0921,
+      "step": 8742
+    },
+    {
+      "epoch": 0.3179619594864894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.59572295591833e-05,
+      "loss": 0.9898,
+      "step": 8743
+    },
+    {
+      "epoch": 0.317998327090228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5956283471156448e-05,
+      "loss": 1.0195,
+      "step": 8744
+    },
+    {
+      "epoch": 0.3180346946939666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5955337300494612e-05,
+      "loss": 1.0762,
+      "step": 8745
+    },
+    {
+      "epoch": 0.3180710622977052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.595439104721092e-05,
+      "loss": 1.0106,
+      "step": 8746
+    },
+    {
+      "epoch": 0.3181074299014438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.59534447113185e-05,
+      "loss": 1.1585,
+      "step": 8747
+    },
+    {
+      "epoch": 0.3181437975051824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5952498292830482e-05,
+      "loss": 1.0582,
+      "step": 8748
+    },
+    {
+      "epoch": 0.318180165108921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5951551791759996e-05,
+      "loss": 1.0996,
+      "step": 8749
+    },
+    {
+      "epoch": 0.31821653271265954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5950605208120172e-05,
+      "loss": 1.1841,
+      "step": 8750
+    },
+    {
+      "epoch": 0.31825290031639814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5949658541924144e-05,
+      "loss": 1.0003,
+      "step": 8751
+    },
+    {
+      "epoch": 0.31828926792013673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5948711793185046e-05,
+      "loss": 0.9797,
+      "step": 8752
+    },
+    {
+      "epoch": 0.3183256355238753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.594776496191601e-05,
+      "loss": 1.0873,
+      "step": 8753
+    },
+    {
+      "epoch": 0.3183620031276139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5946818048130178e-05,
+      "loss": 1.0619,
+      "step": 8754
+    },
+    {
+      "epoch": 0.3183983707313525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5945871051840676e-05,
+      "loss": 1.0635,
+      "step": 8755
+    },
+    {
+      "epoch": 0.3184347383350911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5944923973060655e-05,
+      "loss": 1.0707,
+      "step": 8756
+    },
+    {
+      "epoch": 0.3184711059388297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5943976811803247e-05,
+      "loss": 1.161,
+      "step": 8757
+    },
+    {
+      "epoch": 0.31850747354256825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5943029568081595e-05,
+      "loss": 1.1379,
+      "step": 8758
+    },
+    {
+      "epoch": 0.31854384114630685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5942082241908837e-05,
+      "loss": 1.1803,
+      "step": 8759
+    },
+    {
+      "epoch": 0.31858020875004545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5941134833298124e-05,
+      "loss": 0.9186,
+      "step": 8760
+    },
+    {
+      "epoch": 0.31861657635378404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.594018734226259e-05,
+      "loss": 0.9931,
+      "step": 8761
+    },
+    {
+      "epoch": 0.31865294395752264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5939239768815385e-05,
+      "loss": 1.1008,
+      "step": 8762
+    },
+    {
+      "epoch": 0.31868931156126123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5938292112969656e-05,
+      "loss": 1.1068,
+      "step": 8763
+    },
+    {
+      "epoch": 0.31872567916499983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.593734437473855e-05,
+      "loss": 1.0355,
+      "step": 8764
+    },
+    {
+      "epoch": 0.3187620467687384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5936396554135213e-05,
+      "loss": 1.1005,
+      "step": 8765
+    },
+    {
+      "epoch": 0.318798414372477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.59354486511728e-05,
+      "loss": 0.8314,
+      "step": 8766
+    },
+    {
+      "epoch": 0.31883478197621556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5934500665864453e-05,
+      "loss": 1.1277,
+      "step": 8767
+    },
+    {
+      "epoch": 0.31887114957995416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5933552598223333e-05,
+      "loss": 1.1012,
+      "step": 8768
+    },
+    {
+      "epoch": 0.31890751718369276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5932604448262587e-05,
+      "loss": 1.1825,
+      "step": 8769
+    },
+    {
+      "epoch": 0.31894388478743135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.593165621599537e-05,
+      "loss": 1.0986,
+      "step": 8770
+    },
+    {
+      "epoch": 0.31898025239116995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5930707901434847e-05,
+      "loss": 1.0737,
+      "step": 8771
+    },
+    {
+      "epoch": 0.31901661999490855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5929759504594158e-05,
+      "loss": 1.0384,
+      "step": 8772
+    },
+    {
+      "epoch": 0.31905298759864714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5928811025486472e-05,
+      "loss": 1.0442,
+      "step": 8773
+    },
+    {
+      "epoch": 0.31908935520238574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5927862464124944e-05,
+      "loss": 1.2285,
+      "step": 8774
+    },
+    {
+      "epoch": 0.3191257228061243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5926913820522738e-05,
+      "loss": 1.0692,
+      "step": 8775
+    },
+    {
+      "epoch": 0.3191620904098629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.592596509469301e-05,
+      "loss": 1.073,
+      "step": 8776
+    },
+    {
+      "epoch": 0.31919845801360147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5925016286648924e-05,
+      "loss": 1.0687,
+      "step": 8777
+    },
+    {
+      "epoch": 0.31923482561734007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5924067396403642e-05,
+      "loss": 0.973,
+      "step": 8778
+    },
+    {
+      "epoch": 0.31927119322107866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5923118423970333e-05,
+      "loss": 1.1499,
+      "step": 8779
+    },
+    {
+      "epoch": 0.31930756082481726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5922169369362157e-05,
+      "loss": 1.0654,
+      "step": 8780
+    },
+    {
+      "epoch": 0.31934392842855586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5921220232592283e-05,
+      "loss": 1.2091,
+      "step": 8781
+    },
+    {
+      "epoch": 0.31938029603229445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.592027101367388e-05,
+      "loss": 1.1647,
+      "step": 8782
+    },
+    {
+      "epoch": 0.31941666363603305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5919321712620117e-05,
+      "loss": 1.1464,
+      "step": 8783
+    },
+    {
+      "epoch": 0.3194530312397716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5918372329444168e-05,
+      "loss": 0.8965,
+      "step": 8784
+    },
+    {
+      "epoch": 0.3194893988435102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5917422864159196e-05,
+      "loss": 0.9746,
+      "step": 8785
+    },
+    {
+      "epoch": 0.3195257664472488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5916473316778374e-05,
+      "loss": 1.1059,
+      "step": 8786
+    },
+    {
+      "epoch": 0.3195621340509874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5915523687314887e-05,
+      "loss": 1.2029,
+      "step": 8787
+    },
+    {
+      "epoch": 0.319598501654726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.59145739757819e-05,
+      "loss": 1.1838,
+      "step": 8788
+    },
+    {
+      "epoch": 0.31963486925846457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5913624182192588e-05,
+      "loss": 1.0741,
+      "step": 8789
+    },
+    {
+      "epoch": 0.31967123686220317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5912674306560126e-05,
+      "loss": 1.1055,
+      "step": 8790
+    },
+    {
+      "epoch": 0.31970760446594176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5911724348897704e-05,
+      "loss": 1.1122,
+      "step": 8791
+    },
+    {
+      "epoch": 0.3197439720696803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5910774309218494e-05,
+      "loss": 1.1218,
+      "step": 8792
+    },
+    {
+      "epoch": 0.3197803396734189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5909824187535673e-05,
+      "loss": 1.0882,
+      "step": 8793
+    },
+    {
+      "epoch": 0.3198167072771575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5908873983862428e-05,
+      "loss": 1.0124,
+      "step": 8794
+    },
+    {
+      "epoch": 0.3198530748808961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5907923698211942e-05,
+      "loss": 1.0732,
+      "step": 8795
+    },
+    {
+      "epoch": 0.3198894424846347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5906973330597394e-05,
+      "loss": 1.062,
+      "step": 8796
+    },
+    {
+      "epoch": 0.3199258100883733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5906022881031976e-05,
+      "loss": 0.9888,
+      "step": 8797
+    },
+    {
+      "epoch": 0.3199621776921119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5905072349528865e-05,
+      "loss": 1.0519,
+      "step": 8798
+    },
+    {
+      "epoch": 0.3199985452958505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5904121736101257e-05,
+      "loss": 1.0916,
+      "step": 8799
+    },
+    {
+      "epoch": 0.320034912899589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5903171040762337e-05,
+      "loss": 1.1128,
+      "step": 8800
+    },
+    {
+      "epoch": 0.3200712805033276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590222026352529e-05,
+      "loss": 1.0609,
+      "step": 8801
+    },
+    {
+      "epoch": 0.3201076481070662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5901269404403316e-05,
+      "loss": 1.1668,
+      "step": 8802
+    },
+    {
+      "epoch": 0.3201440157108048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5900318463409598e-05,
+      "loss": 1.129,
+      "step": 8803
+    },
+    {
+      "epoch": 0.3201803833145434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5899367440557335e-05,
+      "loss": 1.1434,
+      "step": 8804
+    },
+    {
+      "epoch": 0.320216750918282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5898416335859718e-05,
+      "loss": 1.2028,
+      "step": 8805
+    },
+    {
+      "epoch": 0.3202531185220206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5897465149329946e-05,
+      "loss": 1.0735,
+      "step": 8806
+    },
+    {
+      "epoch": 0.3202894861257592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.589651388098121e-05,
+      "loss": 1.0978,
+      "step": 8807
+    },
+    {
+      "epoch": 0.3203258537294978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.589556253082671e-05,
+      "loss": 0.9877,
+      "step": 8808
+    },
+    {
+      "epoch": 0.32036222133323633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5894611098879643e-05,
+      "loss": 1.0326,
+      "step": 8809
+    },
+    {
+      "epoch": 0.3203985889369749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.589365958515321e-05,
+      "loss": 1.0528,
+      "step": 8810
+    },
+    {
+      "epoch": 0.3204349565407135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5892707989660617e-05,
+      "loss": 1.1375,
+      "step": 8811
+    },
+    {
+      "epoch": 0.3204713241444521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5891756312415057e-05,
+      "loss": 0.9977,
+      "step": 8812
+    },
+    {
+      "epoch": 0.3205076917481907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.589080455342974e-05,
+      "loss": 1.0997,
+      "step": 8813
+    },
+    {
+      "epoch": 0.3205440593519293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5889852712717866e-05,
+      "loss": 1.0447,
+      "step": 8814
+    },
+    {
+      "epoch": 0.3205804269556679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5888900790292642e-05,
+      "loss": 1.0654,
+      "step": 8815
+    },
+    {
+      "epoch": 0.3206167945594065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5887948786167276e-05,
+      "loss": 1.0992,
+      "step": 8816
+    },
+    {
+      "epoch": 0.32065316216314504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5886996700354974e-05,
+      "loss": 1.0376,
+      "step": 8817
+    },
+    {
+      "epoch": 0.32068952976688364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5886044532868946e-05,
+      "loss": 1.2112,
+      "step": 8818
+    },
+    {
+      "epoch": 0.32072589737062224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.58850922837224e-05,
+      "loss": 1.1349,
+      "step": 8819
+    },
+    {
+      "epoch": 0.32076226497436083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.588413995292855e-05,
+      "loss": 1.1351,
+      "step": 8820
+    },
+    {
+      "epoch": 0.32079863257809943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5883187540500606e-05,
+      "loss": 0.9264,
+      "step": 8821
+    },
+    {
+      "epoch": 0.320835000181838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5882235046451782e-05,
+      "loss": 1.0404,
+      "step": 8822
+    },
+    {
+      "epoch": 0.3208713677855766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5881282470795292e-05,
+      "loss": 1.1489,
+      "step": 8823
+    },
+    {
+      "epoch": 0.3209077353893152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.588032981354435e-05,
+      "loss": 1.109,
+      "step": 8824
+    },
+    {
+      "epoch": 0.3209441029930538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5879377074712185e-05,
+      "loss": 1.0972,
+      "step": 8825
+    },
+    {
+      "epoch": 0.32098047059679236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5878424254311996e-05,
+      "loss": 1.1182,
+      "step": 8826
+    },
+    {
+      "epoch": 0.32101683820053095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.587747135235701e-05,
+      "loss": 1.1218,
+      "step": 8827
+    },
+    {
+      "epoch": 0.32105320580426955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5876518368860452e-05,
+      "loss": 1.1625,
+      "step": 8828
+    },
+    {
+      "epoch": 0.32108957340800814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5875565303835535e-05,
+      "loss": 1.1859,
+      "step": 8829
+    },
+    {
+      "epoch": 0.32112594101174674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5874612157295487e-05,
+      "loss": 1.0371,
+      "step": 8830
+    },
+    {
+      "epoch": 0.32116230861548534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5873658929253536e-05,
+      "loss": 1.0505,
+      "step": 8831
+    },
+    {
+      "epoch": 0.32119867621922393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5872705619722896e-05,
+      "loss": 0.965,
+      "step": 8832
+    },
+    {
+      "epoch": 0.32123504382296253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5871752228716797e-05,
+      "loss": 1.178,
+      "step": 8833
+    },
+    {
+      "epoch": 0.32127141142670107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.587079875624847e-05,
+      "loss": 1.0214,
+      "step": 8834
+    },
+    {
+      "epoch": 0.32130777903043967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5869845202331138e-05,
+      "loss": 1.0867,
+      "step": 8835
+    },
+    {
+      "epoch": 0.32134414663417826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5868891566978032e-05,
+      "loss": 1.04,
+      "step": 8836
+    },
+    {
+      "epoch": 0.32138051423791686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586793785020238e-05,
+      "loss": 0.9664,
+      "step": 8837
+    },
+    {
+      "epoch": 0.32141688184165546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5866984052017417e-05,
+      "loss": 1.0222,
+      "step": 8838
+    },
+    {
+      "epoch": 0.32145324944539405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5866030172436378e-05,
+      "loss": 1.1702,
+      "step": 8839
+    },
+    {
+      "epoch": 0.32148961704913265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586507621147249e-05,
+      "loss": 1.0687,
+      "step": 8840
+    },
+    {
+      "epoch": 0.32152598465287124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586412216913899e-05,
+      "loss": 0.9918,
+      "step": 8841
+    },
+    {
+      "epoch": 0.3215623522566098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586316804544912e-05,
+      "loss": 0.9872,
+      "step": 8842
+    },
+    {
+      "epoch": 0.3215987198603484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586221384041611e-05,
+      "loss": 1.0681,
+      "step": 8843
+    },
+    {
+      "epoch": 0.321635087464087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.58612595540532e-05,
+      "loss": 1.0428,
+      "step": 8844
+    },
+    {
+      "epoch": 0.3216714550678256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586030518637363e-05,
+      "loss": 1.1978,
+      "step": 8845
+    },
+    {
+      "epoch": 0.32170782267156417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585935073739064e-05,
+      "loss": 1.1091,
+      "step": 8846
+    },
+    {
+      "epoch": 0.32174419027530277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585839620711747e-05,
+      "loss": 1.179,
+      "step": 8847
+    },
+    {
+      "epoch": 0.32178055787904136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5857441595567365e-05,
+      "loss": 1.097,
+      "step": 8848
+    },
+    {
+      "epoch": 0.32181692548277996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585648690275357e-05,
+      "loss": 1.0765,
+      "step": 8849
+    },
+    {
+      "epoch": 0.32185329308651855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5855532128689328e-05,
+      "loss": 1.149,
+      "step": 8850
+    },
+    {
+      "epoch": 0.3218896606902571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5854577273387882e-05,
+      "loss": 1.1655,
+      "step": 8851
+    },
+    {
+      "epoch": 0.3219260282939957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585362233686249e-05,
+      "loss": 0.9555,
+      "step": 8852
+    },
+    {
+      "epoch": 0.3219623958977343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585266731912639e-05,
+      "loss": 1.0685,
+      "step": 8853
+    },
+    {
+      "epoch": 0.3219987635014729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585171222019283e-05,
+      "loss": 1.1116,
+      "step": 8854
+    },
+    {
+      "epoch": 0.3220351311052115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5850757040075073e-05,
+      "loss": 1.0951,
+      "step": 8855
+    },
+    {
+      "epoch": 0.3220714987089501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584980177878636e-05,
+      "loss": 1.0929,
+      "step": 8856
+    },
+    {
+      "epoch": 0.3221078663126887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5848846436339945e-05,
+      "loss": 0.9647,
+      "step": 8857
+    },
+    {
+      "epoch": 0.32214423391642727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584789101274909e-05,
+      "loss": 1.0485,
+      "step": 8858
+    },
+    {
+      "epoch": 0.3221806015201658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5846935508027036e-05,
+      "loss": 1.1697,
+      "step": 8859
+    },
+    {
+      "epoch": 0.3222169691239044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5845979922187056e-05,
+      "loss": 1.0851,
+      "step": 8860
+    },
+    {
+      "epoch": 0.322253336727643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584502425524239e-05,
+      "loss": 1.0578,
+      "step": 8861
+    },
+    {
+      "epoch": 0.3222897043313816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5844068507206313e-05,
+      "loss": 1.0596,
+      "step": 8862
+    },
+    {
+      "epoch": 0.3223260719351202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5843112678092074e-05,
+      "loss": 1.287,
+      "step": 8863
+    },
+    {
+      "epoch": 0.3223624395388588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584215676791294e-05,
+      "loss": 1.054,
+      "step": 8864
+    },
+    {
+      "epoch": 0.3223988071425974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584120077668217e-05,
+      "loss": 1.1152,
+      "step": 8865
+    },
+    {
+      "epoch": 0.322435174746336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.584024470441302e-05,
+      "loss": 1.1337,
+      "step": 8866
+    },
+    {
+      "epoch": 0.3224715423500746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5839288551118768e-05,
+      "loss": 0.9518,
+      "step": 8867
+    },
+    {
+      "epoch": 0.3225079099538131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.583833231681267e-05,
+      "loss": 0.9954,
+      "step": 8868
+    },
+    {
+      "epoch": 0.3225442775575517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5837376001507993e-05,
+      "loss": 1.0618,
+      "step": 8869
+    },
+    {
+      "epoch": 0.3225806451612903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5836419605218006e-05,
+      "loss": 0.9926,
+      "step": 8870
+    },
+    {
+      "epoch": 0.3226170127650289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.583546312795598e-05,
+      "loss": 1.1537,
+      "step": 8871
+    },
+    {
+      "epoch": 0.3226533803687675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.583450656973518e-05,
+      "loss": 0.9487,
+      "step": 8872
+    },
+    {
+      "epoch": 0.3226897479725061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.583354993056888e-05,
+      "loss": 1.1591,
+      "step": 8873
+    },
+    {
+      "epoch": 0.3227261155762447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5832593210470353e-05,
+      "loss": 1.148,
+      "step": 8874
+    },
+    {
+      "epoch": 0.3227624831799833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5831636409452868e-05,
+      "loss": 1.0164,
+      "step": 8875
+    },
+    {
+      "epoch": 0.32279885078372184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5830679527529704e-05,
+      "loss": 1.002,
+      "step": 8876
+    },
+    {
+      "epoch": 0.32283521838746043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5829722564714134e-05,
+      "loss": 1.0726,
+      "step": 8877
+    },
+    {
+      "epoch": 0.32287158599119903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5828765521019437e-05,
+      "loss": 1.0756,
+      "step": 8878
+    },
+    {
+      "epoch": 0.3229079535949376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5827808396458883e-05,
+      "loss": 1.1431,
+      "step": 8879
+    },
+    {
+      "epoch": 0.3229443211986762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.582685119104576e-05,
+      "loss": 0.9978,
+      "step": 8880
+    },
+    {
+      "epoch": 0.3229806888024148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.582589390479334e-05,
+      "loss": 1.0836,
+      "step": 8881
+    },
+    {
+      "epoch": 0.3230170564061534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5824936537714913e-05,
+      "loss": 1.1063,
+      "step": 8882
+    },
+    {
+      "epoch": 0.323053424009892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5823979089823757e-05,
+      "loss": 0.9963,
+      "step": 8883
+    },
+    {
+      "epoch": 0.32308979161363055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5823021561133146e-05,
+      "loss": 0.9882,
+      "step": 8884
+    },
+    {
+      "epoch": 0.32312615921736915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5822063951656378e-05,
+      "loss": 1.0999,
+      "step": 8885
+    },
+    {
+      "epoch": 0.32316252682110774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5821106261406737e-05,
+      "loss": 1.1512,
+      "step": 8886
+    },
+    {
+      "epoch": 0.32319889442484634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5820148490397496e-05,
+      "loss": 1.0237,
+      "step": 8887
+    },
+    {
+      "epoch": 0.32323526202858494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.581919063864196e-05,
+      "loss": 1.0493,
+      "step": 8888
+    },
+    {
+      "epoch": 0.32327162963232353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.581823270615341e-05,
+      "loss": 1.1552,
+      "step": 8889
+    },
+    {
+      "epoch": 0.32330799723606213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5817274692945136e-05,
+      "loss": 1.0711,
+      "step": 8890
+    },
+    {
+      "epoch": 0.3233443648398007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5816316599030425e-05,
+      "loss": 0.9534,
+      "step": 8891
+    },
+    {
+      "epoch": 0.3233807324435393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5815358424422575e-05,
+      "loss": 1.0435,
+      "step": 8892
+    },
+    {
+      "epoch": 0.32341710004727786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.581440016913488e-05,
+      "loss": 0.9962,
+      "step": 8893
+    },
+    {
+      "epoch": 0.32345346765101646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.581344183318063e-05,
+      "loss": 1.176,
+      "step": 8894
+    },
+    {
+      "epoch": 0.32348983525475505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5812483416573123e-05,
+      "loss": 1.1974,
+      "step": 8895
+    },
+    {
+      "epoch": 0.32352620285849365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.581152491932566e-05,
+      "loss": 1.1359,
+      "step": 8896
+    },
+    {
+      "epoch": 0.32356257046223225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.581056634145153e-05,
+      "loss": 1.0846,
+      "step": 8897
+    },
+    {
+      "epoch": 0.32359893806597084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5809607682964035e-05,
+      "loss": 1.027,
+      "step": 8898
+    },
+    {
+      "epoch": 0.32363530566970944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5808648943876478e-05,
+      "loss": 1.1362,
+      "step": 8899
+    },
+    {
+      "epoch": 0.32367167327344804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5807690124202156e-05,
+      "loss": 1.1078,
+      "step": 8900
+    },
+    {
+      "epoch": 0.3237080408771866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.580673122395438e-05,
+      "loss": 0.9629,
+      "step": 8901
+    },
+    {
+      "epoch": 0.32374440848092517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5805772243146442e-05,
+      "loss": 1.0765,
+      "step": 8902
+    },
+    {
+      "epoch": 0.32378077608466377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.580481318179165e-05,
+      "loss": 0.9324,
+      "step": 8903
+    },
+    {
+      "epoch": 0.32381714368840236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5803854039903313e-05,
+      "loss": 1.0597,
+      "step": 8904
+    },
+    {
+      "epoch": 0.32385351129214096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5802894817494738e-05,
+      "loss": 1.0047,
+      "step": 8905
+    },
+    {
+      "epoch": 0.32388987889587956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.580193551457923e-05,
+      "loss": 0.9858,
+      "step": 8906
+    },
+    {
+      "epoch": 0.32392624649961815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5800976131170097e-05,
+      "loss": 1.0273,
+      "step": 8907
+    },
+    {
+      "epoch": 0.32396261410335675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5800016667280648e-05,
+      "loss": 1.0154,
+      "step": 8908
+    },
+    {
+      "epoch": 0.32399898170709535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.57990571229242e-05,
+      "loss": 1.0718,
+      "step": 8909
+    },
+    {
+      "epoch": 0.3240353493108339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.579809749811406e-05,
+      "loss": 1.1147,
+      "step": 8910
+    },
+    {
+      "epoch": 0.3240717169145725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5797137792863546e-05,
+      "loss": 1.0383,
+      "step": 8911
+    },
+    {
+      "epoch": 0.3241080845183111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5796178007185973e-05,
+      "loss": 1.0602,
+      "step": 8912
+    },
+    {
+      "epoch": 0.3241444521220497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.579521814109465e-05,
+      "loss": 1.0502,
+      "step": 8913
+    },
+    {
+      "epoch": 0.32418081972578827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5794258194602896e-05,
+      "loss": 1.0632,
+      "step": 8914
+    },
+    {
+      "epoch": 0.32421718732952687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5793298167724034e-05,
+      "loss": 1.1855,
+      "step": 8915
+    },
+    {
+      "epoch": 0.32425355493326546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5792338060471378e-05,
+      "loss": 1.0402,
+      "step": 8916
+    },
+    {
+      "epoch": 0.32428992253700406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.579137787285825e-05,
+      "loss": 1.0208,
+      "step": 8917
+    },
+    {
+      "epoch": 0.3243262901407426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5790417604897968e-05,
+      "loss": 0.9919,
+      "step": 8918
+    },
+    {
+      "epoch": 0.3243626577444812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5789457256603862e-05,
+      "loss": 0.9173,
+      "step": 8919
+    },
+    {
+      "epoch": 0.3243990253482198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.578849682798925e-05,
+      "loss": 1.1374,
+      "step": 8920
+    },
+    {
+      "epoch": 0.3244353929519584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5787536319067454e-05,
+      "loss": 1.0401,
+      "step": 8921
+    },
+    {
+      "epoch": 0.324471760555697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5786575729851808e-05,
+      "loss": 1.1086,
+      "step": 8922
+    },
+    {
+      "epoch": 0.3245081281594356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.578561506035563e-05,
+      "loss": 1.0154,
+      "step": 8923
+    },
+    {
+      "epoch": 0.3245444957631742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5784654310592253e-05,
+      "loss": 1.1338,
+      "step": 8924
+    },
+    {
+      "epoch": 0.3245808633669128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5783693480575005e-05,
+      "loss": 1.0799,
+      "step": 8925
+    },
+    {
+      "epoch": 0.32461723097065137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5782732570317215e-05,
+      "loss": 1.0409,
+      "step": 8926
+    },
+    {
+      "epoch": 0.3246535985743899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5781771579832216e-05,
+      "loss": 1.0376,
+      "step": 8927
+    },
+    {
+      "epoch": 0.3246899661781285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5780810509133338e-05,
+      "loss": 0.9958,
+      "step": 8928
+    },
+    {
+      "epoch": 0.3247263337818671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5779849358233918e-05,
+      "loss": 0.9936,
+      "step": 8929
+    },
+    {
+      "epoch": 0.3247627013856057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577888812714729e-05,
+      "loss": 1.0932,
+      "step": 8930
+    },
+    {
+      "epoch": 0.3247990689893443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5777926815886785e-05,
+      "loss": 1.1352,
+      "step": 8931
+    },
+    {
+      "epoch": 0.3248354365930829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5776965424465746e-05,
+      "loss": 1.0296,
+      "step": 8932
+    },
+    {
+      "epoch": 0.3248718041968215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5776003952897504e-05,
+      "loss": 1.0378,
+      "step": 8933
+    },
+    {
+      "epoch": 0.3249081718005601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5775042401195406e-05,
+      "loss": 1.0729,
+      "step": 8934
+    },
+    {
+      "epoch": 0.3249445394042986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5774080769372786e-05,
+      "loss": 0.9893,
+      "step": 8935
+    },
+    {
+      "epoch": 0.3249809070080372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577311905744299e-05,
+      "loss": 1.0396,
+      "step": 8936
+    },
+    {
+      "epoch": 0.3250172746117758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5772157265419357e-05,
+      "loss": 1.2107,
+      "step": 8937
+    },
+    {
+      "epoch": 0.3250536422155144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577119539331523e-05,
+      "loss": 1.1076,
+      "step": 8938
+    },
+    {
+      "epoch": 0.325090009819253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5770233441143958e-05,
+      "loss": 1.0114,
+      "step": 8939
+    },
+    {
+      "epoch": 0.3251263774229916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5769271408918886e-05,
+      "loss": 1.0897,
+      "step": 8940
+    },
+    {
+      "epoch": 0.3251627450267302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5768309296653355e-05,
+      "loss": 1.0884,
+      "step": 8941
+    },
+    {
+      "epoch": 0.3251991126304688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5767347104360716e-05,
+      "loss": 1.1375,
+      "step": 8942
+    },
+    {
+      "epoch": 0.32523548023420734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.576638483205432e-05,
+      "loss": 1.0473,
+      "step": 8943
+    },
+    {
+      "epoch": 0.32527184783794594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5765422479747515e-05,
+      "loss": 1.0376,
+      "step": 8944
+    },
+    {
+      "epoch": 0.32530821544168453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5764460047453657e-05,
+      "loss": 1.0047,
+      "step": 8945
+    },
+    {
+      "epoch": 0.32534458304542313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.576349753518609e-05,
+      "loss": 1.1586,
+      "step": 8946
+    },
+    {
+      "epoch": 0.3253809506491617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.576253494295817e-05,
+      "loss": 0.9443,
+      "step": 8947
+    },
+    {
+      "epoch": 0.3254173182529003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5761572270783262e-05,
+      "loss": 1.1027,
+      "step": 8948
+    },
+    {
+      "epoch": 0.3254536858566389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5760609518674707e-05,
+      "loss": 1.0589,
+      "step": 8949
+    },
+    {
+      "epoch": 0.3254900534603775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.575964668664587e-05,
+      "loss": 1.0414,
+      "step": 8950
+    },
+    {
+      "epoch": 0.3255264210641161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.575868377471011e-05,
+      "loss": 0.9766,
+      "step": 8951
+    },
+    {
+      "epoch": 0.32556278866785465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5757720782880777e-05,
+      "loss": 0.9611,
+      "step": 8952
+    },
+    {
+      "epoch": 0.32559915627159325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.575675771117124e-05,
+      "loss": 0.9265,
+      "step": 8953
+    },
+    {
+      "epoch": 0.32563552387533184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5755794559594857e-05,
+      "loss": 1.1114,
+      "step": 8954
+    },
+    {
+      "epoch": 0.32567189147907044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.575483132816499e-05,
+      "loss": 0.9921,
+      "step": 8955
+    },
+    {
+      "epoch": 0.32570825908280904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5753868016895008e-05,
+      "loss": 1.1274,
+      "step": 8956
+    },
+    {
+      "epoch": 0.32574462668654763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5752904625798266e-05,
+      "loss": 1.0688,
+      "step": 8957
+    },
+    {
+      "epoch": 0.32578099429028623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5751941154888136e-05,
+      "loss": 1.07,
+      "step": 8958
+    },
+    {
+      "epoch": 0.3258173618940248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5750977604177985e-05,
+      "loss": 0.9645,
+      "step": 8959
+    },
+    {
+      "epoch": 0.32585372949776337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.575001397368118e-05,
+      "loss": 1.0014,
+      "step": 8960
+    },
+    {
+      "epoch": 0.32589009710150196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5749050263411086e-05,
+      "loss": 1.0485,
+      "step": 8961
+    },
+    {
+      "epoch": 0.32592646470524056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.574808647338108e-05,
+      "loss": 1.0186,
+      "step": 8962
+    },
+    {
+      "epoch": 0.32596283230897916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.574712260360453e-05,
+      "loss": 1.0398,
+      "step": 8963
+    },
+    {
+      "epoch": 0.32599919991271775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5746158654094804e-05,
+      "loss": 1.0076,
+      "step": 8964
+    },
+    {
+      "epoch": 0.32603556751645635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.574519462486528e-05,
+      "loss": 1.02,
+      "step": 8965
+    },
+    {
+      "epoch": 0.32607193512019494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.574423051592933e-05,
+      "loss": 1.0956,
+      "step": 8966
+    },
+    {
+      "epoch": 0.32610830272393354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5743266327300333e-05,
+      "loss": 0.9855,
+      "step": 8967
+    },
+    {
+      "epoch": 0.32614467032767214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.574230205899167e-05,
+      "loss": 1.09,
+      "step": 8968
+    },
+    {
+      "epoch": 0.3261810379314107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5741337711016707e-05,
+      "loss": 1.1146,
+      "step": 8969
+    },
+    {
+      "epoch": 0.3262174055351493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5740373283388834e-05,
+      "loss": 1.1148,
+      "step": 8970
+    },
+    {
+      "epoch": 0.32625377313888787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5739408776121425e-05,
+      "loss": 1.0114,
+      "step": 8971
+    },
+    {
+      "epoch": 0.32629014074262647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5738444189227857e-05,
+      "loss": 0.9973,
+      "step": 8972
+    },
+    {
+      "epoch": 0.32632650834636506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5737479522721524e-05,
+      "loss": 1.1426,
+      "step": 8973
+    },
+    {
+      "epoch": 0.32636287595010366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5736514776615797e-05,
+      "loss": 0.9883,
+      "step": 8974
+    },
+    {
+      "epoch": 0.32639924355384226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5735549950924075e-05,
+      "loss": 1.2016,
+      "step": 8975
+    },
+    {
+      "epoch": 0.32643561115758085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5734585045659727e-05,
+      "loss": 1.0948,
+      "step": 8976
+    },
+    {
+      "epoch": 0.3264719787613194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573362006083615e-05,
+      "loss": 1.0598,
+      "step": 8977
+    },
+    {
+      "epoch": 0.326508346365058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573265499646673e-05,
+      "loss": 1.0779,
+      "step": 8978
+    },
+    {
+      "epoch": 0.3265447139687966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573168985256486e-05,
+      "loss": 1.0793,
+      "step": 8979
+    },
+    {
+      "epoch": 0.3265810815725352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573072462914392e-05,
+      "loss": 1.0985,
+      "step": 8980
+    },
+    {
+      "epoch": 0.3266174491762738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.572975932621731e-05,
+      "loss": 1.0899,
+      "step": 8981
+    },
+    {
+      "epoch": 0.3266538167800124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5728793943798416e-05,
+      "loss": 1.0249,
+      "step": 8982
+    },
+    {
+      "epoch": 0.32669018438375097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5727828481900634e-05,
+      "loss": 1.0964,
+      "step": 8983
+    },
+    {
+      "epoch": 0.32672655198748957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.572686294053736e-05,
+      "loss": 1.0113,
+      "step": 8984
+    },
+    {
+      "epoch": 0.3267629195912281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5725897319721988e-05,
+      "loss": 1.0016,
+      "step": 8985
+    },
+    {
+      "epoch": 0.3267992871949667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5724931619467914e-05,
+      "loss": 1.1158,
+      "step": 8986
+    },
+    {
+      "epoch": 0.3268356547987053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5723965839788537e-05,
+      "loss": 0.9228,
+      "step": 8987
+    },
+    {
+      "epoch": 0.3268720224024439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5722999980697256e-05,
+      "loss": 1.0695,
+      "step": 8988
+    },
+    {
+      "epoch": 0.3269083900061825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5722034042207468e-05,
+      "loss": 1.0131,
+      "step": 8989
+    },
+    {
+      "epoch": 0.3269447576099211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5721068024332577e-05,
+      "loss": 0.9487,
+      "step": 8990
+    },
+    {
+      "epoch": 0.3269811252136597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5720101927085985e-05,
+      "loss": 1.1151,
+      "step": 8991
+    },
+    {
+      "epoch": 0.3270174928173983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5719135750481097e-05,
+      "loss": 1.0179,
+      "step": 8992
+    },
+    {
+      "epoch": 0.3270538604211369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.571816949453131e-05,
+      "loss": 1.1423,
+      "step": 8993
+    },
+    {
+      "epoch": 0.3270902280248754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5717203159250035e-05,
+      "loss": 0.959,
+      "step": 8994
+    },
+    {
+      "epoch": 0.327126595628614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.571623674465068e-05,
+      "loss": 1.0274,
+      "step": 8995
+    },
+    {
+      "epoch": 0.3271629632323526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5715270250746648e-05,
+      "loss": 1.0514,
+      "step": 8996
+    },
+    {
+      "epoch": 0.3271993308360912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5714303677551353e-05,
+      "loss": 1.0518,
+      "step": 8997
+    },
+    {
+      "epoch": 0.3272356984398298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.57133370250782e-05,
+      "loss": 1.0271,
+      "step": 8998
+    },
+    {
+      "epoch": 0.3272720660435684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5712370293340605e-05,
+      "loss": 1.0034,
+      "step": 8999
+    },
+    {
+      "epoch": 0.327308433647307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5711403482351973e-05,
+      "loss": 1.0388,
+      "step": 9000
+    },
+    {
+      "epoch": 0.3273448012510456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5710436592125724e-05,
+      "loss": 1.065,
+      "step": 9001
+    },
+    {
+      "epoch": 0.32738116885478413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.570946962267527e-05,
+      "loss": 0.9994,
+      "step": 9002
+    },
+    {
+      "epoch": 0.32741753645852273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.570850257401402e-05,
+      "loss": 1.0702,
+      "step": 9003
+    },
+    {
+      "epoch": 0.3274539040622613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5707535446155402e-05,
+      "loss": 1.1473,
+      "step": 9004
+    },
+    {
+      "epoch": 0.3274902716659999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5706568239112826e-05,
+      "loss": 1.0895,
+      "step": 9005
+    },
+    {
+      "epoch": 0.3275266392697385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.570560095289971e-05,
+      "loss": 1.0456,
+      "step": 9006
+    },
+    {
+      "epoch": 0.3275630068734771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5704633587529478e-05,
+      "loss": 1.0435,
+      "step": 9007
+    },
+    {
+      "epoch": 0.3275993744772157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.570366614301555e-05,
+      "loss": 1.0255,
+      "step": 9008
+    },
+    {
+      "epoch": 0.3276357420809543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5702698619371344e-05,
+      "loss": 1.0012,
+      "step": 9009
+    },
+    {
+      "epoch": 0.3276721096846929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5701731016610286e-05,
+      "loss": 1.0575,
+      "step": 9010
+    },
+    {
+      "epoch": 0.32770847728843144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5700763334745804e-05,
+      "loss": 1.088,
+      "step": 9011
+    },
+    {
+      "epoch": 0.32774484489217004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5699795573791314e-05,
+      "loss": 1.079,
+      "step": 9012
+    },
+    {
+      "epoch": 0.32778121249590864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5698827733760248e-05,
+      "loss": 1.0791,
+      "step": 9013
+    },
+    {
+      "epoch": 0.32781758009964723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5697859814666035e-05,
+      "loss": 1.016,
+      "step": 9014
+    },
+    {
+      "epoch": 0.32785394770338583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.56968918165221e-05,
+      "loss": 0.9715,
+      "step": 9015
+    },
+    {
+      "epoch": 0.3278903153071244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5695923739341872e-05,
+      "loss": 1.0247,
+      "step": 9016
+    },
+    {
+      "epoch": 0.327926682910863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.569495558313879e-05,
+      "loss": 1.1082,
+      "step": 9017
+    },
+    {
+      "epoch": 0.3279630505146016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5693987347926275e-05,
+      "loss": 1.0387,
+      "step": 9018
+    },
+    {
+      "epoch": 0.32799941811834016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5693019033717765e-05,
+      "loss": 1.1028,
+      "step": 9019
+    },
+    {
+      "epoch": 0.32803578572207875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5692050640526692e-05,
+      "loss": 1.0084,
+      "step": 9020
+    },
+    {
+      "epoch": 0.32807215332581735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.569108216836649e-05,
+      "loss": 1.1326,
+      "step": 9021
+    },
+    {
+      "epoch": 0.32810852092955595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.56901136172506e-05,
+      "loss": 0.9221,
+      "step": 9022
+    },
+    {
+      "epoch": 0.32814488853329454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.568914498719246e-05,
+      "loss": 1.0359,
+      "step": 9023
+    },
+    {
+      "epoch": 0.32818125613703314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.56881762782055e-05,
+      "loss": 0.9991,
+      "step": 9024
+    },
+    {
+      "epoch": 0.32821762374077174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.568720749030317e-05,
+      "loss": 1.1824,
+      "step": 9025
+    },
+    {
+      "epoch": 0.32825399134451033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.56862386234989e-05,
+      "loss": 1.0834,
+      "step": 9026
+    },
+    {
+      "epoch": 0.3282903589482489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5685269677806138e-05,
+      "loss": 1.1388,
+      "step": 9027
+    },
+    {
+      "epoch": 0.32832672655198747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.568430065323833e-05,
+      "loss": 1.0466,
+      "step": 9028
+    },
+    {
+      "epoch": 0.32836309415572607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5683331549808907e-05,
+      "loss": 1.1743,
+      "step": 9029
+    },
+    {
+      "epoch": 0.32839946175946466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.568236236753133e-05,
+      "loss": 1.1661,
+      "step": 9030
+    },
+    {
+      "epoch": 0.32843582936320326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5681393106419034e-05,
+      "loss": 1.0579,
+      "step": 9031
+    },
+    {
+      "epoch": 0.32847219696694185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.568042376648547e-05,
+      "loss": 1.193,
+      "step": 9032
+    },
+    {
+      "epoch": 0.32850856457068045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5679454347744086e-05,
+      "loss": 1.1205,
+      "step": 9033
+    },
+    {
+      "epoch": 0.32854493217441905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.567848485020833e-05,
+      "loss": 1.1154,
+      "step": 9034
+    },
+    {
+      "epoch": 0.32858129977815764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5677515273891656e-05,
+      "loss": 1.1538,
+      "step": 9035
+    },
+    {
+      "epoch": 0.3286176673818962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.567654561880751e-05,
+      "loss": 1.1283,
+      "step": 9036
+    },
+    {
+      "epoch": 0.3286540349856348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.567557588496935e-05,
+      "loss": 1.0958,
+      "step": 9037
+    },
+    {
+      "epoch": 0.3286904025893734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5674606072390624e-05,
+      "loss": 1.1191,
+      "step": 9038
+    },
+    {
+      "epoch": 0.328726770193112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5673636181084792e-05,
+      "loss": 1.189,
+      "step": 9039
+    },
+    {
+      "epoch": 0.32876313779685057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.567266621106531e-05,
+      "loss": 1.024,
+      "step": 9040
+    },
+    {
+      "epoch": 0.32879950540058916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5671696162345628e-05,
+      "loss": 1.0532,
+      "step": 9041
+    },
+    {
+      "epoch": 0.32883587300432776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.567072603493921e-05,
+      "loss": 1.0289,
+      "step": 9042
+    },
+    {
+      "epoch": 0.32887224060806636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5669755828859517e-05,
+      "loss": 1.1163,
+      "step": 9043
+    },
+    {
+      "epoch": 0.3289086082118049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5668785544120003e-05,
+      "loss": 1.1418,
+      "step": 9044
+    },
+    {
+      "epoch": 0.3289449758155435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5667815180734138e-05,
+      "loss": 1.1188,
+      "step": 9045
+    },
+    {
+      "epoch": 0.3289813434192821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.566684473871537e-05,
+      "loss": 1.0336,
+      "step": 9046
+    },
+    {
+      "epoch": 0.3290177110230207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5665874218077178e-05,
+      "loss": 1.1156,
+      "step": 9047
+    },
+    {
+      "epoch": 0.3290540786267593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.566490361883302e-05,
+      "loss": 1.1333,
+      "step": 9048
+    },
+    {
+      "epoch": 0.3290904462304979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.566393294099636e-05,
+      "loss": 1.0945,
+      "step": 9049
+    },
+    {
+      "epoch": 0.3291268138342365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5662962184580665e-05,
+      "loss": 1.1404,
+      "step": 9050
+    },
+    {
+      "epoch": 0.32916318143797507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5661991349599407e-05,
+      "loss": 1.1336,
+      "step": 9051
+    },
+    {
+      "epoch": 0.32919954904171367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.566102043606605e-05,
+      "loss": 1.0568,
+      "step": 9052
+    },
+    {
+      "epoch": 0.3292359166454522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5660049443994068e-05,
+      "loss": 0.9834,
+      "step": 9053
+    },
+    {
+      "epoch": 0.3292722842491908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.565907837339693e-05,
+      "loss": 0.9626,
+      "step": 9054
+    },
+    {
+      "epoch": 0.3293086518529294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.565810722428811e-05,
+      "loss": 1.1808,
+      "step": 9055
+    },
+    {
+      "epoch": 0.329345019456668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5657135996681082e-05,
+      "loss": 1.0647,
+      "step": 9056
+    },
+    {
+      "epoch": 0.3293813870604066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5656164690589312e-05,
+      "loss": 1.0623,
+      "step": 9057
+    },
+    {
+      "epoch": 0.3294177546641452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5655193306026285e-05,
+      "loss": 1.0833,
+      "step": 9058
+    },
+    {
+      "epoch": 0.3294541222678838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5654221843005476e-05,
+      "loss": 1.0742,
+      "step": 9059
+    },
+    {
+      "epoch": 0.3294904898716224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5653250301540363e-05,
+      "loss": 1.1704,
+      "step": 9060
+    },
+    {
+      "epoch": 0.3295268574753609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.565227868164442e-05,
+      "loss": 1.0156,
+      "step": 9061
+    },
+    {
+      "epoch": 0.3295632250790995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5651306983331126e-05,
+      "loss": 1.0388,
+      "step": 9062
+    },
+    {
+      "epoch": 0.3295995926828381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5650335206613973e-05,
+      "loss": 1.0811,
+      "step": 9063
+    },
+    {
+      "epoch": 0.3296359602865767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5649363351506434e-05,
+      "loss": 1.0487,
+      "step": 9064
+    },
+    {
+      "epoch": 0.3296723278903153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.564839141802199e-05,
+      "loss": 1.0793,
+      "step": 9065
+    },
+    {
+      "epoch": 0.3297086954940539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5647419406174132e-05,
+      "loss": 0.9594,
+      "step": 9066
+    },
+    {
+      "epoch": 0.3297450630977925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5646447315976342e-05,
+      "loss": 1.0649,
+      "step": 9067
+    },
+    {
+      "epoch": 0.3297814307015311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5645475147442106e-05,
+      "loss": 1.0579,
+      "step": 9068
+    },
+    {
+      "epoch": 0.32981779830526964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5644502900584916e-05,
+      "loss": 1.0578,
+      "step": 9069
+    },
+    {
+      "epoch": 0.32985416590900823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5643530575418252e-05,
+      "loss": 1.1229,
+      "step": 9070
+    },
+    {
+      "epoch": 0.32989053351274683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.564255817195561e-05,
+      "loss": 1.0015,
+      "step": 9071
+    },
+    {
+      "epoch": 0.3299269011164854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5641585690210477e-05,
+      "loss": 1.1503,
+      "step": 9072
+    },
+    {
+      "epoch": 0.329963268720224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.564061313019635e-05,
+      "loss": 1.12,
+      "step": 9073
+    },
+    {
+      "epoch": 0.3299996363239626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5639640491926718e-05,
+      "loss": 1.0622,
+      "step": 9074
+    },
+    {
+      "epoch": 0.3300360039277012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5638667775415078e-05,
+      "loss": 1.2204,
+      "step": 9075
+    },
+    {
+      "epoch": 0.3300723715314398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.563769498067492e-05,
+      "loss": 1.1447,
+      "step": 9076
+    },
+    {
+      "epoch": 0.3301087391351784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.563672210771974e-05,
+      "loss": 1.1313,
+      "step": 9077
+    },
+    {
+      "epoch": 0.33014510673891695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5635749156563043e-05,
+      "loss": 1.0602,
+      "step": 9078
+    },
+    {
+      "epoch": 0.33018147434265555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5634776127218324e-05,
+      "loss": 1.0448,
+      "step": 9079
+    },
+    {
+      "epoch": 0.33021784194639414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.563380301969908e-05,
+      "loss": 1.0773,
+      "step": 9080
+    },
+    {
+      "epoch": 0.33025420955013274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5632829834018808e-05,
+      "loss": 1.0665,
+      "step": 9081
+    },
+    {
+      "epoch": 0.33029057715387133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5631856570191018e-05,
+      "loss": 1.0886,
+      "step": 9082
+    },
+    {
+      "epoch": 0.33032694475760993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5630883228229206e-05,
+      "loss": 1.2589,
+      "step": 9083
+    },
+    {
+      "epoch": 0.3303633123613485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.562990980814688e-05,
+      "loss": 1.0399,
+      "step": 9084
+    },
+    {
+      "epoch": 0.3303996799650871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5628936309957542e-05,
+      "loss": 1.039,
+      "step": 9085
+    },
+    {
+      "epoch": 0.33043604756882566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.56279627336747e-05,
+      "loss": 1.0505,
+      "step": 9086
+    },
+    {
+      "epoch": 0.33047241517256426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.562698907931186e-05,
+      "loss": 0.9852,
+      "step": 9087
+    },
+    {
+      "epoch": 0.33050878277630286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.562601534688253e-05,
+      "loss": 1.0444,
+      "step": 9088
+    },
+    {
+      "epoch": 0.33054515038004145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5625041536400218e-05,
+      "loss": 0.9734,
+      "step": 9089
+    },
+    {
+      "epoch": 0.33058151798378005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.562406764787844e-05,
+      "loss": 1.1241,
+      "step": 9090
+    },
+    {
+      "epoch": 0.33061788558751865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5623093681330698e-05,
+      "loss": 0.9332,
+      "step": 9091
+    },
+    {
+      "epoch": 0.33065425319125724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5622119636770512e-05,
+      "loss": 1.0,
+      "step": 9092
+    },
+    {
+      "epoch": 0.33069062079499584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5621145514211393e-05,
+      "loss": 1.0007,
+      "step": 9093
+    },
+    {
+      "epoch": 0.33072698839873443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5620171313666855e-05,
+      "loss": 1.1246,
+      "step": 9094
+    },
+    {
+      "epoch": 0.330763356002473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5619197035150417e-05,
+      "loss": 0.9667,
+      "step": 9095
+    },
+    {
+      "epoch": 0.33079972360621157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5618222678675587e-05,
+      "loss": 1.018,
+      "step": 9096
+    },
+    {
+      "epoch": 0.33083609120995017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.561724824425589e-05,
+      "loss": 0.9939,
+      "step": 9097
+    },
+    {
+      "epoch": 0.33087245881368876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5616273731904847e-05,
+      "loss": 1.1024,
+      "step": 9098
+    },
+    {
+      "epoch": 0.33090882641742736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.561529914163597e-05,
+      "loss": 1.03,
+      "step": 9099
+    },
+    {
+      "epoch": 0.33094519402116596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5614324473462784e-05,
+      "loss": 1.1209,
+      "step": 9100
+    },
+    {
+      "epoch": 0.33098156162490455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5613349727398816e-05,
+      "loss": 1.156,
+      "step": 9101
+    },
+    {
+      "epoch": 0.33101792922864315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.561237490345758e-05,
+      "loss": 1.1401,
+      "step": 9102
+    },
+    {
+      "epoch": 0.3310542968323817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5611400001652606e-05,
+      "loss": 0.9911,
+      "step": 9103
+    },
+    {
+      "epoch": 0.3310906644361203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5610425021997422e-05,
+      "loss": 1.1271,
+      "step": 9104
+    },
+    {
+      "epoch": 0.3311270320398589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.560944996450555e-05,
+      "loss": 0.9737,
+      "step": 9105
+    },
+    {
+      "epoch": 0.3311633996435975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5608474829190517e-05,
+      "loss": 0.9285,
+      "step": 9106
+    },
+    {
+      "epoch": 0.3311997672473361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5607499616065856e-05,
+      "loss": 1.1257,
+      "step": 9107
+    },
+    {
+      "epoch": 0.33123613485107467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.560652432514509e-05,
+      "loss": 1.0758,
+      "step": 9108
+    },
+    {
+      "epoch": 0.33127250245481327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5605548956441757e-05,
+      "loss": 1.1182,
+      "step": 9109
+    },
+    {
+      "epoch": 0.33130887005855186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5604573509969382e-05,
+      "loss": 1.0752,
+      "step": 9110
+    },
+    {
+      "epoch": 0.33134523766229046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5603597985741503e-05,
+      "loss": 1.0749,
+      "step": 9111
+    },
+    {
+      "epoch": 0.331381605266029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5602622383771654e-05,
+      "loss": 1.0941,
+      "step": 9112
+    },
+    {
+      "epoch": 0.3314179728697676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5601646704073368e-05,
+      "loss": 0.9771,
+      "step": 9113
+    },
+    {
+      "epoch": 0.3314543404735062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.560067094666018e-05,
+      "loss": 1.0627,
+      "step": 9114
+    },
+    {
+      "epoch": 0.3314907080772448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.559969511154563e-05,
+      "loss": 1.0288,
+      "step": 9115
+    },
+    {
+      "epoch": 0.3315270756809834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5598719198743256e-05,
+      "loss": 1.0946,
+      "step": 9116
+    },
+    {
+      "epoch": 0.331563443284722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5597743208266593e-05,
+      "loss": 1.0508,
+      "step": 9117
+    },
+    {
+      "epoch": 0.3315998108884606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5596767140129186e-05,
+      "loss": 1.0587,
+      "step": 9118
+    },
+    {
+      "epoch": 0.3316361784921992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.559579099434458e-05,
+      "loss": 1.095,
+      "step": 9119
+    },
+    {
+      "epoch": 0.3316725460959377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5594814770926312e-05,
+      "loss": 1.0568,
+      "step": 9120
+    },
+    {
+      "epoch": 0.3317089136996763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5593838469887926e-05,
+      "loss": 1.0326,
+      "step": 9121
+    },
+    {
+      "epoch": 0.3317452813034149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5592862091242966e-05,
+      "loss": 0.963,
+      "step": 9122
+    },
+    {
+      "epoch": 0.3317816489071535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5591885635004983e-05,
+      "loss": 1.0612,
+      "step": 9123
+    },
+    {
+      "epoch": 0.3318180165108921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5590909101187516e-05,
+      "loss": 1.182,
+      "step": 9124
+    },
+    {
+      "epoch": 0.3318543841146307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.558993248980412e-05,
+      "loss": 1.0922,
+      "step": 9125
+    },
+    {
+      "epoch": 0.3318907517183693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5588955800868343e-05,
+      "loss": 1.0392,
+      "step": 9126
+    },
+    {
+      "epoch": 0.3319271193221079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5587979034393735e-05,
+      "loss": 1.1722,
+      "step": 9127
+    },
+    {
+      "epoch": 0.33196348692584643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5587002190393844e-05,
+      "loss": 1.125,
+      "step": 9128
+    },
+    {
+      "epoch": 0.331999854529585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5586025268882225e-05,
+      "loss": 1.0578,
+      "step": 9129
+    },
+    {
+      "epoch": 0.3320362221333236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5585048269872434e-05,
+      "loss": 1.0692,
+      "step": 9130
+    },
+    {
+      "epoch": 0.3320725897370622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5584071193378018e-05,
+      "loss": 1.0507,
+      "step": 9131
+    },
+    {
+      "epoch": 0.3321089573408008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5583094039412538e-05,
+      "loss": 1.0338,
+      "step": 9132
+    },
+    {
+      "epoch": 0.3321453249445394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.558211680798955e-05,
+      "loss": 1.1397,
+      "step": 9133
+    },
+    {
+      "epoch": 0.332181692548278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5581139499122613e-05,
+      "loss": 0.9677,
+      "step": 9134
+    },
+    {
+      "epoch": 0.3322180601520166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5580162112825283e-05,
+      "loss": 1.153,
+      "step": 9135
+    },
+    {
+      "epoch": 0.3322544277557552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.557918464911112e-05,
+      "loss": 1.0883,
+      "step": 9136
+    },
+    {
+      "epoch": 0.33229079535949374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5578207107993684e-05,
+      "loss": 1.1232,
+      "step": 9137
+    },
+    {
+      "epoch": 0.33232716296323234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.557722948948654e-05,
+      "loss": 1.1169,
+      "step": 9138
+    },
+    {
+      "epoch": 0.33236353056697093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5576251793603252e-05,
+      "loss": 1.072,
+      "step": 9139
+    },
+    {
+      "epoch": 0.33239989817070953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5575274020357382e-05,
+      "loss": 0.9448,
+      "step": 9140
+    },
+    {
+      "epoch": 0.3324362657744481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5574296169762492e-05,
+      "loss": 0.9845,
+      "step": 9141
+    },
+    {
+      "epoch": 0.3324726333781867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5573318241832157e-05,
+      "loss": 1.0391,
+      "step": 9142
+    },
+    {
+      "epoch": 0.3325090009819253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5572340236579934e-05,
+      "loss": 1.059,
+      "step": 9143
+    },
+    {
+      "epoch": 0.3325453685856639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.55713621540194e-05,
+      "loss": 1.1096,
+      "step": 9144
+    },
+    {
+      "epoch": 0.33258173618940245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5570383994164115e-05,
+      "loss": 0.9856,
+      "step": 9145
+    },
+    {
+      "epoch": 0.33261810379314105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.556940575702766e-05,
+      "loss": 1.1555,
+      "step": 9146
+    },
+    {
+      "epoch": 0.33265447139687965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.55684274426236e-05,
+      "loss": 1.0828,
+      "step": 9147
+    },
+    {
+      "epoch": 0.33269083900061824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5567449050965512e-05,
+      "loss": 1.0932,
+      "step": 9148
+    },
+    {
+      "epoch": 0.33272720660435684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.556647058206697e-05,
+      "loss": 1.0879,
+      "step": 9149
+    },
+    {
+      "epoch": 0.33276357420809544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.556549203594154e-05,
+      "loss": 1.1223,
+      "step": 9150
+    },
+    {
+      "epoch": 0.33279994181183403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5564513412602812e-05,
+      "loss": 1.0031,
+      "step": 9151
+    },
+    {
+      "epoch": 0.33283630941557263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5563534712064346e-05,
+      "loss": 1.1253,
+      "step": 9152
+    },
+    {
+      "epoch": 0.3328726770193112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5562555934339736e-05,
+      "loss": 1.1104,
+      "step": 9153
+    },
+    {
+      "epoch": 0.33290904462304977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.556157707944255e-05,
+      "loss": 0.9945,
+      "step": 9154
+    },
+    {
+      "epoch": 0.33294541222678836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5560598147386378e-05,
+      "loss": 1.087,
+      "step": 9155
+    },
+    {
+      "epoch": 0.33298177983052696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5559619138184797e-05,
+      "loss": 1.0851,
+      "step": 9156
+    },
+    {
+      "epoch": 0.33301814743426555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5558640051851383e-05,
+      "loss": 1.0316,
+      "step": 9157
+    },
+    {
+      "epoch": 0.33305451503800415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5557660888399725e-05,
+      "loss": 1.0619,
+      "step": 9158
+    },
+    {
+      "epoch": 0.33309088264174275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5556681647843413e-05,
+      "loss": 1.0442,
+      "step": 9159
+    },
+    {
+      "epoch": 0.33312725024548134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 1.019,
+      "step": 9160
+    },
+    {
+      "epoch": 0.33316361784921994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5554722935471148e-05,
+      "loss": 1.1546,
+      "step": 9161
+    },
+    {
+      "epoch": 0.3331999854529585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.555374346368237e-05,
+      "loss": 0.985,
+      "step": 9162
+    },
+    {
+      "epoch": 0.3332363530566971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5552763914843285e-05,
+      "loss": 1.0774,
+      "step": 9163
+    },
+    {
+      "epoch": 0.3332727206604357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5551784288967477e-05,
+      "loss": 1.0097,
+      "step": 9164
+    },
+    {
+      "epoch": 0.33330908826417427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5550804586068538e-05,
+      "loss": 1.056,
+      "step": 9165
+    },
+    {
+      "epoch": 0.33334545586791287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5549824806160064e-05,
+      "loss": 1.0554,
+      "step": 9166
+    },
+    {
+      "epoch": 0.33338182347165146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.554884494925564e-05,
+      "loss": 1.0006,
+      "step": 9167
+    },
+    {
+      "epoch": 0.33341819107539006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.554786501536887e-05,
+      "loss": 1.1277,
+      "step": 9168
+    },
+    {
+      "epoch": 0.33345455867912865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5546885004513345e-05,
+      "loss": 1.1864,
+      "step": 9169
+    },
+    {
+      "epoch": 0.3334909262828672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5545904916702658e-05,
+      "loss": 1.0815,
+      "step": 9170
+    },
+    {
+      "epoch": 0.3335272938866058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.554492475195041e-05,
+      "loss": 1.0986,
+      "step": 9171
+    },
+    {
+      "epoch": 0.3335636614903444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5543944510270197e-05,
+      "loss": 1.0386,
+      "step": 9172
+    },
+    {
+      "epoch": 0.333600029094083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.554296419167562e-05,
+      "loss": 1.142,
+      "step": 9173
+    },
+    {
+      "epoch": 0.3336363966978216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.554198379618028e-05,
+      "loss": 1.1518,
+      "step": 9174
+    },
+    {
+      "epoch": 0.3336727643015602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5541003323797776e-05,
+      "loss": 0.9819,
+      "step": 9175
+    },
+    {
+      "epoch": 0.3337091319052988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5540022774541713e-05,
+      "loss": 1.0797,
+      "step": 9176
+    },
+    {
+      "epoch": 0.33374549950903737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5539042148425695e-05,
+      "loss": 1.0916,
+      "step": 9177
+    },
+    {
+      "epoch": 0.33378186711277597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5538061445463327e-05,
+      "loss": 1.041,
+      "step": 9178
+    },
+    {
+      "epoch": 0.3338182347165145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5537080665668214e-05,
+      "loss": 1.0831,
+      "step": 9179
+    },
+    {
+      "epoch": 0.3338546023202531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.553609980905396e-05,
+      "loss": 1.0645,
+      "step": 9180
+    },
+    {
+      "epoch": 0.3338909699239917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.553511887563418e-05,
+      "loss": 1.2223,
+      "step": 9181
+    },
+    {
+      "epoch": 0.3339273375277303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5534137865422475e-05,
+      "loss": 1.1502,
+      "step": 9182
+    },
+    {
+      "epoch": 0.3339637051314689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5533156778432462e-05,
+      "loss": 1.0683,
+      "step": 9183
+    },
+    {
+      "epoch": 0.3340000727352075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5532175614677747e-05,
+      "loss": 1.0831,
+      "step": 9184
+    },
+    {
+      "epoch": 0.3340364403389461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5531194374171944e-05,
+      "loss": 1.1964,
+      "step": 9185
+    },
+    {
+      "epoch": 0.3340728079426847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.553021305692867e-05,
+      "loss": 1.1124,
+      "step": 9186
+    },
+    {
+      "epoch": 0.3341091755464232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5529231662961533e-05,
+      "loss": 1.0271,
+      "step": 9187
+    },
+    {
+      "epoch": 0.3341455431501618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5528250192284155e-05,
+      "loss": 1.0422,
+      "step": 9188
+    },
+    {
+      "epoch": 0.3341819107539004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5527268644910146e-05,
+      "loss": 0.9858,
+      "step": 9189
+    },
+    {
+      "epoch": 0.334218278357639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5526287020853126e-05,
+      "loss": 0.9868,
+      "step": 9190
+    },
+    {
+      "epoch": 0.3342546459613776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5525305320126716e-05,
+      "loss": 1.1893,
+      "step": 9191
+    },
+    {
+      "epoch": 0.3342910135651162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5524323542744536e-05,
+      "loss": 1.0159,
+      "step": 9192
+    },
+    {
+      "epoch": 0.3343273811688548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5523341688720202e-05,
+      "loss": 0.9948,
+      "step": 9193
+    },
+    {
+      "epoch": 0.3343637487725934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.552235975806734e-05,
+      "loss": 1.168,
+      "step": 9194
+    },
+    {
+      "epoch": 0.334400116376332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5521377750799573e-05,
+      "loss": 0.9731,
+      "step": 9195
+    },
+    {
+      "epoch": 0.33443648398007053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5520395666930518e-05,
+      "loss": 1.1658,
+      "step": 9196
+    },
+    {
+      "epoch": 0.3344728515838091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.551941350647381e-05,
+      "loss": 1.0189,
+      "step": 9197
+    },
+    {
+      "epoch": 0.3345092191875477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.551843126944307e-05,
+      "loss": 1.0335,
+      "step": 9198
+    },
+    {
+      "epoch": 0.3345455867912863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5517448955851927e-05,
+      "loss": 1.1537,
+      "step": 9199
+    },
+    {
+      "epoch": 0.3345819543950249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5516466565714007e-05,
+      "loss": 0.9414,
+      "step": 9200
+    },
+    {
+      "epoch": 0.3346183219987635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5515484099042943e-05,
+      "loss": 0.9946,
+      "step": 9201
+    },
+    {
+      "epoch": 0.3346546896025021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5514501555852356e-05,
+      "loss": 0.947,
+      "step": 9202
+    },
+    {
+      "epoch": 0.3346910572062407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5513518936155886e-05,
+      "loss": 1.1439,
+      "step": 9203
+    },
+    {
+      "epoch": 0.33472742480997925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5512536239967168e-05,
+      "loss": 0.9264,
+      "step": 9204
+    },
+    {
+      "epoch": 0.33476379241371784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5511553467299827e-05,
+      "loss": 0.9668,
+      "step": 9205
+    },
+    {
+      "epoch": 0.33480016001745644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5510570618167503e-05,
+      "loss": 1.1318,
+      "step": 9206
+    },
+    {
+      "epoch": 0.33483652762119503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550958769258383e-05,
+      "loss": 1.0206,
+      "step": 9207
+    },
+    {
+      "epoch": 0.33487289522493363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550860469056244e-05,
+      "loss": 1.0562,
+      "step": 9208
+    },
+    {
+      "epoch": 0.3349092628286722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5507621612116987e-05,
+      "loss": 1.094,
+      "step": 9209
+    },
+    {
+      "epoch": 0.3349456304324108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550663845726109e-05,
+      "loss": 1.0838,
+      "step": 9210
+    },
+    {
+      "epoch": 0.3349819980361494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5505655226008397e-05,
+      "loss": 1.0253,
+      "step": 9211
+    },
+    {
+      "epoch": 0.33501836563988796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550467191837255e-05,
+      "loss": 1.0672,
+      "step": 9212
+    },
+    {
+      "epoch": 0.33505473324362656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550368853436719e-05,
+      "loss": 1.0172,
+      "step": 9213
+    },
+    {
+      "epoch": 0.33509110084736515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5502705074005963e-05,
+      "loss": 1.1018,
+      "step": 9214
+    },
+    {
+      "epoch": 0.33512746845110375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550172153730251e-05,
+      "loss": 1.0396,
+      "step": 9215
+    },
+    {
+      "epoch": 0.33516383605484235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5500737924270473e-05,
+      "loss": 0.9601,
+      "step": 9216
+    },
+    {
+      "epoch": 0.33520020365858094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5499754234923505e-05,
+      "loss": 0.994,
+      "step": 9217
+    },
+    {
+      "epoch": 0.33523657126231954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.549877046927525e-05,
+      "loss": 1.0359,
+      "step": 9218
+    },
+    {
+      "epoch": 0.33527293886605813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5497786627339358e-05,
+      "loss": 0.9924,
+      "step": 9219
+    },
+    {
+      "epoch": 0.33530930646979673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5496802709129474e-05,
+      "loss": 1.058,
+      "step": 9220
+    },
+    {
+      "epoch": 0.33534567407353527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5495818714659252e-05,
+      "loss": 1.1091,
+      "step": 9221
+    },
+    {
+      "epoch": 0.33538204167727387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5494834643942343e-05,
+      "loss": 0.9606,
+      "step": 9222
+    },
+    {
+      "epoch": 0.33541840928101246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5493850496992398e-05,
+      "loss": 1.1481,
+      "step": 9223
+    },
+    {
+      "epoch": 0.33545477688475106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5492866273823074e-05,
+      "loss": 1.0936,
+      "step": 9224
+    },
+    {
+      "epoch": 0.33549114448848966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5491881974448025e-05,
+      "loss": 1.0537,
+      "step": 9225
+    },
+    {
+      "epoch": 0.33552751209222825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5490897598880903e-05,
+      "loss": 1.09,
+      "step": 9226
+    },
+    {
+      "epoch": 0.33556387969596685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5489913147135367e-05,
+      "loss": 1.0708,
+      "step": 9227
+    },
+    {
+      "epoch": 0.33560024729970545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.548892861922508e-05,
+      "loss": 1.1391,
+      "step": 9228
+    },
+    {
+      "epoch": 0.335636614903444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.548794401516369e-05,
+      "loss": 1.1165,
+      "step": 9229
+    },
+    {
+      "epoch": 0.3356729825071826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5486959334964868e-05,
+      "loss": 0.991,
+      "step": 9230
+    },
+    {
+      "epoch": 0.3357093501109212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5485974578642266e-05,
+      "loss": 1.0611,
+      "step": 9231
+    },
+    {
+      "epoch": 0.3357457177146598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5484989746209557e-05,
+      "loss": 0.9924,
+      "step": 9232
+    },
+    {
+      "epoch": 0.33578208531839837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.548400483768039e-05,
+      "loss": 0.9953,
+      "step": 9233
+    },
+    {
+      "epoch": 0.33581845292213697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.548301985306844e-05,
+      "loss": 1.1774,
+      "step": 9234
+    },
+    {
+      "epoch": 0.33585482052587556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5482034792387367e-05,
+      "loss": 1.0755,
+      "step": 9235
+    },
+    {
+      "epoch": 0.33589118812961416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.548104965565084e-05,
+      "loss": 1.1417,
+      "step": 9236
+    },
+    {
+      "epoch": 0.33592755573335276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5480064442872528e-05,
+      "loss": 1.0764,
+      "step": 9237
+    },
+    {
+      "epoch": 0.3359639233370913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5479079154066096e-05,
+      "loss": 1.062,
+      "step": 9238
+    },
+    {
+      "epoch": 0.3360002909408299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5478093789245213e-05,
+      "loss": 1.1542,
+      "step": 9239
+    },
+    {
+      "epoch": 0.3360366585445685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5477108348423554e-05,
+      "loss": 1.0367,
+      "step": 9240
+    },
+    {
+      "epoch": 0.3360730261483071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5476122831614785e-05,
+      "loss": 1.1703,
+      "step": 9241
+    },
+    {
+      "epoch": 0.3361093937520457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5475137238832583e-05,
+      "loss": 1.0092,
+      "step": 9242
+    },
+    {
+      "epoch": 0.3361457613557843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5474151570090617e-05,
+      "loss": 1.0203,
+      "step": 9243
+    },
+    {
+      "epoch": 0.3361821289595229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5473165825402565e-05,
+      "loss": 1.1111,
+      "step": 9244
+    },
+    {
+      "epoch": 0.33621849656326147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5472180004782107e-05,
+      "loss": 1.0312,
+      "step": 9245
+    },
+    {
+      "epoch": 0.336254864167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.547119410824291e-05,
+      "loss": 0.9668,
+      "step": 9246
+    },
+    {
+      "epoch": 0.3362912317707386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5470208135798665e-05,
+      "loss": 1.1461,
+      "step": 9247
+    },
+    {
+      "epoch": 0.3363275993744772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5469222087463038e-05,
+      "loss": 1.141,
+      "step": 9248
+    },
+    {
+      "epoch": 0.3363639669782158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5468235963249714e-05,
+      "loss": 1.0739,
+      "step": 9249
+    },
+    {
+      "epoch": 0.3364003345819544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5467249763172373e-05,
+      "loss": 1.0574,
+      "step": 9250
+    },
+    {
+      "epoch": 0.336436702185693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5466263487244702e-05,
+      "loss": 1.1106,
+      "step": 9251
+    },
+    {
+      "epoch": 0.3364730697894316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.546527713548038e-05,
+      "loss": 1.1453,
+      "step": 9252
+    },
+    {
+      "epoch": 0.3365094373931702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.546429070789309e-05,
+      "loss": 1.0645,
+      "step": 9253
+    },
+    {
+      "epoch": 0.3365458049969087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5463304204496524e-05,
+      "loss": 0.9107,
+      "step": 9254
+    },
+    {
+      "epoch": 0.3365821726006473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.546231762530436e-05,
+      "loss": 1.0966,
+      "step": 9255
+    },
+    {
+      "epoch": 0.3366185402043859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5461330970330284e-05,
+      "loss": 1.0941,
+      "step": 9256
+    },
+    {
+      "epoch": 0.3366549078081245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5460344239587998e-05,
+      "loss": 1.0952,
+      "step": 9257
+    },
+    {
+      "epoch": 0.3366912754118631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.545935743309118e-05,
+      "loss": 1.041,
+      "step": 9258
+    },
+    {
+      "epoch": 0.3367276430156017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5458370550853524e-05,
+      "loss": 0.9762,
+      "step": 9259
+    },
+    {
+      "epoch": 0.3367640106193403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5457383592888716e-05,
+      "loss": 1.0059,
+      "step": 9260
+    },
+    {
+      "epoch": 0.3368003782230789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.545639655921046e-05,
+      "loss": 0.9965,
+      "step": 9261
+    },
+    {
+      "epoch": 0.3368367458268175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.545540944983244e-05,
+      "loss": 1.18,
+      "step": 9262
+    },
+    {
+      "epoch": 0.33687311343055604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5454422264768356e-05,
+      "loss": 0.9929,
+      "step": 9263
+    },
+    {
+      "epoch": 0.33690948103429463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54534350040319e-05,
+      "loss": 1.0174,
+      "step": 9264
+    },
+    {
+      "epoch": 0.33694584863803323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5452447667636773e-05,
+      "loss": 1.1175,
+      "step": 9265
+    },
+    {
+      "epoch": 0.3369822162417718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.545146025559667e-05,
+      "loss": 1.0441,
+      "step": 9266
+    },
+    {
+      "epoch": 0.3370185838455104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5450472767925292e-05,
+      "loss": 1.1368,
+      "step": 9267
+    },
+    {
+      "epoch": 0.337054951449249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5449485204636334e-05,
+      "loss": 1.1078,
+      "step": 9268
+    },
+    {
+      "epoch": 0.3370913190529876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5448497565743502e-05,
+      "loss": 0.972,
+      "step": 9269
+    },
+    {
+      "epoch": 0.3371276866567262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5447509851260498e-05,
+      "loss": 1.1149,
+      "step": 9270
+    },
+    {
+      "epoch": 0.33716405426046475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5446522061201023e-05,
+      "loss": 1.1038,
+      "step": 9271
+    },
+    {
+      "epoch": 0.33720042186420335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.544553419557878e-05,
+      "loss": 1.0617,
+      "step": 9272
+    },
+    {
+      "epoch": 0.33723678946794194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5444546254407477e-05,
+      "loss": 0.9597,
+      "step": 9273
+    },
+    {
+      "epoch": 0.33727315707168054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.544355823770082e-05,
+      "loss": 1.0101,
+      "step": 9274
+    },
+    {
+      "epoch": 0.33730952467541914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5442570145472516e-05,
+      "loss": 1.1113,
+      "step": 9275
+    },
+    {
+      "epoch": 0.33734589227915773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5441581977736276e-05,
+      "loss": 0.9981,
+      "step": 9276
+    },
+    {
+      "epoch": 0.33738225988289633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5440593734505803e-05,
+      "loss": 1.1478,
+      "step": 9277
+    },
+    {
+      "epoch": 0.3374186274866349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.543960541579481e-05,
+      "loss": 1.0188,
+      "step": 9278
+    },
+    {
+      "epoch": 0.3374549950903735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5438617021617014e-05,
+      "loss": 1.0704,
+      "step": 9279
+    },
+    {
+      "epoch": 0.33749136269411206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5437628551986123e-05,
+      "loss": 0.8892,
+      "step": 9280
+    },
+    {
+      "epoch": 0.33752773029785066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5436640006915848e-05,
+      "loss": 1.0412,
+      "step": 9281
+    },
+    {
+      "epoch": 0.33756409790158926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5435651386419905e-05,
+      "loss": 1.0453,
+      "step": 9282
+    },
+    {
+      "epoch": 0.33760046550532785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5434662690512015e-05,
+      "loss": 1.171,
+      "step": 9283
+    },
+    {
+      "epoch": 0.33763683310906645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.543367391920589e-05,
+      "loss": 1.0518,
+      "step": 9284
+    },
+    {
+      "epoch": 0.33767320071280504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5432685072515247e-05,
+      "loss": 1.037,
+      "step": 9285
+    },
+    {
+      "epoch": 0.33770956831654364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.543169615045381e-05,
+      "loss": 0.8874,
+      "step": 9286
+    },
+    {
+      "epoch": 0.33774593592028224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5430707153035293e-05,
+      "loss": 1.1092,
+      "step": 9287
+    },
+    {
+      "epoch": 0.3377823035240208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5429718080273422e-05,
+      "loss": 1.0047,
+      "step": 9288
+    },
+    {
+      "epoch": 0.3378186711277594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.542872893218191e-05,
+      "loss": 1.074,
+      "step": 9289
+    },
+    {
+      "epoch": 0.33785503873149797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.542773970877449e-05,
+      "loss": 1.1149,
+      "step": 9290
+    },
+    {
+      "epoch": 0.33789140633523657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5426750410064885e-05,
+      "loss": 1.0161,
+      "step": 9291
+    },
+    {
+      "epoch": 0.33792777393897516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5425761036066816e-05,
+      "loss": 1.0793,
+      "step": 9292
+    },
+    {
+      "epoch": 0.33796414154271376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5424771586794013e-05,
+      "loss": 0.9925,
+      "step": 9293
+    },
+    {
+      "epoch": 0.33800050914645235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54237820622602e-05,
+      "loss": 1.0322,
+      "step": 9294
+    },
+    {
+      "epoch": 0.33803687675019095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.54227924624791e-05,
+      "loss": 1.0233,
+      "step": 9295
+    },
+    {
+      "epoch": 0.33807324435392955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.542180278746445e-05,
+      "loss": 1.0104,
+      "step": 9296
+    },
+    {
+      "epoch": 0.3381096119576681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5420813037229983e-05,
+      "loss": 1.0711,
+      "step": 9297
+    },
+    {
+      "epoch": 0.3381459795614067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.541982321178943e-05,
+      "loss": 1.1351,
+      "step": 9298
+    },
+    {
+      "epoch": 0.3381823471651453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5418833311156512e-05,
+      "loss": 1.0012,
+      "step": 9299
+    },
+    {
+      "epoch": 0.3382187147688839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5417843335344973e-05,
+      "loss": 1.0836,
+      "step": 9300
+    },
+    {
+      "epoch": 0.3382550823726225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5416853284368547e-05,
+      "loss": 1.0762,
+      "step": 9301
+    },
+    {
+      "epoch": 0.33829144997636107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5415863158240965e-05,
+      "loss": 1.0989,
+      "step": 9302
+    },
+    {
+      "epoch": 0.33832781758009967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5414872956975964e-05,
+      "loss": 1.0634,
+      "step": 9303
+    },
+    {
+      "epoch": 0.33836418518383826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5413882680587286e-05,
+      "loss": 1.0126,
+      "step": 9304
+    },
+    {
+      "epoch": 0.3384005527875768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.541289232908867e-05,
+      "loss": 1.0484,
+      "step": 9305
+    },
+    {
+      "epoch": 0.3384369203913154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5411901902493846e-05,
+      "loss": 1.023,
+      "step": 9306
+    },
+    {
+      "epoch": 0.338473287995054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5410911400816565e-05,
+      "loss": 1.0142,
+      "step": 9307
+    },
+    {
+      "epoch": 0.3385096555987926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5409920824070566e-05,
+      "loss": 1.0592,
+      "step": 9308
+    },
+    {
+      "epoch": 0.3385460232025312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.540893017226959e-05,
+      "loss": 0.9838,
+      "step": 9309
+    },
+    {
+      "epoch": 0.3385823908062698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.540793944542738e-05,
+      "loss": 1.0954,
+      "step": 9310
+    },
+    {
+      "epoch": 0.3386187584100084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5406948643557688e-05,
+      "loss": 1.0627,
+      "step": 9311
+    },
+    {
+      "epoch": 0.338655126013747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.540595776667425e-05,
+      "loss": 0.9323,
+      "step": 9312
+    },
+    {
+      "epoch": 0.3386914936174855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5404966814790824e-05,
+      "loss": 0.9886,
+      "step": 9313
+    },
+    {
+      "epoch": 0.3387278612212241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.540397578792115e-05,
+      "loss": 1.1546,
+      "step": 9314
+    },
+    {
+      "epoch": 0.3387642288249627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5402984686078975e-05,
+      "loss": 0.9963,
+      "step": 9315
+    },
+    {
+      "epoch": 0.3388005964287013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5401993509278055e-05,
+      "loss": 1.0559,
+      "step": 9316
+    },
+    {
+      "epoch": 0.3388369640324399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.540100225753214e-05,
+      "loss": 0.9366,
+      "step": 9317
+    },
+    {
+      "epoch": 0.3388733316361785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5400010930854984e-05,
+      "loss": 1.005,
+      "step": 9318
+    },
+    {
+      "epoch": 0.3389096992399171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5399019529260336e-05,
+      "loss": 1.1805,
+      "step": 9319
+    },
+    {
+      "epoch": 0.3389460668436557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.539802805276195e-05,
+      "loss": 1.1617,
+      "step": 9320
+    },
+    {
+      "epoch": 0.3389824344473943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5397036501373588e-05,
+      "loss": 1.1183,
+      "step": 9321
+    },
+    {
+      "epoch": 0.33901880205113283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5396044875109e-05,
+      "loss": 0.8738,
+      "step": 9322
+    },
+    {
+      "epoch": 0.3390551696548714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5395053173981945e-05,
+      "loss": 1.1951,
+      "step": 9323
+    },
+    {
+      "epoch": 0.33909153725861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5394061398006185e-05,
+      "loss": 0.9898,
+      "step": 9324
+    },
+    {
+      "epoch": 0.3391279048623486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5393069547195473e-05,
+      "loss": 0.9116,
+      "step": 9325
+    },
+    {
+      "epoch": 0.3391642724660872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5392077621563573e-05,
+      "loss": 1.1655,
+      "step": 9326
+    },
+    {
+      "epoch": 0.3392006400698258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5391085621124244e-05,
+      "loss": 1.1472,
+      "step": 9327
+    },
+    {
+      "epoch": 0.3392370076735644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5390093545891255e-05,
+      "loss": 0.926,
+      "step": 9328
+    },
+    {
+      "epoch": 0.339273375277303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5389101395878365e-05,
+      "loss": 1.1655,
+      "step": 9329
+    },
+    {
+      "epoch": 0.33930974288104154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5388109171099338e-05,
+      "loss": 1.1009,
+      "step": 9330
+    },
+    {
+      "epoch": 0.33934611048478014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.538711687156794e-05,
+      "loss": 1.0528,
+      "step": 9331
+    },
+    {
+      "epoch": 0.33938247808851874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.538612449729794e-05,
+      "loss": 1.0564,
+      "step": 9332
+    },
+    {
+      "epoch": 0.33941884569225733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5385132048303104e-05,
+      "loss": 1.0475,
+      "step": 9333
+    },
+    {
+      "epoch": 0.33945521329599593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5384139524597205e-05,
+      "loss": 0.965,
+      "step": 9334
+    },
+    {
+      "epoch": 0.3394915808997345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5383146926194005e-05,
+      "loss": 0.9939,
+      "step": 9335
+    },
+    {
+      "epoch": 0.3395279485034731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.538215425310728e-05,
+      "loss": 1.1208,
+      "step": 9336
+    },
+    {
+      "epoch": 0.3395643161072117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.53811615053508e-05,
+      "loss": 1.0473,
+      "step": 9337
+    },
+    {
+      "epoch": 0.3396006837109503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.538016868293834e-05,
+      "loss": 1.087,
+      "step": 9338
+    },
+    {
+      "epoch": 0.33963705131468885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5379175785883672e-05,
+      "loss": 1.0025,
+      "step": 9339
+    },
+    {
+      "epoch": 0.33967341891842745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5378182814200577e-05,
+      "loss": 1.0088,
+      "step": 9340
+    },
+    {
+      "epoch": 0.33970978652216605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.537718976790282e-05,
+      "loss": 1.1149,
+      "step": 9341
+    },
+    {
+      "epoch": 0.33974615412590464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5376196647004188e-05,
+      "loss": 1.1606,
+      "step": 9342
+    },
+    {
+      "epoch": 0.33978252172964324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.537520345151845e-05,
+      "loss": 1.2438,
+      "step": 9343
+    },
+    {
+      "epoch": 0.33981888933338184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5374210181459397e-05,
+      "loss": 1.1209,
+      "step": 9344
+    },
+    {
+      "epoch": 0.33985525693712043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.53732168368408e-05,
+      "loss": 1.0472,
+      "step": 9345
+    },
+    {
+      "epoch": 0.339891624540859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5372223417676444e-05,
+      "loss": 1.05,
+      "step": 9346
+    },
+    {
+      "epoch": 0.33992799214459757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5371229923980113e-05,
+      "loss": 1.074,
+      "step": 9347
+    },
+    {
+      "epoch": 0.33996435974833616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5370236355765584e-05,
+      "loss": 1.0079,
+      "step": 9348
+    },
+    {
+      "epoch": 0.34000072735207476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5369242713046643e-05,
+      "loss": 1.0077,
+      "step": 9349
+    },
+    {
+      "epoch": 0.34003709495581336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5368248995837077e-05,
+      "loss": 1.0743,
+      "step": 9350
+    },
+    {
+      "epoch": 0.34007346255955195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5367255204150673e-05,
+      "loss": 0.9327,
+      "step": 9351
+    },
+    {
+      "epoch": 0.34010983016329055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5366261338001222e-05,
+      "loss": 1.0957,
+      "step": 9352
+    },
+    {
+      "epoch": 0.34014619776702915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.536526739740251e-05,
+      "loss": 1.0546,
+      "step": 9353
+    },
+    {
+      "epoch": 0.34018256537076774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.536427338236832e-05,
+      "loss": 1.0035,
+      "step": 9354
+    },
+    {
+      "epoch": 0.3402189329745063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5363279292912448e-05,
+      "loss": 1.1153,
+      "step": 9355
+    },
+    {
+      "epoch": 0.3402553005782449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5362285129048687e-05,
+      "loss": 1.0808,
+      "step": 9356
+    },
+    {
+      "epoch": 0.3402916681819835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5361290890790826e-05,
+      "loss": 1.1059,
+      "step": 9357
+    },
+    {
+      "epoch": 0.34032803578572207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5360296578152664e-05,
+      "loss": 0.9971,
+      "step": 9358
+    },
+    {
+      "epoch": 0.34036440338946067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5359302191147993e-05,
+      "loss": 1.1583,
+      "step": 9359
+    },
+    {
+      "epoch": 0.34040077099319926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.53583077297906e-05,
+      "loss": 1.2573,
+      "step": 9360
+    },
+    {
+      "epoch": 0.34043713859693786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5357313194094298e-05,
+      "loss": 1.1603,
+      "step": 9361
+    },
+    {
+      "epoch": 0.34047350620067646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5356318584072877e-05,
+      "loss": 0.9869,
+      "step": 9362
+    },
+    {
+      "epoch": 0.34050987380441505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.535532389974013e-05,
+      "loss": 1.1085,
+      "step": 9363
+    },
+    {
+      "epoch": 0.3405462414081536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5354329141109867e-05,
+      "loss": 1.2203,
+      "step": 9364
+    },
+    {
+      "epoch": 0.3405826090118922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.535333430819588e-05,
+      "loss": 0.992,
+      "step": 9365
+    },
+    {
+      "epoch": 0.3406189766156308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5352339401011976e-05,
+      "loss": 1.0603,
+      "step": 9366
+    },
+    {
+      "epoch": 0.3406553442193694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5351344419571957e-05,
+      "loss": 0.9855,
+      "step": 9367
+    },
+    {
+      "epoch": 0.340691711823108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5350349363889625e-05,
+      "loss": 0.893,
+      "step": 9368
+    },
+    {
+      "epoch": 0.3407280794268466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.534935423397879e-05,
+      "loss": 0.9011,
+      "step": 9369
+    },
+    {
+      "epoch": 0.34076444703058517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.534835902985325e-05,
+      "loss": 1.0626,
+      "step": 9370
+    },
+    {
+      "epoch": 0.34080081463432377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.534736375152682e-05,
+      "loss": 0.9697,
+      "step": 9371
+    },
+    {
+      "epoch": 0.3408371822380623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5346368399013306e-05,
+      "loss": 0.9832,
+      "step": 9372
+    },
+    {
+      "epoch": 0.3408735498418009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5345372972326512e-05,
+      "loss": 1.1253,
+      "step": 9373
+    },
+    {
+      "epoch": 0.3409099174455395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.534437747148026e-05,
+      "loss": 0.998,
+      "step": 9374
+    },
+    {
+      "epoch": 0.3409462850492781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5343381896488346e-05,
+      "loss": 1.0346,
+      "step": 9375
+    },
+    {
+      "epoch": 0.3409826526530167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.534238624736459e-05,
+      "loss": 0.9935,
+      "step": 9376
+    },
+    {
+      "epoch": 0.3410190202567553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5341390524122806e-05,
+      "loss": 1.0449,
+      "step": 9377
+    },
+    {
+      "epoch": 0.3410553878604939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5340394726776807e-05,
+      "loss": 1.1537,
+      "step": 9378
+    },
+    {
+      "epoch": 0.3410917554642325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.533939885534041e-05,
+      "loss": 1.0152,
+      "step": 9379
+    },
+    {
+      "epoch": 0.3411281230679711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5338402909827427e-05,
+      "loss": 1.0551,
+      "step": 9380
+    },
+    {
+      "epoch": 0.3411644906717096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.533740689025168e-05,
+      "loss": 1.0501,
+      "step": 9381
+    },
+    {
+      "epoch": 0.3412008582754482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5336410796626987e-05,
+      "loss": 1.0841,
+      "step": 9382
+    },
+    {
+      "epoch": 0.3412372258791868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5335414628967158e-05,
+      "loss": 1.0465,
+      "step": 9383
+    },
+    {
+      "epoch": 0.3412735934829254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5334418387286024e-05,
+      "loss": 0.9616,
+      "step": 9384
+    },
+    {
+      "epoch": 0.341309961086664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.533342207159741e-05,
+      "loss": 0.9646,
+      "step": 9385
+    },
+    {
+      "epoch": 0.3413463286904026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.533242568191512e-05,
+      "loss": 1.048,
+      "step": 9386
+    },
+    {
+      "epoch": 0.3413826962941412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5331429218253e-05,
+      "loss": 1.0064,
+      "step": 9387
+    },
+    {
+      "epoch": 0.3414190638978798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.533043268062486e-05,
+      "loss": 1.11,
+      "step": 9388
+    },
+    {
+      "epoch": 0.34145543150161833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.532943606904453e-05,
+      "loss": 1.1182,
+      "step": 9389
+    },
+    {
+      "epoch": 0.34149179910535693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5328439383525836e-05,
+      "loss": 1.0249,
+      "step": 9390
+    },
+    {
+      "epoch": 0.3415281667090955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5327442624082603e-05,
+      "loss": 1.034,
+      "step": 9391
+    },
+    {
+      "epoch": 0.3415645343128341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5326445790728664e-05,
+      "loss": 1.0359,
+      "step": 9392
+    },
+    {
+      "epoch": 0.3416009019165727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.532544888347785e-05,
+      "loss": 1.1472,
+      "step": 9393
+    },
+    {
+      "epoch": 0.3416372695203113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5324451902343983e-05,
+      "loss": 1.1517,
+      "step": 9394
+    },
+    {
+      "epoch": 0.3416736371240499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5323454847340905e-05,
+      "loss": 0.9885,
+      "step": 9395
+    },
+    {
+      "epoch": 0.3417100047277885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5322457718482443e-05,
+      "loss": 1.1622,
+      "step": 9396
+    },
+    {
+      "epoch": 0.34174637233152705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5321460515782427e-05,
+      "loss": 1.0881,
+      "step": 9397
+    },
+    {
+      "epoch": 0.34178273993526564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.53204632392547e-05,
+      "loss": 1.0429,
+      "step": 9398
+    },
+    {
+      "epoch": 0.34181910753900424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5319465888913098e-05,
+      "loss": 1.182,
+      "step": 9399
+    },
+    {
+      "epoch": 0.34185547514274284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.531846846477145e-05,
+      "loss": 1.0745,
+      "step": 9400
+    },
+    {
+      "epoch": 0.34189184274648143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.53174709668436e-05,
+      "loss": 1.1296,
+      "step": 9401
+    },
+    {
+      "epoch": 0.34192821035022003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.531647339514338e-05,
+      "loss": 1.1287,
+      "step": 9402
+    },
+    {
+      "epoch": 0.3419645779539586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.531547574968464e-05,
+      "loss": 1.0178,
+      "step": 9403
+    },
+    {
+      "epoch": 0.3420009455576972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5314478030481216e-05,
+      "loss": 1.115,
+      "step": 9404
+    },
+    {
+      "epoch": 0.3420373131614358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5313480237546945e-05,
+      "loss": 1.0509,
+      "step": 9405
+    },
+    {
+      "epoch": 0.34207368076517436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5312482370895677e-05,
+      "loss": 0.9883,
+      "step": 9406
+    },
+    {
+      "epoch": 0.34211004836891296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5311484430541254e-05,
+      "loss": 1.0797,
+      "step": 9407
+    },
+    {
+      "epoch": 0.34214641597265155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.531048641649752e-05,
+      "loss": 1.0156,
+      "step": 9408
+    },
+    {
+      "epoch": 0.34218278357639015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5309488328778322e-05,
+      "loss": 1.1182,
+      "step": 9409
+    },
+    {
+      "epoch": 0.34221915118012874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5308490167397505e-05,
+      "loss": 1.1758,
+      "step": 9410
+    },
+    {
+      "epoch": 0.34225551878386734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5307491932368922e-05,
+      "loss": 1.1986,
+      "step": 9411
+    },
+    {
+      "epoch": 0.34229188638760594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5306493623706416e-05,
+      "loss": 1.0177,
+      "step": 9412
+    },
+    {
+      "epoch": 0.34232825399134453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.530549524142384e-05,
+      "loss": 0.9993,
+      "step": 9413
+    },
+    {
+      "epoch": 0.3423646215950831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5304496785535046e-05,
+      "loss": 1.0192,
+      "step": 9414
+    },
+    {
+      "epoch": 0.34240098919882167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5303498256053884e-05,
+      "loss": 1.0826,
+      "step": 9415
+    },
+    {
+      "epoch": 0.34243735680256027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5302499652994208e-05,
+      "loss": 0.9852,
+      "step": 9416
+    },
+    {
+      "epoch": 0.34247372440629886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5301500976369873e-05,
+      "loss": 1.1157,
+      "step": 9417
+    },
+    {
+      "epoch": 0.34251009201003746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5300502226194737e-05,
+      "loss": 1.2057,
+      "step": 9418
+    },
+    {
+      "epoch": 0.34254645961377606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5299503402482653e-05,
+      "loss": 0.8908,
+      "step": 9419
+    },
+    {
+      "epoch": 0.34258282721751465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5298504505247476e-05,
+      "loss": 1.0511,
+      "step": 9420
+    },
+    {
+      "epoch": 0.34261919482125325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5297505534503066e-05,
+      "loss": 1.038,
+      "step": 9421
+    },
+    {
+      "epoch": 0.34265556242499184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5296506490263285e-05,
+      "loss": 1.0708,
+      "step": 9422
+    },
+    {
+      "epoch": 0.3426919300287304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5295507372541985e-05,
+      "loss": 1.0822,
+      "step": 9423
+    },
+    {
+      "epoch": 0.342728297632469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.529450818135304e-05,
+      "loss": 1.1161,
+      "step": 9424
+    },
+    {
+      "epoch": 0.3427646652362076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5293508916710305e-05,
+      "loss": 1.0119,
+      "step": 9425
+    },
+    {
+      "epoch": 0.3428010328399462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.529250957862764e-05,
+      "loss": 1.0251,
+      "step": 9426
+    },
+    {
+      "epoch": 0.34283740044368477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5291510167118918e-05,
+      "loss": 1.0795,
+      "step": 9427
+    },
+    {
+      "epoch": 0.34287376804742337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5290510682198e-05,
+      "loss": 1.044,
+      "step": 9428
+    },
+    {
+      "epoch": 0.34291013565116196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.528951112387875e-05,
+      "loss": 1.1538,
+      "step": 9429
+    },
+    {
+      "epoch": 0.34294650325490056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.528851149217504e-05,
+      "loss": 1.1204,
+      "step": 9430
+    },
+    {
+      "epoch": 0.3429828708586391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.528751178710074e-05,
+      "loss": 1.0179,
+      "step": 9431
+    },
+    {
+      "epoch": 0.3430192384623777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.528651200866971e-05,
+      "loss": 1.0705,
+      "step": 9432
+    },
+    {
+      "epoch": 0.3430556060661163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.528551215689583e-05,
+      "loss": 0.9604,
+      "step": 9433
+    },
+    {
+      "epoch": 0.3430919736698549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5284512231792963e-05,
+      "loss": 1.033,
+      "step": 9434
+    },
+    {
+      "epoch": 0.3431283412735935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5283512233374992e-05,
+      "loss": 1.0459,
+      "step": 9435
+    },
+    {
+      "epoch": 0.3431647088773321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5282512161655785e-05,
+      "loss": 1.0471,
+      "step": 9436
+    },
+    {
+      "epoch": 0.3432010764810707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.528151201664921e-05,
+      "loss": 1.1107,
+      "step": 9437
+    },
+    {
+      "epoch": 0.3432374440848093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5280511798369157e-05,
+      "loss": 1.1534,
+      "step": 9438
+    },
+    {
+      "epoch": 0.3432738116885478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.527951150682949e-05,
+      "loss": 1.1692,
+      "step": 9439
+    },
+    {
+      "epoch": 0.3433101792922864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5278511142044098e-05,
+      "loss": 1.1122,
+      "step": 9440
+    },
+    {
+      "epoch": 0.343346546896025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5277510704026848e-05,
+      "loss": 1.1933,
+      "step": 9441
+    },
+    {
+      "epoch": 0.3433829144997636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5276510192791623e-05,
+      "loss": 1.0491,
+      "step": 9442
+    },
+    {
+      "epoch": 0.3434192821035022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5275509608352308e-05,
+      "loss": 1.1824,
+      "step": 9443
+    },
+    {
+      "epoch": 0.3434556497072408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5274508950722783e-05,
+      "loss": 1.0041,
+      "step": 9444
+    },
+    {
+      "epoch": 0.3434920173109794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5273508219916927e-05,
+      "loss": 1.0519,
+      "step": 9445
+    },
+    {
+      "epoch": 0.343528384914718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.527250741594863e-05,
+      "loss": 1.1416,
+      "step": 9446
+    },
+    {
+      "epoch": 0.3435647525184566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5271506538831774e-05,
+      "loss": 1.2104,
+      "step": 9447
+    },
+    {
+      "epoch": 0.3436011201221951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.527050558858024e-05,
+      "loss": 1.0258,
+      "step": 9448
+    },
+    {
+      "epoch": 0.3436374877259337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5269504565207922e-05,
+      "loss": 1.1036,
+      "step": 9449
+    },
+    {
+      "epoch": 0.3436738553296723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.52685034687287e-05,
+      "loss": 1.0305,
+      "step": 9450
+    },
+    {
+      "epoch": 0.3437102229334109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5267502299156473e-05,
+      "loss": 1.0125,
+      "step": 9451
+    },
+    {
+      "epoch": 0.3437465905371495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.526650105650512e-05,
+      "loss": 1.0548,
+      "step": 9452
+    },
+    {
+      "epoch": 0.3437829581408881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.526549974078854e-05,
+      "loss": 0.9539,
+      "step": 9453
+    },
+    {
+      "epoch": 0.3438193257446267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5264498352020622e-05,
+      "loss": 0.9962,
+      "step": 9454
+    },
+    {
+      "epoch": 0.3438556933483653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.526349689021526e-05,
+      "loss": 1.0,
+      "step": 9455
+    },
+    {
+      "epoch": 0.34389206095210384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5262495355386345e-05,
+      "loss": 1.1435,
+      "step": 9456
+    },
+    {
+      "epoch": 0.34392842855584244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5261493747547776e-05,
+      "loss": 1.1866,
+      "step": 9457
+    },
+    {
+      "epoch": 0.34396479615958103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5260492066713442e-05,
+      "loss": 1.0173,
+      "step": 9458
+    },
+    {
+      "epoch": 0.34400116376331963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5259490312897245e-05,
+      "loss": 1.0775,
+      "step": 9459
+    },
+    {
+      "epoch": 0.3440375313670582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5258488486113086e-05,
+      "loss": 1.1425,
+      "step": 9460
+    },
+    {
+      "epoch": 0.3440738989707968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5257486586374857e-05,
+      "loss": 1.0746,
+      "step": 9461
+    },
+    {
+      "epoch": 0.3441102665745354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5256484613696461e-05,
+      "loss": 1.0601,
+      "step": 9462
+    },
+    {
+      "epoch": 0.344146634178274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.52554825680918e-05,
+      "loss": 1.0002,
+      "step": 9463
+    },
+    {
+      "epoch": 0.3441830017820126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5254480449574773e-05,
+      "loss": 1.0488,
+      "step": 9464
+    },
+    {
+      "epoch": 0.34421936938575115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5253478258159287e-05,
+      "loss": 1.0256,
+      "step": 9465
+    },
+    {
+      "epoch": 0.34425573698948975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.525247599385924e-05,
+      "loss": 1.0102,
+      "step": 9466
+    },
+    {
+      "epoch": 0.34429210459322834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5251473656688544e-05,
+      "loss": 1.1198,
+      "step": 9467
+    },
+    {
+      "epoch": 0.34432847219696694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5250471246661103e-05,
+      "loss": 1.0815,
+      "step": 9468
+    },
+    {
+      "epoch": 0.34436483980070554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.524946876379082e-05,
+      "loss": 0.9362,
+      "step": 9469
+    },
+    {
+      "epoch": 0.34440120740444413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5248466208091607e-05,
+      "loss": 1.1622,
+      "step": 9470
+    },
+    {
+      "epoch": 0.34443757500818273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5247463579577373e-05,
+      "loss": 1.1718,
+      "step": 9471
+    },
+    {
+      "epoch": 0.3444739426119213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5246460878262026e-05,
+      "loss": 0.9979,
+      "step": 9472
+    },
+    {
+      "epoch": 0.34451031021565987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.524545810415948e-05,
+      "loss": 0.9636,
+      "step": 9473
+    },
+    {
+      "epoch": 0.34454667781939846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5244455257283644e-05,
+      "loss": 1.051,
+      "step": 9474
+    },
+    {
+      "epoch": 0.34458304542313706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.524345233764843e-05,
+      "loss": 1.1085,
+      "step": 9475
+    },
+    {
+      "epoch": 0.34461941302687565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5242449345267759e-05,
+      "loss": 1.059,
+      "step": 9476
+    },
+    {
+      "epoch": 0.34465578063061425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5241446280155536e-05,
+      "loss": 1.2362,
+      "step": 9477
+    },
+    {
+      "epoch": 0.34469214823435285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5240443142325687e-05,
+      "loss": 1.075,
+      "step": 9478
+    },
+    {
+      "epoch": 0.34472851583809144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5239439931792124e-05,
+      "loss": 1.0187,
+      "step": 9479
+    },
+    {
+      "epoch": 0.34476488344183004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5238436648568762e-05,
+      "loss": 1.034,
+      "step": 9480
+    },
+    {
+      "epoch": 0.3448012510455686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5237433292669527e-05,
+      "loss": 1.1611,
+      "step": 9481
+    },
+    {
+      "epoch": 0.3448376186493072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5236429864108335e-05,
+      "loss": 1.0605,
+      "step": 9482
+    },
+    {
+      "epoch": 0.3448739862530458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.523542636289911e-05,
+      "loss": 0.9496,
+      "step": 9483
+    },
+    {
+      "epoch": 0.34491035385678437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5234422789055771e-05,
+      "loss": 1.1724,
+      "step": 9484
+    },
+    {
+      "epoch": 0.34494672146052296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5233419142592241e-05,
+      "loss": 0.9601,
+      "step": 9485
+    },
+    {
+      "epoch": 0.34498308906426156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5232415423522446e-05,
+      "loss": 1.0596,
+      "step": 9486
+    },
+    {
+      "epoch": 0.34501945666800016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5231411631860313e-05,
+      "loss": 0.9381,
+      "step": 9487
+    },
+    {
+      "epoch": 0.34505582427173875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5230407767619764e-05,
+      "loss": 1.1156,
+      "step": 9488
+    },
+    {
+      "epoch": 0.34509219187547735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5229403830814731e-05,
+      "loss": 1.0365,
+      "step": 9489
+    },
+    {
+      "epoch": 0.3451285594792159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5228399821459137e-05,
+      "loss": 0.933,
+      "step": 9490
+    },
+    {
+      "epoch": 0.3451649270829545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5227395739566913e-05,
+      "loss": 1.0219,
+      "step": 9491
+    },
+    {
+      "epoch": 0.3452012946866931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5226391585151992e-05,
+      "loss": 1.033,
+      "step": 9492
+    },
+    {
+      "epoch": 0.3452376622904317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5225387358228298e-05,
+      "loss": 1.0283,
+      "step": 9493
+    },
+    {
+      "epoch": 0.3452740298941703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5224383058809773e-05,
+      "loss": 1.0904,
+      "step": 9494
+    },
+    {
+      "epoch": 0.34531039749790887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5223378686910346e-05,
+      "loss": 0.9465,
+      "step": 9495
+    },
+    {
+      "epoch": 0.34534676510164747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.522237424254395e-05,
+      "loss": 1.0999,
+      "step": 9496
+    },
+    {
+      "epoch": 0.34538313270538606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5221369725724522e-05,
+      "loss": 1.0889,
+      "step": 9497
+    },
+    {
+      "epoch": 0.3454195003091246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5220365136465993e-05,
+      "loss": 1.082,
+      "step": 9498
+    },
+    {
+      "epoch": 0.3454558679128632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5219360474782307e-05,
+      "loss": 1.0878,
+      "step": 9499
+    },
+    {
+      "epoch": 0.3454922355166018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.52183557406874e-05,
+      "loss": 0.9195,
+      "step": 9500
+    },
+    {
+      "epoch": 0.3455286031203404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5217350934195212e-05,
+      "loss": 1.1212,
+      "step": 9501
+    },
+    {
+      "epoch": 0.345564970724079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5216346055319679e-05,
+      "loss": 1.0759,
+      "step": 9502
+    },
+    {
+      "epoch": 0.3456013383278176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5215341104074747e-05,
+      "loss": 1.0491,
+      "step": 9503
+    },
+    {
+      "epoch": 0.3456377059315562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5214336080474357e-05,
+      "loss": 1.1406,
+      "step": 9504
+    },
+    {
+      "epoch": 0.3456740735352948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5213330984532455e-05,
+      "loss": 1.1646,
+      "step": 9505
+    },
+    {
+      "epoch": 0.3457104411390334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5212325816262978e-05,
+      "loss": 0.9617,
+      "step": 9506
+    },
+    {
+      "epoch": 0.3457468087427719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5211320575679875e-05,
+      "loss": 1.0163,
+      "step": 9507
+    },
+    {
+      "epoch": 0.3457831763465105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5210315262797097e-05,
+      "loss": 1.0582,
+      "step": 9508
+    },
+    {
+      "epoch": 0.3458195439502491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5209309877628584e-05,
+      "loss": 0.9896,
+      "step": 9509
+    },
+    {
+      "epoch": 0.3458559115539877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5208304420188289e-05,
+      "loss": 1.1649,
+      "step": 9510
+    },
+    {
+      "epoch": 0.3458922791577263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5207298890490161e-05,
+      "loss": 1.0665,
+      "step": 9511
+    },
+    {
+      "epoch": 0.3459286467614649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5206293288548146e-05,
+      "loss": 1.0905,
+      "step": 9512
+    },
+    {
+      "epoch": 0.3459650143652035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5205287614376201e-05,
+      "loss": 1.1248,
+      "step": 9513
+    },
+    {
+      "epoch": 0.3460013819689421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5204281867988276e-05,
+      "loss": 1.0939,
+      "step": 9514
+    },
+    {
+      "epoch": 0.34603774957268063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5203276049398321e-05,
+      "loss": 1.1308,
+      "step": 9515
+    },
+    {
+      "epoch": 0.3460741171764192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5202270158620299e-05,
+      "loss": 1.0952,
+      "step": 9516
+    },
+    {
+      "epoch": 0.3461104847801578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5201264195668153e-05,
+      "loss": 1.1599,
+      "step": 9517
+    },
+    {
+      "epoch": 0.3461468523838964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.520025816055585e-05,
+      "loss": 1.1118,
+      "step": 9518
+    },
+    {
+      "epoch": 0.346183219987635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.519925205329734e-05,
+      "loss": 0.9831,
+      "step": 9519
+    },
+    {
+      "epoch": 0.3462195875913736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5198245873906588e-05,
+      "loss": 0.9556,
+      "step": 9520
+    },
+    {
+      "epoch": 0.3462559551951122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5197239622397545e-05,
+      "loss": 1.1369,
+      "step": 9521
+    },
+    {
+      "epoch": 0.3462923227988508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.519623329878418e-05,
+      "loss": 1.1378,
+      "step": 9522
+    },
+    {
+      "epoch": 0.3463286904025894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.519522690308045e-05,
+      "loss": 1.0817,
+      "step": 9523
+    },
+    {
+      "epoch": 0.34636505800632794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5194220435300317e-05,
+      "loss": 1.1557,
+      "step": 9524
+    },
+    {
+      "epoch": 0.34640142561006654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5193213895457746e-05,
+      "loss": 1.0883,
+      "step": 9525
+    },
+    {
+      "epoch": 0.34643779321380513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5192207283566701e-05,
+      "loss": 1.1021,
+      "step": 9526
+    },
+    {
+      "epoch": 0.34647416081754373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5191200599641147e-05,
+      "loss": 1.1478,
+      "step": 9527
+    },
+    {
+      "epoch": 0.3465105284212823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5190193843695046e-05,
+      "loss": 0.8916,
+      "step": 9528
+    },
+    {
+      "epoch": 0.3465468960250209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5189187015742373e-05,
+      "loss": 1.0001,
+      "step": 9529
+    },
+    {
+      "epoch": 0.3465832636287595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5188180115797093e-05,
+      "loss": 1.0535,
+      "step": 9530
+    },
+    {
+      "epoch": 0.3466196312324981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5187173143873173e-05,
+      "loss": 1.0714,
+      "step": 9531
+    },
+    {
+      "epoch": 0.34665599883623666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5186166099984588e-05,
+      "loss": 1.0764,
+      "step": 9532
+    },
+    {
+      "epoch": 0.34669236643997525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5185158984145301e-05,
+      "loss": 1.0693,
+      "step": 9533
+    },
+    {
+      "epoch": 0.34672873404371385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5184151796369295e-05,
+      "loss": 1.1362,
+      "step": 9534
+    },
+    {
+      "epoch": 0.34676510164745245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.518314453667054e-05,
+      "loss": 1.0417,
+      "step": 9535
+    },
+    {
+      "epoch": 0.34680146925119104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5182137205063003e-05,
+      "loss": 1.1801,
+      "step": 9536
+    },
+    {
+      "epoch": 0.34683783685492964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5181129801560665e-05,
+      "loss": 1.1379,
+      "step": 9537
+    },
+    {
+      "epoch": 0.34687420445866823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5180122326177505e-05,
+      "loss": 1.001,
+      "step": 9538
+    },
+    {
+      "epoch": 0.34691057206240683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5179114778927497e-05,
+      "loss": 1.1331,
+      "step": 9539
+    },
+    {
+      "epoch": 0.34694693966614537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5178107159824618e-05,
+      "loss": 1.0024,
+      "step": 9540
+    },
+    {
+      "epoch": 0.34698330726988397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.517709946888285e-05,
+      "loss": 1.1383,
+      "step": 9541
+    },
+    {
+      "epoch": 0.34701967487362256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5176091706116171e-05,
+      "loss": 1.0435,
+      "step": 9542
+    },
+    {
+      "epoch": 0.34705604247736116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5175083871538566e-05,
+      "loss": 1.0607,
+      "step": 9543
+    },
+    {
+      "epoch": 0.34709241008109976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5174075965164012e-05,
+      "loss": 1.0804,
+      "step": 9544
+    },
+    {
+      "epoch": 0.34712877768483835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5173067987006497e-05,
+      "loss": 1.1012,
+      "step": 9545
+    },
+    {
+      "epoch": 0.34716514528857695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5172059937080003e-05,
+      "loss": 1.0051,
+      "step": 9546
+    },
+    {
+      "epoch": 0.34720151289231554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5171051815398513e-05,
+      "loss": 0.9591,
+      "step": 9547
+    },
+    {
+      "epoch": 0.34723788049605414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.517004362197602e-05,
+      "loss": 0.9924,
+      "step": 9548
+    },
+    {
+      "epoch": 0.3472742480997927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5169035356826504e-05,
+      "loss": 1.0823,
+      "step": 9549
+    },
+    {
+      "epoch": 0.3473106157035313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5168027019963958e-05,
+      "loss": 1.0442,
+      "step": 9550
+    },
+    {
+      "epoch": 0.3473469833072699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5167018611402369e-05,
+      "loss": 1.0633,
+      "step": 9551
+    },
+    {
+      "epoch": 0.34738335091100847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5166010131155728e-05,
+      "loss": 0.9795,
+      "step": 9552
+    },
+    {
+      "epoch": 0.34741971851474707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5165001579238029e-05,
+      "loss": 0.9953,
+      "step": 9553
+    },
+    {
+      "epoch": 0.34745608611848566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5163992955663256e-05,
+      "loss": 1.1204,
+      "step": 9554
+    },
+    {
+      "epoch": 0.34749245372222426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5162984260445411e-05,
+      "loss": 0.9935,
+      "step": 9555
+    },
+    {
+      "epoch": 0.34752882132596286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5161975493598484e-05,
+      "loss": 1.0348,
+      "step": 9556
+    },
+    {
+      "epoch": 0.3475651889297014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5160966655136469e-05,
+      "loss": 1.1995,
+      "step": 9557
+    },
+    {
+      "epoch": 0.34760155653344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5159957745073363e-05,
+      "loss": 1.0222,
+      "step": 9558
+    },
+    {
+      "epoch": 0.3476379241371786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.515894876342317e-05,
+      "loss": 1.0403,
+      "step": 9559
+    },
+    {
+      "epoch": 0.3476742917409172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5157939710199876e-05,
+      "loss": 1.1085,
+      "step": 9560
+    },
+    {
+      "epoch": 0.3477106593446558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5156930585417492e-05,
+      "loss": 1.0544,
+      "step": 9561
+    },
+    {
+      "epoch": 0.3477470269483944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5155921389090011e-05,
+      "loss": 1.0655,
+      "step": 9562
+    },
+    {
+      "epoch": 0.347783394552133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5154912121231435e-05,
+      "loss": 0.9936,
+      "step": 9563
+    },
+    {
+      "epoch": 0.34781976215587157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5153902781855769e-05,
+      "loss": 1.0847,
+      "step": 9564
+    },
+    {
+      "epoch": 0.34785612975961017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5152893370977012e-05,
+      "loss": 1.0776,
+      "step": 9565
+    },
+    {
+      "epoch": 0.3478924973633487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5151883888609171e-05,
+      "loss": 1.0617,
+      "step": 9566
+    },
+    {
+      "epoch": 0.3479288649670873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5150874334766252e-05,
+      "loss": 1.1328,
+      "step": 9567
+    },
+    {
+      "epoch": 0.3479652325708259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5149864709462259e-05,
+      "loss": 0.9787,
+      "step": 9568
+    },
+    {
+      "epoch": 0.3480016001745645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5148855012711198e-05,
+      "loss": 1.0705,
+      "step": 9569
+    },
+    {
+      "epoch": 0.3480379677783031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5147845244527084e-05,
+      "loss": 0.9712,
+      "step": 9570
+    },
+    {
+      "epoch": 0.3480743353820417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5146835404923917e-05,
+      "loss": 1.0351,
+      "step": 9571
+    },
+    {
+      "epoch": 0.3481107029857803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5145825493915713e-05,
+      "loss": 1.1167,
+      "step": 9572
+    },
+    {
+      "epoch": 0.3481470705895189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.514481551151648e-05,
+      "loss": 1.1329,
+      "step": 9573
+    },
+    {
+      "epoch": 0.3481834381932574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5143805457740233e-05,
+      "loss": 1.0034,
+      "step": 9574
+    },
+    {
+      "epoch": 0.348219805796996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5142795332600981e-05,
+      "loss": 1.0501,
+      "step": 9575
+    },
+    {
+      "epoch": 0.3482561734007346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5141785136112742e-05,
+      "loss": 1.0901,
+      "step": 9576
+    },
+    {
+      "epoch": 0.3482925410044732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.514077486828953e-05,
+      "loss": 0.9976,
+      "step": 9577
+    },
+    {
+      "epoch": 0.3483289086082118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5139764529145358e-05,
+      "loss": 1.1417,
+      "step": 9578
+    },
+    {
+      "epoch": 0.3483652762119504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5138754118694245e-05,
+      "loss": 1.0829,
+      "step": 9579
+    },
+    {
+      "epoch": 0.348401643815689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5137743636950215e-05,
+      "loss": 1.0273,
+      "step": 9580
+    },
+    {
+      "epoch": 0.3484380114194276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5136733083927278e-05,
+      "loss": 1.0395,
+      "step": 9581
+    },
+    {
+      "epoch": 0.34847437902316614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5135722459639457e-05,
+      "loss": 1.0641,
+      "step": 9582
+    },
+    {
+      "epoch": 0.34851074662690473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5134711764100774e-05,
+      "loss": 1.0414,
+      "step": 9583
+    },
+    {
+      "epoch": 0.34854711423064333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.513370099732525e-05,
+      "loss": 1.081,
+      "step": 9584
+    },
+    {
+      "epoch": 0.3485834818343819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.513269015932691e-05,
+      "loss": 1.0647,
+      "step": 9585
+    },
+    {
+      "epoch": 0.3486198494381205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5131679250119775e-05,
+      "loss": 1.0948,
+      "step": 9586
+    },
+    {
+      "epoch": 0.3486562170418591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.513066826971787e-05,
+      "loss": 0.8508,
+      "step": 9587
+    },
+    {
+      "epoch": 0.3486925846455977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5129657218135225e-05,
+      "loss": 1.0786,
+      "step": 9588
+    },
+    {
+      "epoch": 0.3487289522493363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5128646095385861e-05,
+      "loss": 1.0901,
+      "step": 9589
+    },
+    {
+      "epoch": 0.3487653198530749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5127634901483812e-05,
+      "loss": 0.9944,
+      "step": 9590
+    },
+    {
+      "epoch": 0.34880168745681345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5126623636443106e-05,
+      "loss": 1.1582,
+      "step": 9591
+    },
+    {
+      "epoch": 0.34883805506055204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5125612300277768e-05,
+      "loss": 1.0355,
+      "step": 9592
+    },
+    {
+      "epoch": 0.34887442266429064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5124600893001831e-05,
+      "loss": 0.9516,
+      "step": 9593
+    },
+    {
+      "epoch": 0.34891079026802924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5123589414629325e-05,
+      "loss": 1.0389,
+      "step": 9594
+    },
+    {
+      "epoch": 0.34894715787176783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5122577865174291e-05,
+      "loss": 0.9556,
+      "step": 9595
+    },
+    {
+      "epoch": 0.34898352547550643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5121566244650754e-05,
+      "loss": 1.0321,
+      "step": 9596
+    },
+    {
+      "epoch": 0.349019893079245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5120554553072753e-05,
+      "loss": 1.1086,
+      "step": 9597
+    },
+    {
+      "epoch": 0.3490562606829836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.511954279045432e-05,
+      "loss": 1.11,
+      "step": 9598
+    },
+    {
+      "epoch": 0.34909262828672216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5118530956809498e-05,
+      "loss": 1.1525,
+      "step": 9599
+    },
+    {
+      "epoch": 0.34912899589046076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5117519052152316e-05,
+      "loss": 1.0637,
+      "step": 9600
+    },
+    {
+      "epoch": 0.34916536349419935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5116507076496825e-05,
+      "loss": 1.16,
+      "step": 9601
+    },
+    {
+      "epoch": 0.34920173109793795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5115495029857052e-05,
+      "loss": 1.1035,
+      "step": 9602
+    },
+    {
+      "epoch": 0.34923809870167655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5114482912247045e-05,
+      "loss": 1.1611,
+      "step": 9603
+    },
+    {
+      "epoch": 0.34927446630541514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5113470723680847e-05,
+      "loss": 1.0752,
+      "step": 9604
+    },
+    {
+      "epoch": 0.34931083390915374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5112458464172495e-05,
+      "loss": 1.0243,
+      "step": 9605
+    },
+    {
+      "epoch": 0.34934720151289234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5111446133736035e-05,
+      "loss": 0.9733,
+      "step": 9606
+    },
+    {
+      "epoch": 0.34938356911663093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5110433732385514e-05,
+      "loss": 1.1916,
+      "step": 9607
+    },
+    {
+      "epoch": 0.3494199367203695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5109421260134974e-05,
+      "loss": 1.0784,
+      "step": 9608
+    },
+    {
+      "epoch": 0.34945630432410807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5108408716998465e-05,
+      "loss": 1.0717,
+      "step": 9609
+    },
+    {
+      "epoch": 0.34949267192784667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.510739610299003e-05,
+      "loss": 1.0498,
+      "step": 9610
+    },
+    {
+      "epoch": 0.34952903953158526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5106383418123725e-05,
+      "loss": 1.0191,
+      "step": 9611
+    },
+    {
+      "epoch": 0.34956540713532386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5105370662413594e-05,
+      "loss": 1.0439,
+      "step": 9612
+    },
+    {
+      "epoch": 0.34960177473906245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5104357835873686e-05,
+      "loss": 1.0154,
+      "step": 9613
+    },
+    {
+      "epoch": 0.34963814234280105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5103344938518056e-05,
+      "loss": 1.0457,
+      "step": 9614
+    },
+    {
+      "epoch": 0.34967450994653965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5102331970360757e-05,
+      "loss": 1.2035,
+      "step": 9615
+    },
+    {
+      "epoch": 0.3497108775502782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5101318931415838e-05,
+      "loss": 1.0768,
+      "step": 9616
+    },
+    {
+      "epoch": 0.3497472451540168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.510030582169736e-05,
+      "loss": 0.9846,
+      "step": 9617
+    },
+    {
+      "epoch": 0.3497836127577554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5099292641219373e-05,
+      "loss": 0.9529,
+      "step": 9618
+    },
+    {
+      "epoch": 0.349819980361494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5098279389995938e-05,
+      "loss": 0.9071,
+      "step": 9619
+    },
+    {
+      "epoch": 0.3498563479652326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5097266068041107e-05,
+      "loss": 1.1047,
+      "step": 9620
+    },
+    {
+      "epoch": 0.34989271556897117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5096252675368942e-05,
+      "loss": 1.0652,
+      "step": 9621
+    },
+    {
+      "epoch": 0.34992908317270977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5095239211993504e-05,
+      "loss": 1.0552,
+      "step": 9622
+    },
+    {
+      "epoch": 0.34996545077644836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.509422567792885e-05,
+      "loss": 0.9098,
+      "step": 9623
+    },
+    {
+      "epoch": 0.3500018183801869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5093212073189043e-05,
+      "loss": 1.105,
+      "step": 9624
+    },
+    {
+      "epoch": 0.3500381859839255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5092198397788145e-05,
+      "loss": 1.0663,
+      "step": 9625
+    },
+    {
+      "epoch": 0.3500745535876641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5091184651740216e-05,
+      "loss": 1.0427,
+      "step": 9626
+    },
+    {
+      "epoch": 0.3501109211914027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5090170835059326e-05,
+      "loss": 1.0635,
+      "step": 9627
+    },
+    {
+      "epoch": 0.3501472887951413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5089156947759538e-05,
+      "loss": 1.0177,
+      "step": 9628
+    },
+    {
+      "epoch": 0.3501836563988799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5088142989854917e-05,
+      "loss": 1.0279,
+      "step": 9629
+    },
+    {
+      "epoch": 0.3502200240026185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5087128961359532e-05,
+      "loss": 1.1252,
+      "step": 9630
+    },
+    {
+      "epoch": 0.3502563916063571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5086114862287451e-05,
+      "loss": 0.946,
+      "step": 9631
+    },
+    {
+      "epoch": 0.35029275921009567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5085100692652738e-05,
+      "loss": 1.1251,
+      "step": 9632
+    },
+    {
+      "epoch": 0.3503291268138342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5084086452469473e-05,
+      "loss": 1.1142,
+      "step": 9633
+    },
+    {
+      "epoch": 0.3503654944175728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5083072141751722e-05,
+      "loss": 1.2268,
+      "step": 9634
+    },
+    {
+      "epoch": 0.3504018620213114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5082057760513553e-05,
+      "loss": 1.0776,
+      "step": 9635
+    },
+    {
+      "epoch": 0.35043822962505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5081043308769045e-05,
+      "loss": 1.0856,
+      "step": 9636
+    },
+    {
+      "epoch": 0.3504745972287886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.508002878653227e-05,
+      "loss": 1.0606,
+      "step": 9637
+    },
+    {
+      "epoch": 0.3505109648325272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5079014193817302e-05,
+      "loss": 0.9896,
+      "step": 9638
+    },
+    {
+      "epoch": 0.3505473324362658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.507799953063822e-05,
+      "loss": 1.0385,
+      "step": 9639
+    },
+    {
+      "epoch": 0.3505837000400044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5076984797009098e-05,
+      "loss": 1.0193,
+      "step": 9640
+    },
+    {
+      "epoch": 0.3506200676437429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5075969992944018e-05,
+      "loss": 0.9245,
+      "step": 9641
+    },
+    {
+      "epoch": 0.3506564352474815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5074955118457055e-05,
+      "loss": 1.1141,
+      "step": 9642
+    },
+    {
+      "epoch": 0.3506928028512201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5073940173562286e-05,
+      "loss": 1.073,
+      "step": 9643
+    },
+    {
+      "epoch": 0.3507291704549587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5072925158273801e-05,
+      "loss": 1.0981,
+      "step": 9644
+    },
+    {
+      "epoch": 0.3507655380586973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5071910072605675e-05,
+      "loss": 1.0374,
+      "step": 9645
+    },
+    {
+      "epoch": 0.3508019056624359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5070894916571993e-05,
+      "loss": 1.0771,
+      "step": 9646
+    },
+    {
+      "epoch": 0.3508382732661745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5069879690186839e-05,
+      "loss": 1.027,
+      "step": 9647
+    },
+    {
+      "epoch": 0.3508746408699131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5068864393464297e-05,
+      "loss": 1.0396,
+      "step": 9648
+    },
+    {
+      "epoch": 0.3509110084736517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5067849026418455e-05,
+      "loss": 1.0275,
+      "step": 9649
+    },
+    {
+      "epoch": 0.35094737607739024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5066833589063396e-05,
+      "loss": 1.0987,
+      "step": 9650
+    },
+    {
+      "epoch": 0.35098374368112883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5065818081413212e-05,
+      "loss": 1.0577,
+      "step": 9651
+    },
+    {
+      "epoch": 0.35102011128486743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5064802503481988e-05,
+      "loss": 0.9753,
+      "step": 9652
+    },
+    {
+      "epoch": 0.351056478888606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5063786855283815e-05,
+      "loss": 0.8957,
+      "step": 9653
+    },
+    {
+      "epoch": 0.3510928464923446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5062771136832784e-05,
+      "loss": 1.0313,
+      "step": 9654
+    },
+    {
+      "epoch": 0.3511292140960832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5061755348142988e-05,
+      "loss": 0.9846,
+      "step": 9655
+    },
+    {
+      "epoch": 0.3511655816998218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5060739489228517e-05,
+      "loss": 1.075,
+      "step": 9656
+    },
+    {
+      "epoch": 0.3512019493035604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5059723560103466e-05,
+      "loss": 1.0052,
+      "step": 9657
+    },
+    {
+      "epoch": 0.35123831690729895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5058707560781928e-05,
+      "loss": 1.207,
+      "step": 9658
+    },
+    {
+      "epoch": 0.35127468451103755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5057691491278003e-05,
+      "loss": 0.9779,
+      "step": 9659
+    },
+    {
+      "epoch": 0.35131105211477615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5056675351605783e-05,
+      "loss": 1.1508,
+      "step": 9660
+    },
+    {
+      "epoch": 0.35134741971851474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5055659141779366e-05,
+      "loss": 0.9367,
+      "step": 9661
+    },
+    {
+      "epoch": 0.35138378732225334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5054642861812854e-05,
+      "loss": 1.0594,
+      "step": 9662
+    },
+    {
+      "epoch": 0.35142015492599193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.505362651172034e-05,
+      "loss": 0.9798,
+      "step": 9663
+    },
+    {
+      "epoch": 0.35145652252973053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5052610091515931e-05,
+      "loss": 0.9578,
+      "step": 9664
+    },
+    {
+      "epoch": 0.3514928901334691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5051593601213728e-05,
+      "loss": 1.0319,
+      "step": 9665
+    },
+    {
+      "epoch": 0.35152925773720767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5050577040827827e-05,
+      "loss": 1.0761,
+      "step": 9666
+    },
+    {
+      "epoch": 0.35156562534094626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5049560410372336e-05,
+      "loss": 0.9291,
+      "step": 9667
+    },
+    {
+      "epoch": 0.35160199294468486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5048543709861361e-05,
+      "loss": 1.0379,
+      "step": 9668
+    },
+    {
+      "epoch": 0.35163836054842346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5047526939309003e-05,
+      "loss": 1.0486,
+      "step": 9669
+    },
+    {
+      "epoch": 0.35167472815216205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5046510098729372e-05,
+      "loss": 1.2183,
+      "step": 9670
+    },
+    {
+      "epoch": 0.35171109575590065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5045493188136573e-05,
+      "loss": 0.9977,
+      "step": 9671
+    },
+    {
+      "epoch": 0.35174746335963925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5044476207544711e-05,
+      "loss": 0.9756,
+      "step": 9672
+    },
+    {
+      "epoch": 0.35178383096337784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5043459156967903e-05,
+      "loss": 1.1598,
+      "step": 9673
+    },
+    {
+      "epoch": 0.35182019856711644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5042442036420255e-05,
+      "loss": 1.0258,
+      "step": 9674
+    },
+    {
+      "epoch": 0.351856566170855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5041424845915874e-05,
+      "loss": 1.0184,
+      "step": 9675
+    },
+    {
+      "epoch": 0.3518929337745936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5040407585468878e-05,
+      "loss": 1.0252,
+      "step": 9676
+    },
+    {
+      "epoch": 0.35192930137833217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5039390255093377e-05,
+      "loss": 0.9635,
+      "step": 9677
+    },
+    {
+      "epoch": 0.35196566898207077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5038372854803492e-05,
+      "loss": 1.118,
+      "step": 9678
+    },
+    {
+      "epoch": 0.35200203658580936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5037355384613327e-05,
+      "loss": 0.9027,
+      "step": 9679
+    },
+    {
+      "epoch": 0.35203840418954796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5036337844537002e-05,
+      "loss": 0.9419,
+      "step": 9680
+    },
+    {
+      "epoch": 0.35207477179328656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5035320234588637e-05,
+      "loss": 0.971,
+      "step": 9681
+    },
+    {
+      "epoch": 0.35211113939702515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.503430255478235e-05,
+      "loss": 1.0579,
+      "step": 9682
+    },
+    {
+      "epoch": 0.3521475070007637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5033284805132255e-05,
+      "loss": 0.9519,
+      "step": 9683
+    },
+    {
+      "epoch": 0.3521838746045023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5032266985652478e-05,
+      "loss": 1.1384,
+      "step": 9684
+    },
+    {
+      "epoch": 0.3522202422082409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5031249096357132e-05,
+      "loss": 1.2487,
+      "step": 9685
+    },
+    {
+      "epoch": 0.3522566098119795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5030231137260344e-05,
+      "loss": 1.1202,
+      "step": 9686
+    },
+    {
+      "epoch": 0.3522929774157181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.502921310837624e-05,
+      "loss": 1.0576,
+      "step": 9687
+    },
+    {
+      "epoch": 0.3523293450194567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5028195009718935e-05,
+      "loss": 0.9092,
+      "step": 9688
+    },
+    {
+      "epoch": 0.35236571262319527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5027176841302563e-05,
+      "loss": 1.0705,
+      "step": 9689
+    },
+    {
+      "epoch": 0.35240208022693387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.502615860314124e-05,
+      "loss": 1.126,
+      "step": 9690
+    },
+    {
+      "epoch": 0.35243844783067246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.50251402952491e-05,
+      "loss": 1.0886,
+      "step": 9691
+    },
+    {
+      "epoch": 0.352474815434411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5024121917640268e-05,
+      "loss": 1.1151,
+      "step": 9692
+    },
+    {
+      "epoch": 0.3525111830381496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5023103470328871e-05,
+      "loss": 1.0304,
+      "step": 9693
+    },
+    {
+      "epoch": 0.3525475506418882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.502208495332904e-05,
+      "loss": 1.0003,
+      "step": 9694
+    },
+    {
+      "epoch": 0.3525839182456268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5021066366654911e-05,
+      "loss": 0.9933,
+      "step": 9695
+    },
+    {
+      "epoch": 0.3526202858493654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5020047710320605e-05,
+      "loss": 1.1597,
+      "step": 9696
+    },
+    {
+      "epoch": 0.352656653453104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5019028984340262e-05,
+      "loss": 1.0329,
+      "step": 9697
+    },
+    {
+      "epoch": 0.3526930210568426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5018010188728012e-05,
+      "loss": 1.1363,
+      "step": 9698
+    },
+    {
+      "epoch": 0.3527293886605812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5016991323497989e-05,
+      "loss": 1.1365,
+      "step": 9699
+    },
+    {
+      "epoch": 0.3527657562643197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5015972388664332e-05,
+      "loss": 1.1001,
+      "step": 9700
+    },
+    {
+      "epoch": 0.3528021238680583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5014953384241173e-05,
+      "loss": 1.0459,
+      "step": 9701
+    },
+    {
+      "epoch": 0.3528384914717969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5013934310242651e-05,
+      "loss": 0.9605,
+      "step": 9702
+    },
+    {
+      "epoch": 0.3528748590755355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5012915166682907e-05,
+      "loss": 1.0685,
+      "step": 9703
+    },
+    {
+      "epoch": 0.3529112266792741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5011895953576074e-05,
+      "loss": 1.0512,
+      "step": 9704
+    },
+    {
+      "epoch": 0.3529475942830127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5010876670936297e-05,
+      "loss": 0.9609,
+      "step": 9705
+    },
+    {
+      "epoch": 0.3529839618867513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5009857318777716e-05,
+      "loss": 0.9591,
+      "step": 9706
+    },
+    {
+      "epoch": 0.3530203294904899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5008837897114474e-05,
+      "loss": 0.9976,
+      "step": 9707
+    },
+    {
+      "epoch": 0.3530566970942285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5007818405960711e-05,
+      "loss": 0.9609,
+      "step": 9708
+    },
+    {
+      "epoch": 0.35309306469796703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.500679884533057e-05,
+      "loss": 1.1416,
+      "step": 9709
+    },
+    {
+      "epoch": 0.3531294323017056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5005779215238203e-05,
+      "loss": 1.0112,
+      "step": 9710
+    },
+    {
+      "epoch": 0.3531657999054442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5004759515697753e-05,
+      "loss": 1.2053,
+      "step": 9711
+    },
+    {
+      "epoch": 0.3532021675091828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.500373974672336e-05,
+      "loss": 1.0643,
+      "step": 9712
+    },
+    {
+      "epoch": 0.3532385351129214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5002719908329183e-05,
+      "loss": 1.1395,
+      "step": 9713
+    },
+    {
+      "epoch": 0.35327490271666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5001700000529362e-05,
+      "loss": 1.0918,
+      "step": 9714
+    },
+    {
+      "epoch": 0.3533112703203986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.500068002333805e-05,
+      "loss": 1.0965,
+      "step": 9715
+    },
+    {
+      "epoch": 0.3533476379241372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4999659976769398e-05,
+      "loss": 1.0237,
+      "step": 9716
+    },
+    {
+      "epoch": 0.35338400552787574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4998639860837559e-05,
+      "loss": 1.1058,
+      "step": 9717
+    },
+    {
+      "epoch": 0.35342037313161434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.499761967555668e-05,
+      "loss": 1.1378,
+      "step": 9718
+    },
+    {
+      "epoch": 0.35345674073535294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4996599420940923e-05,
+      "loss": 1.0646,
+      "step": 9719
+    },
+    {
+      "epoch": 0.35349310833909153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4995579097004436e-05,
+      "loss": 0.9393,
+      "step": 9720
+    },
+    {
+      "epoch": 0.35352947594283013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4994558703761377e-05,
+      "loss": 1.0447,
+      "step": 9721
+    },
+    {
+      "epoch": 0.3535658435465687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4993538241225905e-05,
+      "loss": 0.9805,
+      "step": 9722
+    },
+    {
+      "epoch": 0.3536022111503073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4992517709412171e-05,
+      "loss": 1.0858,
+      "step": 9723
+    },
+    {
+      "epoch": 0.3536385787540459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4991497108334339e-05,
+      "loss": 1.1538,
+      "step": 9724
+    },
+    {
+      "epoch": 0.35367494635778446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4990476438006567e-05,
+      "loss": 1.1016,
+      "step": 9725
+    },
+    {
+      "epoch": 0.35371131396152306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4989455698443015e-05,
+      "loss": 1.0713,
+      "step": 9726
+    },
+    {
+      "epoch": 0.35374768156526165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4988434889657844e-05,
+      "loss": 1.129,
+      "step": 9727
+    },
+    {
+      "epoch": 0.35378404916900025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4987414011665216e-05,
+      "loss": 1.1683,
+      "step": 9728
+    },
+    {
+      "epoch": 0.35382041677273884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4986393064479296e-05,
+      "loss": 1.0534,
+      "step": 9729
+    },
+    {
+      "epoch": 0.35385678437647744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4985372048114247e-05,
+      "loss": 1.0446,
+      "step": 9730
+    },
+    {
+      "epoch": 0.35389315198021604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4984350962584233e-05,
+      "loss": 0.9997,
+      "step": 9731
+    },
+    {
+      "epoch": 0.35392951958395463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4983329807903422e-05,
+      "loss": 1.0319,
+      "step": 9732
+    },
+    {
+      "epoch": 0.35396588718769323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4982308584085978e-05,
+      "loss": 1.0477,
+      "step": 9733
+    },
+    {
+      "epoch": 0.35400225479143177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4981287291146074e-05,
+      "loss": 1.1396,
+      "step": 9734
+    },
+    {
+      "epoch": 0.35403862239517037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4980265929097875e-05,
+      "loss": 1.1845,
+      "step": 9735
+    },
+    {
+      "epoch": 0.35407498999890896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4979244497955554e-05,
+      "loss": 1.1044,
+      "step": 9736
+    },
+    {
+      "epoch": 0.35411135760264756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4978222997733278e-05,
+      "loss": 1.1546,
+      "step": 9737
+    },
+    {
+      "epoch": 0.35414772520638615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4977201428445223e-05,
+      "loss": 0.8775,
+      "step": 9738
+    },
+    {
+      "epoch": 0.35418409281012475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4976179790105559e-05,
+      "loss": 0.9557,
+      "step": 9739
+    },
+    {
+      "epoch": 0.35422046041386335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.497515808272846e-05,
+      "loss": 1.1366,
+      "step": 9740
+    },
+    {
+      "epoch": 0.35425682801760194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4974136306328105e-05,
+      "loss": 1.0072,
+      "step": 9741
+    },
+    {
+      "epoch": 0.3542931956213405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.497311446091866e-05,
+      "loss": 0.9858,
+      "step": 9742
+    },
+    {
+      "epoch": 0.3543295632250791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4972092546514314e-05,
+      "loss": 0.947,
+      "step": 9743
+    },
+    {
+      "epoch": 0.3543659308288177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4971070563129232e-05,
+      "loss": 1.1245,
+      "step": 9744
+    },
+    {
+      "epoch": 0.3544022984325563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4970048510777607e-05,
+      "loss": 1.2832,
+      "step": 9745
+    },
+    {
+      "epoch": 0.35443866603629487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4969026389473606e-05,
+      "loss": 1.1113,
+      "step": 9746
+    },
+    {
+      "epoch": 0.35447503364003347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4968004199231411e-05,
+      "loss": 1.2339,
+      "step": 9747
+    },
+    {
+      "epoch": 0.35451140124377206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4966981940065212e-05,
+      "loss": 1.0477,
+      "step": 9748
+    },
+    {
+      "epoch": 0.35454776884751066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4965959611989183e-05,
+      "loss": 1.0059,
+      "step": 9749
+    },
+    {
+      "epoch": 0.35458413645124925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.496493721501751e-05,
+      "loss": 1.0378,
+      "step": 9750
+    },
+    {
+      "epoch": 0.3546205040549878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.496391474916438e-05,
+      "loss": 0.9956,
+      "step": 9751
+    },
+    {
+      "epoch": 0.3546568716587264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4962892214443972e-05,
+      "loss": 1.0076,
+      "step": 9752
+    },
+    {
+      "epoch": 0.354693239262465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4961869610870477e-05,
+      "loss": 1.066,
+      "step": 9753
+    },
+    {
+      "epoch": 0.3547296068662036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4960846938458082e-05,
+      "loss": 1.1094,
+      "step": 9754
+    },
+    {
+      "epoch": 0.3547659744699422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4959824197220974e-05,
+      "loss": 1.1462,
+      "step": 9755
+    },
+    {
+      "epoch": 0.3548023420736808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4958801387173343e-05,
+      "loss": 0.9634,
+      "step": 9756
+    },
+    {
+      "epoch": 0.3548387096774194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4957778508329377e-05,
+      "loss": 1.0043,
+      "step": 9757
+    },
+    {
+      "epoch": 0.35487507728115797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4956755560703267e-05,
+      "loss": 1.1732,
+      "step": 9758
+    },
+    {
+      "epoch": 0.3549114448848965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4955732544309209e-05,
+      "loss": 1.1564,
+      "step": 9759
+    },
+    {
+      "epoch": 0.3549478124886351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4954709459161391e-05,
+      "loss": 0.9516,
+      "step": 9760
+    },
+    {
+      "epoch": 0.3549841800923737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.495368630527401e-05,
+      "loss": 1.0757,
+      "step": 9761
+    },
+    {
+      "epoch": 0.3550205476961123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4952663082661263e-05,
+      "loss": 1.0646,
+      "step": 9762
+    },
+    {
+      "epoch": 0.3550569152998509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.495163979133734e-05,
+      "loss": 1.091,
+      "step": 9763
+    },
+    {
+      "epoch": 0.3550932829035895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4950616431316442e-05,
+      "loss": 1.0954,
+      "step": 9764
+    },
+    {
+      "epoch": 0.3551296505073281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.494959300261276e-05,
+      "loss": 1.0052,
+      "step": 9765
+    },
+    {
+      "epoch": 0.3551660181110667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4948569505240504e-05,
+      "loss": 1.0347,
+      "step": 9766
+    },
+    {
+      "epoch": 0.3552023857148052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4947545939213866e-05,
+      "loss": 0.9694,
+      "step": 9767
+    },
+    {
+      "epoch": 0.3552387533185438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4946522304547044e-05,
+      "loss": 1.0193,
+      "step": 9768
+    },
+    {
+      "epoch": 0.3552751209222824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4945498601254244e-05,
+      "loss": 1.0815,
+      "step": 9769
+    },
+    {
+      "epoch": 0.355311488526021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.494447482934967e-05,
+      "loss": 0.9986,
+      "step": 9770
+    },
+    {
+      "epoch": 0.3553478561297596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4943450988847522e-05,
+      "loss": 1.1099,
+      "step": 9771
+    },
+    {
+      "epoch": 0.3553842237334982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4942427079762008e-05,
+      "loss": 0.9795,
+      "step": 9772
+    },
+    {
+      "epoch": 0.3554205913372368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4941403102107325e-05,
+      "loss": 1.0694,
+      "step": 9773
+    },
+    {
+      "epoch": 0.3554569589409754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4940379055897689e-05,
+      "loss": 1.006,
+      "step": 9774
+    },
+    {
+      "epoch": 0.355493326544714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4939354941147302e-05,
+      "loss": 1.0956,
+      "step": 9775
+    },
+    {
+      "epoch": 0.35552969414845254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4938330757870372e-05,
+      "loss": 1.0179,
+      "step": 9776
+    },
+    {
+      "epoch": 0.35556606175219113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.493730650608111e-05,
+      "loss": 1.0329,
+      "step": 9777
+    },
+    {
+      "epoch": 0.35560242935592973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4936282185793726e-05,
+      "loss": 1.1098,
+      "step": 9778
+    },
+    {
+      "epoch": 0.3556387969596683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4935257797022431e-05,
+      "loss": 1.0839,
+      "step": 9779
+    },
+    {
+      "epoch": 0.3556751645634069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4934233339781436e-05,
+      "loss": 1.126,
+      "step": 9780
+    },
+    {
+      "epoch": 0.3557115321671455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4933208814084954e-05,
+      "loss": 0.9387,
+      "step": 9781
+    },
+    {
+      "epoch": 0.3557478997708841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4932184219947196e-05,
+      "loss": 1.1107,
+      "step": 9782
+    },
+    {
+      "epoch": 0.3557842673746227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4931159557382387e-05,
+      "loss": 0.9492,
+      "step": 9783
+    },
+    {
+      "epoch": 0.35582063497836125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.493013482640473e-05,
+      "loss": 1.0134,
+      "step": 9784
+    },
+    {
+      "epoch": 0.35585700258209985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4929110027028448e-05,
+      "loss": 1.1207,
+      "step": 9785
+    },
+    {
+      "epoch": 0.35589337018583844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4928085159267758e-05,
+      "loss": 1.0904,
+      "step": 9786
+    },
+    {
+      "epoch": 0.35592973778957704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4927060223136878e-05,
+      "loss": 1.0859,
+      "step": 9787
+    },
+    {
+      "epoch": 0.35596610539331563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.492603521865003e-05,
+      "loss": 0.9354,
+      "step": 9788
+    },
+    {
+      "epoch": 0.35600247299705423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.492501014582143e-05,
+      "loss": 1.1976,
+      "step": 9789
+    },
+    {
+      "epoch": 0.3560388406007928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4923985004665302e-05,
+      "loss": 1.0524,
+      "step": 9790
+    },
+    {
+      "epoch": 0.3560752082045314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.492295979519587e-05,
+      "loss": 1.0495,
+      "step": 9791
+    },
+    {
+      "epoch": 0.35611157580827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4921934517427355e-05,
+      "loss": 1.0238,
+      "step": 9792
+    },
+    {
+      "epoch": 0.35614794341200856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.492090917137398e-05,
+      "loss": 1.1132,
+      "step": 9793
+    },
+    {
+      "epoch": 0.35618431101574716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4919883757049975e-05,
+      "loss": 1.0428,
+      "step": 9794
+    },
+    {
+      "epoch": 0.35622067861948575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4918858274469562e-05,
+      "loss": 1.0403,
+      "step": 9795
+    },
+    {
+      "epoch": 0.35625704622322435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.491783272364697e-05,
+      "loss": 0.9911,
+      "step": 9796
+    },
+    {
+      "epoch": 0.35629341382696295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4916807104596428e-05,
+      "loss": 1.0051,
+      "step": 9797
+    },
+    {
+      "epoch": 0.35632978143070154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4915781417332157e-05,
+      "loss": 1.0944,
+      "step": 9798
+    },
+    {
+      "epoch": 0.35636614903444014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4914755661868401e-05,
+      "loss": 1.0759,
+      "step": 9799
+    },
+    {
+      "epoch": 0.35640251663817873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.491372983821938e-05,
+      "loss": 1.0605,
+      "step": 9800
+    },
+    {
+      "epoch": 0.3564388842419173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4912703946399328e-05,
+      "loss": 1.0698,
+      "step": 9801
+    },
+    {
+      "epoch": 0.35647525184565587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4911677986422483e-05,
+      "loss": 1.0846,
+      "step": 9802
+    },
+    {
+      "epoch": 0.35651161944939447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.491065195830307e-05,
+      "loss": 1.06,
+      "step": 9803
+    },
+    {
+      "epoch": 0.35654798705313306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.490962586205533e-05,
+      "loss": 1.096,
+      "step": 9804
+    },
+    {
+      "epoch": 0.35658435465687166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4908599697693497e-05,
+      "loss": 1.0123,
+      "step": 9805
+    },
+    {
+      "epoch": 0.35662072226061026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4907573465231808e-05,
+      "loss": 0.9509,
+      "step": 9806
+    },
+    {
+      "epoch": 0.35665708986434885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4906547164684502e-05,
+      "loss": 1.12,
+      "step": 9807
+    },
+    {
+      "epoch": 0.35669345746808745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.490552079606581e-05,
+      "loss": 1.0498,
+      "step": 9808
+    },
+    {
+      "epoch": 0.356729825071826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4904494359389982e-05,
+      "loss": 1.0967,
+      "step": 9809
+    },
+    {
+      "epoch": 0.3567661926755646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4903467854671253e-05,
+      "loss": 1.0179,
+      "step": 9810
+    },
+    {
+      "epoch": 0.3568025602793032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4902441281923863e-05,
+      "loss": 1.006,
+      "step": 9811
+    },
+    {
+      "epoch": 0.3568389278830418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4901414641162055e-05,
+      "loss": 1.0458,
+      "step": 9812
+    },
+    {
+      "epoch": 0.3568752954867804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4900387932400079e-05,
+      "loss": 0.923,
+      "step": 9813
+    },
+    {
+      "epoch": 0.35691166309051897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4899361155652165e-05,
+      "loss": 0.9798,
+      "step": 9814
+    },
+    {
+      "epoch": 0.35694803069425757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4898334310932572e-05,
+      "loss": 1.1334,
+      "step": 9815
+    },
+    {
+      "epoch": 0.35698439829799616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4897307398255537e-05,
+      "loss": 1.0843,
+      "step": 9816
+    },
+    {
+      "epoch": 0.35702076590173476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.489628041763531e-05,
+      "loss": 1.1715,
+      "step": 9817
+    },
+    {
+      "epoch": 0.3570571335054733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4895253369086142e-05,
+      "loss": 1.0426,
+      "step": 9818
+    },
+    {
+      "epoch": 0.3570935011092119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4894226252622277e-05,
+      "loss": 0.9997,
+      "step": 9819
+    },
+    {
+      "epoch": 0.3571298687129505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4893199068257966e-05,
+      "loss": 1.0323,
+      "step": 9820
+    },
+    {
+      "epoch": 0.3571662363166891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4892171816007464e-05,
+      "loss": 1.0861,
+      "step": 9821
+    },
+    {
+      "epoch": 0.3572026039204277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4891144495885015e-05,
+      "loss": 1.0596,
+      "step": 9822
+    },
+    {
+      "epoch": 0.3572389715241663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.489011710790488e-05,
+      "loss": 1.037,
+      "step": 9823
+    },
+    {
+      "epoch": 0.3572753391279049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4889089652081304e-05,
+      "loss": 1.123,
+      "step": 9824
+    },
+    {
+      "epoch": 0.3573117067316435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4888062128428548e-05,
+      "loss": 0.902,
+      "step": 9825
+    },
+    {
+      "epoch": 0.357348074335382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4887034536960865e-05,
+      "loss": 1.0422,
+      "step": 9826
+    },
+    {
+      "epoch": 0.3573844419391206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4886006877692512e-05,
+      "loss": 1.0268,
+      "step": 9827
+    },
+    {
+      "epoch": 0.3574208095428592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4884979150637744e-05,
+      "loss": 1.0413,
+      "step": 9828
+    },
+    {
+      "epoch": 0.3574571771465978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4883951355810824e-05,
+      "loss": 1.0619,
+      "step": 9829
+    },
+    {
+      "epoch": 0.3574935447503364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4882923493226005e-05,
+      "loss": 1.057,
+      "step": 9830
+    },
+    {
+      "epoch": 0.357529912354075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4881895562897555e-05,
+      "loss": 1.1101,
+      "step": 9831
+    },
+    {
+      "epoch": 0.3575662799578136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4880867564839725e-05,
+      "loss": 1.055,
+      "step": 9832
+    },
+    {
+      "epoch": 0.3576026475615522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4879839499066786e-05,
+      "loss": 1.0535,
+      "step": 9833
+    },
+    {
+      "epoch": 0.3576390151652908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4878811365592997e-05,
+      "loss": 1.136,
+      "step": 9834
+    },
+    {
+      "epoch": 0.3576753827690293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4877783164432623e-05,
+      "loss": 1.0104,
+      "step": 9835
+    },
+    {
+      "epoch": 0.3577117503727679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4876754895599928e-05,
+      "loss": 1.0499,
+      "step": 9836
+    },
+    {
+      "epoch": 0.3577481179765065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4875726559109177e-05,
+      "loss": 1.0728,
+      "step": 9837
+    },
+    {
+      "epoch": 0.3577844855802451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4874698154974638e-05,
+      "loss": 0.9448,
+      "step": 9838
+    },
+    {
+      "epoch": 0.3578208531839837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4873669683210578e-05,
+      "loss": 1.0926,
+      "step": 9839
+    },
+    {
+      "epoch": 0.3578572207877223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4872641143831271e-05,
+      "loss": 1.1347,
+      "step": 9840
+    },
+    {
+      "epoch": 0.3578935883914609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4871612536850977e-05,
+      "loss": 1.0492,
+      "step": 9841
+    },
+    {
+      "epoch": 0.3579299559951995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4870583862283972e-05,
+      "loss": 1.1183,
+      "step": 9842
+    },
+    {
+      "epoch": 0.35796632359893804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4869555120144523e-05,
+      "loss": 0.9294,
+      "step": 9843
+    },
+    {
+      "epoch": 0.35800269120267664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4868526310446909e-05,
+      "loss": 1.0111,
+      "step": 9844
+    },
+    {
+      "epoch": 0.35803905880641523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.48674974332054e-05,
+      "loss": 1.0697,
+      "step": 9845
+    },
+    {
+      "epoch": 0.35807542641015383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.486646848843427e-05,
+      "loss": 1.0394,
+      "step": 9846
+    },
+    {
+      "epoch": 0.3581117940138924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4865439476147794e-05,
+      "loss": 1.051,
+      "step": 9847
+    },
+    {
+      "epoch": 0.358148161617631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4864410396360247e-05,
+      "loss": 1.1339,
+      "step": 9848
+    },
+    {
+      "epoch": 0.3581845292213696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4863381249085909e-05,
+      "loss": 0.9327,
+      "step": 9849
+    },
+    {
+      "epoch": 0.3582208968251082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4862352034339057e-05,
+      "loss": 1.0068,
+      "step": 9850
+    },
+    {
+      "epoch": 0.35825726442884676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4861322752133966e-05,
+      "loss": 1.0713,
+      "step": 9851
+    },
+    {
+      "epoch": 0.35829363203258535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4860293402484925e-05,
+      "loss": 0.9068,
+      "step": 9852
+    },
+    {
+      "epoch": 0.35832999963632395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4859263985406205e-05,
+      "loss": 0.9533,
+      "step": 9853
+    },
+    {
+      "epoch": 0.35836636724006254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.485823450091209e-05,
+      "loss": 1.1058,
+      "step": 9854
+    },
+    {
+      "epoch": 0.35840273484380114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4857204949016868e-05,
+      "loss": 1.0291,
+      "step": 9855
+    },
+    {
+      "epoch": 0.35843910244753974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4856175329734819e-05,
+      "loss": 1.0897,
+      "step": 9856
+    },
+    {
+      "epoch": 0.35847547005127833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4855145643080222e-05,
+      "loss": 1.1052,
+      "step": 9857
+    },
+    {
+      "epoch": 0.35851183765501693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4854115889067372e-05,
+      "loss": 1.0285,
+      "step": 9858
+    },
+    {
+      "epoch": 0.3585482052587555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4853086067710549e-05,
+      "loss": 1.0476,
+      "step": 9859
+    },
+    {
+      "epoch": 0.35858457286249407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4852056179024046e-05,
+      "loss": 1.1112,
+      "step": 9860
+    },
+    {
+      "epoch": 0.35862094046623266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4851026223022147e-05,
+      "loss": 1.0034,
+      "step": 9861
+    },
+    {
+      "epoch": 0.35865730806997126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4849996199719138e-05,
+      "loss": 0.9229,
+      "step": 9862
+    },
+    {
+      "epoch": 0.35869367567370986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4848966109129318e-05,
+      "loss": 1.0782,
+      "step": 9863
+    },
+    {
+      "epoch": 0.35873004327744845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4847935951266968e-05,
+      "loss": 1.0781,
+      "step": 9864
+    },
+    {
+      "epoch": 0.35876641088118705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4846905726146388e-05,
+      "loss": 1.0305,
+      "step": 9865
+    },
+    {
+      "epoch": 0.35880277848492564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4845875433781869e-05,
+      "loss": 1.07,
+      "step": 9866
+    },
+    {
+      "epoch": 0.35883914608866424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.48448450741877e-05,
+      "loss": 1.1256,
+      "step": 9867
+    },
+    {
+      "epoch": 0.3588755136924028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4843814647378181e-05,
+      "loss": 1.0837,
+      "step": 9868
+    },
+    {
+      "epoch": 0.3589118812961414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4842784153367608e-05,
+      "loss": 1.0831,
+      "step": 9869
+    },
+    {
+      "epoch": 0.35894824889988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4841753592170273e-05,
+      "loss": 1.1241,
+      "step": 9870
+    },
+    {
+      "epoch": 0.35898461650361857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.484072296380048e-05,
+      "loss": 1.0018,
+      "step": 9871
+    },
+    {
+      "epoch": 0.35902098410735717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4839692268272523e-05,
+      "loss": 1.0475,
+      "step": 9872
+    },
+    {
+      "epoch": 0.35905735171109576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.48386615056007e-05,
+      "loss": 1.1271,
+      "step": 9873
+    },
+    {
+      "epoch": 0.35909371931483436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4837630675799315e-05,
+      "loss": 1.1403,
+      "step": 9874
+    },
+    {
+      "epoch": 0.35913008691857295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4836599778882667e-05,
+      "loss": 1.1314,
+      "step": 9875
+    },
+    {
+      "epoch": 0.35916645452231155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4835568814865062e-05,
+      "loss": 1.0922,
+      "step": 9876
+    },
+    {
+      "epoch": 0.3592028221260501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.48345377837608e-05,
+      "loss": 1.0809,
+      "step": 9877
+    },
+    {
+      "epoch": 0.3592391897297887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4833506685584182e-05,
+      "loss": 0.97,
+      "step": 9878
+    },
+    {
+      "epoch": 0.3592755573335273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4832475520349523e-05,
+      "loss": 1.16,
+      "step": 9879
+    },
+    {
+      "epoch": 0.3593119249372659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4831444288071117e-05,
+      "loss": 0.9849,
+      "step": 9880
+    },
+    {
+      "epoch": 0.3593482925410045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4830412988763278e-05,
+      "loss": 1.1107,
+      "step": 9881
+    },
+    {
+      "epoch": 0.3593846601447431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4829381622440315e-05,
+      "loss": 0.9733,
+      "step": 9882
+    },
+    {
+      "epoch": 0.35942102774848167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.482835018911653e-05,
+      "loss": 1.0343,
+      "step": 9883
+    },
+    {
+      "epoch": 0.35945739535222027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4827318688806238e-05,
+      "loss": 1.0567,
+      "step": 9884
+    },
+    {
+      "epoch": 0.3594937629559588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4826287121523748e-05,
+      "loss": 1.0015,
+      "step": 9885
+    },
+    {
+      "epoch": 0.3595301305596974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4825255487283372e-05,
+      "loss": 1.0547,
+      "step": 9886
+    },
+    {
+      "epoch": 0.359566498163436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4824223786099424e-05,
+      "loss": 1.046,
+      "step": 9887
+    },
+    {
+      "epoch": 0.3596028657671746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4823192017986211e-05,
+      "loss": 1.0471,
+      "step": 9888
+    },
+    {
+      "epoch": 0.3596392333709132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4822160182958057e-05,
+      "loss": 1.0884,
+      "step": 9889
+    },
+    {
+      "epoch": 0.3596756009746518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4821128281029272e-05,
+      "loss": 0.9288,
+      "step": 9890
+    },
+    {
+      "epoch": 0.3597119685783904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4820096312214167e-05,
+      "loss": 0.9706,
+      "step": 9891
+    },
+    {
+      "epoch": 0.359748336182129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4819064276527069e-05,
+      "loss": 1.1406,
+      "step": 9892
+    },
+    {
+      "epoch": 0.3597847037858676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.481803217398229e-05,
+      "loss": 1.1076,
+      "step": 9893
+    },
+    {
+      "epoch": 0.3598210713896061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4817000004594151e-05,
+      "loss": 1.2193,
+      "step": 9894
+    },
+    {
+      "epoch": 0.3598574389933447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4815967768376973e-05,
+      "loss": 1.1952,
+      "step": 9895
+    },
+    {
+      "epoch": 0.3598938065970833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4814935465345075e-05,
+      "loss": 1.1522,
+      "step": 9896
+    },
+    {
+      "epoch": 0.3599301742008219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4813903095512774e-05,
+      "loss": 1.0657,
+      "step": 9897
+    },
+    {
+      "epoch": 0.3599665418045605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4812870658894405e-05,
+      "loss": 1.0294,
+      "step": 9898
+    },
+    {
+      "epoch": 0.3600029094082991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4811838155504278e-05,
+      "loss": 1.1132,
+      "step": 9899
+    },
+    {
+      "epoch": 0.3600392770120377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4810805585356727e-05,
+      "loss": 1.1288,
+      "step": 9900
+    },
+    {
+      "epoch": 0.3600756446157763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4809772948466075e-05,
+      "loss": 1.1768,
+      "step": 9901
+    },
+    {
+      "epoch": 0.36011201221951483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4808740244846644e-05,
+      "loss": 0.9713,
+      "step": 9902
+    },
+    {
+      "epoch": 0.36014837982325343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4807707474512766e-05,
+      "loss": 1.0454,
+      "step": 9903
+    },
+    {
+      "epoch": 0.360184747426992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4806674637478769e-05,
+      "loss": 1.1084,
+      "step": 9904
+    },
+    {
+      "epoch": 0.3602211150307306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.480564173375898e-05,
+      "loss": 1.0546,
+      "step": 9905
+    },
+    {
+      "epoch": 0.3602574826344692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4804608763367731e-05,
+      "loss": 1.0732,
+      "step": 9906
+    },
+    {
+      "epoch": 0.3602938502382078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4803575726319353e-05,
+      "loss": 1.1401,
+      "step": 9907
+    },
+    {
+      "epoch": 0.3603302178419464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4802542622628176e-05,
+      "loss": 1.0508,
+      "step": 9908
+    },
+    {
+      "epoch": 0.360366585445685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4801509452308536e-05,
+      "loss": 1.0462,
+      "step": 9909
+    },
+    {
+      "epoch": 0.36040295304942355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4800476215374762e-05,
+      "loss": 1.1529,
+      "step": 9910
+    },
+    {
+      "epoch": 0.36043932065316214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4799442911841192e-05,
+      "loss": 1.0915,
+      "step": 9911
+    },
+    {
+      "epoch": 0.36047568825690074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4798409541722163e-05,
+      "loss": 1.1171,
+      "step": 9912
+    },
+    {
+      "epoch": 0.36051205586063934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4797376105032007e-05,
+      "loss": 0.9939,
+      "step": 9913
+    },
+    {
+      "epoch": 0.36054842346437793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4796342601785068e-05,
+      "loss": 0.9316,
+      "step": 9914
+    },
+    {
+      "epoch": 0.36058479106811653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4795309031995675e-05,
+      "loss": 1.102,
+      "step": 9915
+    },
+    {
+      "epoch": 0.3606211586718551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4794275395678177e-05,
+      "loss": 1.0642,
+      "step": 9916
+    },
+    {
+      "epoch": 0.3606575262755937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4793241692846911e-05,
+      "loss": 1.0475,
+      "step": 9917
+    },
+    {
+      "epoch": 0.3606938938793323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4792207923516215e-05,
+      "loss": 1.0567,
+      "step": 9918
+    },
+    {
+      "epoch": 0.36073026148307086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4791174087700433e-05,
+      "loss": 1.0371,
+      "step": 9919
+    },
+    {
+      "epoch": 0.36076662908680945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4790140185413908e-05,
+      "loss": 0.9869,
+      "step": 9920
+    },
+    {
+      "epoch": 0.36080299669054805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4789106216670987e-05,
+      "loss": 1.0365,
+      "step": 9921
+    },
+    {
+      "epoch": 0.36083936429428665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4788072181486011e-05,
+      "loss": 1.1031,
+      "step": 9922
+    },
+    {
+      "epoch": 0.36087573189802524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4787038079873323e-05,
+      "loss": 1.1078,
+      "step": 9923
+    },
+    {
+      "epoch": 0.36091209950176384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.478600391184728e-05,
+      "loss": 1.1207,
+      "step": 9924
+    },
+    {
+      "epoch": 0.36094846710550244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4784969677422218e-05,
+      "loss": 1.1153,
+      "step": 9925
+    },
+    {
+      "epoch": 0.36098483470924103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4783935376612493e-05,
+      "loss": 1.0352,
+      "step": 9926
+    },
+    {
+      "epoch": 0.36102120231297957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4782901009432454e-05,
+      "loss": 0.9773,
+      "step": 9927
+    },
+    {
+      "epoch": 0.36105756991671817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4781866575896446e-05,
+      "loss": 1.033,
+      "step": 9928
+    },
+    {
+      "epoch": 0.36109393752045676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4780832076018824e-05,
+      "loss": 1.0794,
+      "step": 9929
+    },
+    {
+      "epoch": 0.36113030512419536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4779797509813942e-05,
+      "loss": 1.0375,
+      "step": 9930
+    },
+    {
+      "epoch": 0.36116667272793396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.477876287729615e-05,
+      "loss": 1.0609,
+      "step": 9931
+    },
+    {
+      "epoch": 0.36120304033167255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4777728178479804e-05,
+      "loss": 1.1063,
+      "step": 9932
+    },
+    {
+      "epoch": 0.36123940793541115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4776693413379257e-05,
+      "loss": 1.2128,
+      "step": 9933
+    },
+    {
+      "epoch": 0.36127577553914975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4775658582008867e-05,
+      "loss": 1.0307,
+      "step": 9934
+    },
+    {
+      "epoch": 0.36131214314288834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.477462368438299e-05,
+      "loss": 0.9973,
+      "step": 9935
+    },
+    {
+      "epoch": 0.3613485107466269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4773588720515984e-05,
+      "loss": 0.9457,
+      "step": 9936
+    },
+    {
+      "epoch": 0.3613848783503655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.477255369042221e-05,
+      "loss": 0.9463,
+      "step": 9937
+    },
+    {
+      "epoch": 0.3614212459541041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4771518594116021e-05,
+      "loss": 1.0225,
+      "step": 9938
+    },
+    {
+      "epoch": 0.36145761355784267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4770483431611781e-05,
+      "loss": 1.1159,
+      "step": 9939
+    },
+    {
+      "epoch": 0.36149398116158127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4769448202923853e-05,
+      "loss": 1.0589,
+      "step": 9940
+    },
+    {
+      "epoch": 0.36153034876531986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4768412908066601e-05,
+      "loss": 1.0037,
+      "step": 9941
+    },
+    {
+      "epoch": 0.36156671636905846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.476737754705438e-05,
+      "loss": 0.9682,
+      "step": 9942
+    },
+    {
+      "epoch": 0.36160308397279706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4766342119901562e-05,
+      "loss": 1.099,
+      "step": 9943
+    },
+    {
+      "epoch": 0.3616394515765356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4765306626622512e-05,
+      "loss": 0.9607,
+      "step": 9944
+    },
+    {
+      "epoch": 0.3616758191802742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4764271067231591e-05,
+      "loss": 1.0853,
+      "step": 9945
+    },
+    {
+      "epoch": 0.3617121867840128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.476323544174317e-05,
+      "loss": 1.0243,
+      "step": 9946
+    },
+    {
+      "epoch": 0.3617485543877514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.476219975017161e-05,
+      "loss": 0.9552,
+      "step": 9947
+    },
+    {
+      "epoch": 0.36178492199149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.476116399253129e-05,
+      "loss": 1.1105,
+      "step": 9948
+    },
+    {
+      "epoch": 0.3618212895952286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4760128168836576e-05,
+      "loss": 0.9145,
+      "step": 9949
+    },
+    {
+      "epoch": 0.3618576571989672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4759092279101834e-05,
+      "loss": 0.9886,
+      "step": 9950
+    },
+    {
+      "epoch": 0.36189402480270577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4758056323341443e-05,
+      "loss": 1.2296,
+      "step": 9951
+    },
+    {
+      "epoch": 0.3619303924064443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.475702030156977e-05,
+      "loss": 0.9674,
+      "step": 9952
+    },
+    {
+      "epoch": 0.3619667600101829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4755984213801187e-05,
+      "loss": 1.118,
+      "step": 9953
+    },
+    {
+      "epoch": 0.3620031276139215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4754948060050072e-05,
+      "loss": 0.9946,
+      "step": 9954
+    },
+    {
+      "epoch": 0.3620394952176601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4753911840330802e-05,
+      "loss": 1.12,
+      "step": 9955
+    },
+    {
+      "epoch": 0.3620758628213987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4752875554657746e-05,
+      "loss": 1.0388,
+      "step": 9956
+    },
+    {
+      "epoch": 0.3621122304251373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4751839203045288e-05,
+      "loss": 0.9774,
+      "step": 9957
+    },
+    {
+      "epoch": 0.3621485980288759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4750802785507801e-05,
+      "loss": 1.0507,
+      "step": 9958
+    },
+    {
+      "epoch": 0.3621849656326145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.474976630205967e-05,
+      "loss": 1.1244,
+      "step": 9959
+    },
+    {
+      "epoch": 0.3622213332363531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4748729752715264e-05,
+      "loss": 0.9333,
+      "step": 9960
+    },
+    {
+      "epoch": 0.3622577008400916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4747693137488974e-05,
+      "loss": 1.0466,
+      "step": 9961
+    },
+    {
+      "epoch": 0.3622940684438302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.474665645639518e-05,
+      "loss": 1.1171,
+      "step": 9962
+    },
+    {
+      "epoch": 0.3623304360475688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4745619709448256e-05,
+      "loss": 1.1168,
+      "step": 9963
+    },
+    {
+      "epoch": 0.3623668036513074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4744582896662598e-05,
+      "loss": 1.0424,
+      "step": 9964
+    },
+    {
+      "epoch": 0.362403171255046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.474354601805258e-05,
+      "loss": 1.1577,
+      "step": 9965
+    },
+    {
+      "epoch": 0.3624395388587846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4742509073632594e-05,
+      "loss": 1.004,
+      "step": 9966
+    },
+    {
+      "epoch": 0.3624759064625232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.474147206341702e-05,
+      "loss": 0.9374,
+      "step": 9967
+    },
+    {
+      "epoch": 0.3625122740662618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.474043498742025e-05,
+      "loss": 1.0593,
+      "step": 9968
+    },
+    {
+      "epoch": 0.36254864167000034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4739397845656671e-05,
+      "loss": 0.991,
+      "step": 9969
+    },
+    {
+      "epoch": 0.36258500927373893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4738360638140672e-05,
+      "loss": 1.1226,
+      "step": 9970
+    },
+    {
+      "epoch": 0.36262137687747753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.473732336488664e-05,
+      "loss": 0.9917,
+      "step": 9971
+    },
+    {
+      "epoch": 0.3626577444812161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.473628602590897e-05,
+      "loss": 1.1505,
+      "step": 9972
+    },
+    {
+      "epoch": 0.3626941120849547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.473524862122205e-05,
+      "loss": 0.9417,
+      "step": 9973
+    },
+    {
+      "epoch": 0.3627304796886933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4734211150840271e-05,
+      "loss": 1.0204,
+      "step": 9974
+    },
+    {
+      "epoch": 0.3627668472924319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4733173614778035e-05,
+      "loss": 1.0392,
+      "step": 9975
+    },
+    {
+      "epoch": 0.3628032148961705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4732136013049728e-05,
+      "loss": 1.1571,
+      "step": 9976
+    },
+    {
+      "epoch": 0.3628395824999091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4731098345669748e-05,
+      "loss": 1.1365,
+      "step": 9977
+    },
+    {
+      "epoch": 0.36287595010364765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.473006061265249e-05,
+      "loss": 0.9503,
+      "step": 9978
+    },
+    {
+      "epoch": 0.36291231770738624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.472902281401235e-05,
+      "loss": 1.0882,
+      "step": 9979
+    },
+    {
+      "epoch": 0.36294868531112484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4727984949763732e-05,
+      "loss": 0.9622,
+      "step": 9980
+    },
+    {
+      "epoch": 0.36298505291486344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4726947019921029e-05,
+      "loss": 0.9503,
+      "step": 9981
+    },
+    {
+      "epoch": 0.36302142051860203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4725909024498643e-05,
+      "loss": 1.0093,
+      "step": 9982
+    },
+    {
+      "epoch": 0.36305778812234063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4724870963510975e-05,
+      "loss": 1.072,
+      "step": 9983
+    },
+    {
+      "epoch": 0.3630941557260792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4723832836972427e-05,
+      "loss": 0.931,
+      "step": 9984
+    },
+    {
+      "epoch": 0.3631305233298178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4722794644897399e-05,
+      "loss": 0.9253,
+      "step": 9985
+    },
+    {
+      "epoch": 0.36316689093355636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4721756387300297e-05,
+      "loss": 1.2171,
+      "step": 9986
+    },
+    {
+      "epoch": 0.36320325853729496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4720718064195523e-05,
+      "loss": 0.9966,
+      "step": 9987
+    },
+    {
+      "epoch": 0.36323962614103356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4719679675597485e-05,
+      "loss": 0.9623,
+      "step": 9988
+    },
+    {
+      "epoch": 0.36327599374477215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4718641221520588e-05,
+      "loss": 0.9612,
+      "step": 9989
+    },
+    {
+      "epoch": 0.36331236134851075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4717602701979236e-05,
+      "loss": 1.0072,
+      "step": 9990
+    },
+    {
+      "epoch": 0.36334872895224934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4716564116987844e-05,
+      "loss": 1.1159,
+      "step": 9991
+    },
+    {
+      "epoch": 0.36338509655598794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4715525466560816e-05,
+      "loss": 0.9727,
+      "step": 9992
+    },
+    {
+      "epoch": 0.36342146415972654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.471448675071256e-05,
+      "loss": 1.0445,
+      "step": 9993
+    },
+    {
+      "epoch": 0.3634578317634651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4713447969457493e-05,
+      "loss": 1.0645,
+      "step": 9994
+    },
+    {
+      "epoch": 0.3634941993672037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.471240912281002e-05,
+      "loss": 1.2202,
+      "step": 9995
+    },
+    {
+      "epoch": 0.36353056697094227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.471137021078456e-05,
+      "loss": 1.0367,
+      "step": 9996
+    },
+    {
+      "epoch": 0.36356693457468087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.471033123339552e-05,
+      "loss": 0.992,
+      "step": 9997
+    },
+    {
+      "epoch": 0.36360330217841946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4709292190657316e-05,
+      "loss": 1.1252,
+      "step": 9998
+    },
+    {
+      "epoch": 0.36363966978215806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4708253082584366e-05,
+      "loss": 1.0884,
+      "step": 9999
+    },
+    {
+      "epoch": 0.36367603738589666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4707213909191085e-05,
+      "loss": 1.1943,
+      "step": 10000
+    },
+    {
+      "epoch": 0.36371240498963525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4706174670491888e-05,
+      "loss": 1.0031,
+      "step": 10001
+    },
+    {
+      "epoch": 0.36374877259337385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4705135366501197e-05,
+      "loss": 0.9596,
+      "step": 10002
+    },
+    {
+      "epoch": 0.3637851401971124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4704095997233427e-05,
+      "loss": 1.1547,
+      "step": 10003
+    },
+    {
+      "epoch": 0.363821507800851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4703056562703e-05,
+      "loss": 1.0504,
+      "step": 10004
+    },
+    {
+      "epoch": 0.3638578754045896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4702017062924337e-05,
+      "loss": 0.9815,
+      "step": 10005
+    },
+    {
+      "epoch": 0.3638942430083282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4700977497911854e-05,
+      "loss": 0.9641,
+      "step": 10006
+    },
+    {
+      "epoch": 0.3639306106120668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4699937867679983e-05,
+      "loss": 1.068,
+      "step": 10007
+    },
+    {
+      "epoch": 0.36396697821580537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.469889817224314e-05,
+      "loss": 1.0303,
+      "step": 10008
+    },
+    {
+      "epoch": 0.36400334581954397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4697858411615753e-05,
+      "loss": 1.093,
+      "step": 10009
+    },
+    {
+      "epoch": 0.36403971342328256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4696818585812244e-05,
+      "loss": 1.0883,
+      "step": 10010
+    },
+    {
+      "epoch": 0.3640760810270211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4695778694847045e-05,
+      "loss": 0.9134,
+      "step": 10011
+    },
+    {
+      "epoch": 0.3641124486307597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4694738738734575e-05,
+      "loss": 0.9833,
+      "step": 10012
+    },
+    {
+      "epoch": 0.3641488162344983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4693698717489269e-05,
+      "loss": 1.0411,
+      "step": 10013
+    },
+    {
+      "epoch": 0.3641851838382369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.469265863112555e-05,
+      "loss": 0.9352,
+      "step": 10014
+    },
+    {
+      "epoch": 0.3642215514419755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4691618479657854e-05,
+      "loss": 0.9115,
+      "step": 10015
+    },
+    {
+      "epoch": 0.3642579190457141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4690578263100606e-05,
+      "loss": 1.0811,
+      "step": 10016
+    },
+    {
+      "epoch": 0.3642942866494527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4689537981468238e-05,
+      "loss": 1.1861,
+      "step": 10017
+    },
+    {
+      "epoch": 0.3643306542531913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4688497634775185e-05,
+      "loss": 1.097,
+      "step": 10018
+    },
+    {
+      "epoch": 0.3643670218569299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.468745722303588e-05,
+      "loss": 1.1773,
+      "step": 10019
+    },
+    {
+      "epoch": 0.3644033894606684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4686416746264755e-05,
+      "loss": 1.1772,
+      "step": 10020
+    },
+    {
+      "epoch": 0.364439757064407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4685376204476252e-05,
+      "loss": 0.953,
+      "step": 10021
+    },
+    {
+      "epoch": 0.3644761246681456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4684335597684796e-05,
+      "loss": 0.9637,
+      "step": 10022
+    },
+    {
+      "epoch": 0.3645124922718842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4683294925904833e-05,
+      "loss": 1.0436,
+      "step": 10023
+    },
+    {
+      "epoch": 0.3645488598756228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4682254189150798e-05,
+      "loss": 0.9712,
+      "step": 10024
+    },
+    {
+      "epoch": 0.3645852274793614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4681213387437125e-05,
+      "loss": 0.9618,
+      "step": 10025
+    },
+    {
+      "epoch": 0.3646215950831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4680172520778262e-05,
+      "loss": 0.9349,
+      "step": 10026
+    },
+    {
+      "epoch": 0.3646579626868386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4679131589188647e-05,
+      "loss": 0.9617,
+      "step": 10027
+    },
+    {
+      "epoch": 0.36469433029057713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4678090592682714e-05,
+      "loss": 0.9901,
+      "step": 10028
+    },
+    {
+      "epoch": 0.3647306978943157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4677049531274917e-05,
+      "loss": 1.1413,
+      "step": 10029
+    },
+    {
+      "epoch": 0.3647670654980543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4676008404979691e-05,
+      "loss": 0.9394,
+      "step": 10030
+    },
+    {
+      "epoch": 0.3648034331017929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4674967213811484e-05,
+      "loss": 0.9284,
+      "step": 10031
+    },
+    {
+      "epoch": 0.3648398007055315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.467392595778474e-05,
+      "loss": 1.0384,
+      "step": 10032
+    },
+    {
+      "epoch": 0.3648761683092701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4672884636913905e-05,
+      "loss": 1.127,
+      "step": 10033
+    },
+    {
+      "epoch": 0.3649125359130087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4671843251213425e-05,
+      "loss": 1.1158,
+      "step": 10034
+    },
+    {
+      "epoch": 0.3649489035167473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4670801800697744e-05,
+      "loss": 1.0596,
+      "step": 10035
+    },
+    {
+      "epoch": 0.36498527112048584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4669760285381323e-05,
+      "loss": 1.033,
+      "step": 10036
+    },
+    {
+      "epoch": 0.36502163872422444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4668718705278599e-05,
+      "loss": 1.0597,
+      "step": 10037
+    },
+    {
+      "epoch": 0.36505800632796304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4667677060404026e-05,
+      "loss": 1.0244,
+      "step": 10038
+    },
+    {
+      "epoch": 0.36509437393170163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4666635350772058e-05,
+      "loss": 0.9938,
+      "step": 10039
+    },
+    {
+      "epoch": 0.36513074153544023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4665593576397145e-05,
+      "loss": 0.9888,
+      "step": 10040
+    },
+    {
+      "epoch": 0.3651671091391788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4664551737293742e-05,
+      "loss": 1.1035,
+      "step": 10041
+    },
+    {
+      "epoch": 0.3652034767429174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4663509833476301e-05,
+      "loss": 1.022,
+      "step": 10042
+    },
+    {
+      "epoch": 0.365239844346656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4662467864959278e-05,
+      "loss": 0.9807,
+      "step": 10043
+    },
+    {
+      "epoch": 0.3652762119503946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4661425831757128e-05,
+      "loss": 1.0158,
+      "step": 10044
+    },
+    {
+      "epoch": 0.36531257955413315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4660383733884309e-05,
+      "loss": 1.0698,
+      "step": 10045
+    },
+    {
+      "epoch": 0.36534894715787175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4659341571355277e-05,
+      "loss": 1.1578,
+      "step": 10046
+    },
+    {
+      "epoch": 0.36538531476161035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4658299344184491e-05,
+      "loss": 1.0743,
+      "step": 10047
+    },
+    {
+      "epoch": 0.36542168236534894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4657257052386413e-05,
+      "loss": 1.0248,
+      "step": 10048
+    },
+    {
+      "epoch": 0.36545804996908754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4656214695975499e-05,
+      "loss": 1.143,
+      "step": 10049
+    },
+    {
+      "epoch": 0.36549441757282614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4655172274966214e-05,
+      "loss": 1.0538,
+      "step": 10050
+    },
+    {
+      "epoch": 0.36553078517656473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.465412978937302e-05,
+      "loss": 0.9943,
+      "step": 10051
+    },
+    {
+      "epoch": 0.36556715278030333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4653087239210378e-05,
+      "loss": 1.0683,
+      "step": 10052
+    },
+    {
+      "epoch": 0.36560352038404187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.465204462449275e-05,
+      "loss": 1.0635,
+      "step": 10053
+    },
+    {
+      "epoch": 0.36563988798778047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4651001945234602e-05,
+      "loss": 1.0252,
+      "step": 10054
+    },
+    {
+      "epoch": 0.36567625559151906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4649959201450403e-05,
+      "loss": 1.0292,
+      "step": 10055
+    },
+    {
+      "epoch": 0.36571262319525766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.464891639315462e-05,
+      "loss": 0.9747,
+      "step": 10056
+    },
+    {
+      "epoch": 0.36574899079899625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4647873520361713e-05,
+      "loss": 1.0502,
+      "step": 10057
+    },
+    {
+      "epoch": 0.36578535840273485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.464683058308616e-05,
+      "loss": 1.131,
+      "step": 10058
+    },
+    {
+      "epoch": 0.36582172600647345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4645787581342421e-05,
+      "loss": 1.0663,
+      "step": 10059
+    },
+    {
+      "epoch": 0.36585809361021204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4644744515144975e-05,
+      "loss": 0.9591,
+      "step": 10060
+    },
+    {
+      "epoch": 0.36589446121395064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4643701384508288e-05,
+      "loss": 1.1482,
+      "step": 10061
+    },
+    {
+      "epoch": 0.3659308288176892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.464265818944683e-05,
+      "loss": 1.0186,
+      "step": 10062
+    },
+    {
+      "epoch": 0.3659671964214278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4641614929975081e-05,
+      "loss": 1.0266,
+      "step": 10063
+    },
+    {
+      "epoch": 0.3660035640251664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4640571606107508e-05,
+      "loss": 1.0223,
+      "step": 10064
+    },
+    {
+      "epoch": 0.36603993162890497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4639528217858588e-05,
+      "loss": 1.0504,
+      "step": 10065
+    },
+    {
+      "epoch": 0.36607629923264356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4638484765242797e-05,
+      "loss": 1.0961,
+      "step": 10066
+    },
+    {
+      "epoch": 0.36611266683638216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.463744124827461e-05,
+      "loss": 1.082,
+      "step": 10067
+    },
+    {
+      "epoch": 0.36614903444012076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4636397666968508e-05,
+      "loss": 1.013,
+      "step": 10068
+    },
+    {
+      "epoch": 0.36618540204385935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4635354021338964e-05,
+      "loss": 1.0366,
+      "step": 10069
+    },
+    {
+      "epoch": 0.3662217696475979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4634310311400457e-05,
+      "loss": 1.0992,
+      "step": 10070
+    },
+    {
+      "epoch": 0.3662581372513365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4633266537167476e-05,
+      "loss": 1.1339,
+      "step": 10071
+    },
+    {
+      "epoch": 0.3662945048550751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4632222698654495e-05,
+      "loss": 0.9206,
+      "step": 10072
+    },
+    {
+      "epoch": 0.3663308724588137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4631178795875994e-05,
+      "loss": 1.0092,
+      "step": 10073
+    },
+    {
+      "epoch": 0.3663672400625523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4630134828846456e-05,
+      "loss": 1.167,
+      "step": 10074
+    },
+    {
+      "epoch": 0.3664036076662909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4629090797580368e-05,
+      "loss": 0.9283,
+      "step": 10075
+    },
+    {
+      "epoch": 0.36643997527002947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4628046702092214e-05,
+      "loss": 0.9109,
+      "step": 10076
+    },
+    {
+      "epoch": 0.36647634287376807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.462700254239648e-05,
+      "loss": 1.0616,
+      "step": 10077
+    },
+    {
+      "epoch": 0.36651271047750666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4625958318507646e-05,
+      "loss": 0.8329,
+      "step": 10078
+    },
+    {
+      "epoch": 0.3665490780812452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4624914030440207e-05,
+      "loss": 1.2462,
+      "step": 10079
+    },
+    {
+      "epoch": 0.3665854456849838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4623869678208648e-05,
+      "loss": 1.0698,
+      "step": 10080
+    },
+    {
+      "epoch": 0.3666218132887224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4622825261827456e-05,
+      "loss": 1.026,
+      "step": 10081
+    },
+    {
+      "epoch": 0.366658180892461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4621780781311122e-05,
+      "loss": 1.0832,
+      "step": 10082
+    },
+    {
+      "epoch": 0.3666945484961996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.462073623667414e-05,
+      "loss": 1.0648,
+      "step": 10083
+    },
+    {
+      "epoch": 0.3667309160999382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4619691627930994e-05,
+      "loss": 1.0509,
+      "step": 10084
+    },
+    {
+      "epoch": 0.3667672837036768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4618646955096186e-05,
+      "loss": 0.9797,
+      "step": 10085
+    },
+    {
+      "epoch": 0.3668036513074154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.46176022181842e-05,
+      "loss": 1.102,
+      "step": 10086
+    },
+    {
+      "epoch": 0.3668400189111539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4616557417209538e-05,
+      "loss": 1.0733,
+      "step": 10087
+    },
+    {
+      "epoch": 0.3668763865148925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4615512552186693e-05,
+      "loss": 1.0991,
+      "step": 10088
+    },
+    {
+      "epoch": 0.3669127541186311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4614467623130155e-05,
+      "loss": 1.1159,
+      "step": 10089
+    },
+    {
+      "epoch": 0.3669491217223697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4613422630054433e-05,
+      "loss": 1.087,
+      "step": 10090
+    },
+    {
+      "epoch": 0.3669854893261083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.461237757297401e-05,
+      "loss": 0.899,
+      "step": 10091
+    },
+    {
+      "epoch": 0.3670218569298469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4611332451903396e-05,
+      "loss": 1.1872,
+      "step": 10092
+    },
+    {
+      "epoch": 0.3670582245335855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4610287266857089e-05,
+      "loss": 0.9607,
+      "step": 10093
+    },
+    {
+      "epoch": 0.3670945921373241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4609242017849583e-05,
+      "loss": 1.1217,
+      "step": 10094
+    },
+    {
+      "epoch": 0.36713095974106263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4608196704895385e-05,
+      "loss": 1.1299,
+      "step": 10095
+    },
+    {
+      "epoch": 0.36716732734480123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4607151328008996e-05,
+      "loss": 1.0042,
+      "step": 10096
+    },
+    {
+      "epoch": 0.3672036949485398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.460610588720492e-05,
+      "loss": 0.9014,
+      "step": 10097
+    },
+    {
+      "epoch": 0.3672400625522784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.460506038249766e-05,
+      "loss": 1.0644,
+      "step": 10098
+    },
+    {
+      "epoch": 0.367276430156017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.460401481390172e-05,
+      "loss": 1.0352,
+      "step": 10099
+    },
+    {
+      "epoch": 0.3673127977597556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4602969181431607e-05,
+      "loss": 1.0651,
+      "step": 10100
+    },
+    {
+      "epoch": 0.3673491653634942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4601923485101828e-05,
+      "loss": 1.0378,
+      "step": 10101
+    },
+    {
+      "epoch": 0.3673855329672328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4600877724926888e-05,
+      "loss": 1.0435,
+      "step": 10102
+    },
+    {
+      "epoch": 0.3674219005709714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.45998319009213e-05,
+      "loss": 0.9702,
+      "step": 10103
+    },
+    {
+      "epoch": 0.36745826817470995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4598786013099568e-05,
+      "loss": 0.9651,
+      "step": 10104
+    },
+    {
+      "epoch": 0.36749463577844854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4597740061476207e-05,
+      "loss": 1.0091,
+      "step": 10105
+    },
+    {
+      "epoch": 0.36753100338218714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4596694046065725e-05,
+      "loss": 1.0547,
+      "step": 10106
+    },
+    {
+      "epoch": 0.36756737098592573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4595647966882636e-05,
+      "loss": 0.9988,
+      "step": 10107
+    },
+    {
+      "epoch": 0.36760373858966433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4594601823941447e-05,
+      "loss": 0.9364,
+      "step": 10108
+    },
+    {
+      "epoch": 0.3676401061934029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4593555617256685e-05,
+      "loss": 0.949,
+      "step": 10109
+    },
+    {
+      "epoch": 0.3676764737971415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.459250934684285e-05,
+      "loss": 1.0485,
+      "step": 10110
+    },
+    {
+      "epoch": 0.3677128414008801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4591463012714466e-05,
+      "loss": 0.9747,
+      "step": 10111
+    },
+    {
+      "epoch": 0.36774920900461866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4590416614886049e-05,
+      "loss": 0.9698,
+      "step": 10112
+    },
+    {
+      "epoch": 0.36778557660835726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.458937015337211e-05,
+      "loss": 0.9926,
+      "step": 10113
+    },
+    {
+      "epoch": 0.36782194421209585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4588323628187176e-05,
+      "loss": 1.0029,
+      "step": 10114
+    },
+    {
+      "epoch": 0.36785831181583445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4587277039345757e-05,
+      "loss": 1.0899,
+      "step": 10115
+    },
+    {
+      "epoch": 0.36789467941957305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4586230386862381e-05,
+      "loss": 0.9905,
+      "step": 10116
+    },
+    {
+      "epoch": 0.36793104702331164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4585183670751563e-05,
+      "loss": 0.9914,
+      "step": 10117
+    },
+    {
+      "epoch": 0.36796741462705024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4584136891027827e-05,
+      "loss": 1.0635,
+      "step": 10118
+    },
+    {
+      "epoch": 0.36800378223078883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4583090047705699e-05,
+      "loss": 0.9604,
+      "step": 10119
+    },
+    {
+      "epoch": 0.36804014983452743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4582043140799698e-05,
+      "loss": 0.9821,
+      "step": 10120
+    },
+    {
+      "epoch": 0.36807651743826597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4580996170324347e-05,
+      "loss": 1.0443,
+      "step": 10121
+    },
+    {
+      "epoch": 0.36811288504200457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4579949136294176e-05,
+      "loss": 1.1298,
+      "step": 10122
+    },
+    {
+      "epoch": 0.36814925264574316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4578902038723708e-05,
+      "loss": 1.0229,
+      "step": 10123
+    },
+    {
+      "epoch": 0.36818562024948176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4577854877627471e-05,
+      "loss": 1.0138,
+      "step": 10124
+    },
+    {
+      "epoch": 0.36822198785322036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4576807653019993e-05,
+      "loss": 1.0413,
+      "step": 10125
+    },
+    {
+      "epoch": 0.36825835545695895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4575760364915801e-05,
+      "loss": 1.0312,
+      "step": 10126
+    },
+    {
+      "epoch": 0.36829472306069755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4574713013329427e-05,
+      "loss": 1.1069,
+      "step": 10127
+    },
+    {
+      "epoch": 0.36833109066443614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4573665598275403e-05,
+      "loss": 1.0596,
+      "step": 10128
+    },
+    {
+      "epoch": 0.3683674582681747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4572618119768256e-05,
+      "loss": 1.0423,
+      "step": 10129
+    },
+    {
+      "epoch": 0.3684038258719133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.457157057782252e-05,
+      "loss": 1.0417,
+      "step": 10130
+    },
+    {
+      "epoch": 0.3684401934756519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4570522972452727e-05,
+      "loss": 0.8721,
+      "step": 10131
+    },
+    {
+      "epoch": 0.3684765610793905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4569475303673416e-05,
+      "loss": 1.0133,
+      "step": 10132
+    },
+    {
+      "epoch": 0.36851292868312907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4568427571499118e-05,
+      "loss": 1.0451,
+      "step": 10133
+    },
+    {
+      "epoch": 0.36854929628686767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4567379775944365e-05,
+      "loss": 1.2389,
+      "step": 10134
+    },
+    {
+      "epoch": 0.36858566389060626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4566331917023701e-05,
+      "loss": 0.9722,
+      "step": 10135
+    },
+    {
+      "epoch": 0.36862203149434486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.456528399475166e-05,
+      "loss": 1.0963,
+      "step": 10136
+    },
+    {
+      "epoch": 0.3686583990980834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4564236009142781e-05,
+      "loss": 1.0038,
+      "step": 10137
+    },
+    {
+      "epoch": 0.368694766701822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4563187960211602e-05,
+      "loss": 1.2043,
+      "step": 10138
+    },
+    {
+      "epoch": 0.3687311343055606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4562139847972667e-05,
+      "loss": 1.0737,
+      "step": 10139
+    },
+    {
+      "epoch": 0.3687675019092992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4561091672440513e-05,
+      "loss": 0.9289,
+      "step": 10140
+    },
+    {
+      "epoch": 0.3688038695130378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4560043433629683e-05,
+      "loss": 1.0269,
+      "step": 10141
+    },
+    {
+      "epoch": 0.3688402371167764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4558995131554719e-05,
+      "loss": 1.0973,
+      "step": 10142
+    },
+    {
+      "epoch": 0.368876604720515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4557946766230169e-05,
+      "loss": 0.9999,
+      "step": 10143
+    },
+    {
+      "epoch": 0.3689129723242536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4556898337670573e-05,
+      "loss": 1.0073,
+      "step": 10144
+    },
+    {
+      "epoch": 0.36894933992799217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4555849845890477e-05,
+      "loss": 1.1075,
+      "step": 10145
+    },
+    {
+      "epoch": 0.3689857075317307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.455480129090443e-05,
+      "loss": 1.1535,
+      "step": 10146
+    },
+    {
+      "epoch": 0.3690220751354693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4553752672726975e-05,
+      "loss": 1.1067,
+      "step": 10147
+    },
+    {
+      "epoch": 0.3690584427392079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4552703991372666e-05,
+      "loss": 1.1344,
+      "step": 10148
+    },
+    {
+      "epoch": 0.3690948103429465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4551655246856048e-05,
+      "loss": 1.009,
+      "step": 10149
+    },
+    {
+      "epoch": 0.3691311779466851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4550606439191668e-05,
+      "loss": 1.1323,
+      "step": 10150
+    },
+    {
+      "epoch": 0.3691675455504237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4549557568394084e-05,
+      "loss": 1.217,
+      "step": 10151
+    },
+    {
+      "epoch": 0.3692039131541623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4548508634477842e-05,
+      "loss": 1.1247,
+      "step": 10152
+    },
+    {
+      "epoch": 0.3692402807579009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4547459637457494e-05,
+      "loss": 1.0109,
+      "step": 10153
+    },
+    {
+      "epoch": 0.3692766483616394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4546410577347599e-05,
+      "loss": 1.1804,
+      "step": 10154
+    },
+    {
+      "epoch": 0.369313015965378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4545361454162708e-05,
+      "loss": 0.9511,
+      "step": 10155
+    },
+    {
+      "epoch": 0.3693493835691166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4544312267917373e-05,
+      "loss": 1.0376,
+      "step": 10156
+    },
+    {
+      "epoch": 0.3693857511728552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4543263018626155e-05,
+      "loss": 0.9532,
+      "step": 10157
+    },
+    {
+      "epoch": 0.3694221187765938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4542213706303607e-05,
+      "loss": 1.0732,
+      "step": 10158
+    },
+    {
+      "epoch": 0.3694584863803324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4541164330964291e-05,
+      "loss": 1.0627,
+      "step": 10159
+    },
+    {
+      "epoch": 0.369494853984071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.454011489262276e-05,
+      "loss": 0.9729,
+      "step": 10160
+    },
+    {
+      "epoch": 0.3695312215878096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4539065391293577e-05,
+      "loss": 1.0636,
+      "step": 10161
+    },
+    {
+      "epoch": 0.3695675891915482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4538015826991302e-05,
+      "loss": 1.0213,
+      "step": 10162
+    },
+    {
+      "epoch": 0.36960395679528674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4536966199730498e-05,
+      "loss": 1.1588,
+      "step": 10163
+    },
+    {
+      "epoch": 0.36964032439902533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4535916509525723e-05,
+      "loss": 1.154,
+      "step": 10164
+    },
+    {
+      "epoch": 0.36967669200276393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.453486675639154e-05,
+      "loss": 0.9681,
+      "step": 10165
+    },
+    {
+      "epoch": 0.3697130596065025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.453381694034252e-05,
+      "loss": 1.0862,
+      "step": 10166
+    },
+    {
+      "epoch": 0.3697494272102411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.453276706139322e-05,
+      "loss": 1.0814,
+      "step": 10167
+    },
+    {
+      "epoch": 0.3697857948139797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.453171711955821e-05,
+      "loss": 1.0443,
+      "step": 10168
+    },
+    {
+      "epoch": 0.3698221624177183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4530667114852051e-05,
+      "loss": 1.0067,
+      "step": 10169
+    },
+    {
+      "epoch": 0.3698585300214569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4529617047289318e-05,
+      "loss": 1.1746,
+      "step": 10170
+    },
+    {
+      "epoch": 0.36989489762519545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4528566916884572e-05,
+      "loss": 0.912,
+      "step": 10171
+    },
+    {
+      "epoch": 0.36993126522893405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4527516723652385e-05,
+      "loss": 0.9847,
+      "step": 10172
+    },
+    {
+      "epoch": 0.36996763283267264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.452646646760733e-05,
+      "loss": 1.0321,
+      "step": 10173
+    },
+    {
+      "epoch": 0.37000400043641124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4525416148763972e-05,
+      "loss": 0.9827,
+      "step": 10174
+    },
+    {
+      "epoch": 0.37004036804014984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4524365767136888e-05,
+      "loss": 0.9763,
+      "step": 10175
+    },
+    {
+      "epoch": 0.37007673564388843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4523315322740647e-05,
+      "loss": 1.1241,
+      "step": 10176
+    },
+    {
+      "epoch": 0.37011310324762703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4522264815589822e-05,
+      "loss": 1.1027,
+      "step": 10177
+    },
+    {
+      "epoch": 0.3701494708513656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4521214245698992e-05,
+      "loss": 1.0004,
+      "step": 10178
+    },
+    {
+      "epoch": 0.37018583845510417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4520163613082728e-05,
+      "loss": 0.959,
+      "step": 10179
+    },
+    {
+      "epoch": 0.37022220605884276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4519112917755607e-05,
+      "loss": 1.1396,
+      "step": 10180
+    },
+    {
+      "epoch": 0.37025857366258136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4518062159732206e-05,
+      "loss": 1.0593,
+      "step": 10181
+    },
+    {
+      "epoch": 0.37029494126631995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4517011339027102e-05,
+      "loss": 1.1092,
+      "step": 10182
+    },
+    {
+      "epoch": 0.37033130887005855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4515960455654876e-05,
+      "loss": 0.9698,
+      "step": 10183
+    },
+    {
+      "epoch": 0.37036767647379715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4514909509630105e-05,
+      "loss": 1.1044,
+      "step": 10184
+    },
+    {
+      "epoch": 0.37040404407753574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.451385850096737e-05,
+      "loss": 1.075,
+      "step": 10185
+    },
+    {
+      "epoch": 0.37044041168127434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4512807429681258e-05,
+      "loss": 1.0845,
+      "step": 10186
+    },
+    {
+      "epoch": 0.37047677928501294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.451175629578634e-05,
+      "loss": 1.109,
+      "step": 10187
+    },
+    {
+      "epoch": 0.3705131468887515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4510705099297205e-05,
+      "loss": 1.0865,
+      "step": 10188
+    },
+    {
+      "epoch": 0.3705495144924901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.450965384022844e-05,
+      "loss": 1.0541,
+      "step": 10189
+    },
+    {
+      "epoch": 0.37058588209622867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4508602518594622e-05,
+      "loss": 1.1296,
+      "step": 10190
+    },
+    {
+      "epoch": 0.37062224969996727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4507551134410343e-05,
+      "loss": 0.9994,
+      "step": 10191
+    },
+    {
+      "epoch": 0.37065861730370586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4506499687690188e-05,
+      "loss": 1.0752,
+      "step": 10192
+    },
+    {
+      "epoch": 0.37069498490744446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.450544817844874e-05,
+      "loss": 1.0841,
+      "step": 10193
+    },
+    {
+      "epoch": 0.37073135251118305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4504396606700595e-05,
+      "loss": 0.9634,
+      "step": 10194
+    },
+    {
+      "epoch": 0.37076772011492165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4503344972460335e-05,
+      "loss": 1.1936,
+      "step": 10195
+    },
+    {
+      "epoch": 0.3708040877186602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4502293275742556e-05,
+      "loss": 0.9771,
+      "step": 10196
+    },
+    {
+      "epoch": 0.3708404553223988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4501241516561843e-05,
+      "loss": 1.0459,
+      "step": 10197
+    },
+    {
+      "epoch": 0.3708768229261374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4500189694932791e-05,
+      "loss": 1.0591,
+      "step": 10198
+    },
+    {
+      "epoch": 0.370913190529876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4499137810869995e-05,
+      "loss": 0.9497,
+      "step": 10199
+    },
+    {
+      "epoch": 0.3709495581336146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4498085864388042e-05,
+      "loss": 1.1403,
+      "step": 10200
+    },
+    {
+      "epoch": 0.3709859257373532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.449703385550153e-05,
+      "loss": 1.0472,
+      "step": 10201
+    },
+    {
+      "epoch": 0.37102229334109177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4495981784225056e-05,
+      "loss": 1.0143,
+      "step": 10202
+    },
+    {
+      "epoch": 0.37105866094483037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4494929650573211e-05,
+      "loss": 1.089,
+      "step": 10203
+    },
+    {
+      "epoch": 0.37109502854856896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4493877454560594e-05,
+      "loss": 0.9898,
+      "step": 10204
+    },
+    {
+      "epoch": 0.3711313961523075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4492825196201809e-05,
+      "loss": 1.0433,
+      "step": 10205
+    },
+    {
+      "epoch": 0.3711677637560461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4491772875511443e-05,
+      "loss": 1.0136,
+      "step": 10206
+    },
+    {
+      "epoch": 0.3712041313597847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4490720492504105e-05,
+      "loss": 1.0059,
+      "step": 10207
+    },
+    {
+      "epoch": 0.3712404989635233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.448966804719439e-05,
+      "loss": 1.0013,
+      "step": 10208
+    },
+    {
+      "epoch": 0.3712768665672619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4488615539596898e-05,
+      "loss": 1.2731,
+      "step": 10209
+    },
+    {
+      "epoch": 0.3713132341710005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4487562969726238e-05,
+      "loss": 1.0082,
+      "step": 10210
+    },
+    {
+      "epoch": 0.3713496017747391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4486510337597009e-05,
+      "loss": 1.0521,
+      "step": 10211
+    },
+    {
+      "epoch": 0.3713859693784777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4485457643223811e-05,
+      "loss": 1.1996,
+      "step": 10212
+    },
+    {
+      "epoch": 0.3714223369822162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4484404886621255e-05,
+      "loss": 1.0002,
+      "step": 10213
+    },
+    {
+      "epoch": 0.3714587045859548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.448335206780394e-05,
+      "loss": 1.0596,
+      "step": 10214
+    },
+    {
+      "epoch": 0.3714950721896934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.448229918678648e-05,
+      "loss": 1.0172,
+      "step": 10215
+    },
+    {
+      "epoch": 0.371531439793432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4481246243583477e-05,
+      "loss": 0.9627,
+      "step": 10216
+    },
+    {
+      "epoch": 0.3715678073971706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.448019323820954e-05,
+      "loss": 1.0221,
+      "step": 10217
+    },
+    {
+      "epoch": 0.3716041750009092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.447914017067928e-05,
+      "loss": 0.9727,
+      "step": 10218
+    },
+    {
+      "epoch": 0.3716405426046478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4478087041007305e-05,
+      "loss": 0.954,
+      "step": 10219
+    },
+    {
+      "epoch": 0.3716769102083864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4477033849208223e-05,
+      "loss": 1.041,
+      "step": 10220
+    },
+    {
+      "epoch": 0.37171327781212493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4475980595296653e-05,
+      "loss": 1.0097,
+      "step": 10221
+    },
+    {
+      "epoch": 0.3717496454158635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4474927279287201e-05,
+      "loss": 1.1106,
+      "step": 10222
+    },
+    {
+      "epoch": 0.3717860130196021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.447387390119448e-05,
+      "loss": 1.0804,
+      "step": 10223
+    },
+    {
+      "epoch": 0.3718223806233407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.447282046103311e-05,
+      "loss": 1.0757,
+      "step": 10224
+    },
+    {
+      "epoch": 0.3718587482270793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.44717669588177e-05,
+      "loss": 0.9804,
+      "step": 10225
+    },
+    {
+      "epoch": 0.3718951158308179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4470713394562867e-05,
+      "loss": 1.0514,
+      "step": 10226
+    },
+    {
+      "epoch": 0.3719314834345565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4469659768283233e-05,
+      "loss": 1.0266,
+      "step": 10227
+    },
+    {
+      "epoch": 0.3719678510382951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4468606079993406e-05,
+      "loss": 1.0624,
+      "step": 10228
+    },
+    {
+      "epoch": 0.3720042186420337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4467552329708012e-05,
+      "loss": 1.0028,
+      "step": 10229
+    },
+    {
+      "epoch": 0.37204058624577224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4466498517441667e-05,
+      "loss": 1.0596,
+      "step": 10230
+    },
+    {
+      "epoch": 0.37207695384951084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4465444643208994e-05,
+      "loss": 1.137,
+      "step": 10231
+    },
+    {
+      "epoch": 0.37211332145324943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.446439070702461e-05,
+      "loss": 1.1349,
+      "step": 10232
+    },
+    {
+      "epoch": 0.37214968905698803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4463336708903142e-05,
+      "loss": 1.1138,
+      "step": 10233
+    },
+    {
+      "epoch": 0.3721860566607266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4462282648859208e-05,
+      "loss": 1.0618,
+      "step": 10234
+    },
+    {
+      "epoch": 0.3722224242644652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4461228526907434e-05,
+      "loss": 1.1089,
+      "step": 10235
+    },
+    {
+      "epoch": 0.3722587918682038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4460174343062443e-05,
+      "loss": 0.9384,
+      "step": 10236
+    },
+    {
+      "epoch": 0.3722951594719424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4459120097338861e-05,
+      "loss": 0.9777,
+      "step": 10237
+    },
+    {
+      "epoch": 0.37233152707568096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4458065789751317e-05,
+      "loss": 1.0643,
+      "step": 10238
+    },
+    {
+      "epoch": 0.37236789467941955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.445701142031443e-05,
+      "loss": 1.1262,
+      "step": 10239
+    },
+    {
+      "epoch": 0.37240426228315815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4455956989042839e-05,
+      "loss": 1.0451,
+      "step": 10240
+    },
+    {
+      "epoch": 0.37244062988689675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4454902495951162e-05,
+      "loss": 1.0676,
+      "step": 10241
+    },
+    {
+      "epoch": 0.37247699749063534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4453847941054037e-05,
+      "loss": 0.931,
+      "step": 10242
+    },
+    {
+      "epoch": 0.37251336509437394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4452793324366093e-05,
+      "loss": 1.0428,
+      "step": 10243
+    },
+    {
+      "epoch": 0.37254973269811253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4451738645901957e-05,
+      "loss": 1.0422,
+      "step": 10244
+    },
+    {
+      "epoch": 0.37258610030185113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4450683905676261e-05,
+      "loss": 0.9581,
+      "step": 10245
+    },
+    {
+      "epoch": 0.3726224679055897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4449629103703642e-05,
+      "loss": 1.0498,
+      "step": 10246
+    },
+    {
+      "epoch": 0.37265883550932827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4448574239998733e-05,
+      "loss": 0.9634,
+      "step": 10247
+    },
+    {
+      "epoch": 0.37269520311306686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4447519314576168e-05,
+      "loss": 0.9394,
+      "step": 10248
+    },
+    {
+      "epoch": 0.37273157071680546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4446464327450582e-05,
+      "loss": 1.0107,
+      "step": 10249
+    },
+    {
+      "epoch": 0.37276793832054406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4445409278636613e-05,
+      "loss": 0.9226,
+      "step": 10250
+    },
+    {
+      "epoch": 0.37280430592428265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4444354168148897e-05,
+      "loss": 1.1489,
+      "step": 10251
+    },
+    {
+      "epoch": 0.37284067352802125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.444329899600207e-05,
+      "loss": 0.9366,
+      "step": 10252
+    },
+    {
+      "epoch": 0.37287704113175985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.444224376221078e-05,
+      "loss": 1.0591,
+      "step": 10253
+    },
+    {
+      "epoch": 0.37291340873549844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4441188466789658e-05,
+      "loss": 1.0897,
+      "step": 10254
+    },
+    {
+      "epoch": 0.372949776339237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4440133109753345e-05,
+      "loss": 1.0152,
+      "step": 10255
+    },
+    {
+      "epoch": 0.3729861439429756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4439077691116485e-05,
+      "loss": 1.1561,
+      "step": 10256
+    },
+    {
+      "epoch": 0.3730225115467142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4438022210893721e-05,
+      "loss": 1.0452,
+      "step": 10257
+    },
+    {
+      "epoch": 0.37305887915045277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4436966669099697e-05,
+      "loss": 1.1101,
+      "step": 10258
+    },
+    {
+      "epoch": 0.37309524675419137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4435911065749056e-05,
+      "loss": 1.0266,
+      "step": 10259
+    },
+    {
+      "epoch": 0.37313161435792996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4434855400856438e-05,
+      "loss": 1.0399,
+      "step": 10260
+    },
+    {
+      "epoch": 0.37316798196166856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4433799674436497e-05,
+      "loss": 1.0533,
+      "step": 10261
+    },
+    {
+      "epoch": 0.37320434956540716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4432743886503877e-05,
+      "loss": 1.0912,
+      "step": 10262
+    },
+    {
+      "epoch": 0.37324071716914575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4431688037073226e-05,
+      "loss": 1.1698,
+      "step": 10263
+    },
+    {
+      "epoch": 0.3732770847728843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4430632126159187e-05,
+      "loss": 1.0669,
+      "step": 10264
+    },
+    {
+      "epoch": 0.3733134523766229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4429576153776417e-05,
+      "loss": 0.9395,
+      "step": 10265
+    },
+    {
+      "epoch": 0.3733498199803615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.442852011993956e-05,
+      "loss": 1.0638,
+      "step": 10266
+    },
+    {
+      "epoch": 0.3733861875841001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4427464024663272e-05,
+      "loss": 1.1556,
+      "step": 10267
+    },
+    {
+      "epoch": 0.3734225551878387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.44264078679622e-05,
+      "loss": 1.1954,
+      "step": 10268
+    },
+    {
+      "epoch": 0.3734589227915773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4425351649851e-05,
+      "loss": 1.0226,
+      "step": 10269
+    },
+    {
+      "epoch": 0.37349529039531587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4424295370344324e-05,
+      "loss": 1.0997,
+      "step": 10270
+    },
+    {
+      "epoch": 0.37353165799905447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4423239029456829e-05,
+      "loss": 1.105,
+      "step": 10271
+    },
+    {
+      "epoch": 0.373568025602793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4422182627203167e-05,
+      "loss": 1.0305,
+      "step": 10272
+    },
+    {
+      "epoch": 0.3736043932065316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4421126163597996e-05,
+      "loss": 1.0407,
+      "step": 10273
+    },
+    {
+      "epoch": 0.3736407608102702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.442006963865597e-05,
+      "loss": 1.1055,
+      "step": 10274
+    },
+    {
+      "epoch": 0.3736771284140088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4419013052391753e-05,
+      "loss": 1.0089,
+      "step": 10275
+    },
+    {
+      "epoch": 0.3737134960177474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4417956404819994e-05,
+      "loss": 1.0392,
+      "step": 10276
+    },
+    {
+      "epoch": 0.373749863621486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4416899695955364e-05,
+      "loss": 1.032,
+      "step": 10277
+    },
+    {
+      "epoch": 0.3737862312252246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4415842925812515e-05,
+      "loss": 0.8761,
+      "step": 10278
+    },
+    {
+      "epoch": 0.3738225988289632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4414786094406111e-05,
+      "loss": 1.1179,
+      "step": 10279
+    },
+    {
+      "epoch": 0.3738589664327017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4413729201750812e-05,
+      "loss": 1.0153,
+      "step": 10280
+    },
+    {
+      "epoch": 0.3738953340364403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4412672247861285e-05,
+      "loss": 1.0698,
+      "step": 10281
+    },
+    {
+      "epoch": 0.3739317016401789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4411615232752191e-05,
+      "loss": 1.1292,
+      "step": 10282
+    },
+    {
+      "epoch": 0.3739680692439175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4410558156438194e-05,
+      "loss": 1.0713,
+      "step": 10283
+    },
+    {
+      "epoch": 0.3740044368476561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.440950101893396e-05,
+      "loss": 1.0559,
+      "step": 10284
+    },
+    {
+      "epoch": 0.3740408044513947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4408443820254157e-05,
+      "loss": 1.0681,
+      "step": 10285
+    },
+    {
+      "epoch": 0.3740771720551333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4407386560413448e-05,
+      "loss": 1.0853,
+      "step": 10286
+    },
+    {
+      "epoch": 0.3741135396588719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4406329239426506e-05,
+      "loss": 1.1082,
+      "step": 10287
+    },
+    {
+      "epoch": 0.3741499072626105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4405271857307999e-05,
+      "loss": 1.1793,
+      "step": 10288
+    },
+    {
+      "epoch": 0.37418627486634903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.440421441407259e-05,
+      "loss": 1.0775,
+      "step": 10289
+    },
+    {
+      "epoch": 0.37422264247008763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4403156909734962e-05,
+      "loss": 0.9832,
+      "step": 10290
+    },
+    {
+      "epoch": 0.3742590100738262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4402099344309775e-05,
+      "loss": 0.9975,
+      "step": 10291
+    },
+    {
+      "epoch": 0.3742953776775648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4401041717811704e-05,
+      "loss": 0.9641,
+      "step": 10292
+    },
+    {
+      "epoch": 0.3743317452813034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4399984030255425e-05,
+      "loss": 1.1007,
+      "step": 10293
+    },
+    {
+      "epoch": 0.374368112885042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4398926281655612e-05,
+      "loss": 1.0356,
+      "step": 10294
+    },
+    {
+      "epoch": 0.3744044804887806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4397868472026934e-05,
+      "loss": 1.0507,
+      "step": 10295
+    },
+    {
+      "epoch": 0.3744408480925192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4396810601384075e-05,
+      "loss": 0.9992,
+      "step": 10296
+    },
+    {
+      "epoch": 0.37447721569625775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4395752669741705e-05,
+      "loss": 1.1122,
+      "step": 10297
+    },
+    {
+      "epoch": 0.37451358329999634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4394694677114504e-05,
+      "loss": 0.9329,
+      "step": 10298
+    },
+    {
+      "epoch": 0.37454995090373494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.439363662351715e-05,
+      "loss": 1.0241,
+      "step": 10299
+    },
+    {
+      "epoch": 0.37458631850747354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4392578508964319e-05,
+      "loss": 1.061,
+      "step": 10300
+    },
+    {
+      "epoch": 0.37462268611121213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4391520333470699e-05,
+      "loss": 1.0031,
+      "step": 10301
+    },
+    {
+      "epoch": 0.37465905371495073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.439046209705096e-05,
+      "loss": 1.0729,
+      "step": 10302
+    },
+    {
+      "epoch": 0.3746954213186893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.438940379971979e-05,
+      "loss": 1.0262,
+      "step": 10303
+    },
+    {
+      "epoch": 0.3747317889224279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.438834544149187e-05,
+      "loss": 0.9731,
+      "step": 10304
+    },
+    {
+      "epoch": 0.3747681565261665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4387287022381883e-05,
+      "loss": 0.9879,
+      "step": 10305
+    },
+    {
+      "epoch": 0.37480452412990506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4386228542404517e-05,
+      "loss": 1.1876,
+      "step": 10306
+    },
+    {
+      "epoch": 0.37484089173364366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4385170001574448e-05,
+      "loss": 1.1002,
+      "step": 10307
+    },
+    {
+      "epoch": 0.37487725933738225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4384111399906369e-05,
+      "loss": 0.9561,
+      "step": 10308
+    },
+    {
+      "epoch": 0.37491362694112085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4383052737414966e-05,
+      "loss": 1.0672,
+      "step": 10309
+    },
+    {
+      "epoch": 0.37494999454485944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4381994014114924e-05,
+      "loss": 1.0896,
+      "step": 10310
+    },
+    {
+      "epoch": 0.37498636214859804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4380935230020932e-05,
+      "loss": 0.942,
+      "step": 10311
+    },
+    {
+      "epoch": 0.37502272975233664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4379876385147679e-05,
+      "loss": 0.9893,
+      "step": 10312
+    },
+    {
+      "epoch": 0.37505909735607523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4378817479509854e-05,
+      "loss": 1.064,
+      "step": 10313
+    },
+    {
+      "epoch": 0.3750954649598138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4377758513122153e-05,
+      "loss": 1.0621,
+      "step": 10314
+    },
+    {
+      "epoch": 0.37513183256355237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4376699485999261e-05,
+      "loss": 0.9463,
+      "step": 10315
+    },
+    {
+      "epoch": 0.37516820016729097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4375640398155875e-05,
+      "loss": 1.0073,
+      "step": 10316
+    },
+    {
+      "epoch": 0.37520456777102956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4374581249606685e-05,
+      "loss": 1.026,
+      "step": 10317
+    },
+    {
+      "epoch": 0.37524093537476816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4373522040366388e-05,
+      "loss": 1.0483,
+      "step": 10318
+    },
+    {
+      "epoch": 0.37527730297850675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4372462770449676e-05,
+      "loss": 1.1615,
+      "step": 10319
+    },
+    {
+      "epoch": 0.37531367058224535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4371403439871253e-05,
+      "loss": 1.1082,
+      "step": 10320
+    },
+    {
+      "epoch": 0.37535003818598395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4370344048645804e-05,
+      "loss": 0.9903,
+      "step": 10321
+    },
+    {
+      "epoch": 0.3753864057897225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4369284596788031e-05,
+      "loss": 0.9343,
+      "step": 10322
+    },
+    {
+      "epoch": 0.3754227733934611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4368225084312637e-05,
+      "loss": 1.0634,
+      "step": 10323
+    },
+    {
+      "epoch": 0.3754591409971997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4367165511234316e-05,
+      "loss": 0.9012,
+      "step": 10324
+    },
+    {
+      "epoch": 0.3754955086009383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4366105877567768e-05,
+      "loss": 0.9861,
+      "step": 10325
+    },
+    {
+      "epoch": 0.3755318762046769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.43650461833277e-05,
+      "loss": 1.1518,
+      "step": 10326
+    },
+    {
+      "epoch": 0.37556824380841547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4363986428528805e-05,
+      "loss": 1.095,
+      "step": 10327
+    },
+    {
+      "epoch": 0.37560461141215407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4362926613185792e-05,
+      "loss": 1.0234,
+      "step": 10328
+    },
+    {
+      "epoch": 0.37564097901589266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4361866737313362e-05,
+      "loss": 1.0627,
+      "step": 10329
+    },
+    {
+      "epoch": 0.37567734661963126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.436080680092622e-05,
+      "loss": 1.0018,
+      "step": 10330
+    },
+    {
+      "epoch": 0.3757137142233698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4359746804039073e-05,
+      "loss": 1.1372,
+      "step": 10331
+    },
+    {
+      "epoch": 0.3757500818271084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.435868674666662e-05,
+      "loss": 0.9075,
+      "step": 10332
+    },
+    {
+      "epoch": 0.375786449430847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4357626628823576e-05,
+      "loss": 0.9849,
+      "step": 10333
+    },
+    {
+      "epoch": 0.3758228170345856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4356566450524642e-05,
+      "loss": 0.9551,
+      "step": 10334
+    },
+    {
+      "epoch": 0.3758591846383242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4355506211784534e-05,
+      "loss": 1.1763,
+      "step": 10335
+    },
+    {
+      "epoch": 0.3758955522420628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4354445912617954e-05,
+      "loss": 1.0155,
+      "step": 10336
+    },
+    {
+      "epoch": 0.3759319198458014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4353385553039618e-05,
+      "loss": 1.023,
+      "step": 10337
+    },
+    {
+      "epoch": 0.37596828744954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4352325133064229e-05,
+      "loss": 0.9782,
+      "step": 10338
+    },
+    {
+      "epoch": 0.3760046550532785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4351264652706509e-05,
+      "loss": 1.0958,
+      "step": 10339
+    },
+    {
+      "epoch": 0.3760410226570171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4350204111981164e-05,
+      "loss": 1.0745,
+      "step": 10340
+    },
+    {
+      "epoch": 0.3760773902607557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4349143510902908e-05,
+      "loss": 0.9575,
+      "step": 10341
+    },
+    {
+      "epoch": 0.3761137578644943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4348082849486456e-05,
+      "loss": 1.0707,
+      "step": 10342
+    },
+    {
+      "epoch": 0.3761501254682329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4347022127746526e-05,
+      "loss": 0.9074,
+      "step": 10343
+    },
+    {
+      "epoch": 0.3761864930719715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.434596134569783e-05,
+      "loss": 1.1119,
+      "step": 10344
+    },
+    {
+      "epoch": 0.3762228606757101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4344900503355086e-05,
+      "loss": 1.1191,
+      "step": 10345
+    },
+    {
+      "epoch": 0.3762592282794487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4343839600733013e-05,
+      "loss": 1.1038,
+      "step": 10346
+    },
+    {
+      "epoch": 0.3762955958831873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.434277863784633e-05,
+      "loss": 1.1012,
+      "step": 10347
+    },
+    {
+      "epoch": 0.3763319634869258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4341717614709752e-05,
+      "loss": 1.0747,
+      "step": 10348
+    },
+    {
+      "epoch": 0.3763683310906644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4340656531338004e-05,
+      "loss": 1.0834,
+      "step": 10349
+    },
+    {
+      "epoch": 0.376404698694403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4339595387745807e-05,
+      "loss": 1.0856,
+      "step": 10350
+    },
+    {
+      "epoch": 0.3764410662981416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4338534183947876e-05,
+      "loss": 1.1412,
+      "step": 10351
+    },
+    {
+      "epoch": 0.3764774339018802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4337472919958943e-05,
+      "loss": 1.2374,
+      "step": 10352
+    },
+    {
+      "epoch": 0.3765138015056188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4336411595793724e-05,
+      "loss": 1.1572,
+      "step": 10353
+    },
+    {
+      "epoch": 0.3765501691093574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4335350211466949e-05,
+      "loss": 0.9557,
+      "step": 10354
+    },
+    {
+      "epoch": 0.376586536713096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4334288766993343e-05,
+      "loss": 1.0025,
+      "step": 10355
+    },
+    {
+      "epoch": 0.37662290431683454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4333227262387626e-05,
+      "loss": 1.2736,
+      "step": 10356
+    },
+    {
+      "epoch": 0.37665927192057314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.433216569766453e-05,
+      "loss": 1.042,
+      "step": 10357
+    },
+    {
+      "epoch": 0.37669563952431173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4331104072838784e-05,
+      "loss": 1.1968,
+      "step": 10358
+    },
+    {
+      "epoch": 0.37673200712805033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4330042387925111e-05,
+      "loss": 1.059,
+      "step": 10359
+    },
+    {
+      "epoch": 0.3767683747317889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4328980642938244e-05,
+      "loss": 0.9432,
+      "step": 10360
+    },
+    {
+      "epoch": 0.3768047423355275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4327918837892911e-05,
+      "loss": 1.0646,
+      "step": 10361
+    },
+    {
+      "epoch": 0.3768411099392661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4326856972803848e-05,
+      "loss": 0.9618,
+      "step": 10362
+    },
+    {
+      "epoch": 0.3768774775430047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4325795047685782e-05,
+      "loss": 0.9449,
+      "step": 10363
+    },
+    {
+      "epoch": 0.37691384514674325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4324733062553446e-05,
+      "loss": 1.0444,
+      "step": 10364
+    },
+    {
+      "epoch": 0.37695021275048185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4323671017421576e-05,
+      "loss": 1.0976,
+      "step": 10365
+    },
+    {
+      "epoch": 0.37698658035422045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4322608912304907e-05,
+      "loss": 1.1246,
+      "step": 10366
+    },
+    {
+      "epoch": 0.37702294795795904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4321546747218169e-05,
+      "loss": 1.1545,
+      "step": 10367
+    },
+    {
+      "epoch": 0.37705931556169764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4320484522176104e-05,
+      "loss": 1.1115,
+      "step": 10368
+    },
+    {
+      "epoch": 0.37709568316543624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4319422237193443e-05,
+      "loss": 0.9708,
+      "step": 10369
+    },
+    {
+      "epoch": 0.37713205076917483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.431835989228493e-05,
+      "loss": 0.9991,
+      "step": 10370
+    },
+    {
+      "epoch": 0.3771684183729134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.43172974874653e-05,
+      "loss": 0.9611,
+      "step": 10371
+    },
+    {
+      "epoch": 0.377204785976652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.431623502274929e-05,
+      "loss": 0.9381,
+      "step": 10372
+    },
+    {
+      "epoch": 0.37724115358039056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4315172498151647e-05,
+      "loss": 0.9605,
+      "step": 10373
+    },
+    {
+      "epoch": 0.37727752118412916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4314109913687104e-05,
+      "loss": 1.1582,
+      "step": 10374
+    },
+    {
+      "epoch": 0.37731388878786776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4313047269370408e-05,
+      "loss": 1.0435,
+      "step": 10375
+    },
+    {
+      "epoch": 0.37735025639160635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4311984565216304e-05,
+      "loss": 1.1014,
+      "step": 10376
+    },
+    {
+      "epoch": 0.37738662399534495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.431092180123953e-05,
+      "loss": 1.0707,
+      "step": 10377
+    },
+    {
+      "epoch": 0.37742299159908355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4309858977454832e-05,
+      "loss": 1.0346,
+      "step": 10378
+    },
+    {
+      "epoch": 0.37745935920282214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4308796093876955e-05,
+      "loss": 1.0756,
+      "step": 10379
+    },
+    {
+      "epoch": 0.37749572680656074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4307733150520647e-05,
+      "loss": 1.0314,
+      "step": 10380
+    },
+    {
+      "epoch": 0.3775320944102993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4306670147400652e-05,
+      "loss": 1.0962,
+      "step": 10381
+    },
+    {
+      "epoch": 0.3775684620140379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4305607084531721e-05,
+      "loss": 0.9441,
+      "step": 10382
+    },
+    {
+      "epoch": 0.37760482961777647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4304543961928597e-05,
+      "loss": 1.1675,
+      "step": 10383
+    },
+    {
+      "epoch": 0.37764119722151507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4303480779606036e-05,
+      "loss": 0.9293,
+      "step": 10384
+    },
+    {
+      "epoch": 0.37767756482525366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4302417537578783e-05,
+      "loss": 1.0892,
+      "step": 10385
+    },
+    {
+      "epoch": 0.37771393242899226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4301354235861594e-05,
+      "loss": 1.088,
+      "step": 10386
+    },
+    {
+      "epoch": 0.37775030003273086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4300290874469215e-05,
+      "loss": 1.0273,
+      "step": 10387
+    },
+    {
+      "epoch": 0.37778666763646945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.42992274534164e-05,
+      "loss": 1.1132,
+      "step": 10388
+    },
+    {
+      "epoch": 0.37782303524020805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4298163972717909e-05,
+      "loss": 1.1012,
+      "step": 10389
+    },
+    {
+      "epoch": 0.3778594028439466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4297100432388487e-05,
+      "loss": 1.0261,
+      "step": 10390
+    },
+    {
+      "epoch": 0.3778957704476852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4296036832442895e-05,
+      "loss": 1.0854,
+      "step": 10391
+    },
+    {
+      "epoch": 0.3779321380514238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4294973172895888e-05,
+      "loss": 1.017,
+      "step": 10392
+    },
+    {
+      "epoch": 0.3779685056551624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.429390945376222e-05,
+      "loss": 1.0786,
+      "step": 10393
+    },
+    {
+      "epoch": 0.378004873258901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.429284567505665e-05,
+      "loss": 1.0529,
+      "step": 10394
+    },
+    {
+      "epoch": 0.37804124086263957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4291781836793939e-05,
+      "loss": 1.0387,
+      "step": 10395
+    },
+    {
+      "epoch": 0.37807760846637817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4290717938988845e-05,
+      "loss": 0.9995,
+      "step": 10396
+    },
+    {
+      "epoch": 0.37811397607011676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4289653981656127e-05,
+      "loss": 1.0589,
+      "step": 10397
+    },
+    {
+      "epoch": 0.3781503436738553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4288589964810546e-05,
+      "loss": 1.099,
+      "step": 10398
+    },
+    {
+      "epoch": 0.3781867112775939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4287525888466862e-05,
+      "loss": 1.1208,
+      "step": 10399
+    },
+    {
+      "epoch": 0.3782230788813325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4286461752639839e-05,
+      "loss": 0.9808,
+      "step": 10400
+    },
+    {
+      "epoch": 0.3782594464850711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4285397557344243e-05,
+      "loss": 1.1832,
+      "step": 10401
+    },
+    {
+      "epoch": 0.3782958140888097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4284333302594838e-05,
+      "loss": 1.0983,
+      "step": 10402
+    },
+    {
+      "epoch": 0.3783321816925483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4283268988406385e-05,
+      "loss": 1.0905,
+      "step": 10403
+    },
+    {
+      "epoch": 0.3783685492962869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4282204614793648e-05,
+      "loss": 1.0862,
+      "step": 10404
+    },
+    {
+      "epoch": 0.3784049169000255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4281140181771403e-05,
+      "loss": 1.1442,
+      "step": 10405
+    },
+    {
+      "epoch": 0.378441284503764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.428007568935441e-05,
+      "loss": 1.2277,
+      "step": 10406
+    },
+    {
+      "epoch": 0.3784776521075026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.427901113755744e-05,
+      "loss": 1.144,
+      "step": 10407
+    },
+    {
+      "epoch": 0.3785140197112412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4277946526395261e-05,
+      "loss": 1.1018,
+      "step": 10408
+    },
+    {
+      "epoch": 0.3785503873149798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4276881855882648e-05,
+      "loss": 1.086,
+      "step": 10409
+    },
+    {
+      "epoch": 0.3785867549187184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4275817126034358e-05,
+      "loss": 1.1801,
+      "step": 10410
+    },
+    {
+      "epoch": 0.378623122522457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.427475233686518e-05,
+      "loss": 1.143,
+      "step": 10411
+    },
+    {
+      "epoch": 0.3786594901261956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4273687488389873e-05,
+      "loss": 0.9718,
+      "step": 10412
+    },
+    {
+      "epoch": 0.3786958577299342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4272622580623221e-05,
+      "loss": 0.978,
+      "step": 10413
+    },
+    {
+      "epoch": 0.3787322253336728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4271557613579989e-05,
+      "loss": 1.0789,
+      "step": 10414
+    },
+    {
+      "epoch": 0.37876859293741133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4270492587274957e-05,
+      "loss": 0.9379,
+      "step": 10415
+    },
+    {
+      "epoch": 0.3788049605411499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4269427501722902e-05,
+      "loss": 0.9134,
+      "step": 10416
+    },
+    {
+      "epoch": 0.3788413281448885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4268362356938595e-05,
+      "loss": 1.0678,
+      "step": 10417
+    },
+    {
+      "epoch": 0.3788776957486271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4267297152936816e-05,
+      "loss": 1.0382,
+      "step": 10418
+    },
+    {
+      "epoch": 0.3789140633523657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4266231889732345e-05,
+      "loss": 1.0071,
+      "step": 10419
+    },
+    {
+      "epoch": 0.3789504309561043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4265166567339956e-05,
+      "loss": 1.1732,
+      "step": 10420
+    },
+    {
+      "epoch": 0.3789867985598429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4264101185774438e-05,
+      "loss": 1.0649,
+      "step": 10421
+    },
+    {
+      "epoch": 0.3790231661635815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4263035745050561e-05,
+      "loss": 0.9866,
+      "step": 10422
+    },
+    {
+      "epoch": 0.37905953376732004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4261970245183112e-05,
+      "loss": 0.9478,
+      "step": 10423
+    },
+    {
+      "epoch": 0.37909590137105864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4260904686186876e-05,
+      "loss": 1.0303,
+      "step": 10424
+    },
+    {
+      "epoch": 0.37913226897479724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4259839068076628e-05,
+      "loss": 1.1195,
+      "step": 10425
+    },
+    {
+      "epoch": 0.37916863657853583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4258773390867163e-05,
+      "loss": 1.0086,
+      "step": 10426
+    },
+    {
+      "epoch": 0.37920500418227443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4257707654573257e-05,
+      "loss": 1.0108,
+      "step": 10427
+    },
+    {
+      "epoch": 0.379241371786013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4256641859209697e-05,
+      "loss": 0.9081,
+      "step": 10428
+    },
+    {
+      "epoch": 0.3792777393897516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4255576004791271e-05,
+      "loss": 0.9863,
+      "step": 10429
+    },
+    {
+      "epoch": 0.3793141069934902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4254510091332767e-05,
+      "loss": 1.0565,
+      "step": 10430
+    },
+    {
+      "epoch": 0.3793504745972288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4253444118848968e-05,
+      "loss": 0.9895,
+      "step": 10431
+    },
+    {
+      "epoch": 0.37938684220096736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4252378087354673e-05,
+      "loss": 0.9505,
+      "step": 10432
+    },
+    {
+      "epoch": 0.37942320980470595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4251311996864662e-05,
+      "loss": 1.0099,
+      "step": 10433
+    },
+    {
+      "epoch": 0.37945957740844455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4250245847393727e-05,
+      "loss": 0.9757,
+      "step": 10434
+    },
+    {
+      "epoch": 0.37949594501218314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4249179638956668e-05,
+      "loss": 1.1117,
+      "step": 10435
+    },
+    {
+      "epoch": 0.37953231261592174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4248113371568262e-05,
+      "loss": 0.927,
+      "step": 10436
+    },
+    {
+      "epoch": 0.37956868021966034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4247047045243314e-05,
+      "loss": 1.0139,
+      "step": 10437
+    },
+    {
+      "epoch": 0.37960504782339893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4245980659996615e-05,
+      "loss": 1.0426,
+      "step": 10438
+    },
+    {
+      "epoch": 0.37964141542713753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4244914215842955e-05,
+      "loss": 1.0994,
+      "step": 10439
+    },
+    {
+      "epoch": 0.37967778303087607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4243847712797137e-05,
+      "loss": 1.0722,
+      "step": 10440
+    },
+    {
+      "epoch": 0.37971415063461467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.424278115087395e-05,
+      "loss": 1.0721,
+      "step": 10441
+    },
+    {
+      "epoch": 0.37975051823835326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4241714530088197e-05,
+      "loss": 1.0755,
+      "step": 10442
+    },
+    {
+      "epoch": 0.37978688584209186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4240647850454671e-05,
+      "loss": 1.0036,
+      "step": 10443
+    },
+    {
+      "epoch": 0.37982325344583046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4239581111988172e-05,
+      "loss": 1.0366,
+      "step": 10444
+    },
+    {
+      "epoch": 0.37985962104956905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4238514314703505e-05,
+      "loss": 0.9867,
+      "step": 10445
+    },
+    {
+      "epoch": 0.37989598865330765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4237447458615461e-05,
+      "loss": 1.0322,
+      "step": 10446
+    },
+    {
+      "epoch": 0.37993235625704624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4236380543738847e-05,
+      "loss": 1.0776,
+      "step": 10447
+    },
+    {
+      "epoch": 0.37996872386078484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4235313570088466e-05,
+      "loss": 1.1182,
+      "step": 10448
+    },
+    {
+      "epoch": 0.3800050914645234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4234246537679114e-05,
+      "loss": 1.0303,
+      "step": 10449
+    },
+    {
+      "epoch": 0.380041459068262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.42331794465256e-05,
+      "loss": 1.0056,
+      "step": 10450
+    },
+    {
+      "epoch": 0.3800778266720006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4232112296642729e-05,
+      "loss": 1.0788,
+      "step": 10451
+    },
+    {
+      "epoch": 0.38011419427573917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4231045088045303e-05,
+      "loss": 0.969,
+      "step": 10452
+    },
+    {
+      "epoch": 0.38015056187947777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.422997782074813e-05,
+      "loss": 1.0028,
+      "step": 10453
+    },
+    {
+      "epoch": 0.38018692948321636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.422891049476602e-05,
+      "loss": 0.8113,
+      "step": 10454
+    },
+    {
+      "epoch": 0.38022329708695496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4227843110113772e-05,
+      "loss": 1.1071,
+      "step": 10455
+    },
+    {
+      "epoch": 0.38025966469069356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.42267756668062e-05,
+      "loss": 1.0473,
+      "step": 10456
+    },
+    {
+      "epoch": 0.3802960322944321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4225708164858114e-05,
+      "loss": 1.053,
+      "step": 10457
+    },
+    {
+      "epoch": 0.3803323998981707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4224640604284321e-05,
+      "loss": 1.0266,
+      "step": 10458
+    },
+    {
+      "epoch": 0.3803687675019093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4223572985099637e-05,
+      "loss": 1.0812,
+      "step": 10459
+    },
+    {
+      "epoch": 0.3804051351056479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4222505307318866e-05,
+      "loss": 1.1388,
+      "step": 10460
+    },
+    {
+      "epoch": 0.3804415027093865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4221437570956826e-05,
+      "loss": 1.0235,
+      "step": 10461
+    },
+    {
+      "epoch": 0.3804778703131251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.422036977602833e-05,
+      "loss": 1.0333,
+      "step": 10462
+    },
+    {
+      "epoch": 0.3805142379168637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4219301922548192e-05,
+      "loss": 1.1531,
+      "step": 10463
+    },
+    {
+      "epoch": 0.38055060552060227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4218234010531227e-05,
+      "loss": 1.0241,
+      "step": 10464
+    },
+    {
+      "epoch": 0.3805869731243408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4217166039992248e-05,
+      "loss": 1.0568,
+      "step": 10465
+    },
+    {
+      "epoch": 0.3806233407280794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4216098010946073e-05,
+      "loss": 0.8475,
+      "step": 10466
+    },
+    {
+      "epoch": 0.380659708331818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4215029923407522e-05,
+      "loss": 0.997,
+      "step": 10467
+    },
+    {
+      "epoch": 0.3806960759355566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4213961777391407e-05,
+      "loss": 1.0754,
+      "step": 10468
+    },
+    {
+      "epoch": 0.3807324435392952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4212893572912557e-05,
+      "loss": 0.9716,
+      "step": 10469
+    },
+    {
+      "epoch": 0.3807688111430338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4211825309985783e-05,
+      "loss": 0.9941,
+      "step": 10470
+    },
+    {
+      "epoch": 0.3808051787467724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4210756988625906e-05,
+      "loss": 1.0117,
+      "step": 10471
+    },
+    {
+      "epoch": 0.380841546350511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4209688608847754e-05,
+      "loss": 1.149,
+      "step": 10472
+    },
+    {
+      "epoch": 0.3808779139542496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4208620170666146e-05,
+      "loss": 1.0783,
+      "step": 10473
+    },
+    {
+      "epoch": 0.3809142815579881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4207551674095903e-05,
+      "loss": 1.0082,
+      "step": 10474
+    },
+    {
+      "epoch": 0.3809506491617267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.420648311915185e-05,
+      "loss": 0.9912,
+      "step": 10475
+    },
+    {
+      "epoch": 0.3809870167654653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.420541450584881e-05,
+      "loss": 1.1389,
+      "step": 10476
+    },
+    {
+      "epoch": 0.3810233843692039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4204345834201613e-05,
+      "loss": 1.1097,
+      "step": 10477
+    },
+    {
+      "epoch": 0.3810597519729425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4203277104225084e-05,
+      "loss": 1.0317,
+      "step": 10478
+    },
+    {
+      "epoch": 0.3810961195766811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4202208315934044e-05,
+      "loss": 0.9197,
+      "step": 10479
+    },
+    {
+      "epoch": 0.3811324871804197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.420113946934333e-05,
+      "loss": 1.2234,
+      "step": 10480
+    },
+    {
+      "epoch": 0.3811688547841583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4200070564467766e-05,
+      "loss": 1.2643,
+      "step": 10481
+    },
+    {
+      "epoch": 0.38120522238789684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.419900160132218e-05,
+      "loss": 0.9954,
+      "step": 10482
+    },
+    {
+      "epoch": 0.38124158999163543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4197932579921408e-05,
+      "loss": 1.1101,
+      "step": 10483
+    },
+    {
+      "epoch": 0.38127795759537403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4196863500280271e-05,
+      "loss": 0.9202,
+      "step": 10484
+    },
+    {
+      "epoch": 0.3813143251991126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4195794362413615e-05,
+      "loss": 1.1284,
+      "step": 10485
+    },
+    {
+      "epoch": 0.3813506928028512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4194725166336261e-05,
+      "loss": 1.0476,
+      "step": 10486
+    },
+    {
+      "epoch": 0.3813870604065898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4193655912063048e-05,
+      "loss": 1.1121,
+      "step": 10487
+    },
+    {
+      "epoch": 0.3814234280103284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4192586599608808e-05,
+      "loss": 0.935,
+      "step": 10488
+    },
+    {
+      "epoch": 0.381459795614067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.419151722898838e-05,
+      "loss": 0.9942,
+      "step": 10489
+    },
+    {
+      "epoch": 0.3814961632178056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4190447800216593e-05,
+      "loss": 1.0443,
+      "step": 10490
+    },
+    {
+      "epoch": 0.38153253082154415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4189378313308293e-05,
+      "loss": 1.0582,
+      "step": 10491
+    },
+    {
+      "epoch": 0.38156889842528274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4188308768278312e-05,
+      "loss": 1.1138,
+      "step": 10492
+    },
+    {
+      "epoch": 0.38160526602902134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.418723916514149e-05,
+      "loss": 1.0644,
+      "step": 10493
+    },
+    {
+      "epoch": 0.38164163363275994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4186169503912663e-05,
+      "loss": 1.1107,
+      "step": 10494
+    },
+    {
+      "epoch": 0.38167800123649853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4185099784606673e-05,
+      "loss": 1.0453,
+      "step": 10495
+    },
+    {
+      "epoch": 0.38171436884023713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4184030007238363e-05,
+      "loss": 1.0423,
+      "step": 10496
+    },
+    {
+      "epoch": 0.3817507364439757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4182960171822572e-05,
+      "loss": 1.0308,
+      "step": 10497
+    },
+    {
+      "epoch": 0.3817871040477143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4181890278374144e-05,
+      "loss": 1.0172,
+      "step": 10498
+    },
+    {
+      "epoch": 0.38182347165145286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4180820326907922e-05,
+      "loss": 1.1144,
+      "step": 10499
+    },
+    {
+      "epoch": 0.38185983925519146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4179750317438749e-05,
+      "loss": 1.1007,
+      "step": 10500
+    },
+    {
+      "epoch": 0.38189620685893005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4178680249981472e-05,
+      "loss": 1.0592,
+      "step": 10501
+    },
+    {
+      "epoch": 0.38193257446266865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4177610124550935e-05,
+      "loss": 1.113,
+      "step": 10502
+    },
+    {
+      "epoch": 0.38196894206640725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4176539941161985e-05,
+      "loss": 0.9325,
+      "step": 10503
+    },
+    {
+      "epoch": 0.38200530967014584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4175469699829467e-05,
+      "loss": 1.0461,
+      "step": 10504
+    },
+    {
+      "epoch": 0.38204167727388444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4174399400568233e-05,
+      "loss": 0.9352,
+      "step": 10505
+    },
+    {
+      "epoch": 0.38207804487762304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.417332904339313e-05,
+      "loss": 1.0877,
+      "step": 10506
+    },
+    {
+      "epoch": 0.3821144124813616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4172258628319006e-05,
+      "loss": 0.9059,
+      "step": 10507
+    },
+    {
+      "epoch": 0.3821507800851002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4171188155360713e-05,
+      "loss": 1.0994,
+      "step": 10508
+    },
+    {
+      "epoch": 0.38218714768883877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4170117624533104e-05,
+      "loss": 1.1516,
+      "step": 10509
+    },
+    {
+      "epoch": 0.38222351529257736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4169047035851032e-05,
+      "loss": 1.1335,
+      "step": 10510
+    },
+    {
+      "epoch": 0.38225988289631596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4167976389329345e-05,
+      "loss": 1.0528,
+      "step": 10511
+    },
+    {
+      "epoch": 0.38229625050005456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4166905684982898e-05,
+      "loss": 1.0227,
+      "step": 10512
+    },
+    {
+      "epoch": 0.38233261810379315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4165834922826547e-05,
+      "loss": 1.161,
+      "step": 10513
+    },
+    {
+      "epoch": 0.38236898570753175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.416476410287515e-05,
+      "loss": 1.0689,
+      "step": 10514
+    },
+    {
+      "epoch": 0.38240535331127035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4163693225143558e-05,
+      "loss": 1.0508,
+      "step": 10515
+    },
+    {
+      "epoch": 0.3824417209150089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4162622289646629e-05,
+      "loss": 1.1056,
+      "step": 10516
+    },
+    {
+      "epoch": 0.3824780885187475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4161551296399223e-05,
+      "loss": 1.1147,
+      "step": 10517
+    },
+    {
+      "epoch": 0.3825144561224861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4160480245416197e-05,
+      "loss": 0.9883,
+      "step": 10518
+    },
+    {
+      "epoch": 0.3825508237262247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.415940913671241e-05,
+      "loss": 1.0674,
+      "step": 10519
+    },
+    {
+      "epoch": 0.38258719132996327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4158337970302722e-05,
+      "loss": 1.0945,
+      "step": 10520
+    },
+    {
+      "epoch": 0.38262355893370187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4157266746201996e-05,
+      "loss": 1.096,
+      "step": 10521
+    },
+    {
+      "epoch": 0.38265992653744046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4156195464425092e-05,
+      "loss": 1.2102,
+      "step": 10522
+    },
+    {
+      "epoch": 0.38269629414117906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4155124124986874e-05,
+      "loss": 1.119,
+      "step": 10523
+    },
+    {
+      "epoch": 0.3827326617449176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4154052727902201e-05,
+      "loss": 1.1056,
+      "step": 10524
+    },
+    {
+      "epoch": 0.3827690293486562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4152981273185942e-05,
+      "loss": 1.0849,
+      "step": 10525
+    },
+    {
+      "epoch": 0.3828053969523948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4151909760852963e-05,
+      "loss": 1.0088,
+      "step": 10526
+    },
+    {
+      "epoch": 0.3828417645561334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4150838190918122e-05,
+      "loss": 0.9884,
+      "step": 10527
+    },
+    {
+      "epoch": 0.382878132159872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4149766563396292e-05,
+      "loss": 1.0263,
+      "step": 10528
+    },
+    {
+      "epoch": 0.3829144997636106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.414869487830234e-05,
+      "loss": 1.0002,
+      "step": 10529
+    },
+    {
+      "epoch": 0.3829508673673492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4147623135651132e-05,
+      "loss": 1.0348,
+      "step": 10530
+    },
+    {
+      "epoch": 0.3829872349710878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.414655133545754e-05,
+      "loss": 1.1112,
+      "step": 10531
+    },
+    {
+      "epoch": 0.38302360257482637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4145479477736426e-05,
+      "loss": 1.1515,
+      "step": 10532
+    },
+    {
+      "epoch": 0.3830599701785649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4144407562502668e-05,
+      "loss": 1.0922,
+      "step": 10533
+    },
+    {
+      "epoch": 0.3830963377823035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.414333558977114e-05,
+      "loss": 0.9727,
+      "step": 10534
+    },
+    {
+      "epoch": 0.3831327053860421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4142263559556705e-05,
+      "loss": 1.0911,
+      "step": 10535
+    },
+    {
+      "epoch": 0.3831690729897807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.414119147187424e-05,
+      "loss": 1.0097,
+      "step": 10536
+    },
+    {
+      "epoch": 0.3832054405935193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.414011932673862e-05,
+      "loss": 1.0183,
+      "step": 10537
+    },
+    {
+      "epoch": 0.3832418081972579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4139047124164715e-05,
+      "loss": 1.0765,
+      "step": 10538
+    },
+    {
+      "epoch": 0.3832781758009965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.413797486416741e-05,
+      "loss": 0.9807,
+      "step": 10539
+    },
+    {
+      "epoch": 0.3833145434047351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4136902546761569e-05,
+      "loss": 1.1034,
+      "step": 10540
+    },
+    {
+      "epoch": 0.3833509110084736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4135830171962077e-05,
+      "loss": 0.9832,
+      "step": 10541
+    },
+    {
+      "epoch": 0.3833872786122122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.413475773978381e-05,
+      "loss": 1.0796,
+      "step": 10542
+    },
+    {
+      "epoch": 0.3834236462159508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4133685250241644e-05,
+      "loss": 0.971,
+      "step": 10543
+    },
+    {
+      "epoch": 0.3834600138196894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4132612703350464e-05,
+      "loss": 0.98,
+      "step": 10544
+    },
+    {
+      "epoch": 0.383496381423428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4131540099125143e-05,
+      "loss": 1.0802,
+      "step": 10545
+    },
+    {
+      "epoch": 0.3835327490271666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4130467437580564e-05,
+      "loss": 1.0221,
+      "step": 10546
+    },
+    {
+      "epoch": 0.3835691166309052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4129394718731612e-05,
+      "loss": 0.9985,
+      "step": 10547
+    },
+    {
+      "epoch": 0.3836054842346438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4128321942593165e-05,
+      "loss": 1.1544,
+      "step": 10548
+    },
+    {
+      "epoch": 0.38364185183838234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.412724910918011e-05,
+      "loss": 1.1075,
+      "step": 10549
+    },
+    {
+      "epoch": 0.38367821944212094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4126176218507332e-05,
+      "loss": 1.0978,
+      "step": 10550
+    },
+    {
+      "epoch": 0.38371458704585953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4125103270589708e-05,
+      "loss": 1.0734,
+      "step": 10551
+    },
+    {
+      "epoch": 0.38375095464959813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4124030265442132e-05,
+      "loss": 1.1742,
+      "step": 10552
+    },
+    {
+      "epoch": 0.3837873222533367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4122957203079489e-05,
+      "loss": 1.0401,
+      "step": 10553
+    },
+    {
+      "epoch": 0.3838236898570753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.412188408351666e-05,
+      "loss": 1.0654,
+      "step": 10554
+    },
+    {
+      "epoch": 0.3838600574608139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4120810906768541e-05,
+      "loss": 1.0775,
+      "step": 10555
+    },
+    {
+      "epoch": 0.3838964250645525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4119737672850013e-05,
+      "loss": 1.0559,
+      "step": 10556
+    },
+    {
+      "epoch": 0.3839327926682911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4118664381775973e-05,
+      "loss": 0.9161,
+      "step": 10557
+    },
+    {
+      "epoch": 0.38396916027202965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.411759103356131e-05,
+      "loss": 1.1375,
+      "step": 10558
+    },
+    {
+      "epoch": 0.38400552787576825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4116517628220909e-05,
+      "loss": 1.0947,
+      "step": 10559
+    },
+    {
+      "epoch": 0.38404189547950685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4115444165769671e-05,
+      "loss": 1.0671,
+      "step": 10560
+    },
+    {
+      "epoch": 0.38407826308324544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4114370646222484e-05,
+      "loss": 1.0524,
+      "step": 10561
+    },
+    {
+      "epoch": 0.38411463068698404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4113297069594239e-05,
+      "loss": 0.9831,
+      "step": 10562
+    },
+    {
+      "epoch": 0.38415099829072263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4112223435899834e-05,
+      "loss": 1.0345,
+      "step": 10563
+    },
+    {
+      "epoch": 0.38418736589446123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4111149745154164e-05,
+      "loss": 0.9719,
+      "step": 10564
+    },
+    {
+      "epoch": 0.3842237334981998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4110075997372124e-05,
+      "loss": 1.0946,
+      "step": 10565
+    },
+    {
+      "epoch": 0.38426010110193837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4109002192568611e-05,
+      "loss": 1.0223,
+      "step": 10566
+    },
+    {
+      "epoch": 0.38429646870567696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.410792833075852e-05,
+      "loss": 1.1097,
+      "step": 10567
+    },
+    {
+      "epoch": 0.38433283630941556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4106854411956757e-05,
+      "loss": 1.0105,
+      "step": 10568
+    },
+    {
+      "epoch": 0.38436920391315416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4105780436178212e-05,
+      "loss": 0.9329,
+      "step": 10569
+    },
+    {
+      "epoch": 0.38440557151689275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4104706403437792e-05,
+      "loss": 0.884,
+      "step": 10570
+    },
+    {
+      "epoch": 0.38444193912063135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4103632313750391e-05,
+      "loss": 1.1349,
+      "step": 10571
+    },
+    {
+      "epoch": 0.38447830672436994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4102558167130915e-05,
+      "loss": 1.0379,
+      "step": 10572
+    },
+    {
+      "epoch": 0.38451467432810854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4101483963594266e-05,
+      "loss": 1.0527,
+      "step": 10573
+    },
+    {
+      "epoch": 0.38455104193184714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4100409703155347e-05,
+      "loss": 0.9881,
+      "step": 10574
+    },
+    {
+      "epoch": 0.3845874095355857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4099335385829059e-05,
+      "loss": 1.0318,
+      "step": 10575
+    },
+    {
+      "epoch": 0.3846237771393243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.409826101163031e-05,
+      "loss": 0.986,
+      "step": 10576
+    },
+    {
+      "epoch": 0.38466014474306287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4097186580574003e-05,
+      "loss": 1.0165,
+      "step": 10577
+    },
+    {
+      "epoch": 0.38469651234680147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4096112092675045e-05,
+      "loss": 1.0636,
+      "step": 10578
+    },
+    {
+      "epoch": 0.38473287995054006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4095037547948345e-05,
+      "loss": 1.1303,
+      "step": 10579
+    },
+    {
+      "epoch": 0.38476924755427866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4093962946408808e-05,
+      "loss": 1.1564,
+      "step": 10580
+    },
+    {
+      "epoch": 0.38480561515801726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4092888288071344e-05,
+      "loss": 0.9432,
+      "step": 10581
+    },
+    {
+      "epoch": 0.38484198276175585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4091813572950863e-05,
+      "loss": 1.0977,
+      "step": 10582
+    },
+    {
+      "epoch": 0.3848783503654944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4090738801062271e-05,
+      "loss": 1.1096,
+      "step": 10583
+    },
+    {
+      "epoch": 0.384914717969233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4089663972420487e-05,
+      "loss": 1.1056,
+      "step": 10584
+    },
+    {
+      "epoch": 0.3849510855729716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4088589087040414e-05,
+      "loss": 1.0963,
+      "step": 10585
+    },
+    {
+      "epoch": 0.3849874531767102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4087514144936968e-05,
+      "loss": 1.0734,
+      "step": 10586
+    },
+    {
+      "epoch": 0.3850238207804488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4086439146125068e-05,
+      "loss": 1.1046,
+      "step": 10587
+    },
+    {
+      "epoch": 0.3850601883841874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.408536409061962e-05,
+      "loss": 1.072,
+      "step": 10588
+    },
+    {
+      "epoch": 0.38509655598792597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.408428897843554e-05,
+      "loss": 1.0475,
+      "step": 10589
+    },
+    {
+      "epoch": 0.38513292359166457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4083213809587748e-05,
+      "loss": 1.1655,
+      "step": 10590
+    },
+    {
+      "epoch": 0.3851692911954031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4082138584091155e-05,
+      "loss": 1.0105,
+      "step": 10591
+    },
+    {
+      "epoch": 0.3852056587991417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.408106330196068e-05,
+      "loss": 1.0753,
+      "step": 10592
+    },
+    {
+      "epoch": 0.3852420264028803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4079987963211246e-05,
+      "loss": 1.1458,
+      "step": 10593
+    },
+    {
+      "epoch": 0.3852783940066189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4078912567857764e-05,
+      "loss": 1.0047,
+      "step": 10594
+    },
+    {
+      "epoch": 0.3853147616103575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4077837115915158e-05,
+      "loss": 1.0829,
+      "step": 10595
+    },
+    {
+      "epoch": 0.3853511292140961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4076761607398348e-05,
+      "loss": 1.0704,
+      "step": 10596
+    },
+    {
+      "epoch": 0.3853874968178347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4075686042322257e-05,
+      "loss": 1.0037,
+      "step": 10597
+    },
+    {
+      "epoch": 0.3854238644215733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4074610420701805e-05,
+      "loss": 1.0677,
+      "step": 10598
+    },
+    {
+      "epoch": 0.3854602320253119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4073534742551909e-05,
+      "loss": 1.1097,
+      "step": 10599
+    },
+    {
+      "epoch": 0.3854965996290504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4072459007887502e-05,
+      "loss": 0.98,
+      "step": 10600
+    },
+    {
+      "epoch": 0.385532967232789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4071383216723504e-05,
+      "loss": 1.0927,
+      "step": 10601
+    },
+    {
+      "epoch": 0.3855693348365276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4070307369074841e-05,
+      "loss": 1.0626,
+      "step": 10602
+    },
+    {
+      "epoch": 0.3856057024402662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4069231464956436e-05,
+      "loss": 1.0724,
+      "step": 10603
+    },
+    {
+      "epoch": 0.3856420700440048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4068155504383223e-05,
+      "loss": 1.1346,
+      "step": 10604
+    },
+    {
+      "epoch": 0.3856784376477434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4067079487370119e-05,
+      "loss": 1.0673,
+      "step": 10605
+    },
+    {
+      "epoch": 0.385714805251482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4066003413932059e-05,
+      "loss": 1.1298,
+      "step": 10606
+    },
+    {
+      "epoch": 0.3857511728552206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4064927284083973e-05,
+      "loss": 1.0299,
+      "step": 10607
+    },
+    {
+      "epoch": 0.38578754045895913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4063851097840786e-05,
+      "loss": 1.0419,
+      "step": 10608
+    },
+    {
+      "epoch": 0.38582390806269773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4062774855217431e-05,
+      "loss": 1.0936,
+      "step": 10609
+    },
+    {
+      "epoch": 0.3858602756664363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.406169855622884e-05,
+      "loss": 0.9668,
+      "step": 10610
+    },
+    {
+      "epoch": 0.3858966432701749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4060622200889944e-05,
+      "loss": 1.0344,
+      "step": 10611
+    },
+    {
+      "epoch": 0.3859330108739135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4059545789215676e-05,
+      "loss": 1.1132,
+      "step": 10612
+    },
+    {
+      "epoch": 0.3859693784776521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4058469321220968e-05,
+      "loss": 0.9997,
+      "step": 10613
+    },
+    {
+      "epoch": 0.3860057460813907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4057392796920762e-05,
+      "loss": 1.0862,
+      "step": 10614
+    },
+    {
+      "epoch": 0.3860421136851293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4056316216329982e-05,
+      "loss": 1.048,
+      "step": 10615
+    },
+    {
+      "epoch": 0.3860784812888679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4055239579463571e-05,
+      "loss": 1.1067,
+      "step": 10616
+    },
+    {
+      "epoch": 0.38611484889260644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4054162886336466e-05,
+      "loss": 1.0729,
+      "step": 10617
+    },
+    {
+      "epoch": 0.38615121649634504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.40530861369636e-05,
+      "loss": 1.0093,
+      "step": 10618
+    },
+    {
+      "epoch": 0.38618758410008364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4052009331359919e-05,
+      "loss": 0.9521,
+      "step": 10619
+    },
+    {
+      "epoch": 0.38622395170382223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4050932469540355e-05,
+      "loss": 1.0255,
+      "step": 10620
+    },
+    {
+      "epoch": 0.38626031930756083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4049855551519848e-05,
+      "loss": 1.08,
+      "step": 10621
+    },
+    {
+      "epoch": 0.3862966869112994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4048778577313345e-05,
+      "loss": 1.0786,
+      "step": 10622
+    },
+    {
+      "epoch": 0.386333054515038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4047701546935781e-05,
+      "loss": 1.0316,
+      "step": 10623
+    },
+    {
+      "epoch": 0.3863694221187766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4046624460402103e-05,
+      "loss": 0.9244,
+      "step": 10624
+    },
+    {
+      "epoch": 0.38640578972251516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4045547317727252e-05,
+      "loss": 1.1448,
+      "step": 10625
+    },
+    {
+      "epoch": 0.38644215732625375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4044470118926174e-05,
+      "loss": 1.0603,
+      "step": 10626
+    },
+    {
+      "epoch": 0.38647852492999235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.404339286401381e-05,
+      "loss": 1.0887,
+      "step": 10627
+    },
+    {
+      "epoch": 0.38651489253373095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4042315553005104e-05,
+      "loss": 1.0221,
+      "step": 10628
+    },
+    {
+      "epoch": 0.38655126013746954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4041238185915007e-05,
+      "loss": 1.045,
+      "step": 10629
+    },
+    {
+      "epoch": 0.38658762774120814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4040160762758463e-05,
+      "loss": 1.1159,
+      "step": 10630
+    },
+    {
+      "epoch": 0.38662399534494674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4039083283550421e-05,
+      "loss": 1.0625,
+      "step": 10631
+    },
+    {
+      "epoch": 0.38666036294868533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.403800574830583e-05,
+      "loss": 1.0219,
+      "step": 10632
+    },
+    {
+      "epoch": 0.38669673055242393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4036928157039638e-05,
+      "loss": 1.109,
+      "step": 10633
+    },
+    {
+      "epoch": 0.38673309815616247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4035850509766795e-05,
+      "loss": 1.097,
+      "step": 10634
+    },
+    {
+      "epoch": 0.38676946575990107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4034772806502253e-05,
+      "loss": 1.0976,
+      "step": 10635
+    },
+    {
+      "epoch": 0.38680583336363966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4033695047260963e-05,
+      "loss": 0.9767,
+      "step": 10636
+    },
+    {
+      "epoch": 0.38684220096737826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4032617232057874e-05,
+      "loss": 1.0439,
+      "step": 10637
+    },
+    {
+      "epoch": 0.38687856857111685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4031539360907946e-05,
+      "loss": 1.0237,
+      "step": 10638
+    },
+    {
+      "epoch": 0.38691493617485545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4030461433826127e-05,
+      "loss": 1.1549,
+      "step": 10639
+    },
+    {
+      "epoch": 0.38695130377859405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4029383450827379e-05,
+      "loss": 1.0516,
+      "step": 10640
+    },
+    {
+      "epoch": 0.38698767138233264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4028305411926648e-05,
+      "loss": 1.0183,
+      "step": 10641
+    },
+    {
+      "epoch": 0.3870240389860712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4027227317138896e-05,
+      "loss": 1.2068,
+      "step": 10642
+    },
+    {
+      "epoch": 0.3870604065898098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.402614916647908e-05,
+      "loss": 0.9485,
+      "step": 10643
+    },
+    {
+      "epoch": 0.3870967741935484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4025070959962155e-05,
+      "loss": 0.9483,
+      "step": 10644
+    },
+    {
+      "epoch": 0.387133141797287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4023992697603084e-05,
+      "loss": 1.0014,
+      "step": 10645
+    },
+    {
+      "epoch": 0.38716950940102557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.402291437941682e-05,
+      "loss": 1.0898,
+      "step": 10646
+    },
+    {
+      "epoch": 0.38720587700476417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4021836005418328e-05,
+      "loss": 1.1152,
+      "step": 10647
+    },
+    {
+      "epoch": 0.38724224460850276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.402075757562257e-05,
+      "loss": 1.0325,
+      "step": 10648
+    },
+    {
+      "epoch": 0.38727861221224136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4019679090044503e-05,
+      "loss": 1.0981,
+      "step": 10649
+    },
+    {
+      "epoch": 0.3873149798159799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.401860054869909e-05,
+      "loss": 0.9706,
+      "step": 10650
+    },
+    {
+      "epoch": 0.3873513474197185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.40175219516013e-05,
+      "loss": 1.0722,
+      "step": 10651
+    },
+    {
+      "epoch": 0.3873877150234571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4016443298766089e-05,
+      "loss": 1.0279,
+      "step": 10652
+    },
+    {
+      "epoch": 0.3874240826271957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4015364590208429e-05,
+      "loss": 1.0804,
+      "step": 10653
+    },
+    {
+      "epoch": 0.3874604502309343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4014285825943282e-05,
+      "loss": 1.1408,
+      "step": 10654
+    },
+    {
+      "epoch": 0.3874968178346729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4013207005985612e-05,
+      "loss": 0.9409,
+      "step": 10655
+    },
+    {
+      "epoch": 0.3875331854384115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.401212813035039e-05,
+      "loss": 1.041,
+      "step": 10656
+    },
+    {
+      "epoch": 0.3875695530421501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4011049199052586e-05,
+      "loss": 1.0578,
+      "step": 10657
+    },
+    {
+      "epoch": 0.38760592064588867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4009970212107163e-05,
+      "loss": 1.084,
+      "step": 10658
+    },
+    {
+      "epoch": 0.3876422882496272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.400889116952909e-05,
+      "loss": 1.1061,
+      "step": 10659
+    },
+    {
+      "epoch": 0.3876786558533658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4007812071333345e-05,
+      "loss": 1.0124,
+      "step": 10660
+    },
+    {
+      "epoch": 0.3877150234571044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.400673291753489e-05,
+      "loss": 1.044,
+      "step": 10661
+    },
+    {
+      "epoch": 0.387751391060843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4005653708148702e-05,
+      "loss": 1.0158,
+      "step": 10662
+    },
+    {
+      "epoch": 0.3877877586645816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4004574443189751e-05,
+      "loss": 1.0268,
+      "step": 10663
+    },
+    {
+      "epoch": 0.3878241262683202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4003495122673012e-05,
+      "loss": 1.1468,
+      "step": 10664
+    },
+    {
+      "epoch": 0.3878604938720588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4002415746613462e-05,
+      "loss": 1.035,
+      "step": 10665
+    },
+    {
+      "epoch": 0.3878968614757974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4001336315026066e-05,
+      "loss": 1.0082,
+      "step": 10666
+    },
+    {
+      "epoch": 0.3879332290795359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4000256827925812e-05,
+      "loss": 0.9903,
+      "step": 10667
+    },
+    {
+      "epoch": 0.3879695966832745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3999177285327665e-05,
+      "loss": 0.9554,
+      "step": 10668
+    },
+    {
+      "epoch": 0.3880059642870131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3998097687246608e-05,
+      "loss": 1.145,
+      "step": 10669
+    },
+    {
+      "epoch": 0.3880423318907517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3997018033697621e-05,
+      "loss": 1.0162,
+      "step": 10670
+    },
+    {
+      "epoch": 0.3880786994944903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3995938324695677e-05,
+      "loss": 0.8907,
+      "step": 10671
+    },
+    {
+      "epoch": 0.3881150670982289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3994858560255759e-05,
+      "loss": 1.102,
+      "step": 10672
+    },
+    {
+      "epoch": 0.3881514347019675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3993778740392847e-05,
+      "loss": 1.0768,
+      "step": 10673
+    },
+    {
+      "epoch": 0.3881878023057061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.399269886512192e-05,
+      "loss": 1.1906,
+      "step": 10674
+    },
+    {
+      "epoch": 0.3882241699094447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3991618934457963e-05,
+      "loss": 1.0697,
+      "step": 10675
+    },
+    {
+      "epoch": 0.38826053751318323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3990538948415957e-05,
+      "loss": 1.0916,
+      "step": 10676
+    },
+    {
+      "epoch": 0.38829690511692183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3989458907010882e-05,
+      "loss": 0.8887,
+      "step": 10677
+    },
+    {
+      "epoch": 0.3883332727206604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3988378810257727e-05,
+      "loss": 0.983,
+      "step": 10678
+    },
+    {
+      "epoch": 0.388369640324399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3987298658171473e-05,
+      "loss": 1.1359,
+      "step": 10679
+    },
+    {
+      "epoch": 0.3884060079281376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.398621845076711e-05,
+      "loss": 1.1424,
+      "step": 10680
+    },
+    {
+      "epoch": 0.3884423755318762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3985138188059622e-05,
+      "loss": 0.9644,
+      "step": 10681
+    },
+    {
+      "epoch": 0.3884787431356148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3984057870063991e-05,
+      "loss": 1.0465,
+      "step": 10682
+    },
+    {
+      "epoch": 0.3885151107393534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3982977496795215e-05,
+      "loss": 0.9491,
+      "step": 10683
+    },
+    {
+      "epoch": 0.38855147834309195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3981897068268278e-05,
+      "loss": 1.105,
+      "step": 10684
+    },
+    {
+      "epoch": 0.38858784594683055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3980816584498166e-05,
+      "loss": 0.9257,
+      "step": 10685
+    },
+    {
+      "epoch": 0.38862421355056914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3979736045499872e-05,
+      "loss": 0.9266,
+      "step": 10686
+    },
+    {
+      "epoch": 0.38866058115430774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3978655451288386e-05,
+      "loss": 0.9624,
+      "step": 10687
+    },
+    {
+      "epoch": 0.38869694875804633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3977574801878702e-05,
+      "loss": 1.0249,
+      "step": 10688
+    },
+    {
+      "epoch": 0.38873331636178493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3976494097285811e-05,
+      "loss": 1.0648,
+      "step": 10689
+    },
+    {
+      "epoch": 0.3887696839655235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3975413337524705e-05,
+      "loss": 1.061,
+      "step": 10690
+    },
+    {
+      "epoch": 0.3888060515692621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.397433252261038e-05,
+      "loss": 1.1249,
+      "step": 10691
+    },
+    {
+      "epoch": 0.38884241917300066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3973251652557831e-05,
+      "loss": 0.9343,
+      "step": 10692
+    },
+    {
+      "epoch": 0.38887878677673926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.397217072738205e-05,
+      "loss": 1.0718,
+      "step": 10693
+    },
+    {
+      "epoch": 0.38891515438047786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3971089747098039e-05,
+      "loss": 1.0963,
+      "step": 10694
+    },
+    {
+      "epoch": 0.38895152198421645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3970008711720791e-05,
+      "loss": 0.9462,
+      "step": 10695
+    },
+    {
+      "epoch": 0.38898788958795505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3968927621265305e-05,
+      "loss": 1.0976,
+      "step": 10696
+    },
+    {
+      "epoch": 0.38902425719169365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3967846475746578e-05,
+      "loss": 1.0391,
+      "step": 10697
+    },
+    {
+      "epoch": 0.38906062479543224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3966765275179613e-05,
+      "loss": 1.0477,
+      "step": 10698
+    },
+    {
+      "epoch": 0.38909699239917084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3965684019579406e-05,
+      "loss": 1.0653,
+      "step": 10699
+    },
+    {
+      "epoch": 0.38913336000290943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3964602708960962e-05,
+      "loss": 0.9228,
+      "step": 10700
+    },
+    {
+      "epoch": 0.389169727606648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3963521343339278e-05,
+      "loss": 1.004,
+      "step": 10701
+    },
+    {
+      "epoch": 0.38920609521038657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.396243992272936e-05,
+      "loss": 0.9839,
+      "step": 10702
+    },
+    {
+      "epoch": 0.38924246281412517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3961358447146211e-05,
+      "loss": 1.1516,
+      "step": 10703
+    },
+    {
+      "epoch": 0.38927883041786376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3960276916604834e-05,
+      "loss": 1.1384,
+      "step": 10704
+    },
+    {
+      "epoch": 0.38931519802160236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3959195331120232e-05,
+      "loss": 1.0551,
+      "step": 10705
+    },
+    {
+      "epoch": 0.38935156562534096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3958113690707413e-05,
+      "loss": 1.0863,
+      "step": 10706
+    },
+    {
+      "epoch": 0.38938793322907955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3957031995381383e-05,
+      "loss": 1.0322,
+      "step": 10707
+    },
+    {
+      "epoch": 0.38942430083281815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3955950245157145e-05,
+      "loss": 1.045,
+      "step": 10708
+    },
+    {
+      "epoch": 0.3894606684365567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3954868440049713e-05,
+      "loss": 1.126,
+      "step": 10709
+    },
+    {
+      "epoch": 0.3894970360402953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3953786580074094e-05,
+      "loss": 1.0147,
+      "step": 10710
+    },
+    {
+      "epoch": 0.3895334036440339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3952704665245293e-05,
+      "loss": 1.0117,
+      "step": 10711
+    },
+    {
+      "epoch": 0.3895697712477725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3951622695578324e-05,
+      "loss": 1.1029,
+      "step": 10712
+    },
+    {
+      "epoch": 0.3896061388515111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3950540671088198e-05,
+      "loss": 0.9886,
+      "step": 10713
+    },
+    {
+      "epoch": 0.38964250645524967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3949458591789922e-05,
+      "loss": 0.9395,
+      "step": 10714
+    },
+    {
+      "epoch": 0.38967887405898827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3948376457698513e-05,
+      "loss": 0.9963,
+      "step": 10715
+    },
+    {
+      "epoch": 0.38971524166272686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3947294268828985e-05,
+      "loss": 0.9939,
+      "step": 10716
+    },
+    {
+      "epoch": 0.38975160926646546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3946212025196347e-05,
+      "loss": 0.9807,
+      "step": 10717
+    },
+    {
+      "epoch": 0.389787976870204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3945129726815618e-05,
+      "loss": 1.0659,
+      "step": 10718
+    },
+    {
+      "epoch": 0.3898243444739426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.394404737370181e-05,
+      "loss": 1.1264,
+      "step": 10719
+    },
+    {
+      "epoch": 0.3898607120776812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.394296496586994e-05,
+      "loss": 1.0363,
+      "step": 10720
+    },
+    {
+      "epoch": 0.3898970796814198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3941882503335026e-05,
+      "loss": 1.0167,
+      "step": 10721
+    },
+    {
+      "epoch": 0.3899334472851584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3940799986112087e-05,
+      "loss": 1.0401,
+      "step": 10722
+    },
+    {
+      "epoch": 0.389969814888897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3939717414216138e-05,
+      "loss": 1.0007,
+      "step": 10723
+    },
+    {
+      "epoch": 0.3900061824926356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.39386347876622e-05,
+      "loss": 1.1008,
+      "step": 10724
+    },
+    {
+      "epoch": 0.3900425500963742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3937552106465291e-05,
+      "loss": 1.0684,
+      "step": 10725
+    },
+    {
+      "epoch": 0.3900789177001127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3936469370640435e-05,
+      "loss": 1.1624,
+      "step": 10726
+    },
+    {
+      "epoch": 0.3901152853038513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.393538658020265e-05,
+      "loss": 1.2521,
+      "step": 10727
+    },
+    {
+      "epoch": 0.3901516529075899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.393430373516696e-05,
+      "loss": 0.9793,
+      "step": 10728
+    },
+    {
+      "epoch": 0.3901880205113285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.393322083554839e-05,
+      "loss": 0.9771,
+      "step": 10729
+    },
+    {
+      "epoch": 0.3902243881150671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3932137881361957e-05,
+      "loss": 1.0371,
+      "step": 10730
+    },
+    {
+      "epoch": 0.3902607557188057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3931054872622696e-05,
+      "loss": 1.0163,
+      "step": 10731
+    },
+    {
+      "epoch": 0.3902971233225443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3929971809345622e-05,
+      "loss": 1.0369,
+      "step": 10732
+    },
+    {
+      "epoch": 0.3903334909262829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3928888691545766e-05,
+      "loss": 1.0385,
+      "step": 10733
+    },
+    {
+      "epoch": 0.39036985853002143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3927805519238154e-05,
+      "loss": 1.0809,
+      "step": 10734
+    },
+    {
+      "epoch": 0.39040622613376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.392672229243781e-05,
+      "loss": 1.2225,
+      "step": 10735
+    },
+    {
+      "epoch": 0.3904425937374986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3925639011159768e-05,
+      "loss": 1.0783,
+      "step": 10736
+    },
+    {
+      "epoch": 0.3904789613412372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3924555675419055e-05,
+      "loss": 1.0583,
+      "step": 10737
+    },
+    {
+      "epoch": 0.3905153289449758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.39234722852307e-05,
+      "loss": 1.0675,
+      "step": 10738
+    },
+    {
+      "epoch": 0.3905516965487144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3922388840609733e-05,
+      "loss": 0.9148,
+      "step": 10739
+    },
+    {
+      "epoch": 0.390588064152453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3921305341571188e-05,
+      "loss": 1.1066,
+      "step": 10740
+    },
+    {
+      "epoch": 0.3906244317561916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3920221788130095e-05,
+      "loss": 1.2365,
+      "step": 10741
+    },
+    {
+      "epoch": 0.3906607993599302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3919138180301486e-05,
+      "loss": 1.0453,
+      "step": 10742
+    },
+    {
+      "epoch": 0.39069716696366874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3918054518100393e-05,
+      "loss": 0.9367,
+      "step": 10743
+    },
+    {
+      "epoch": 0.39073353456740734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3916970801541854e-05,
+      "loss": 0.9539,
+      "step": 10744
+    },
+    {
+      "epoch": 0.39076990217114593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3915887030640905e-05,
+      "loss": 0.9401,
+      "step": 10745
+    },
+    {
+      "epoch": 0.39080626977488453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3914803205412575e-05,
+      "loss": 1.1049,
+      "step": 10746
+    },
+    {
+      "epoch": 0.3908426373786231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3913719325871907e-05,
+      "loss": 1.102,
+      "step": 10747
+    },
+    {
+      "epoch": 0.3908790049823617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3912635392033938e-05,
+      "loss": 1.153,
+      "step": 10748
+    },
+    {
+      "epoch": 0.3909153725861003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3911551403913702e-05,
+      "loss": 1.0952,
+      "step": 10749
+    },
+    {
+      "epoch": 0.3909517401898389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3910467361526238e-05,
+      "loss": 0.9802,
+      "step": 10750
+    },
+    {
+      "epoch": 0.39098810779357746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.390938326488659e-05,
+      "loss": 1.0176,
+      "step": 10751
+    },
+    {
+      "epoch": 0.39102447539731605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3908299114009797e-05,
+      "loss": 0.9841,
+      "step": 10752
+    },
+    {
+      "epoch": 0.39106084300105465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3907214908910897e-05,
+      "loss": 1.098,
+      "step": 10753
+    },
+    {
+      "epoch": 0.39109721060479324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3906130649604933e-05,
+      "loss": 1.0718,
+      "step": 10754
+    },
+    {
+      "epoch": 0.39113357820853184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3905046336106952e-05,
+      "loss": 0.9441,
+      "step": 10755
+    },
+    {
+      "epoch": 0.39116994581227044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3903961968431992e-05,
+      "loss": 1.0119,
+      "step": 10756
+    },
+    {
+      "epoch": 0.39120631341600903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3902877546595097e-05,
+      "loss": 0.9328,
+      "step": 10757
+    },
+    {
+      "epoch": 0.39124268101974763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3901793070611316e-05,
+      "loss": 1.1044,
+      "step": 10758
+    },
+    {
+      "epoch": 0.3912790486234862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.390070854049569e-05,
+      "loss": 0.9803,
+      "step": 10759
+    },
+    {
+      "epoch": 0.39131541622722477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3899623956263271e-05,
+      "loss": 0.9744,
+      "step": 10760
+    },
+    {
+      "epoch": 0.39135178383096336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3898539317929101e-05,
+      "loss": 1.0472,
+      "step": 10761
+    },
+    {
+      "epoch": 0.39138815143470196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3897454625508224e-05,
+      "loss": 1.04,
+      "step": 10762
+    },
+    {
+      "epoch": 0.39142451903844055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3896369879015698e-05,
+      "loss": 0.9877,
+      "step": 10763
+    },
+    {
+      "epoch": 0.39146088664217915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3895285078466569e-05,
+      "loss": 0.9973,
+      "step": 10764
+    },
+    {
+      "epoch": 0.39149725424591775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3894200223875885e-05,
+      "loss": 1.0897,
+      "step": 10765
+    },
+    {
+      "epoch": 0.39153362184965634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3893115315258699e-05,
+      "loss": 0.9277,
+      "step": 10766
+    },
+    {
+      "epoch": 0.39156998945339494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3892030352630059e-05,
+      "loss": 1.1458,
+      "step": 10767
+    },
+    {
+      "epoch": 0.3916063570571335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3890945336005022e-05,
+      "loss": 1.1139,
+      "step": 10768
+    },
+    {
+      "epoch": 0.3916427246608721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.388986026539864e-05,
+      "loss": 1.0299,
+      "step": 10769
+    },
+    {
+      "epoch": 0.3916790922646107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3888775140825963e-05,
+      "loss": 1.1561,
+      "step": 10770
+    },
+    {
+      "epoch": 0.39171545986834927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3887689962302052e-05,
+      "loss": 0.9966,
+      "step": 10771
+    },
+    {
+      "epoch": 0.39175182747208787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3886604729841957e-05,
+      "loss": 1.1249,
+      "step": 10772
+    },
+    {
+      "epoch": 0.39178819507582646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3885519443460733e-05,
+      "loss": 0.8843,
+      "step": 10773
+    },
+    {
+      "epoch": 0.39182456267956506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3884434103173443e-05,
+      "loss": 1.1595,
+      "step": 10774
+    },
+    {
+      "epoch": 0.39186093028330365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.388334870899514e-05,
+      "loss": 0.9237,
+      "step": 10775
+    },
+    {
+      "epoch": 0.3918972978870422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3882263260940883e-05,
+      "loss": 1.0881,
+      "step": 10776
+    },
+    {
+      "epoch": 0.3919336654907808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3881177759025732e-05,
+      "loss": 1.1591,
+      "step": 10777
+    },
+    {
+      "epoch": 0.3919700330945194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3880092203264744e-05,
+      "loss": 1.0461,
+      "step": 10778
+    },
+    {
+      "epoch": 0.392006400698258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3879006593672986e-05,
+      "loss": 1.0269,
+      "step": 10779
+    },
+    {
+      "epoch": 0.3920427683019966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3877920930265511e-05,
+      "loss": 1.0649,
+      "step": 10780
+    },
+    {
+      "epoch": 0.3920791359057352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3876835213057389e-05,
+      "loss": 1.0316,
+      "step": 10781
+    },
+    {
+      "epoch": 0.3921155035094738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3875749442063677e-05,
+      "loss": 1.0374,
+      "step": 10782
+    },
+    {
+      "epoch": 0.39215187111321237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3874663617299441e-05,
+      "loss": 1.1092,
+      "step": 10783
+    },
+    {
+      "epoch": 0.39218823871695097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3873577738779744e-05,
+      "loss": 1.064,
+      "step": 10784
+    },
+    {
+      "epoch": 0.3922246063206895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3872491806519653e-05,
+      "loss": 1.039,
+      "step": 10785
+    },
+    {
+      "epoch": 0.3922609739244281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3871405820534229e-05,
+      "loss": 0.9837,
+      "step": 10786
+    },
+    {
+      "epoch": 0.3922973415281667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3870319780838548e-05,
+      "loss": 0.8651,
+      "step": 10787
+    },
+    {
+      "epoch": 0.3923337091319053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3869233687447667e-05,
+      "loss": 0.9798,
+      "step": 10788
+    },
+    {
+      "epoch": 0.3923700767356439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.386814754037666e-05,
+      "loss": 1.0915,
+      "step": 10789
+    },
+    {
+      "epoch": 0.3924064443393825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3867061339640593e-05,
+      "loss": 1.1284,
+      "step": 10790
+    },
+    {
+      "epoch": 0.3924428119431211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3865975085254538e-05,
+      "loss": 1.0724,
+      "step": 10791
+    },
+    {
+      "epoch": 0.3924791795468597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.386488877723356e-05,
+      "loss": 1.1698,
+      "step": 10792
+    },
+    {
+      "epoch": 0.3925155471505982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3863802415592738e-05,
+      "loss": 1.2076,
+      "step": 10793
+    },
+    {
+      "epoch": 0.3925519147543368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3862716000347137e-05,
+      "loss": 1.0134,
+      "step": 10794
+    },
+    {
+      "epoch": 0.3925882823580754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3861629531511833e-05,
+      "loss": 1.0425,
+      "step": 10795
+    },
+    {
+      "epoch": 0.392624649961814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3860543009101899e-05,
+      "loss": 0.9814,
+      "step": 10796
+    },
+    {
+      "epoch": 0.3926610175655526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3859456433132406e-05,
+      "loss": 0.9372,
+      "step": 10797
+    },
+    {
+      "epoch": 0.3926973851692912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3858369803618432e-05,
+      "loss": 0.9699,
+      "step": 10798
+    },
+    {
+      "epoch": 0.3927337527730298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.385728312057505e-05,
+      "loss": 1.0291,
+      "step": 10799
+    },
+    {
+      "epoch": 0.3927701203767684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3856196384017342e-05,
+      "loss": 0.9601,
+      "step": 10800
+    },
+    {
+      "epoch": 0.392806487980507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3855109593960379e-05,
+      "loss": 0.9482,
+      "step": 10801
+    },
+    {
+      "epoch": 0.39284285558424553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3854022750419236e-05,
+      "loss": 0.9536,
+      "step": 10802
+    },
+    {
+      "epoch": 0.39287922318798413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3852935853408998e-05,
+      "loss": 1.0617,
+      "step": 10803
+    },
+    {
+      "epoch": 0.3929155907917227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3851848902944743e-05,
+      "loss": 1.1155,
+      "step": 10804
+    },
+    {
+      "epoch": 0.3929519583954613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3850761899041546e-05,
+      "loss": 1.0439,
+      "step": 10805
+    },
+    {
+      "epoch": 0.3929883259991999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3849674841714493e-05,
+      "loss": 1.0194,
+      "step": 10806
+    },
+    {
+      "epoch": 0.3930246936029385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3848587730978662e-05,
+      "loss": 0.9578,
+      "step": 10807
+    },
+    {
+      "epoch": 0.3930610612066771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3847500566849142e-05,
+      "loss": 1.0201,
+      "step": 10808
+    },
+    {
+      "epoch": 0.3930974288104157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3846413349341006e-05,
+      "loss": 0.9935,
+      "step": 10809
+    },
+    {
+      "epoch": 0.39313379641415425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3845326078469341e-05,
+      "loss": 1.1435,
+      "step": 10810
+    },
+    {
+      "epoch": 0.39317016401789284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3844238754249235e-05,
+      "loss": 1.0124,
+      "step": 10811
+    },
+    {
+      "epoch": 0.39320653162163144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3843151376695771e-05,
+      "loss": 0.975,
+      "step": 10812
+    },
+    {
+      "epoch": 0.39324289922537004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3842063945824031e-05,
+      "loss": 0.8908,
+      "step": 10813
+    },
+    {
+      "epoch": 0.39327926682910863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3840976461649109e-05,
+      "loss": 1.0738,
+      "step": 10814
+    },
+    {
+      "epoch": 0.3933156344328472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3839888924186087e-05,
+      "loss": 1.078,
+      "step": 10815
+    },
+    {
+      "epoch": 0.3933520020365858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3838801333450054e-05,
+      "loss": 1.0309,
+      "step": 10816
+    },
+    {
+      "epoch": 0.3933883696403244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3837713689456098e-05,
+      "loss": 1.0183,
+      "step": 10817
+    },
+    {
+      "epoch": 0.393424737244063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.383662599221931e-05,
+      "loss": 1.114,
+      "step": 10818
+    },
+    {
+      "epoch": 0.39346110484780156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3835538241754783e-05,
+      "loss": 1.0157,
+      "step": 10819
+    },
+    {
+      "epoch": 0.39349747245154015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3834450438077603e-05,
+      "loss": 1.1516,
+      "step": 10820
+    },
+    {
+      "epoch": 0.39353384005527875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3833362581202861e-05,
+      "loss": 0.9148,
+      "step": 10821
+    },
+    {
+      "epoch": 0.39357020765901735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3832274671145653e-05,
+      "loss": 1.0439,
+      "step": 10822
+    },
+    {
+      "epoch": 0.39360657526275594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3831186707921071e-05,
+      "loss": 1.0907,
+      "step": 10823
+    },
+    {
+      "epoch": 0.39364294286649454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3830098691544211e-05,
+      "loss": 1.0966,
+      "step": 10824
+    },
+    {
+      "epoch": 0.39367931047023313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3829010622030166e-05,
+      "loss": 1.0269,
+      "step": 10825
+    },
+    {
+      "epoch": 0.39371567807397173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.382792249939403e-05,
+      "loss": 1.0263,
+      "step": 10826
+    },
+    {
+      "epoch": 0.39375204567771027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 1.0415,
+      "step": 10827
+    },
+    {
+      "epoch": 0.39378841328144887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3825746094815875e-05,
+      "loss": 0.9521,
+      "step": 10828
+    },
+    {
+      "epoch": 0.39382478088518746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3824657812904047e-05,
+      "loss": 1.0676,
+      "step": 10829
+    },
+    {
+      "epoch": 0.39386114848892606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3823569477930524e-05,
+      "loss": 1.0644,
+      "step": 10830
+    },
+    {
+      "epoch": 0.39389751609266466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3822481089910396e-05,
+      "loss": 1.0674,
+      "step": 10831
+    },
+    {
+      "epoch": 0.39393388369640325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3821392648858765e-05,
+      "loss": 0.9426,
+      "step": 10832
+    },
+    {
+      "epoch": 0.39397025130014185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3820304154790736e-05,
+      "loss": 1.137,
+      "step": 10833
+    },
+    {
+      "epoch": 0.39400661890388045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3819215607721404e-05,
+      "loss": 1.0067,
+      "step": 10834
+    },
+    {
+      "epoch": 0.394042986507619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3818127007665876e-05,
+      "loss": 1.1392,
+      "step": 10835
+    },
+    {
+      "epoch": 0.3940793541113576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3817038354639253e-05,
+      "loss": 1.0669,
+      "step": 10836
+    },
+    {
+      "epoch": 0.3941157217150962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3815949648656642e-05,
+      "loss": 1.0207,
+      "step": 10837
+    },
+    {
+      "epoch": 0.3941520893188348,
+      "grad_norm": 0.0,
+      "learning_rate": 1.381486088973314e-05,
+      "loss": 0.9824,
+      "step": 10838
+    },
+    {
+      "epoch": 0.39418845692257337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3813772077883854e-05,
+      "loss": 0.9984,
+      "step": 10839
+    },
+    {
+      "epoch": 0.39422482452631197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3812683213123894e-05,
+      "loss": 1.0152,
+      "step": 10840
+    },
+    {
+      "epoch": 0.39426119213005056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3811594295468365e-05,
+      "loss": 1.2214,
+      "step": 10841
+    },
+    {
+      "epoch": 0.39429755973378916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3810505324932369e-05,
+      "loss": 1.1192,
+      "step": 10842
+    },
+    {
+      "epoch": 0.39433392733752776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3809416301531023e-05,
+      "loss": 1.0874,
+      "step": 10843
+    },
+    {
+      "epoch": 0.3943702949412663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3808327225279427e-05,
+      "loss": 1.005,
+      "step": 10844
+    },
+    {
+      "epoch": 0.3944066625450049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3807238096192695e-05,
+      "loss": 0.8677,
+      "step": 10845
+    },
+    {
+      "epoch": 0.3944430301487435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3806148914285938e-05,
+      "loss": 0.9218,
+      "step": 10846
+    },
+    {
+      "epoch": 0.3944793977524821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3805059679574261e-05,
+      "loss": 1.1188,
+      "step": 10847
+    },
+    {
+      "epoch": 0.3945157653562207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3803970392072783e-05,
+      "loss": 1.0334,
+      "step": 10848
+    },
+    {
+      "epoch": 0.3945521329599593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3802881051796613e-05,
+      "loss": 1.0364,
+      "step": 10849
+    },
+    {
+      "epoch": 0.3945885005636979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3801791658760861e-05,
+      "loss": 1.0071,
+      "step": 10850
+    },
+    {
+      "epoch": 0.39462486816743647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3800702212980646e-05,
+      "loss": 1.1875,
+      "step": 10851
+    },
+    {
+      "epoch": 0.394661235771175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.379961271447108e-05,
+      "loss": 1.0976,
+      "step": 10852
+    },
+    {
+      "epoch": 0.3946976033749136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3798523163247279e-05,
+      "loss": 0.9648,
+      "step": 10853
+    },
+    {
+      "epoch": 0.3947339709786522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3797433559324357e-05,
+      "loss": 0.9967,
+      "step": 10854
+    },
+    {
+      "epoch": 0.3947703385823908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3796343902717435e-05,
+      "loss": 1.0297,
+      "step": 10855
+    },
+    {
+      "epoch": 0.3948067061861294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3795254193441628e-05,
+      "loss": 0.9354,
+      "step": 10856
+    },
+    {
+      "epoch": 0.394843073789868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3794164431512052e-05,
+      "loss": 1.1586,
+      "step": 10857
+    },
+    {
+      "epoch": 0.3948794413936066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3793074616943824e-05,
+      "loss": 1.0475,
+      "step": 10858
+    },
+    {
+      "epoch": 0.3949158089973452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3791984749752073e-05,
+      "loss": 0.9619,
+      "step": 10859
+    },
+    {
+      "epoch": 0.3949521766010838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.379089482995191e-05,
+      "loss": 1.0336,
+      "step": 10860
+    },
+    {
+      "epoch": 0.3949885442048223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3789804857558462e-05,
+      "loss": 1.0641,
+      "step": 10861
+    },
+    {
+      "epoch": 0.3950249118085609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3788714832586848e-05,
+      "loss": 1.016,
+      "step": 10862
+    },
+    {
+      "epoch": 0.3950612794122995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3787624755052192e-05,
+      "loss": 0.9493,
+      "step": 10863
+    },
+    {
+      "epoch": 0.3950976470160381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3786534624969617e-05,
+      "loss": 0.9568,
+      "step": 10864
+    },
+    {
+      "epoch": 0.3951340146197767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3785444442354243e-05,
+      "loss": 0.9532,
+      "step": 10865
+    },
+    {
+      "epoch": 0.3951703822235153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.37843542072212e-05,
+      "loss": 1.1095,
+      "step": 10866
+    },
+    {
+      "epoch": 0.3952067498272539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3783263919585611e-05,
+      "loss": 1.0587,
+      "step": 10867
+    },
+    {
+      "epoch": 0.3952431174309925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3782173579462606e-05,
+      "loss": 1.2252,
+      "step": 10868
+    },
+    {
+      "epoch": 0.39527948503473104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3781083186867306e-05,
+      "loss": 1.0025,
+      "step": 10869
+    },
+    {
+      "epoch": 0.39531585263846963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3779992741814843e-05,
+      "loss": 1.0584,
+      "step": 10870
+    },
+    {
+      "epoch": 0.39535222024220823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3778902244320343e-05,
+      "loss": 1.0208,
+      "step": 10871
+    },
+    {
+      "epoch": 0.3953885878459468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3777811694398935e-05,
+      "loss": 0.9738,
+      "step": 10872
+    },
+    {
+      "epoch": 0.3954249554496854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.377672109206575e-05,
+      "loss": 1.1365,
+      "step": 10873
+    },
+    {
+      "epoch": 0.395461323053424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.377563043733592e-05,
+      "loss": 1.2016,
+      "step": 10874
+    },
+    {
+      "epoch": 0.3954976906571626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3774539730224575e-05,
+      "loss": 1.0833,
+      "step": 10875
+    },
+    {
+      "epoch": 0.3955340582609012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3773448970746845e-05,
+      "loss": 0.9963,
+      "step": 10876
+    },
+    {
+      "epoch": 0.39557042586463975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3772358158917864e-05,
+      "loss": 0.9365,
+      "step": 10877
+    },
+    {
+      "epoch": 0.39560679346837835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3771267294752766e-05,
+      "loss": 1.0225,
+      "step": 10878
+    },
+    {
+      "epoch": 0.39564316107211694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3770176378266684e-05,
+      "loss": 1.0057,
+      "step": 10879
+    },
+    {
+      "epoch": 0.39567952867585554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3769085409474757e-05,
+      "loss": 1.049,
+      "step": 10880
+    },
+    {
+      "epoch": 0.39571589627959414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3767994388392116e-05,
+      "loss": 0.9758,
+      "step": 10881
+    },
+    {
+      "epoch": 0.39575226388333273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3766903315033896e-05,
+      "loss": 0.9379,
+      "step": 10882
+    },
+    {
+      "epoch": 0.39578863148707133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3765812189415241e-05,
+      "loss": 1.1,
+      "step": 10883
+    },
+    {
+      "epoch": 0.3958249990908099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3764721011551286e-05,
+      "loss": 1.0775,
+      "step": 10884
+    },
+    {
+      "epoch": 0.3958613666945485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3763629781457163e-05,
+      "loss": 0.9647,
+      "step": 10885
+    },
+    {
+      "epoch": 0.39589773429828706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3762538499148021e-05,
+      "loss": 1.014,
+      "step": 10886
+    },
+    {
+      "epoch": 0.39593410190202566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3761447164638995e-05,
+      "loss": 1.0277,
+      "step": 10887
+    },
+    {
+      "epoch": 0.39597046950576426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3760355777945225e-05,
+      "loss": 1.0057,
+      "step": 10888
+    },
+    {
+      "epoch": 0.39600683710950285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3759264339081855e-05,
+      "loss": 1.0317,
+      "step": 10889
+    },
+    {
+      "epoch": 0.39604320471324145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3758172848064023e-05,
+      "loss": 1.0851,
+      "step": 10890
+    },
+    {
+      "epoch": 0.39607957231698004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.375708130490688e-05,
+      "loss": 1.0965,
+      "step": 10891
+    },
+    {
+      "epoch": 0.39611593992071864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3755989709625563e-05,
+      "loss": 1.0114,
+      "step": 10892
+    },
+    {
+      "epoch": 0.39615230752445724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3754898062235216e-05,
+      "loss": 1.0248,
+      "step": 10893
+    },
+    {
+      "epoch": 0.3961886751281958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3753806362750987e-05,
+      "loss": 0.931,
+      "step": 10894
+    },
+    {
+      "epoch": 0.3962250427319344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.375271461118802e-05,
+      "loss": 0.9595,
+      "step": 10895
+    },
+    {
+      "epoch": 0.39626141033567297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3751622807561465e-05,
+      "loss": 1.0568,
+      "step": 10896
+    },
+    {
+      "epoch": 0.39629777793941157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3750530951886463e-05,
+      "loss": 1.0367,
+      "step": 10897
+    },
+    {
+      "epoch": 0.39633414554315016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3749439044178166e-05,
+      "loss": 1.0496,
+      "step": 10898
+    },
+    {
+      "epoch": 0.39637051314688876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3748347084451723e-05,
+      "loss": 0.8891,
+      "step": 10899
+    },
+    {
+      "epoch": 0.39640688075062736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3747255072722281e-05,
+      "loss": 1.0394,
+      "step": 10900
+    },
+    {
+      "epoch": 0.39644324835436595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3746163009004991e-05,
+      "loss": 1.0496,
+      "step": 10901
+    },
+    {
+      "epoch": 0.39647961595810455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3745070893315007e-05,
+      "loss": 0.9721,
+      "step": 10902
+    },
+    {
+      "epoch": 0.3965159835618431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3743978725667477e-05,
+      "loss": 1.1057,
+      "step": 10903
+    },
+    {
+      "epoch": 0.3965523511655817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3742886506077553e-05,
+      "loss": 1.1432,
+      "step": 10904
+    },
+    {
+      "epoch": 0.3965887187693203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.374179423456039e-05,
+      "loss": 1.1046,
+      "step": 10905
+    },
+    {
+      "epoch": 0.3966250863730589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.374070191113114e-05,
+      "loss": 1.076,
+      "step": 10906
+    },
+    {
+      "epoch": 0.3966614539767975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.373960953580496e-05,
+      "loss": 1.1824,
+      "step": 10907
+    },
+    {
+      "epoch": 0.39669782158053607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3738517108597002e-05,
+      "loss": 1.0018,
+      "step": 10908
+    },
+    {
+      "epoch": 0.39673418918427467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3737424629522424e-05,
+      "loss": 1.0674,
+      "step": 10909
+    },
+    {
+      "epoch": 0.39677055678801326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3736332098596383e-05,
+      "loss": 1.0765,
+      "step": 10910
+    },
+    {
+      "epoch": 0.3968069243917518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3735239515834034e-05,
+      "loss": 1.0126,
+      "step": 10911
+    },
+    {
+      "epoch": 0.3968432919954904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3734146881250536e-05,
+      "loss": 1.0459,
+      "step": 10912
+    },
+    {
+      "epoch": 0.396879659599229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3733054194861049e-05,
+      "loss": 1.0436,
+      "step": 10913
+    },
+    {
+      "epoch": 0.3969160272029676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3731961456680734e-05,
+      "loss": 0.9304,
+      "step": 10914
+    },
+    {
+      "epoch": 0.3969523948067062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3730868666724746e-05,
+      "loss": 1.0376,
+      "step": 10915
+    },
+    {
+      "epoch": 0.3969887624104448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.372977582500825e-05,
+      "loss": 0.966,
+      "step": 10916
+    },
+    {
+      "epoch": 0.3970251300141834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3728682931546402e-05,
+      "loss": 1.0257,
+      "step": 10917
+    },
+    {
+      "epoch": 0.397061497617922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3727589986354372e-05,
+      "loss": 1.1285,
+      "step": 10918
+    },
+    {
+      "epoch": 0.3970978652216605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.372649698944732e-05,
+      "loss": 1.0636,
+      "step": 10919
+    },
+    {
+      "epoch": 0.3971342328253991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.372540394084041e-05,
+      "loss": 1.038,
+      "step": 10920
+    },
+    {
+      "epoch": 0.3971706004291377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3724310840548804e-05,
+      "loss": 1.138,
+      "step": 10921
+    },
+    {
+      "epoch": 0.3972069680328763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.372321768858767e-05,
+      "loss": 1.0551,
+      "step": 10922
+    },
+    {
+      "epoch": 0.3972433356366149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3722124484972173e-05,
+      "loss": 1.1055,
+      "step": 10923
+    },
+    {
+      "epoch": 0.3972797032403535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3721031229717481e-05,
+      "loss": 1.039,
+      "step": 10924
+    },
+    {
+      "epoch": 0.3973160708440921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3719937922838757e-05,
+      "loss": 1.0751,
+      "step": 10925
+    },
+    {
+      "epoch": 0.3973524384478307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3718844564351175e-05,
+      "loss": 1.1815,
+      "step": 10926
+    },
+    {
+      "epoch": 0.3973888060515693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3717751154269901e-05,
+      "loss": 1.1171,
+      "step": 10927
+    },
+    {
+      "epoch": 0.39742517365530783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3716657692610102e-05,
+      "loss": 0.9895,
+      "step": 10928
+    },
+    {
+      "epoch": 0.3974615412590464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3715564179386953e-05,
+      "loss": 1.0478,
+      "step": 10929
+    },
+    {
+      "epoch": 0.397497908862785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3714470614615623e-05,
+      "loss": 0.9846,
+      "step": 10930
+    },
+    {
+      "epoch": 0.3975342764665236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3713376998311281e-05,
+      "loss": 1.0144,
+      "step": 10931
+    },
+    {
+      "epoch": 0.3975706440702622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3712283330489104e-05,
+      "loss": 1.0428,
+      "step": 10932
+    },
+    {
+      "epoch": 0.3976070116740008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3711189611164263e-05,
+      "loss": 1.0656,
+      "step": 10933
+    },
+    {
+      "epoch": 0.3976433792777394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3710095840351932e-05,
+      "loss": 0.905,
+      "step": 10934
+    },
+    {
+      "epoch": 0.397679746881478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3709002018067283e-05,
+      "loss": 1.0725,
+      "step": 10935
+    },
+    {
+      "epoch": 0.39771611448521654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3707908144325495e-05,
+      "loss": 1.1052,
+      "step": 10936
+    },
+    {
+      "epoch": 0.39775248208895514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3706814219141742e-05,
+      "loss": 0.8645,
+      "step": 10937
+    },
+    {
+      "epoch": 0.39778884969269374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.37057202425312e-05,
+      "loss": 0.9986,
+      "step": 10938
+    },
+    {
+      "epoch": 0.39782521729643233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.370462621450905e-05,
+      "loss": 1.0707,
+      "step": 10939
+    },
+    {
+      "epoch": 0.39786158490017093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3703532135090465e-05,
+      "loss": 1.1212,
+      "step": 10940
+    },
+    {
+      "epoch": 0.3978979525039095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3702438004290625e-05,
+      "loss": 1.0945,
+      "step": 10941
+    },
+    {
+      "epoch": 0.3979343201076481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3701343822124713e-05,
+      "loss": 1.0956,
+      "step": 10942
+    },
+    {
+      "epoch": 0.3979706877113867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3700249588607908e-05,
+      "loss": 0.9737,
+      "step": 10943
+    },
+    {
+      "epoch": 0.3980070553151253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3699155303755388e-05,
+      "loss": 1.0726,
+      "step": 10944
+    },
+    {
+      "epoch": 0.39804342291886385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3698060967582337e-05,
+      "loss": 1.0303,
+      "step": 10945
+    },
+    {
+      "epoch": 0.39807979052260245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3696966580103936e-05,
+      "loss": 1.0261,
+      "step": 10946
+    },
+    {
+      "epoch": 0.39811615812634105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.369587214133537e-05,
+      "loss": 1.1541,
+      "step": 10947
+    },
+    {
+      "epoch": 0.39815252573007964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3694777651291822e-05,
+      "loss": 1.0601,
+      "step": 10948
+    },
+    {
+      "epoch": 0.39818889333381824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3693683109988474e-05,
+      "loss": 0.9584,
+      "step": 10949
+    },
+    {
+      "epoch": 0.39822526093755684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3692588517440517e-05,
+      "loss": 1.1362,
+      "step": 10950
+    },
+    {
+      "epoch": 0.39826162854129543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3691493873663133e-05,
+      "loss": 1.1259,
+      "step": 10951
+    },
+    {
+      "epoch": 0.39829799614503403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3690399178671508e-05,
+      "loss": 1.1506,
+      "step": 10952
+    },
+    {
+      "epoch": 0.39833436374877257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3689304432480832e-05,
+      "loss": 1.0696,
+      "step": 10953
+    },
+    {
+      "epoch": 0.39837073135251116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3688209635106288e-05,
+      "loss": 0.9759,
+      "step": 10954
+    },
+    {
+      "epoch": 0.39840709895624976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3687114786563072e-05,
+      "loss": 0.9579,
+      "step": 10955
+    },
+    {
+      "epoch": 0.39844346655998836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3686019886866367e-05,
+      "loss": 1.1394,
+      "step": 10956
+    },
+    {
+      "epoch": 0.39847983416372695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3684924936031365e-05,
+      "loss": 1.0767,
+      "step": 10957
+    },
+    {
+      "epoch": 0.39851620176746555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.368382993407326e-05,
+      "loss": 1.1801,
+      "step": 10958
+    },
+    {
+      "epoch": 0.39855256937120415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3682734881007242e-05,
+      "loss": 1.1161,
+      "step": 10959
+    },
+    {
+      "epoch": 0.39858893697494274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3681639776848502e-05,
+      "loss": 0.9519,
+      "step": 10960
+    },
+    {
+      "epoch": 0.3986253045786813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3680544621612231e-05,
+      "loss": 1.1594,
+      "step": 10961
+    },
+    {
+      "epoch": 0.3986616721824199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3679449415313626e-05,
+      "loss": 1.0214,
+      "step": 10962
+    },
+    {
+      "epoch": 0.3986980397861585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3678354157967884e-05,
+      "loss": 0.9908,
+      "step": 10963
+    },
+    {
+      "epoch": 0.39873440738989707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3677258849590196e-05,
+      "loss": 1.0611,
+      "step": 10964
+    },
+    {
+      "epoch": 0.39877077499363567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3676163490195754e-05,
+      "loss": 1.0219,
+      "step": 10965
+    },
+    {
+      "epoch": 0.39880714259737426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3675068079799765e-05,
+      "loss": 0.9747,
+      "step": 10966
+    },
+    {
+      "epoch": 0.39884351020111286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3673972618417421e-05,
+      "loss": 0.9326,
+      "step": 10967
+    },
+    {
+      "epoch": 0.39887987780485146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3672877106063915e-05,
+      "loss": 1.0484,
+      "step": 10968
+    },
+    {
+      "epoch": 0.39891624540859005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3671781542754455e-05,
+      "loss": 1.0105,
+      "step": 10969
+    },
+    {
+      "epoch": 0.3989526130123286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3670685928504232e-05,
+      "loss": 0.9332,
+      "step": 10970
+    },
+    {
+      "epoch": 0.3989889806160672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3669590263328453e-05,
+      "loss": 0.933,
+      "step": 10971
+    },
+    {
+      "epoch": 0.3990253482198058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3668494547242315e-05,
+      "loss": 1.0131,
+      "step": 10972
+    },
+    {
+      "epoch": 0.3990617158235444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3667398780261017e-05,
+      "loss": 0.8788,
+      "step": 10973
+    },
+    {
+      "epoch": 0.399098083427283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3666302962399767e-05,
+      "loss": 1.0391,
+      "step": 10974
+    },
+    {
+      "epoch": 0.3991344510310216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3665207093673765e-05,
+      "loss": 1.0514,
+      "step": 10975
+    },
+    {
+      "epoch": 0.39917081863476017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3664111174098214e-05,
+      "loss": 1.1211,
+      "step": 10976
+    },
+    {
+      "epoch": 0.39920718623849877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3663015203688319e-05,
+      "loss": 1.0441,
+      "step": 10977
+    },
+    {
+      "epoch": 0.3992435538422373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3661919182459285e-05,
+      "loss": 1.065,
+      "step": 10978
+    },
+    {
+      "epoch": 0.3992799214459759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3660823110426318e-05,
+      "loss": 1.0735,
+      "step": 10979
+    },
+    {
+      "epoch": 0.3993162890497145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3659726987604626e-05,
+      "loss": 1.0514,
+      "step": 10980
+    },
+    {
+      "epoch": 0.3993526566534531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3658630814009413e-05,
+      "loss": 1.1957,
+      "step": 10981
+    },
+    {
+      "epoch": 0.3993890242571917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3657534589655888e-05,
+      "loss": 1.0096,
+      "step": 10982
+    },
+    {
+      "epoch": 0.3994253918609303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3656438314559262e-05,
+      "loss": 1.1144,
+      "step": 10983
+    },
+    {
+      "epoch": 0.3994617594646689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.365534198873474e-05,
+      "loss": 1.0915,
+      "step": 10984
+    },
+    {
+      "epoch": 0.3994981270684075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3654245612197534e-05,
+      "loss": 1.0604,
+      "step": 10985
+    },
+    {
+      "epoch": 0.3995344946721461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3653149184962857e-05,
+      "loss": 1.0522,
+      "step": 10986
+    },
+    {
+      "epoch": 0.3995708622758846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3652052707045916e-05,
+      "loss": 1.0165,
+      "step": 10987
+    },
+    {
+      "epoch": 0.3996072298796232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3650956178461927e-05,
+      "loss": 1.0408,
+      "step": 10988
+    },
+    {
+      "epoch": 0.3996435974833618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3649859599226099e-05,
+      "loss": 0.9958,
+      "step": 10989
+    },
+    {
+      "epoch": 0.3996799650871004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3648762969353651e-05,
+      "loss": 1.0964,
+      "step": 10990
+    },
+    {
+      "epoch": 0.399716332690839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3647666288859794e-05,
+      "loss": 1.0036,
+      "step": 10991
+    },
+    {
+      "epoch": 0.3997527002945776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3646569557759736e-05,
+      "loss": 0.9699,
+      "step": 10992
+    },
+    {
+      "epoch": 0.3997890678983162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3645472776068707e-05,
+      "loss": 1.0324,
+      "step": 10993
+    },
+    {
+      "epoch": 0.3998254355020548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.364437594380191e-05,
+      "loss": 1.1147,
+      "step": 10994
+    },
+    {
+      "epoch": 0.39986180310579333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3643279060974571e-05,
+      "loss": 1.0779,
+      "step": 10995
+    },
+    {
+      "epoch": 0.39989817070953193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3642182127601902e-05,
+      "loss": 1.0019,
+      "step": 10996
+    },
+    {
+      "epoch": 0.3999345383132705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3641085143699122e-05,
+      "loss": 0.9399,
+      "step": 10997
+    },
+    {
+      "epoch": 0.3999709059170091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3639988109281455e-05,
+      "loss": 0.9155,
+      "step": 10998
+    },
+    {
+      "epoch": 0.4000072735207477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3638891024364118e-05,
+      "loss": 0.9863,
+      "step": 10999
+    },
+    {
+      "epoch": 0.4000436411244863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3637793888962328e-05,
+      "loss": 1.0577,
+      "step": 11000
+    },
+    {
+      "epoch": 0.4000800087282249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.363669670309131e-05,
+      "loss": 1.0041,
+      "step": 11001
+    },
+    {
+      "epoch": 0.4001163763319635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3635599466766285e-05,
+      "loss": 1.0453,
+      "step": 11002
+    },
+    {
+      "epoch": 0.4001527439357021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3634502180002475e-05,
+      "loss": 0.9827,
+      "step": 11003
+    },
+    {
+      "epoch": 0.40018911153944065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3633404842815106e-05,
+      "loss": 1.0431,
+      "step": 11004
+    },
+    {
+      "epoch": 0.40022547914317924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3632307455219398e-05,
+      "loss": 0.971,
+      "step": 11005
+    },
+    {
+      "epoch": 0.40026184674691784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3631210017230578e-05,
+      "loss": 1.0081,
+      "step": 11006
+    },
+    {
+      "epoch": 0.40029821435065643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3630112528863872e-05,
+      "loss": 0.9895,
+      "step": 11007
+    },
+    {
+      "epoch": 0.40033458195439503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3629014990134504e-05,
+      "loss": 1.0608,
+      "step": 11008
+    },
+    {
+      "epoch": 0.4003709495581336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3627917401057702e-05,
+      "loss": 0.9242,
+      "step": 11009
+    },
+    {
+      "epoch": 0.4004073171618722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3626819761648696e-05,
+      "loss": 0.9698,
+      "step": 11010
+    },
+    {
+      "epoch": 0.4004436847656108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3625722071922712e-05,
+      "loss": 0.9417,
+      "step": 11011
+    },
+    {
+      "epoch": 0.40048005236934936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3624624331894975e-05,
+      "loss": 1.0521,
+      "step": 11012
+    },
+    {
+      "epoch": 0.40051641997308796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3623526541580719e-05,
+      "loss": 0.9482,
+      "step": 11013
+    },
+    {
+      "epoch": 0.40055278757682655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3622428700995172e-05,
+      "loss": 0.9955,
+      "step": 11014
+    },
+    {
+      "epoch": 0.40058915518056515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3621330810153568e-05,
+      "loss": 1.0293,
+      "step": 11015
+    },
+    {
+      "epoch": 0.40062552278430374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3620232869071137e-05,
+      "loss": 1.1807,
+      "step": 11016
+    },
+    {
+      "epoch": 0.40066189038804234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3619134877763112e-05,
+      "loss": 1.136,
+      "step": 11017
+    },
+    {
+      "epoch": 0.40069825799178094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3618036836244723e-05,
+      "loss": 0.9439,
+      "step": 11018
+    },
+    {
+      "epoch": 0.40073462559551953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3616938744531207e-05,
+      "loss": 1.0957,
+      "step": 11019
+    },
+    {
+      "epoch": 0.4007709931992581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.36158406026378e-05,
+      "loss": 1.0245,
+      "step": 11020
+    },
+    {
+      "epoch": 0.40080736080299667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3614742410579731e-05,
+      "loss": 1.0257,
+      "step": 11021
+    },
+    {
+      "epoch": 0.40084372840673527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3613644168372244e-05,
+      "loss": 0.9875,
+      "step": 11022
+    },
+    {
+      "epoch": 0.40088009601047386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.361254587603057e-05,
+      "loss": 1.0291,
+      "step": 11023
+    },
+    {
+      "epoch": 0.40091646361421246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3611447533569947e-05,
+      "loss": 1.0298,
+      "step": 11024
+    },
+    {
+      "epoch": 0.40095283121795106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3610349141005612e-05,
+      "loss": 0.94,
+      "step": 11025
+    },
+    {
+      "epoch": 0.40098919882168965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3609250698352807e-05,
+      "loss": 1.0841,
+      "step": 11026
+    },
+    {
+      "epoch": 0.40102556642542825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3608152205626769e-05,
+      "loss": 1.0991,
+      "step": 11027
+    },
+    {
+      "epoch": 0.40106193402916684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3607053662842738e-05,
+      "loss": 1.0787,
+      "step": 11028
+    },
+    {
+      "epoch": 0.4010983016329054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3605955070015955e-05,
+      "loss": 1.0429,
+      "step": 11029
+    },
+    {
+      "epoch": 0.401134669236644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3604856427161663e-05,
+      "loss": 1.0201,
+      "step": 11030
+    },
+    {
+      "epoch": 0.4011710368403826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3603757734295102e-05,
+      "loss": 0.9299,
+      "step": 11031
+    },
+    {
+      "epoch": 0.4012074044441212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3602658991431514e-05,
+      "loss": 1.0622,
+      "step": 11032
+    },
+    {
+      "epoch": 0.40124377204785977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3601560198586145e-05,
+      "loss": 0.9238,
+      "step": 11033
+    },
+    {
+      "epoch": 0.40128013965159837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3600461355774237e-05,
+      "loss": 1.0136,
+      "step": 11034
+    },
+    {
+      "epoch": 0.40131650725533696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3599362463011038e-05,
+      "loss": 0.9897,
+      "step": 11035
+    },
+    {
+      "epoch": 0.40135287485907556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3598263520311794e-05,
+      "loss": 1.0798,
+      "step": 11036
+    },
+    {
+      "epoch": 0.4013892424628141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3597164527691744e-05,
+      "loss": 1.0143,
+      "step": 11037
+    },
+    {
+      "epoch": 0.4014256100665527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3596065485166141e-05,
+      "loss": 0.9768,
+      "step": 11038
+    },
+    {
+      "epoch": 0.4014619776702913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3594966392750233e-05,
+      "loss": 0.9596,
+      "step": 11039
+    },
+    {
+      "epoch": 0.4014983452740299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3593867250459267e-05,
+      "loss": 0.9936,
+      "step": 11040
+    },
+    {
+      "epoch": 0.4015347128777685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3592768058308492e-05,
+      "loss": 1.1751,
+      "step": 11041
+    },
+    {
+      "epoch": 0.4015710804815071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3591668816313158e-05,
+      "loss": 1.0276,
+      "step": 11042
+    },
+    {
+      "epoch": 0.4016074480852457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.359056952448851e-05,
+      "loss": 0.9445,
+      "step": 11043
+    },
+    {
+      "epoch": 0.4016438156889843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.358947018284981e-05,
+      "loss": 0.9831,
+      "step": 11044
+    },
+    {
+      "epoch": 0.40168018329272287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.35883707914123e-05,
+      "loss": 0.8852,
+      "step": 11045
+    },
+    {
+      "epoch": 0.4017165508964614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.358727135019124e-05,
+      "loss": 1.144,
+      "step": 11046
+    },
+    {
+      "epoch": 0.4017529185002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.358617185920188e-05,
+      "loss": 1.1128,
+      "step": 11047
+    },
+    {
+      "epoch": 0.4017892861039386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3585072318459472e-05,
+      "loss": 0.9487,
+      "step": 11048
+    },
+    {
+      "epoch": 0.4018256537076772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3583972727979274e-05,
+      "loss": 1.2028,
+      "step": 11049
+    },
+    {
+      "epoch": 0.4018620213114158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3582873087776536e-05,
+      "loss": 0.944,
+      "step": 11050
+    },
+    {
+      "epoch": 0.4018983889151544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3581773397866517e-05,
+      "loss": 1.022,
+      "step": 11051
+    },
+    {
+      "epoch": 0.401934756518893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3580673658264479e-05,
+      "loss": 1.1403,
+      "step": 11052
+    },
+    {
+      "epoch": 0.4019711241226316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3579573868985669e-05,
+      "loss": 0.9603,
+      "step": 11053
+    },
+    {
+      "epoch": 0.4020074917263701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3578474030045352e-05,
+      "loss": 1.0643,
+      "step": 11054
+    },
+    {
+      "epoch": 0.4020438593301087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3577374141458785e-05,
+      "loss": 1.0477,
+      "step": 11055
+    },
+    {
+      "epoch": 0.4020802269338473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3576274203241227e-05,
+      "loss": 1.0616,
+      "step": 11056
+    },
+    {
+      "epoch": 0.4021165945375859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.357517421540794e-05,
+      "loss": 1.0009,
+      "step": 11057
+    },
+    {
+      "epoch": 0.4021529621413245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3574074177974184e-05,
+      "loss": 0.9939,
+      "step": 11058
+    },
+    {
+      "epoch": 0.4021893297450631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3572974090955217e-05,
+      "loss": 1.0031,
+      "step": 11059
+    },
+    {
+      "epoch": 0.4022256973488017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3571873954366305e-05,
+      "loss": 1.0221,
+      "step": 11060
+    },
+    {
+      "epoch": 0.4022620649525403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.357077376822271e-05,
+      "loss": 1.0552,
+      "step": 11061
+    },
+    {
+      "epoch": 0.40229843255627884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3569673532539696e-05,
+      "loss": 1.0406,
+      "step": 11062
+    },
+    {
+      "epoch": 0.40233480016001744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3568573247332527e-05,
+      "loss": 1.0166,
+      "step": 11063
+    },
+    {
+      "epoch": 0.40237116776375603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3567472912616464e-05,
+      "loss": 1.0278,
+      "step": 11064
+    },
+    {
+      "epoch": 0.40240753536749463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3566372528406777e-05,
+      "loss": 1.0574,
+      "step": 11065
+    },
+    {
+      "epoch": 0.4024439029712332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3565272094718733e-05,
+      "loss": 1.0567,
+      "step": 11066
+    },
+    {
+      "epoch": 0.4024802705749718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3564171611567599e-05,
+      "loss": 1.0836,
+      "step": 11067
+    },
+    {
+      "epoch": 0.4025166381787104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3563071078968637e-05,
+      "loss": 1.1231,
+      "step": 11068
+    },
+    {
+      "epoch": 0.402553005782449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3561970496937118e-05,
+      "loss": 0.9456,
+      "step": 11069
+    },
+    {
+      "epoch": 0.4025893733861876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3560869865488316e-05,
+      "loss": 1.1215,
+      "step": 11070
+    },
+    {
+      "epoch": 0.40262574098992615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3559769184637494e-05,
+      "loss": 1.0594,
+      "step": 11071
+    },
+    {
+      "epoch": 0.40266210859366475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3558668454399925e-05,
+      "loss": 0.9741,
+      "step": 11072
+    },
+    {
+      "epoch": 0.40269847619740334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3557567674790883e-05,
+      "loss": 0.9774,
+      "step": 11073
+    },
+    {
+      "epoch": 0.40273484380114194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3556466845825637e-05,
+      "loss": 0.9982,
+      "step": 11074
+    },
+    {
+      "epoch": 0.40277121140488054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3555365967519455e-05,
+      "loss": 1.1175,
+      "step": 11075
+    },
+    {
+      "epoch": 0.40280757900861913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.355426503988762e-05,
+      "loss": 1.0691,
+      "step": 11076
+    },
+    {
+      "epoch": 0.40284394661235773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3553164062945396e-05,
+      "loss": 0.9843,
+      "step": 11077
+    },
+    {
+      "epoch": 0.4028803142160963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3552063036708066e-05,
+      "loss": 0.9355,
+      "step": 11078
+    },
+    {
+      "epoch": 0.40291668181983487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3550961961190897e-05,
+      "loss": 1.0058,
+      "step": 11079
+    },
+    {
+      "epoch": 0.40295304942357346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3549860836409171e-05,
+      "loss": 0.9612,
+      "step": 11080
+    },
+    {
+      "epoch": 0.40298941702731206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3548759662378163e-05,
+      "loss": 0.8873,
+      "step": 11081
+    },
+    {
+      "epoch": 0.40302578463105065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3547658439113151e-05,
+      "loss": 0.9701,
+      "step": 11082
+    },
+    {
+      "epoch": 0.40306215223478925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3546557166629408e-05,
+      "loss": 1.0486,
+      "step": 11083
+    },
+    {
+      "epoch": 0.40309851983852785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3545455844942219e-05,
+      "loss": 0.9846,
+      "step": 11084
+    },
+    {
+      "epoch": 0.40313488744226644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.354435447406686e-05,
+      "loss": 1.0404,
+      "step": 11085
+    },
+    {
+      "epoch": 0.40317125504600504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3543253054018612e-05,
+      "loss": 1.0356,
+      "step": 11086
+    },
+    {
+      "epoch": 0.40320762264974364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3542151584812755e-05,
+      "loss": 0.8895,
+      "step": 11087
+    },
+    {
+      "epoch": 0.4032439902534822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3541050066464568e-05,
+      "loss": 0.9852,
+      "step": 11088
+    },
+    {
+      "epoch": 0.4032803578572208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3539948498989337e-05,
+      "loss": 1.0964,
+      "step": 11089
+    },
+    {
+      "epoch": 0.40331672546095937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3538846882402342e-05,
+      "loss": 0.8998,
+      "step": 11090
+    },
+    {
+      "epoch": 0.40335309306469797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3537745216718869e-05,
+      "loss": 0.9162,
+      "step": 11091
+    },
+    {
+      "epoch": 0.40338946066843656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.35366435019542e-05,
+      "loss": 1.0201,
+      "step": 11092
+    },
+    {
+      "epoch": 0.40342582827217516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3535541738123621e-05,
+      "loss": 1.0359,
+      "step": 11093
+    },
+    {
+      "epoch": 0.40346219587591375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3534439925242415e-05,
+      "loss": 1.0312,
+      "step": 11094
+    },
+    {
+      "epoch": 0.40349856347965235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.353333806332587e-05,
+      "loss": 1.0097,
+      "step": 11095
+    },
+    {
+      "epoch": 0.4035349310833909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3532236152389273e-05,
+      "loss": 1.1202,
+      "step": 11096
+    },
+    {
+      "epoch": 0.4035712986871295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3531134192447913e-05,
+      "loss": 1.1198,
+      "step": 11097
+    },
+    {
+      "epoch": 0.4036076662908681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3530032183517074e-05,
+      "loss": 1.0934,
+      "step": 11098
+    },
+    {
+      "epoch": 0.4036440338946067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3528930125612047e-05,
+      "loss": 1.1698,
+      "step": 11099
+    },
+    {
+      "epoch": 0.4036804014983453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3527828018748122e-05,
+      "loss": 1.0346,
+      "step": 11100
+    },
+    {
+      "epoch": 0.40371676910208387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3526725862940586e-05,
+      "loss": 0.9334,
+      "step": 11101
+    },
+    {
+      "epoch": 0.40375313670582247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3525623658204737e-05,
+      "loss": 1.1366,
+      "step": 11102
+    },
+    {
+      "epoch": 0.40378950430956106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.352452140455586e-05,
+      "loss": 1.0474,
+      "step": 11103
+    },
+    {
+      "epoch": 0.4038258719132996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3523419102009247e-05,
+      "loss": 0.9605,
+      "step": 11104
+    },
+    {
+      "epoch": 0.4038622395170382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3522316750580197e-05,
+      "loss": 1.0694,
+      "step": 11105
+    },
+    {
+      "epoch": 0.4038986071207768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3521214350283996e-05,
+      "loss": 1.065,
+      "step": 11106
+    },
+    {
+      "epoch": 0.4039349747245154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3520111901135943e-05,
+      "loss": 1.1333,
+      "step": 11107
+    },
+    {
+      "epoch": 0.403971342328254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3519009403151335e-05,
+      "loss": 1.0464,
+      "step": 11108
+    },
+    {
+      "epoch": 0.4040077099319926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3517906856345459e-05,
+      "loss": 0.9373,
+      "step": 11109
+    },
+    {
+      "epoch": 0.4040440775357312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3516804260733621e-05,
+      "loss": 1.1393,
+      "step": 11110
+    },
+    {
+      "epoch": 0.4040804451394698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3515701616331111e-05,
+      "loss": 1.1243,
+      "step": 11111
+    },
+    {
+      "epoch": 0.4041168127432084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3514598923153229e-05,
+      "loss": 1.03,
+      "step": 11112
+    },
+    {
+      "epoch": 0.4041531803469469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3513496181215273e-05,
+      "loss": 1.0195,
+      "step": 11113
+    },
+    {
+      "epoch": 0.4041895479506855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3512393390532545e-05,
+      "loss": 1.0405,
+      "step": 11114
+    },
+    {
+      "epoch": 0.4042259155544241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.351129055112034e-05,
+      "loss": 0.9998,
+      "step": 11115
+    },
+    {
+      "epoch": 0.4042622831581627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3510187662993963e-05,
+      "loss": 0.9706,
+      "step": 11116
+    },
+    {
+      "epoch": 0.4042986507619013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3509084726168708e-05,
+      "loss": 0.9939,
+      "step": 11117
+    },
+    {
+      "epoch": 0.4043350183656399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3507981740659886e-05,
+      "loss": 1.0816,
+      "step": 11118
+    },
+    {
+      "epoch": 0.4043713859693785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.350687870648279e-05,
+      "loss": 0.9206,
+      "step": 11119
+    },
+    {
+      "epoch": 0.4044077535731171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3505775623652731e-05,
+      "loss": 1.0596,
+      "step": 11120
+    },
+    {
+      "epoch": 0.40444412117685563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.350467249218501e-05,
+      "loss": 1.0451,
+      "step": 11121
+    },
+    {
+      "epoch": 0.4044804887805942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3503569312094928e-05,
+      "loss": 1.0728,
+      "step": 11122
+    },
+    {
+      "epoch": 0.4045168563843328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3502466083397794e-05,
+      "loss": 1.0295,
+      "step": 11123
+    },
+    {
+      "epoch": 0.4045532239880714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3501362806108913e-05,
+      "loss": 1.0247,
+      "step": 11124
+    },
+    {
+      "epoch": 0.40458959159181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.350025948024359e-05,
+      "loss": 0.9827,
+      "step": 11125
+    },
+    {
+      "epoch": 0.4046259591955486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3499156105817133e-05,
+      "loss": 1.1318,
+      "step": 11126
+    },
+    {
+      "epoch": 0.4046623267992872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3498052682844849e-05,
+      "loss": 1.2144,
+      "step": 11127
+    },
+    {
+      "epoch": 0.4046986944030258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3496949211342047e-05,
+      "loss": 1.0829,
+      "step": 11128
+    },
+    {
+      "epoch": 0.4047350620067644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.349584569132404e-05,
+      "loss": 1.0676,
+      "step": 11129
+    },
+    {
+      "epoch": 0.40477142961050294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.349474212280613e-05,
+      "loss": 1.0576,
+      "step": 11130
+    },
+    {
+      "epoch": 0.40480779721424154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.349363850580363e-05,
+      "loss": 1.0481,
+      "step": 11131
+    },
+    {
+      "epoch": 0.40484416481798013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3492534840331858e-05,
+      "loss": 0.9097,
+      "step": 11132
+    },
+    {
+      "epoch": 0.40488053242171873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3491431126406114e-05,
+      "loss": 0.9198,
+      "step": 11133
+    },
+    {
+      "epoch": 0.4049169000254573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3490327364041722e-05,
+      "loss": 0.979,
+      "step": 11134
+    },
+    {
+      "epoch": 0.4049532676291959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3489223553253987e-05,
+      "loss": 1.027,
+      "step": 11135
+    },
+    {
+      "epoch": 0.4049896352329345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3488119694058225e-05,
+      "loss": 1.0209,
+      "step": 11136
+    },
+    {
+      "epoch": 0.4050260028366731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3487015786469755e-05,
+      "loss": 0.9906,
+      "step": 11137
+    },
+    {
+      "epoch": 0.40506237044041166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3485911830503886e-05,
+      "loss": 1.1001,
+      "step": 11138
+    },
+    {
+      "epoch": 0.40509873804415025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3484807826175936e-05,
+      "loss": 1.002,
+      "step": 11139
+    },
+    {
+      "epoch": 0.40513510564788885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3483703773501225e-05,
+      "loss": 1.0871,
+      "step": 11140
+    },
+    {
+      "epoch": 0.40517147325162745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3482599672495062e-05,
+      "loss": 0.9562,
+      "step": 11141
+    },
+    {
+      "epoch": 0.40520784085536604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3481495523172772e-05,
+      "loss": 0.9373,
+      "step": 11142
+    },
+    {
+      "epoch": 0.40524420845910464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.348039132554967e-05,
+      "loss": 1.1602,
+      "step": 11143
+    },
+    {
+      "epoch": 0.40528057606284323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.347928707964108e-05,
+      "loss": 1.0642,
+      "step": 11144
+    },
+    {
+      "epoch": 0.40531694366658183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3478182785462317e-05,
+      "loss": 1.1011,
+      "step": 11145
+    },
+    {
+      "epoch": 0.40535331127032037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.34770784430287e-05,
+      "loss": 1.0577,
+      "step": 11146
+    },
+    {
+      "epoch": 0.40538967887405897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3475974052355554e-05,
+      "loss": 1.0847,
+      "step": 11147
+    },
+    {
+      "epoch": 0.40542604647779756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3474869613458202e-05,
+      "loss": 1.0328,
+      "step": 11148
+    },
+    {
+      "epoch": 0.40546241408153616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3473765126351958e-05,
+      "loss": 1.0203,
+      "step": 11149
+    },
+    {
+      "epoch": 0.40549878168527476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3472660591052158e-05,
+      "loss": 0.9605,
+      "step": 11150
+    },
+    {
+      "epoch": 0.40553514928901335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3471556007574117e-05,
+      "loss": 0.9488,
+      "step": 11151
+    },
+    {
+      "epoch": 0.40557151689275195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3470451375933163e-05,
+      "loss": 0.9196,
+      "step": 11152
+    },
+    {
+      "epoch": 0.40560788449649054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3469346696144618e-05,
+      "loss": 1.0044,
+      "step": 11153
+    },
+    {
+      "epoch": 0.40564425210022914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3468241968223814e-05,
+      "loss": 1.0609,
+      "step": 11154
+    },
+    {
+      "epoch": 0.4056806197039677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3467137192186068e-05,
+      "loss": 1.0957,
+      "step": 11155
+    },
+    {
+      "epoch": 0.4057169873077063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3466032368046717e-05,
+      "loss": 1.0034,
+      "step": 11156
+    },
+    {
+      "epoch": 0.4057533549114449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3464927495821086e-05,
+      "loss": 1.0224,
+      "step": 11157
+    },
+    {
+      "epoch": 0.40578972251518347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3463822575524495e-05,
+      "loss": 1.1402,
+      "step": 11158
+    },
+    {
+      "epoch": 0.40582609011892207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3462717607172287e-05,
+      "loss": 1.0298,
+      "step": 11159
+    },
+    {
+      "epoch": 0.40586245772266066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.346161259077978e-05,
+      "loss": 0.9841,
+      "step": 11160
+    },
+    {
+      "epoch": 0.40589882532639926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3460507526362311e-05,
+      "loss": 1.0855,
+      "step": 11161
+    },
+    {
+      "epoch": 0.40593519293013786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3459402413935214e-05,
+      "loss": 1.0454,
+      "step": 11162
+    },
+    {
+      "epoch": 0.4059715605338764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3458297253513815e-05,
+      "loss": 1.0474,
+      "step": 11163
+    },
+    {
+      "epoch": 0.406007928137615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3457192045113448e-05,
+      "loss": 1.1234,
+      "step": 11164
+    },
+    {
+      "epoch": 0.4060442957413536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3456086788749443e-05,
+      "loss": 1.0726,
+      "step": 11165
+    },
+    {
+      "epoch": 0.4060806633450922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.345498148443714e-05,
+      "loss": 0.9942,
+      "step": 11166
+    },
+    {
+      "epoch": 0.4061170309488308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3453876132191873e-05,
+      "loss": 1.0425,
+      "step": 11167
+    },
+    {
+      "epoch": 0.4061533985525694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.345277073202897e-05,
+      "loss": 0.992,
+      "step": 11168
+    },
+    {
+      "epoch": 0.406189766156308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3451665283963777e-05,
+      "loss": 1.1054,
+      "step": 11169
+    },
+    {
+      "epoch": 0.40622613376004657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3450559788011623e-05,
+      "loss": 1.0329,
+      "step": 11170
+    },
+    {
+      "epoch": 0.40626250136378517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3449454244187847e-05,
+      "loss": 1.1099,
+      "step": 11171
+    },
+    {
+      "epoch": 0.4062988689675237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.344834865250779e-05,
+      "loss": 0.9995,
+      "step": 11172
+    },
+    {
+      "epoch": 0.4063352365712623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3447243012986784e-05,
+      "loss": 0.8399,
+      "step": 11173
+    },
+    {
+      "epoch": 0.4063716041750009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3446137325640176e-05,
+      "loss": 1.0328,
+      "step": 11174
+    },
+    {
+      "epoch": 0.4064079717787395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3445031590483302e-05,
+      "loss": 1.0262,
+      "step": 11175
+    },
+    {
+      "epoch": 0.4064443393824781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.34439258075315e-05,
+      "loss": 1.0212,
+      "step": 11176
+    },
+    {
+      "epoch": 0.4064807069862167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3442819976800114e-05,
+      "loss": 1.0652,
+      "step": 11177
+    },
+    {
+      "epoch": 0.4065170745899553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3441714098304488e-05,
+      "loss": 1.1365,
+      "step": 11178
+    },
+    {
+      "epoch": 0.4065534421936939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3440608172059959e-05,
+      "loss": 0.9496,
+      "step": 11179
+    },
+    {
+      "epoch": 0.4065898097974324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3439502198081875e-05,
+      "loss": 0.9573,
+      "step": 11180
+    },
+    {
+      "epoch": 0.406626177401171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.343839617638558e-05,
+      "loss": 1.0897,
+      "step": 11181
+    },
+    {
+      "epoch": 0.4066625450049096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3437290106986414e-05,
+      "loss": 1.1935,
+      "step": 11182
+    },
+    {
+      "epoch": 0.4066989126086482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3436183989899725e-05,
+      "loss": 1.0255,
+      "step": 11183
+    },
+    {
+      "epoch": 0.4067352802123868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3435077825140859e-05,
+      "loss": 1.0228,
+      "step": 11184
+    },
+    {
+      "epoch": 0.4067716478161254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3433971612725161e-05,
+      "loss": 1.1184,
+      "step": 11185
+    },
+    {
+      "epoch": 0.406808015419864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.343286535266798e-05,
+      "loss": 1.1313,
+      "step": 11186
+    },
+    {
+      "epoch": 0.4068443830236026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.343175904498466e-05,
+      "loss": 1.0317,
+      "step": 11187
+    },
+    {
+      "epoch": 0.40688075062734114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3430652689690555e-05,
+      "loss": 1.0944,
+      "step": 11188
+    },
+    {
+      "epoch": 0.40691711823107973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3429546286801009e-05,
+      "loss": 1.1008,
+      "step": 11189
+    },
+    {
+      "epoch": 0.40695348583481833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3428439836331377e-05,
+      "loss": 0.9683,
+      "step": 11190
+    },
+    {
+      "epoch": 0.4069898534385569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3427333338297008e-05,
+      "loss": 1.1398,
+      "step": 11191
+    },
+    {
+      "epoch": 0.4070262210422955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3426226792713246e-05,
+      "loss": 1.1398,
+      "step": 11192
+    },
+    {
+      "epoch": 0.4070625886460341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3425120199595455e-05,
+      "loss": 0.9706,
+      "step": 11193
+    },
+    {
+      "epoch": 0.4070989562497727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3424013558958978e-05,
+      "loss": 1.0486,
+      "step": 11194
+    },
+    {
+      "epoch": 0.4071353238535113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3422906870819169e-05,
+      "loss": 0.9506,
+      "step": 11195
+    },
+    {
+      "epoch": 0.4071716914572499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3421800135191386e-05,
+      "loss": 1.0128,
+      "step": 11196
+    },
+    {
+      "epoch": 0.40720805906098845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3420693352090979e-05,
+      "loss": 1.0484,
+      "step": 11197
+    },
+    {
+      "epoch": 0.40724442666472704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3419586521533306e-05,
+      "loss": 1.0851,
+      "step": 11198
+    },
+    {
+      "epoch": 0.40728079426846564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3418479643533722e-05,
+      "loss": 1.0245,
+      "step": 11199
+    },
+    {
+      "epoch": 0.40731716187220424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3417372718107582e-05,
+      "loss": 1.0282,
+      "step": 11200
+    },
+    {
+      "epoch": 0.40735352947594283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3416265745270248e-05,
+      "loss": 1.025,
+      "step": 11201
+    },
+    {
+      "epoch": 0.40738989707968143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3415158725037072e-05,
+      "loss": 1.0703,
+      "step": 11202
+    },
+    {
+      "epoch": 0.40742626468342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3414051657423411e-05,
+      "loss": 1.1084,
+      "step": 11203
+    },
+    {
+      "epoch": 0.4074626322871586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.341294454244463e-05,
+      "loss": 1.101,
+      "step": 11204
+    },
+    {
+      "epoch": 0.40749899989089716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3411837380116085e-05,
+      "loss": 1.0689,
+      "step": 11205
+    },
+    {
+      "epoch": 0.40753536749463576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3410730170453137e-05,
+      "loss": 1.026,
+      "step": 11206
+    },
+    {
+      "epoch": 0.40757173509837435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3409622913471147e-05,
+      "loss": 0.9914,
+      "step": 11207
+    },
+    {
+      "epoch": 0.40760810270211295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3408515609185477e-05,
+      "loss": 1.0632,
+      "step": 11208
+    },
+    {
+      "epoch": 0.40764447030585155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.340740825761149e-05,
+      "loss": 1.0676,
+      "step": 11209
+    },
+    {
+      "epoch": 0.40768083790959014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3406300858764547e-05,
+      "loss": 0.9538,
+      "step": 11210
+    },
+    {
+      "epoch": 0.40771720551332874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3405193412660011e-05,
+      "loss": 1.0536,
+      "step": 11211
+    },
+    {
+      "epoch": 0.40775357311706734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.340408591931325e-05,
+      "loss": 1.0696,
+      "step": 11212
+    },
+    {
+      "epoch": 0.40778994072080593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3402978378739625e-05,
+      "loss": 0.969,
+      "step": 11213
+    },
+    {
+      "epoch": 0.4078263083245445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3401870790954505e-05,
+      "loss": 0.8713,
+      "step": 11214
+    },
+    {
+      "epoch": 0.40786267592828307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3400763155973253e-05,
+      "loss": 1.0853,
+      "step": 11215
+    },
+    {
+      "epoch": 0.40789904353202167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3399655473811237e-05,
+      "loss": 1.102,
+      "step": 11216
+    },
+    {
+      "epoch": 0.40793541113576026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3398547744483824e-05,
+      "loss": 1.0761,
+      "step": 11217
+    },
+    {
+      "epoch": 0.40797177873949886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3397439968006387e-05,
+      "loss": 1.0479,
+      "step": 11218
+    },
+    {
+      "epoch": 0.40800814634323745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3396332144394286e-05,
+      "loss": 0.9836,
+      "step": 11219
+    },
+    {
+      "epoch": 0.40804451394697605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3395224273662902e-05,
+      "loss": 0.9679,
+      "step": 11220
+    },
+    {
+      "epoch": 0.40808088155071465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.339411635582759e-05,
+      "loss": 0.9942,
+      "step": 11221
+    },
+    {
+      "epoch": 0.4081172491544532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3393008390903734e-05,
+      "loss": 1.1184,
+      "step": 11222
+    },
+    {
+      "epoch": 0.4081536167581918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3391900378906702e-05,
+      "loss": 1.1378,
+      "step": 11223
+    },
+    {
+      "epoch": 0.4081899843619304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3390792319851859e-05,
+      "loss": 0.9849,
+      "step": 11224
+    },
+    {
+      "epoch": 0.408226351965669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.338968421375459e-05,
+      "loss": 1.026,
+      "step": 11225
+    },
+    {
+      "epoch": 0.4082627195694076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3388576060630258e-05,
+      "loss": 0.9761,
+      "step": 11226
+    },
+    {
+      "epoch": 0.40829908717314617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.338746786049424e-05,
+      "loss": 0.9857,
+      "step": 11227
+    },
+    {
+      "epoch": 0.40833545477688477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3386359613361915e-05,
+      "loss": 1.1252,
+      "step": 11228
+    },
+    {
+      "epoch": 0.40837182238062336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3385251319248654e-05,
+      "loss": 0.9362,
+      "step": 11229
+    },
+    {
+      "epoch": 0.40840818998436196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3384142978169832e-05,
+      "loss": 1.0182,
+      "step": 11230
+    },
+    {
+      "epoch": 0.4084445575881005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.338303459014083e-05,
+      "loss": 1.0286,
+      "step": 11231
+    },
+    {
+      "epoch": 0.4084809251918391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3381926155177021e-05,
+      "loss": 1.1699,
+      "step": 11232
+    },
+    {
+      "epoch": 0.4085172927955777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3380817673293785e-05,
+      "loss": 1.1017,
+      "step": 11233
+    },
+    {
+      "epoch": 0.4085536603993163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3379709144506504e-05,
+      "loss": 1.0129,
+      "step": 11234
+    },
+    {
+      "epoch": 0.4085900280030549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3378600568830548e-05,
+      "loss": 0.9608,
+      "step": 11235
+    },
+    {
+      "epoch": 0.4086263956067935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3377491946281306e-05,
+      "loss": 1.0797,
+      "step": 11236
+    },
+    {
+      "epoch": 0.4086627632105321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3376383276874157e-05,
+      "loss": 1.1357,
+      "step": 11237
+    },
+    {
+      "epoch": 0.4086991308142707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3375274560624476e-05,
+      "loss": 1.0247,
+      "step": 11238
+    },
+    {
+      "epoch": 0.4087354984180092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3374165797547654e-05,
+      "loss": 0.9515,
+      "step": 11239
+    },
+    {
+      "epoch": 0.4087718660217478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3373056987659065e-05,
+      "loss": 1.0779,
+      "step": 11240
+    },
+    {
+      "epoch": 0.4088082336254864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.33719481309741e-05,
+      "loss": 1.0477,
+      "step": 11241
+    },
+    {
+      "epoch": 0.408844601229225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3370839227508136e-05,
+      "loss": 0.9282,
+      "step": 11242
+    },
+    {
+      "epoch": 0.4088809688329636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3369730277276561e-05,
+      "loss": 1.0048,
+      "step": 11243
+    },
+    {
+      "epoch": 0.4089173364367022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.336862128029476e-05,
+      "loss": 1.0343,
+      "step": 11244
+    },
+    {
+      "epoch": 0.4089537040404408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3367512236578115e-05,
+      "loss": 0.9261,
+      "step": 11245
+    },
+    {
+      "epoch": 0.4089900716441794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.336640314614202e-05,
+      "loss": 1.0238,
+      "step": 11246
+    },
+    {
+      "epoch": 0.40902643924791793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3365294009001857e-05,
+      "loss": 1.1214,
+      "step": 11247
+    },
+    {
+      "epoch": 0.4090628068516565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3364184825173012e-05,
+      "loss": 1.0025,
+      "step": 11248
+    },
+    {
+      "epoch": 0.4090991744553951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3363075594670877e-05,
+      "loss": 0.9661,
+      "step": 11249
+    },
+    {
+      "epoch": 0.4091355420591337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3361966317510841e-05,
+      "loss": 1.1147,
+      "step": 11250
+    },
+    {
+      "epoch": 0.4091719096628723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3360856993708289e-05,
+      "loss": 0.9752,
+      "step": 11251
+    },
+    {
+      "epoch": 0.4092082772666109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3359747623278619e-05,
+      "loss": 1.0557,
+      "step": 11252
+    },
+    {
+      "epoch": 0.4092446448703495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3358638206237213e-05,
+      "loss": 1.0972,
+      "step": 11253
+    },
+    {
+      "epoch": 0.4092810124740881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.335752874259947e-05,
+      "loss": 1.0173,
+      "step": 11254
+    },
+    {
+      "epoch": 0.4093173800778267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.335641923238078e-05,
+      "loss": 0.9311,
+      "step": 11255
+    },
+    {
+      "epoch": 0.40935374768156524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3355309675596533e-05,
+      "loss": 1.1591,
+      "step": 11256
+    },
+    {
+      "epoch": 0.40939011528530383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3354200072262124e-05,
+      "loss": 1.0616,
+      "step": 11257
+    },
+    {
+      "epoch": 0.40942648288904243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.335309042239295e-05,
+      "loss": 1.1138,
+      "step": 11258
+    },
+    {
+      "epoch": 0.409462850492781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3351980726004406e-05,
+      "loss": 1.0287,
+      "step": 11259
+    },
+    {
+      "epoch": 0.4094992180965196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3350870983111883e-05,
+      "loss": 1.0233,
+      "step": 11260
+    },
+    {
+      "epoch": 0.4095355857002582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.334976119373078e-05,
+      "loss": 0.9311,
+      "step": 11261
+    },
+    {
+      "epoch": 0.4095719533039968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3348651357876491e-05,
+      "loss": 1.0748,
+      "step": 11262
+    },
+    {
+      "epoch": 0.4096083209077354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3347541475564418e-05,
+      "loss": 0.9834,
+      "step": 11263
+    },
+    {
+      "epoch": 0.40964468851147395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3346431546809955e-05,
+      "loss": 0.9883,
+      "step": 11264
+    },
+    {
+      "epoch": 0.40968105611521255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3345321571628502e-05,
+      "loss": 0.9933,
+      "step": 11265
+    },
+    {
+      "epoch": 0.40971742371895115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3344211550035463e-05,
+      "loss": 1.1669,
+      "step": 11266
+    },
+    {
+      "epoch": 0.40975379132268974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3343101482046227e-05,
+      "loss": 1.1493,
+      "step": 11267
+    },
+    {
+      "epoch": 0.40979015892642834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3341991367676203e-05,
+      "loss": 1.0511,
+      "step": 11268
+    },
+    {
+      "epoch": 0.40982652653016693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3340881206940794e-05,
+      "loss": 0.9686,
+      "step": 11269
+    },
+    {
+      "epoch": 0.40986289413390553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3339770999855394e-05,
+      "loss": 1.1001,
+      "step": 11270
+    },
+    {
+      "epoch": 0.4098992617376441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3338660746435412e-05,
+      "loss": 1.1343,
+      "step": 11271
+    },
+    {
+      "epoch": 0.4099356293413827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3337550446696248e-05,
+      "loss": 0.9781,
+      "step": 11272
+    },
+    {
+      "epoch": 0.40997199694512126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3336440100653309e-05,
+      "loss": 1.0854,
+      "step": 11273
+    },
+    {
+      "epoch": 0.41000836454885986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3335329708321995e-05,
+      "loss": 1.0931,
+      "step": 11274
+    },
+    {
+      "epoch": 0.41004473215259846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3334219269717714e-05,
+      "loss": 1.0482,
+      "step": 11275
+    },
+    {
+      "epoch": 0.41008109975633705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3333108784855875e-05,
+      "loss": 1.0255,
+      "step": 11276
+    },
+    {
+      "epoch": 0.41011746736007565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3331998253751877e-05,
+      "loss": 1.0501,
+      "step": 11277
+    },
+    {
+      "epoch": 0.41015383496381425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.333088767642113e-05,
+      "loss": 1.0195,
+      "step": 11278
+    },
+    {
+      "epoch": 0.41019020256755284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3329777052879047e-05,
+      "loss": 1.1287,
+      "step": 11279
+    },
+    {
+      "epoch": 0.41022657017129144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3328666383141027e-05,
+      "loss": 0.9964,
+      "step": 11280
+    },
+    {
+      "epoch": 0.41026293777503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3327555667222485e-05,
+      "loss": 1.0046,
+      "step": 11281
+    },
+    {
+      "epoch": 0.4102993053787686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3326444905138832e-05,
+      "loss": 1.0468,
+      "step": 11282
+    },
+    {
+      "epoch": 0.41033567298250717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.332533409690547e-05,
+      "loss": 1.028,
+      "step": 11283
+    },
+    {
+      "epoch": 0.41037204058624577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.332422324253782e-05,
+      "loss": 0.8829,
+      "step": 11284
+    },
+    {
+      "epoch": 0.41040840818998436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3323112342051288e-05,
+      "loss": 1.0819,
+      "step": 11285
+    },
+    {
+      "epoch": 0.41044477579372296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3322001395461284e-05,
+      "loss": 0.9289,
+      "step": 11286
+    },
+    {
+      "epoch": 0.41048114339746156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3320890402783229e-05,
+      "loss": 0.9247,
+      "step": 11287
+    },
+    {
+      "epoch": 0.41051751100120015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3319779364032526e-05,
+      "loss": 1.1286,
+      "step": 11288
+    },
+    {
+      "epoch": 0.4105538786049387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3318668279224598e-05,
+      "loss": 1.1035,
+      "step": 11289
+    },
+    {
+      "epoch": 0.4105902462086773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3317557148374856e-05,
+      "loss": 1.0543,
+      "step": 11290
+    },
+    {
+      "epoch": 0.4106266138124159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3316445971498714e-05,
+      "loss": 1.0233,
+      "step": 11291
+    },
+    {
+      "epoch": 0.4106629814161545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3315334748611592e-05,
+      "loss": 1.1169,
+      "step": 11292
+    },
+    {
+      "epoch": 0.4106993490198931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3314223479728905e-05,
+      "loss": 1.0042,
+      "step": 11293
+    },
+    {
+      "epoch": 0.4107357166236317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3313112164866065e-05,
+      "loss": 1.0334,
+      "step": 11294
+    },
+    {
+      "epoch": 0.41077208422737027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3312000804038496e-05,
+      "loss": 1.0709,
+      "step": 11295
+    },
+    {
+      "epoch": 0.41080845183110887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3310889397261617e-05,
+      "loss": 0.9858,
+      "step": 11296
+    },
+    {
+      "epoch": 0.41084481943484746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3309777944550845e-05,
+      "loss": 0.9964,
+      "step": 11297
+    },
+    {
+      "epoch": 0.410881187038586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.33086664459216e-05,
+      "loss": 1.0326,
+      "step": 11298
+    },
+    {
+      "epoch": 0.4109175546423246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.33075549013893e-05,
+      "loss": 0.9295,
+      "step": 11299
+    },
+    {
+      "epoch": 0.4109539222460632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.330644331096937e-05,
+      "loss": 1.032,
+      "step": 11300
+    },
+    {
+      "epoch": 0.4109902898498018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.330533167467723e-05,
+      "loss": 1.07,
+      "step": 11301
+    },
+    {
+      "epoch": 0.4110266574535404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3304219992528302e-05,
+      "loss": 1.0068,
+      "step": 11302
+    },
+    {
+      "epoch": 0.411063025057279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3303108264538012e-05,
+      "loss": 0.9858,
+      "step": 11303
+    },
+    {
+      "epoch": 0.4110993926610176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3301996490721778e-05,
+      "loss": 0.9882,
+      "step": 11304
+    },
+    {
+      "epoch": 0.4111357602647562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3300884671095032e-05,
+      "loss": 1.1532,
+      "step": 11305
+    },
+    {
+      "epoch": 0.4111721278684947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3299772805673192e-05,
+      "loss": 1.1717,
+      "step": 11306
+    },
+    {
+      "epoch": 0.4112084954722333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3298660894471684e-05,
+      "loss": 0.8898,
+      "step": 11307
+    },
+    {
+      "epoch": 0.4112448630759719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.329754893750594e-05,
+      "loss": 1.0193,
+      "step": 11308
+    },
+    {
+      "epoch": 0.4112812306797105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3296436934791382e-05,
+      "loss": 1.017,
+      "step": 11309
+    },
+    {
+      "epoch": 0.4113175982834491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3295324886343437e-05,
+      "loss": 1.0607,
+      "step": 11310
+    },
+    {
+      "epoch": 0.4113539658871877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3294212792177535e-05,
+      "loss": 1.0258,
+      "step": 11311
+    },
+    {
+      "epoch": 0.4113903334909263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3293100652309105e-05,
+      "loss": 1.1775,
+      "step": 11312
+    },
+    {
+      "epoch": 0.4114267010946649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3291988466753577e-05,
+      "loss": 1.0793,
+      "step": 11313
+    },
+    {
+      "epoch": 0.4114630686984035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3290876235526377e-05,
+      "loss": 1.0312,
+      "step": 11314
+    },
+    {
+      "epoch": 0.41149943630214203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3289763958642938e-05,
+      "loss": 1.0784,
+      "step": 11315
+    },
+    {
+      "epoch": 0.4115358039058806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3288651636118697e-05,
+      "loss": 1.003,
+      "step": 11316
+    },
+    {
+      "epoch": 0.4115721715096192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3287539267969076e-05,
+      "loss": 1.0332,
+      "step": 11317
+    },
+    {
+      "epoch": 0.4116085391133578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3286426854209513e-05,
+      "loss": 1.0667,
+      "step": 11318
+    },
+    {
+      "epoch": 0.4116449067170964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3285314394855441e-05,
+      "loss": 1.0483,
+      "step": 11319
+    },
+    {
+      "epoch": 0.411681274320835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3284201889922293e-05,
+      "loss": 1.035,
+      "step": 11320
+    },
+    {
+      "epoch": 0.4117176419245736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3283089339425504e-05,
+      "loss": 1.088,
+      "step": 11321
+    },
+    {
+      "epoch": 0.4117540095283122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3281976743380508e-05,
+      "loss": 1.1747,
+      "step": 11322
+    },
+    {
+      "epoch": 0.41179037713205074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3280864101802737e-05,
+      "loss": 1.0431,
+      "step": 11323
+    },
+    {
+      "epoch": 0.41182674473578934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3279751414707636e-05,
+      "loss": 1.1244,
+      "step": 11324
+    },
+    {
+      "epoch": 0.41186311233952794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.327863868211064e-05,
+      "loss": 1.072,
+      "step": 11325
+    },
+    {
+      "epoch": 0.41189947994326653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3277525904027178e-05,
+      "loss": 1.0462,
+      "step": 11326
+    },
+    {
+      "epoch": 0.41193584754700513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3276413080472697e-05,
+      "loss": 0.9701,
+      "step": 11327
+    },
+    {
+      "epoch": 0.4119722151507437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3275300211462632e-05,
+      "loss": 0.8613,
+      "step": 11328
+    },
+    {
+      "epoch": 0.4120085827544823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3274187297012423e-05,
+      "loss": 0.9927,
+      "step": 11329
+    },
+    {
+      "epoch": 0.4120449503582209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3273074337137515e-05,
+      "loss": 1.0865,
+      "step": 11330
+    },
+    {
+      "epoch": 0.41208131796195946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.327196133185334e-05,
+      "loss": 1.1136,
+      "step": 11331
+    },
+    {
+      "epoch": 0.41211768556569806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3270848281175348e-05,
+      "loss": 1.1279,
+      "step": 11332
+    },
+    {
+      "epoch": 0.41215405316943665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3269735185118973e-05,
+      "loss": 1.0529,
+      "step": 11333
+    },
+    {
+      "epoch": 0.41219042077317525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3268622043699662e-05,
+      "loss": 0.9718,
+      "step": 11334
+    },
+    {
+      "epoch": 0.41222678837691384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3267508856932863e-05,
+      "loss": 1.0557,
+      "step": 11335
+    },
+    {
+      "epoch": 0.41226315598065244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3266395624834008e-05,
+      "loss": 1.0246,
+      "step": 11336
+    },
+    {
+      "epoch": 0.41229952358439104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3265282347418551e-05,
+      "loss": 1.0934,
+      "step": 11337
+    },
+    {
+      "epoch": 0.41233589118812963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3264169024701933e-05,
+      "loss": 1.1638,
+      "step": 11338
+    },
+    {
+      "epoch": 0.41237225879186823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3263055656699601e-05,
+      "loss": 1.0361,
+      "step": 11339
+    },
+    {
+      "epoch": 0.41240862639560677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3261942243427004e-05,
+      "loss": 0.9525,
+      "step": 11340
+    },
+    {
+      "epoch": 0.41244499399934537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3260828784899584e-05,
+      "loss": 1.0107,
+      "step": 11341
+    },
+    {
+      "epoch": 0.41248136160308396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.325971528113279e-05,
+      "loss": 0.8508,
+      "step": 11342
+    },
+    {
+      "epoch": 0.41251772920682256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3258601732142076e-05,
+      "loss": 1.0765,
+      "step": 11343
+    },
+    {
+      "epoch": 0.41255409681056116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.325748813794288e-05,
+      "loss": 1.105,
+      "step": 11344
+    },
+    {
+      "epoch": 0.41259046441429975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3256374498550661e-05,
+      "loss": 0.9893,
+      "step": 11345
+    },
+    {
+      "epoch": 0.41262683201803835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3255260813980868e-05,
+      "loss": 1.0542,
+      "step": 11346
+    },
+    {
+      "epoch": 0.41266319962177694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3254147084248947e-05,
+      "loss": 1.1514,
+      "step": 11347
+    },
+    {
+      "epoch": 0.4126995672255155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3253033309370356e-05,
+      "loss": 1.0382,
+      "step": 11348
+    },
+    {
+      "epoch": 0.4127359348292541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.325191948936054e-05,
+      "loss": 1.0601,
+      "step": 11349
+    },
+    {
+      "epoch": 0.4127723024329927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3250805624234953e-05,
+      "loss": 0.9212,
+      "step": 11350
+    },
+    {
+      "epoch": 0.4128086700367313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3249691714009054e-05,
+      "loss": 0.8856,
+      "step": 11351
+    },
+    {
+      "epoch": 0.41284503764046987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3248577758698293e-05,
+      "loss": 1.0694,
+      "step": 11352
+    },
+    {
+      "epoch": 0.41288140524420847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3247463758318122e-05,
+      "loss": 0.9999,
+      "step": 11353
+    },
+    {
+      "epoch": 0.41291777284794706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3246349712884004e-05,
+      "loss": 1.0424,
+      "step": 11354
+    },
+    {
+      "epoch": 0.41295414045168566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3245235622411386e-05,
+      "loss": 1.009,
+      "step": 11355
+    },
+    {
+      "epoch": 0.41299050805542425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.324412148691573e-05,
+      "loss": 1.027,
+      "step": 11356
+    },
+    {
+      "epoch": 0.4130268756591628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3243007306412487e-05,
+      "loss": 0.9349,
+      "step": 11357
+    },
+    {
+      "epoch": 0.4130632432629014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.324189308091712e-05,
+      "loss": 1.0206,
+      "step": 11358
+    },
+    {
+      "epoch": 0.41309961086664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.324077881044509e-05,
+      "loss": 1.0507,
+      "step": 11359
+    },
+    {
+      "epoch": 0.4131359784703786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3239664495011848e-05,
+      "loss": 1.0176,
+      "step": 11360
+    },
+    {
+      "epoch": 0.4131723460741172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3238550134632858e-05,
+      "loss": 0.9557,
+      "step": 11361
+    },
+    {
+      "epoch": 0.4132087136778558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3237435729323584e-05,
+      "loss": 1.0062,
+      "step": 11362
+    },
+    {
+      "epoch": 0.4132450812815944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3236321279099479e-05,
+      "loss": 1.0995,
+      "step": 11363
+    },
+    {
+      "epoch": 0.41328144888533297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.323520678397601e-05,
+      "loss": 1.0031,
+      "step": 11364
+    },
+    {
+      "epoch": 0.4133178164890715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3234092243968636e-05,
+      "loss": 0.9958,
+      "step": 11365
+    },
+    {
+      "epoch": 0.4133541840928101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3232977659092818e-05,
+      "loss": 0.9131,
+      "step": 11366
+    },
+    {
+      "epoch": 0.4133905516965487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3231863029364025e-05,
+      "loss": 1.1273,
+      "step": 11367
+    },
+    {
+      "epoch": 0.4134269193002873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3230748354797717e-05,
+      "loss": 1.0865,
+      "step": 11368
+    },
+    {
+      "epoch": 0.4134632869040259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3229633635409358e-05,
+      "loss": 1.0261,
+      "step": 11369
+    },
+    {
+      "epoch": 0.4134996545077645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3228518871214418e-05,
+      "loss": 0.9713,
+      "step": 11370
+    },
+    {
+      "epoch": 0.4135360221115031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3227404062228355e-05,
+      "loss": 0.9473,
+      "step": 11371
+    },
+    {
+      "epoch": 0.4135723897152417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3226289208466645e-05,
+      "loss": 1.1301,
+      "step": 11372
+    },
+    {
+      "epoch": 0.4136087573189802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3225174309944748e-05,
+      "loss": 0.9825,
+      "step": 11373
+    },
+    {
+      "epoch": 0.4136451249227188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3224059366678132e-05,
+      "loss": 1.0725,
+      "step": 11374
+    },
+    {
+      "epoch": 0.4136814925264574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3222944378682268e-05,
+      "loss": 0.9365,
+      "step": 11375
+    },
+    {
+      "epoch": 0.413717860130196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.322182934597262e-05,
+      "loss": 1.0687,
+      "step": 11376
+    },
+    {
+      "epoch": 0.4137542277339346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3220714268564664e-05,
+      "loss": 0.9334,
+      "step": 11377
+    },
+    {
+      "epoch": 0.4137905953376732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3219599146473869e-05,
+      "loss": 0.9653,
+      "step": 11378
+    },
+    {
+      "epoch": 0.4138269629414118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3218483979715699e-05,
+      "loss": 1.1663,
+      "step": 11379
+    },
+    {
+      "epoch": 0.4138633305451504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3217368768305633e-05,
+      "loss": 1.0683,
+      "step": 11380
+    },
+    {
+      "epoch": 0.413899698148889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3216253512259144e-05,
+      "loss": 0.9949,
+      "step": 11381
+    },
+    {
+      "epoch": 0.41393606575262754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3215138211591694e-05,
+      "loss": 1.0127,
+      "step": 11382
+    },
+    {
+      "epoch": 0.41397243335636613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3214022866318765e-05,
+      "loss": 1.0637,
+      "step": 11383
+    },
+    {
+      "epoch": 0.41400880096010473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3212907476455832e-05,
+      "loss": 0.9711,
+      "step": 11384
+    },
+    {
+      "epoch": 0.4140451685638433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3211792042018361e-05,
+      "loss": 1.1157,
+      "step": 11385
+    },
+    {
+      "epoch": 0.4140815361675819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3210676563021836e-05,
+      "loss": 0.9841,
+      "step": 11386
+    },
+    {
+      "epoch": 0.4141179037713205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3209561039481727e-05,
+      "loss": 0.9803,
+      "step": 11387
+    },
+    {
+      "epoch": 0.4141542713750591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3208445471413514e-05,
+      "loss": 0.9525,
+      "step": 11388
+    },
+    {
+      "epoch": 0.4141906389787977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3207329858832671e-05,
+      "loss": 1.1229,
+      "step": 11389
+    },
+    {
+      "epoch": 0.41422700658253625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3206214201754678e-05,
+      "loss": 1.0214,
+      "step": 11390
+    },
+    {
+      "epoch": 0.41426337418627485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3205098500195013e-05,
+      "loss": 1.0269,
+      "step": 11391
+    },
+    {
+      "epoch": 0.41429974179001344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.320398275416915e-05,
+      "loss": 1.0532,
+      "step": 11392
+    },
+    {
+      "epoch": 0.41433610939375204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3202866963692578e-05,
+      "loss": 0.9994,
+      "step": 11393
+    },
+    {
+      "epoch": 0.41437247699749064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3201751128780767e-05,
+      "loss": 1.0683,
+      "step": 11394
+    },
+    {
+      "epoch": 0.41440884460122923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3200635249449203e-05,
+      "loss": 1.0802,
+      "step": 11395
+    },
+    {
+      "epoch": 0.41444521220496783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3199519325713364e-05,
+      "loss": 1.1096,
+      "step": 11396
+    },
+    {
+      "epoch": 0.4144815798087064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3198403357588739e-05,
+      "loss": 1.0914,
+      "step": 11397
+    },
+    {
+      "epoch": 0.414517947412445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3197287345090799e-05,
+      "loss": 1.0192,
+      "step": 11398
+    },
+    {
+      "epoch": 0.41455431501618356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3196171288235036e-05,
+      "loss": 0.9875,
+      "step": 11399
+    },
+    {
+      "epoch": 0.41459068261992216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.319505518703693e-05,
+      "loss": 0.9328,
+      "step": 11400
+    },
+    {
+      "epoch": 0.41462705022366075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.319393904151197e-05,
+      "loss": 1.0568,
+      "step": 11401
+    },
+    {
+      "epoch": 0.41466341782739935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3192822851675633e-05,
+      "loss": 0.9915,
+      "step": 11402
+    },
+    {
+      "epoch": 0.41469978543113795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3191706617543411e-05,
+      "loss": 1.0254,
+      "step": 11403
+    },
+    {
+      "epoch": 0.41473615303487654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3190590339130789e-05,
+      "loss": 0.9853,
+      "step": 11404
+    },
+    {
+      "epoch": 0.41477252063861514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3189474016453253e-05,
+      "loss": 1.0545,
+      "step": 11405
+    },
+    {
+      "epoch": 0.41480888824235373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3188357649526288e-05,
+      "loss": 0.9892,
+      "step": 11406
+    },
+    {
+      "epoch": 0.4148452558460923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3187241238365387e-05,
+      "loss": 1.0842,
+      "step": 11407
+    },
+    {
+      "epoch": 0.41488162344983087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3186124782986035e-05,
+      "loss": 1.0931,
+      "step": 11408
+    },
+    {
+      "epoch": 0.41491799105356947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.318500828340372e-05,
+      "loss": 1.0295,
+      "step": 11409
+    },
+    {
+      "epoch": 0.41495435865730806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3183891739633936e-05,
+      "loss": 1.0441,
+      "step": 11410
+    },
+    {
+      "epoch": 0.41499072626104666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3182775151692171e-05,
+      "loss": 0.9779,
+      "step": 11411
+    },
+    {
+      "epoch": 0.41502709386478526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3181658519593918e-05,
+      "loss": 1.0124,
+      "step": 11412
+    },
+    {
+      "epoch": 0.41506346146852385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3180541843354668e-05,
+      "loss": 1.1248,
+      "step": 11413
+    },
+    {
+      "epoch": 0.41509982907226245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3179425122989909e-05,
+      "loss": 1.0424,
+      "step": 11414
+    },
+    {
+      "epoch": 0.41513619667600105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3178308358515139e-05,
+      "loss": 1.0185,
+      "step": 11415
+    },
+    {
+      "epoch": 0.4151725642797396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.317719154994585e-05,
+      "loss": 1.0919,
+      "step": 11416
+    },
+    {
+      "epoch": 0.4152089318834782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3176074697297536e-05,
+      "loss": 1.1103,
+      "step": 11417
+    },
+    {
+      "epoch": 0.4152452994872168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3174957800585693e-05,
+      "loss": 1.0311,
+      "step": 11418
+    },
+    {
+      "epoch": 0.4152816670909554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3173840859825813e-05,
+      "loss": 1.052,
+      "step": 11419
+    },
+    {
+      "epoch": 0.41531803469469397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3172723875033395e-05,
+      "loss": 1.1389,
+      "step": 11420
+    },
+    {
+      "epoch": 0.41535440229843257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3171606846223935e-05,
+      "loss": 1.0509,
+      "step": 11421
+    },
+    {
+      "epoch": 0.41539076990217116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3170489773412928e-05,
+      "loss": 1.1272,
+      "step": 11422
+    },
+    {
+      "epoch": 0.41542713750590976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3169372656615878e-05,
+      "loss": 1.174,
+      "step": 11423
+    },
+    {
+      "epoch": 0.4154635051096483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3168255495848277e-05,
+      "loss": 1.0879,
+      "step": 11424
+    },
+    {
+      "epoch": 0.4154998727133869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3167138291125626e-05,
+      "loss": 1.1342,
+      "step": 11425
+    },
+    {
+      "epoch": 0.4155362403171255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3166021042463423e-05,
+      "loss": 1.076,
+      "step": 11426
+    },
+    {
+      "epoch": 0.4155726079208641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3164903749877172e-05,
+      "loss": 0.9189,
+      "step": 11427
+    },
+    {
+      "epoch": 0.4156089755246027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.316378641338237e-05,
+      "loss": 1.0703,
+      "step": 11428
+    },
+    {
+      "epoch": 0.4156453431283413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3162669032994524e-05,
+      "loss": 1.0745,
+      "step": 11429
+    },
+    {
+      "epoch": 0.4156817107320799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.316155160872913e-05,
+      "loss": 0.9508,
+      "step": 11430
+    },
+    {
+      "epoch": 0.4157180783358185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3160434140601694e-05,
+      "loss": 1.0599,
+      "step": 11431
+    },
+    {
+      "epoch": 0.415754445939557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.315931662862772e-05,
+      "loss": 1.0406,
+      "step": 11432
+    },
+    {
+      "epoch": 0.4157908135432956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3158199072822707e-05,
+      "loss": 0.9577,
+      "step": 11433
+    },
+    {
+      "epoch": 0.4158271811470342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3157081473202165e-05,
+      "loss": 1.1695,
+      "step": 11434
+    },
+    {
+      "epoch": 0.4158635487507728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3155963829781593e-05,
+      "loss": 1.1863,
+      "step": 11435
+    },
+    {
+      "epoch": 0.4158999163545114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3154846142576505e-05,
+      "loss": 1.0568,
+      "step": 11436
+    },
+    {
+      "epoch": 0.41593628395825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3153728411602402e-05,
+      "loss": 0.9907,
+      "step": 11437
+    },
+    {
+      "epoch": 0.4159726515619886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.315261063687479e-05,
+      "loss": 1.1642,
+      "step": 11438
+    },
+    {
+      "epoch": 0.4160090191657272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3151492818409179e-05,
+      "loss": 1.0126,
+      "step": 11439
+    },
+    {
+      "epoch": 0.4160453867694658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3150374956221077e-05,
+      "loss": 1.1351,
+      "step": 11440
+    },
+    {
+      "epoch": 0.4160817543732043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3149257050325992e-05,
+      "loss": 1.0983,
+      "step": 11441
+    },
+    {
+      "epoch": 0.4161181219769429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3148139100739435e-05,
+      "loss": 1.0872,
+      "step": 11442
+    },
+    {
+      "epoch": 0.4161544895806815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3147021107476911e-05,
+      "loss": 1.0098,
+      "step": 11443
+    },
+    {
+      "epoch": 0.4161908571844201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3145903070553938e-05,
+      "loss": 0.9952,
+      "step": 11444
+    },
+    {
+      "epoch": 0.4162272247881587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3144784989986023e-05,
+      "loss": 1.0123,
+      "step": 11445
+    },
+    {
+      "epoch": 0.4162635923918973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3143666865788675e-05,
+      "loss": 0.9345,
+      "step": 11446
+    },
+    {
+      "epoch": 0.4162999599956359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3142548697977413e-05,
+      "loss": 1.1274,
+      "step": 11447
+    },
+    {
+      "epoch": 0.4163363275993745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3141430486567746e-05,
+      "loss": 1.063,
+      "step": 11448
+    },
+    {
+      "epoch": 0.41637269520311304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3140312231575186e-05,
+      "loss": 1.0345,
+      "step": 11449
+    },
+    {
+      "epoch": 0.41640906280685164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3139193933015254e-05,
+      "loss": 1.0464,
+      "step": 11450
+    },
+    {
+      "epoch": 0.41644543041059023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3138075590903454e-05,
+      "loss": 1.0276,
+      "step": 11451
+    },
+    {
+      "epoch": 0.41648179801432883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3136957205255311e-05,
+      "loss": 1.03,
+      "step": 11452
+    },
+    {
+      "epoch": 0.4165181656180674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3135838776086338e-05,
+      "loss": 1.058,
+      "step": 11453
+    },
+    {
+      "epoch": 0.416554533221806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3134720303412048e-05,
+      "loss": 0.9612,
+      "step": 11454
+    },
+    {
+      "epoch": 0.4165909008255446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3133601787247964e-05,
+      "loss": 1.0532,
+      "step": 11455
+    },
+    {
+      "epoch": 0.4166272684292832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3132483227609603e-05,
+      "loss": 1.033,
+      "step": 11456
+    },
+    {
+      "epoch": 0.4166636360330218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3131364624512477e-05,
+      "loss": 1.0548,
+      "step": 11457
+    },
+    {
+      "epoch": 0.41670000363676035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3130245977972111e-05,
+      "loss": 0.944,
+      "step": 11458
+    },
+    {
+      "epoch": 0.41673637124049895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3129127288004024e-05,
+      "loss": 1.0342,
+      "step": 11459
+    },
+    {
+      "epoch": 0.41677273884423754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3128008554623737e-05,
+      "loss": 1.1335,
+      "step": 11460
+    },
+    {
+      "epoch": 0.41680910644797614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3126889777846767e-05,
+      "loss": 0.955,
+      "step": 11461
+    },
+    {
+      "epoch": 0.41684547405171474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3125770957688637e-05,
+      "loss": 0.9893,
+      "step": 11462
+    },
+    {
+      "epoch": 0.41688184165545333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3124652094164873e-05,
+      "loss": 0.9652,
+      "step": 11463
+    },
+    {
+      "epoch": 0.41691820925919193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3123533187290995e-05,
+      "loss": 1.1193,
+      "step": 11464
+    },
+    {
+      "epoch": 0.4169545768629305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3122414237082521e-05,
+      "loss": 1.0042,
+      "step": 11465
+    },
+    {
+      "epoch": 0.41699094446666907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3121295243554983e-05,
+      "loss": 1.0764,
+      "step": 11466
+    },
+    {
+      "epoch": 0.41702731207040766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3120176206723903e-05,
+      "loss": 0.9301,
+      "step": 11467
+    },
+    {
+      "epoch": 0.41706367967414626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3119057126604802e-05,
+      "loss": 1.1256,
+      "step": 11468
+    },
+    {
+      "epoch": 0.41710004727788486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3117938003213214e-05,
+      "loss": 0.95,
+      "step": 11469
+    },
+    {
+      "epoch": 0.41713641488162345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3116818836564656e-05,
+      "loss": 0.9359,
+      "step": 11470
+    },
+    {
+      "epoch": 0.41717278248536205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3115699626674659e-05,
+      "loss": 1.0977,
+      "step": 11471
+    },
+    {
+      "epoch": 0.41720915008910064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.311458037355875e-05,
+      "loss": 1.0644,
+      "step": 11472
+    },
+    {
+      "epoch": 0.41724551769283924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3113461077232456e-05,
+      "loss": 1.0448,
+      "step": 11473
+    },
+    {
+      "epoch": 0.4172818852965778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.311234173771131e-05,
+      "loss": 1.0916,
+      "step": 11474
+    },
+    {
+      "epoch": 0.4173182529003164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3111222355010835e-05,
+      "loss": 1.0094,
+      "step": 11475
+    },
+    {
+      "epoch": 0.417354620504055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3110102929146566e-05,
+      "loss": 1.1648,
+      "step": 11476
+    },
+    {
+      "epoch": 0.41739098810779357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3108983460134033e-05,
+      "loss": 0.9636,
+      "step": 11477
+    },
+    {
+      "epoch": 0.41742735571153217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3107863947988764e-05,
+      "loss": 1.1317,
+      "step": 11478
+    },
+    {
+      "epoch": 0.41746372331527076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3106744392726292e-05,
+      "loss": 1.0618,
+      "step": 11479
+    },
+    {
+      "epoch": 0.41750009091900936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3105624794362151e-05,
+      "loss": 0.9215,
+      "step": 11480
+    },
+    {
+      "epoch": 0.41753645852274796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3104505152911871e-05,
+      "loss": 1.1681,
+      "step": 11481
+    },
+    {
+      "epoch": 0.41757282612648655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3103385468390986e-05,
+      "loss": 1.0801,
+      "step": 11482
+    },
+    {
+      "epoch": 0.4176091937302251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.310226574081503e-05,
+      "loss": 0.9992,
+      "step": 11483
+    },
+    {
+      "epoch": 0.4176455613339637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3101145970199541e-05,
+      "loss": 1.0032,
+      "step": 11484
+    },
+    {
+      "epoch": 0.4176819289377023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.310002615656005e-05,
+      "loss": 1.0104,
+      "step": 11485
+    },
+    {
+      "epoch": 0.4177182965414409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3098906299912094e-05,
+      "loss": 1.159,
+      "step": 11486
+    },
+    {
+      "epoch": 0.4177546641451795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3097786400271213e-05,
+      "loss": 0.9976,
+      "step": 11487
+    },
+    {
+      "epoch": 0.4177910317489181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3096666457652938e-05,
+      "loss": 0.9964,
+      "step": 11488
+    },
+    {
+      "epoch": 0.41782739935265667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3095546472072811e-05,
+      "loss": 1.0667,
+      "step": 11489
+    },
+    {
+      "epoch": 0.41786376695639527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3094426443546369e-05,
+      "loss": 1.0307,
+      "step": 11490
+    },
+    {
+      "epoch": 0.4179001345601338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3093306372089148e-05,
+      "loss": 1.0122,
+      "step": 11491
+    },
+    {
+      "epoch": 0.4179365021638724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3092186257716693e-05,
+      "loss": 0.9604,
+      "step": 11492
+    },
+    {
+      "epoch": 0.417972869767611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3091066100444537e-05,
+      "loss": 1.0231,
+      "step": 11493
+    },
+    {
+      "epoch": 0.4180092373713496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3089945900288226e-05,
+      "loss": 1.0036,
+      "step": 11494
+    },
+    {
+      "epoch": 0.4180456049750882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3088825657263302e-05,
+      "loss": 0.9726,
+      "step": 11495
+    },
+    {
+      "epoch": 0.4180819725788268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3087705371385304e-05,
+      "loss": 1.0443,
+      "step": 11496
+    },
+    {
+      "epoch": 0.4181183401825654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3086585042669773e-05,
+      "loss": 0.9307,
+      "step": 11497
+    },
+    {
+      "epoch": 0.418154707786304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3085464671132253e-05,
+      "loss": 1.0636,
+      "step": 11498
+    },
+    {
+      "epoch": 0.4181910753900426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.308434425678829e-05,
+      "loss": 1.0656,
+      "step": 11499
+    },
+    {
+      "epoch": 0.4182274429937811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3083223799653429e-05,
+      "loss": 1.0878,
+      "step": 11500
+    },
+    {
+      "epoch": 0.4182638105975197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3082103299743208e-05,
+      "loss": 1.1482,
+      "step": 11501
+    },
+    {
+      "epoch": 0.4183001782012583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3080982757073178e-05,
+      "loss": 1.0422,
+      "step": 11502
+    },
+    {
+      "epoch": 0.4183365458049969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3079862171658883e-05,
+      "loss": 0.9986,
+      "step": 11503
+    },
+    {
+      "epoch": 0.4183729134087355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3078741543515873e-05,
+      "loss": 0.9779,
+      "step": 11504
+    },
+    {
+      "epoch": 0.4184092810124741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3077620872659688e-05,
+      "loss": 1.0323,
+      "step": 11505
+    },
+    {
+      "epoch": 0.4184456486162127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3076500159105885e-05,
+      "loss": 1.1323,
+      "step": 11506
+    },
+    {
+      "epoch": 0.4184820162199513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3075379402870006e-05,
+      "loss": 0.9035,
+      "step": 11507
+    },
+    {
+      "epoch": 0.41851838382368983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.30742586039676e-05,
+      "loss": 0.9044,
+      "step": 11508
+    },
+    {
+      "epoch": 0.41855475142742843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3073137762414217e-05,
+      "loss": 0.9571,
+      "step": 11509
+    },
+    {
+      "epoch": 0.418591119031167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3072016878225408e-05,
+      "loss": 1.1346,
+      "step": 11510
+    },
+    {
+      "epoch": 0.4186274866349056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3070895951416724e-05,
+      "loss": 0.9741,
+      "step": 11511
+    },
+    {
+      "epoch": 0.4186638542386442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3069774982003716e-05,
+      "loss": 1.0455,
+      "step": 11512
+    },
+    {
+      "epoch": 0.4187002218423828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3068653970001935e-05,
+      "loss": 1.0736,
+      "step": 11513
+    },
+    {
+      "epoch": 0.4187365894461214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3067532915426933e-05,
+      "loss": 1.1961,
+      "step": 11514
+    },
+    {
+      "epoch": 0.41877295704986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3066411818294264e-05,
+      "loss": 1.1005,
+      "step": 11515
+    },
+    {
+      "epoch": 0.41880932465359855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3065290678619485e-05,
+      "loss": 0.8969,
+      "step": 11516
+    },
+    {
+      "epoch": 0.41884569225733714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3064169496418144e-05,
+      "loss": 1.1202,
+      "step": 11517
+    },
+    {
+      "epoch": 0.41888205986107574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3063048271705799e-05,
+      "loss": 1.0578,
+      "step": 11518
+    },
+    {
+      "epoch": 0.41891842746481434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3061927004498006e-05,
+      "loss": 1.0831,
+      "step": 11519
+    },
+    {
+      "epoch": 0.41895479506855293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.306080569481032e-05,
+      "loss": 1.0399,
+      "step": 11520
+    },
+    {
+      "epoch": 0.41899116267229153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3059684342658294e-05,
+      "loss": 0.9391,
+      "step": 11521
+    },
+    {
+      "epoch": 0.4190275302760301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3058562948057495e-05,
+      "loss": 1.008,
+      "step": 11522
+    },
+    {
+      "epoch": 0.4190638978797687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3057441511023471e-05,
+      "loss": 1.0407,
+      "step": 11523
+    },
+    {
+      "epoch": 0.4191002654835073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3056320031571783e-05,
+      "loss": 1.0303,
+      "step": 11524
+    },
+    {
+      "epoch": 0.41913663308724586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3055198509717995e-05,
+      "loss": 1.0526,
+      "step": 11525
+    },
+    {
+      "epoch": 0.41917300069098445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3054076945477658e-05,
+      "loss": 0.9995,
+      "step": 11526
+    },
+    {
+      "epoch": 0.41920936829472305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3052955338866343e-05,
+      "loss": 0.9952,
+      "step": 11527
+    },
+    {
+      "epoch": 0.41924573589846165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.30518336898996e-05,
+      "loss": 1.1086,
+      "step": 11528
+    },
+    {
+      "epoch": 0.41928210350220024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3050711998592993e-05,
+      "loss": 0.9985,
+      "step": 11529
+    },
+    {
+      "epoch": 0.41931847110593884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3049590264962086e-05,
+      "loss": 1.0148,
+      "step": 11530
+    },
+    {
+      "epoch": 0.41935483870967744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.304846848902244e-05,
+      "loss": 1.1133,
+      "step": 11531
+    },
+    {
+      "epoch": 0.41939120631341603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3047346670789622e-05,
+      "loss": 1.0435,
+      "step": 11532
+    },
+    {
+      "epoch": 0.4194275739171546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3046224810279193e-05,
+      "loss": 1.0926,
+      "step": 11533
+    },
+    {
+      "epoch": 0.41946394152089317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3045102907506712e-05,
+      "loss": 1.0598,
+      "step": 11534
+    },
+    {
+      "epoch": 0.41950030912463177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3043980962487752e-05,
+      "loss": 1.0772,
+      "step": 11535
+    },
+    {
+      "epoch": 0.41953667672837036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3042858975237877e-05,
+      "loss": 1.0645,
+      "step": 11536
+    },
+    {
+      "epoch": 0.41957304433210896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3041736945772648e-05,
+      "loss": 1.0392,
+      "step": 11537
+    },
+    {
+      "epoch": 0.41960941193584755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3040614874107636e-05,
+      "loss": 0.9734,
+      "step": 11538
+    },
+    {
+      "epoch": 0.41964577953958615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3039492760258405e-05,
+      "loss": 1.0172,
+      "step": 11539
+    },
+    {
+      "epoch": 0.41968214714332475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3038370604240524e-05,
+      "loss": 0.9033,
+      "step": 11540
+    },
+    {
+      "epoch": 0.41971851474706334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3037248406069564e-05,
+      "loss": 1.1658,
+      "step": 11541
+    },
+    {
+      "epoch": 0.4197548823508019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.303612616576109e-05,
+      "loss": 1.0228,
+      "step": 11542
+    },
+    {
+      "epoch": 0.4197912499545405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3035003883330675e-05,
+      "loss": 1.0806,
+      "step": 11543
+    },
+    {
+      "epoch": 0.4198276175582791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3033881558793887e-05,
+      "loss": 1.0979,
+      "step": 11544
+    },
+    {
+      "epoch": 0.41986398516201767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3032759192166293e-05,
+      "loss": 1.0824,
+      "step": 11545
+    },
+    {
+      "epoch": 0.41990035276575627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3031636783463473e-05,
+      "loss": 1.1038,
+      "step": 11546
+    },
+    {
+      "epoch": 0.41993672036949486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3030514332700993e-05,
+      "loss": 1.0789,
+      "step": 11547
+    },
+    {
+      "epoch": 0.41997308797323346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3029391839894426e-05,
+      "loss": 0.9499,
+      "step": 11548
+    },
+    {
+      "epoch": 0.42000945557697206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3028269305059345e-05,
+      "loss": 0.9347,
+      "step": 11549
+    },
+    {
+      "epoch": 0.4200458231807106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3027146728211321e-05,
+      "loss": 0.9899,
+      "step": 11550
+    },
+    {
+      "epoch": 0.4200821907844492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3026024109365936e-05,
+      "loss": 1.2617,
+      "step": 11551
+    },
+    {
+      "epoch": 0.4201185583881878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3024901448538758e-05,
+      "loss": 1.0056,
+      "step": 11552
+    },
+    {
+      "epoch": 0.4201549259919264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3023778745745361e-05,
+      "loss": 1.0725,
+      "step": 11553
+    },
+    {
+      "epoch": 0.420191293595665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3022656001001328e-05,
+      "loss": 1.0812,
+      "step": 11554
+    },
+    {
+      "epoch": 0.4202276611994036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3021533214322228e-05,
+      "loss": 1.0044,
+      "step": 11555
+    },
+    {
+      "epoch": 0.4202640288031422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3020410385723645e-05,
+      "loss": 1.1301,
+      "step": 11556
+    },
+    {
+      "epoch": 0.42030039640688077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3019287515221155e-05,
+      "loss": 1.0673,
+      "step": 11557
+    },
+    {
+      "epoch": 0.4203367640106193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.301816460283033e-05,
+      "loss": 0.939,
+      "step": 11558
+    },
+    {
+      "epoch": 0.4203731316143579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3017041648566754e-05,
+      "loss": 1.0898,
+      "step": 11559
+    },
+    {
+      "epoch": 0.4204094992180965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.301591865244601e-05,
+      "loss": 1.0749,
+      "step": 11560
+    },
+    {
+      "epoch": 0.4204458668218351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3014795614483671e-05,
+      "loss": 1.0373,
+      "step": 11561
+    },
+    {
+      "epoch": 0.4204822344255737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3013672534695318e-05,
+      "loss": 1.2045,
+      "step": 11562
+    },
+    {
+      "epoch": 0.4205186020293123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3012549413096539e-05,
+      "loss": 1.073,
+      "step": 11563
+    },
+    {
+      "epoch": 0.4205549696330509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3011426249702907e-05,
+      "loss": 1.0853,
+      "step": 11564
+    },
+    {
+      "epoch": 0.4205913372367895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3010303044530012e-05,
+      "loss": 1.097,
+      "step": 11565
+    },
+    {
+      "epoch": 0.4206277048405281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.300917979759343e-05,
+      "loss": 0.8423,
+      "step": 11566
+    },
+    {
+      "epoch": 0.4206640724442666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3008056508908751e-05,
+      "loss": 1.0475,
+      "step": 11567
+    },
+    {
+      "epoch": 0.4207004400480052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3006933178491554e-05,
+      "loss": 1.057,
+      "step": 11568
+    },
+    {
+      "epoch": 0.4207368076517438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3005809806357424e-05,
+      "loss": 0.9963,
+      "step": 11569
+    },
+    {
+      "epoch": 0.4207731752554824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.300468639252195e-05,
+      "loss": 0.9787,
+      "step": 11570
+    },
+    {
+      "epoch": 0.420809542859221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3003562937000714e-05,
+      "loss": 0.9821,
+      "step": 11571
+    },
+    {
+      "epoch": 0.4208459104629596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3002439439809304e-05,
+      "loss": 1.0738,
+      "step": 11572
+    },
+    {
+      "epoch": 0.4208822780666982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3001315900963308e-05,
+      "loss": 1.0216,
+      "step": 11573
+    },
+    {
+      "epoch": 0.4209186456704368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3000192320478308e-05,
+      "loss": 0.9921,
+      "step": 11574
+    },
+    {
+      "epoch": 0.42095501327417534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2999068698369901e-05,
+      "loss": 1.1014,
+      "step": 11575
+    },
+    {
+      "epoch": 0.42099138087791393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2997945034653669e-05,
+      "loss": 1.0259,
+      "step": 11576
+    },
+    {
+      "epoch": 0.42102774848165253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2996821329345204e-05,
+      "loss": 1.0846,
+      "step": 11577
+    },
+    {
+      "epoch": 0.4210641160853911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2995697582460094e-05,
+      "loss": 1.1276,
+      "step": 11578
+    },
+    {
+      "epoch": 0.4211004836891297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2994573794013932e-05,
+      "loss": 1.1031,
+      "step": 11579
+    },
+    {
+      "epoch": 0.4211368512928683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2993449964022304e-05,
+      "loss": 1.009,
+      "step": 11580
+    },
+    {
+      "epoch": 0.4211732188966069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.299232609250081e-05,
+      "loss": 0.9002,
+      "step": 11581
+    },
+    {
+      "epoch": 0.4212095865003455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.299120217946503e-05,
+      "loss": 1.0683,
+      "step": 11582
+    },
+    {
+      "epoch": 0.4212459541040841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.299007822493057e-05,
+      "loss": 0.946,
+      "step": 11583
+    },
+    {
+      "epoch": 0.42128232170782265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2988954228913019e-05,
+      "loss": 0.9102,
+      "step": 11584
+    },
+    {
+      "epoch": 0.42131868931156125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2987830191427962e-05,
+      "loss": 0.9697,
+      "step": 11585
+    },
+    {
+      "epoch": 0.42135505691529984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2986706112491004e-05,
+      "loss": 1.0354,
+      "step": 11586
+    },
+    {
+      "epoch": 0.42139142451903844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2985581992117732e-05,
+      "loss": 0.9344,
+      "step": 11587
+    },
+    {
+      "epoch": 0.42142779212277703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2984457830323752e-05,
+      "loss": 1.1605,
+      "step": 11588
+    },
+    {
+      "epoch": 0.42146415972651563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.298333362712465e-05,
+      "loss": 1.1448,
+      "step": 11589
+    },
+    {
+      "epoch": 0.4215005273302542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2982209382536026e-05,
+      "loss": 1.0233,
+      "step": 11590
+    },
+    {
+      "epoch": 0.4215368949339928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.298108509657348e-05,
+      "loss": 1.1133,
+      "step": 11591
+    },
+    {
+      "epoch": 0.42157326253773136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2979960769252609e-05,
+      "loss": 1.0898,
+      "step": 11592
+    },
+    {
+      "epoch": 0.42160963014146996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2978836400589008e-05,
+      "loss": 0.9867,
+      "step": 11593
+    },
+    {
+      "epoch": 0.42164599774520856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.297771199059828e-05,
+      "loss": 0.9626,
+      "step": 11594
+    },
+    {
+      "epoch": 0.42168236534894715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2976587539296021e-05,
+      "loss": 1.0684,
+      "step": 11595
+    },
+    {
+      "epoch": 0.42171873295268575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2975463046697835e-05,
+      "loss": 1.0176,
+      "step": 11596
+    },
+    {
+      "epoch": 0.42175510055642434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.297433851281932e-05,
+      "loss": 0.9907,
+      "step": 11597
+    },
+    {
+      "epoch": 0.42179146816016294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2973213937676076e-05,
+      "loss": 1.0183,
+      "step": 11598
+    },
+    {
+      "epoch": 0.42182783576390154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2972089321283712e-05,
+      "loss": 0.9174,
+      "step": 11599
+    },
+    {
+      "epoch": 0.42186420336764013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2970964663657824e-05,
+      "loss": 0.9518,
+      "step": 11600
+    },
+    {
+      "epoch": 0.4219005709713787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2969839964814014e-05,
+      "loss": 1.0037,
+      "step": 11601
+    },
+    {
+      "epoch": 0.42193693857511727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.296871522476789e-05,
+      "loss": 1.0765,
+      "step": 11602
+    },
+    {
+      "epoch": 0.42197330617885587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2967590443535056e-05,
+      "loss": 0.962,
+      "step": 11603
+    },
+    {
+      "epoch": 0.42200967378259446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2966465621131113e-05,
+      "loss": 0.961,
+      "step": 11604
+    },
+    {
+      "epoch": 0.42204604138633306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2965340757571669e-05,
+      "loss": 1.0642,
+      "step": 11605
+    },
+    {
+      "epoch": 0.42208240899007166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2964215852872326e-05,
+      "loss": 0.9357,
+      "step": 11606
+    },
+    {
+      "epoch": 0.42211877659381025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2963090907048698e-05,
+      "loss": 0.9241,
+      "step": 11607
+    },
+    {
+      "epoch": 0.42215514419754885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2961965920116386e-05,
+      "loss": 1.1352,
+      "step": 11608
+    },
+    {
+      "epoch": 0.4221915118012874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2960840892090998e-05,
+      "loss": 0.9734,
+      "step": 11609
+    },
+    {
+      "epoch": 0.422227879405026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2959715822988148e-05,
+      "loss": 1.083,
+      "step": 11610
+    },
+    {
+      "epoch": 0.4222642470087646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2958590712823437e-05,
+      "loss": 1.1362,
+      "step": 11611
+    },
+    {
+      "epoch": 0.4223006146125032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2957465561612477e-05,
+      "loss": 1.0645,
+      "step": 11612
+    },
+    {
+      "epoch": 0.4223369822162418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.295634036937088e-05,
+      "loss": 1.1088,
+      "step": 11613
+    },
+    {
+      "epoch": 0.42237334981998037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2955215136114253e-05,
+      "loss": 1.1317,
+      "step": 11614
+    },
+    {
+      "epoch": 0.42240971742371897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.295408986185821e-05,
+      "loss": 1.1323,
+      "step": 11615
+    },
+    {
+      "epoch": 0.42244608502745756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2952964546618363e-05,
+      "loss": 1.0388,
+      "step": 11616
+    },
+    {
+      "epoch": 0.4224824526311961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2951839190410318e-05,
+      "loss": 0.9691,
+      "step": 11617
+    },
+    {
+      "epoch": 0.4225188202349347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2950713793249696e-05,
+      "loss": 1.0269,
+      "step": 11618
+    },
+    {
+      "epoch": 0.4225551878386733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2949588355152107e-05,
+      "loss": 1.0707,
+      "step": 11619
+    },
+    {
+      "epoch": 0.4225915554424119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2948462876133161e-05,
+      "loss": 0.9106,
+      "step": 11620
+    },
+    {
+      "epoch": 0.4226279230461505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.294733735620848e-05,
+      "loss": 1.0706,
+      "step": 11621
+    },
+    {
+      "epoch": 0.4226642906498891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2946211795393675e-05,
+      "loss": 1.0411,
+      "step": 11622
+    },
+    {
+      "epoch": 0.4227006582536277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2945086193704359e-05,
+      "loss": 0.965,
+      "step": 11623
+    },
+    {
+      "epoch": 0.4227370258573663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2943960551156152e-05,
+      "loss": 0.9941,
+      "step": 11624
+    },
+    {
+      "epoch": 0.4227733934611049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2942834867764669e-05,
+      "loss": 1.0573,
+      "step": 11625
+    },
+    {
+      "epoch": 0.4228097610648434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2941709143545526e-05,
+      "loss": 1.0055,
+      "step": 11626
+    },
+    {
+      "epoch": 0.422846128668582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2940583378514345e-05,
+      "loss": 1.0337,
+      "step": 11627
+    },
+    {
+      "epoch": 0.4228824962723206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.293945757268674e-05,
+      "loss": 1.1384,
+      "step": 11628
+    },
+    {
+      "epoch": 0.4229188638760592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2938331726078332e-05,
+      "loss": 1.1094,
+      "step": 11629
+    },
+    {
+      "epoch": 0.4229552314797978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2937205838704739e-05,
+      "loss": 1.0212,
+      "step": 11630
+    },
+    {
+      "epoch": 0.4229915990835364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2936079910581585e-05,
+      "loss": 1.015,
+      "step": 11631
+    },
+    {
+      "epoch": 0.423027966687275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2934953941724485e-05,
+      "loss": 1.082,
+      "step": 11632
+    },
+    {
+      "epoch": 0.4230643342910136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2933827932149066e-05,
+      "loss": 1.0305,
+      "step": 11633
+    },
+    {
+      "epoch": 0.42310070189475213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2932701881870947e-05,
+      "loss": 1.061,
+      "step": 11634
+    },
+    {
+      "epoch": 0.4231370694984907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.293157579090575e-05,
+      "loss": 0.9313,
+      "step": 11635
+    },
+    {
+      "epoch": 0.4231734371022293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2930449659269095e-05,
+      "loss": 0.9688,
+      "step": 11636
+    },
+    {
+      "epoch": 0.4232098047059679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2929323486976612e-05,
+      "loss": 1.0291,
+      "step": 11637
+    },
+    {
+      "epoch": 0.4232461723097065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2928197274043921e-05,
+      "loss": 0.9013,
+      "step": 11638
+    },
+    {
+      "epoch": 0.4232825399134451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2927071020486648e-05,
+      "loss": 1.0516,
+      "step": 11639
+    },
+    {
+      "epoch": 0.4233189075171837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2925944726320416e-05,
+      "loss": 1.0313,
+      "step": 11640
+    },
+    {
+      "epoch": 0.4233552751209223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.292481839156085e-05,
+      "loss": 1.0597,
+      "step": 11641
+    },
+    {
+      "epoch": 0.4233916427246609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2923692016223587e-05,
+      "loss": 0.9829,
+      "step": 11642
+    },
+    {
+      "epoch": 0.42342801032839944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.292256560032424e-05,
+      "loss": 1.0013,
+      "step": 11643
+    },
+    {
+      "epoch": 0.42346437793213804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2921439143878442e-05,
+      "loss": 0.8734,
+      "step": 11644
+    },
+    {
+      "epoch": 0.42350074553587663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2920312646901822e-05,
+      "loss": 0.9692,
+      "step": 11645
+    },
+    {
+      "epoch": 0.42353711313961523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2919186109410004e-05,
+      "loss": 1.1238,
+      "step": 11646
+    },
+    {
+      "epoch": 0.4235734807433538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2918059531418626e-05,
+      "loss": 1.0644,
+      "step": 11647
+    },
+    {
+      "epoch": 0.4236098483470924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.291693291294331e-05,
+      "loss": 1.0889,
+      "step": 11648
+    },
+    {
+      "epoch": 0.423646215950831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2915806253999686e-05,
+      "loss": 1.0244,
+      "step": 11649
+    },
+    {
+      "epoch": 0.4236825835545696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2914679554603389e-05,
+      "loss": 0.943,
+      "step": 11650
+    },
+    {
+      "epoch": 0.42371895115830815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2913552814770047e-05,
+      "loss": 1.143,
+      "step": 11651
+    },
+    {
+      "epoch": 0.42375531876204675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2912426034515295e-05,
+      "loss": 1.0728,
+      "step": 11652
+    },
+    {
+      "epoch": 0.42379168636578535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2911299213854764e-05,
+      "loss": 1.1472,
+      "step": 11653
+    },
+    {
+      "epoch": 0.42382805396952394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2910172352804087e-05,
+      "loss": 0.97,
+      "step": 11654
+    },
+    {
+      "epoch": 0.42386442157326254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2909045451378897e-05,
+      "loss": 1.0482,
+      "step": 11655
+    },
+    {
+      "epoch": 0.42390078917700114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.290791850959483e-05,
+      "loss": 1.078,
+      "step": 11656
+    },
+    {
+      "epoch": 0.42393715678073973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2906791527467517e-05,
+      "loss": 1.0219,
+      "step": 11657
+    },
+    {
+      "epoch": 0.42397352438447833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2905664505012599e-05,
+      "loss": 1.0187,
+      "step": 11658
+    },
+    {
+      "epoch": 0.42400989198821687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2904537442245706e-05,
+      "loss": 1.0217,
+      "step": 11659
+    },
+    {
+      "epoch": 0.42404625959195547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2903410339182477e-05,
+      "loss": 1.0766,
+      "step": 11660
+    },
+    {
+      "epoch": 0.42408262719569406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2902283195838554e-05,
+      "loss": 0.9391,
+      "step": 11661
+    },
+    {
+      "epoch": 0.42411899479943266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2901156012229562e-05,
+      "loss": 1.1436,
+      "step": 11662
+    },
+    {
+      "epoch": 0.42415536240317125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.290002878837115e-05,
+      "loss": 0.9528,
+      "step": 11663
+    },
+    {
+      "epoch": 0.42419173000690985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2898901524278955e-05,
+      "loss": 0.9418,
+      "step": 11664
+    },
+    {
+      "epoch": 0.42422809761064845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2897774219968611e-05,
+      "loss": 1.0477,
+      "step": 11665
+    },
+    {
+      "epoch": 0.42426446521438704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2896646875455763e-05,
+      "loss": 1.083,
+      "step": 11666
+    },
+    {
+      "epoch": 0.42430083281812564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.289551949075605e-05,
+      "loss": 1.2042,
+      "step": 11667
+    },
+    {
+      "epoch": 0.4243372004218642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.289439206588511e-05,
+      "loss": 1.0468,
+      "step": 11668
+    },
+    {
+      "epoch": 0.4243735680256028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2893264600858588e-05,
+      "loss": 1.003,
+      "step": 11669
+    },
+    {
+      "epoch": 0.4244099356293414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2892137095692124e-05,
+      "loss": 1.0249,
+      "step": 11670
+    },
+    {
+      "epoch": 0.42444630323307997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2891009550401364e-05,
+      "loss": 1.0891,
+      "step": 11671
+    },
+    {
+      "epoch": 0.42448267083681857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2889881965001945e-05,
+      "loss": 1.1183,
+      "step": 11672
+    },
+    {
+      "epoch": 0.42451903844055716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2888754339509513e-05,
+      "loss": 1.0005,
+      "step": 11673
+    },
+    {
+      "epoch": 0.42455540604429576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2887626673939718e-05,
+      "loss": 1.0406,
+      "step": 11674
+    },
+    {
+      "epoch": 0.42459177364803435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2886498968308196e-05,
+      "loss": 1.004,
+      "step": 11675
+    },
+    {
+      "epoch": 0.4246281412517729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2885371222630595e-05,
+      "loss": 1.0316,
+      "step": 11676
+    },
+    {
+      "epoch": 0.4246645088555115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2884243436922566e-05,
+      "loss": 0.9856,
+      "step": 11677
+    },
+    {
+      "epoch": 0.4247008764592501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2883115611199752e-05,
+      "loss": 1.0058,
+      "step": 11678
+    },
+    {
+      "epoch": 0.4247372440629887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2881987745477796e-05,
+      "loss": 1.1162,
+      "step": 11679
+    },
+    {
+      "epoch": 0.4247736116667273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2880859839772352e-05,
+      "loss": 1.138,
+      "step": 11680
+    },
+    {
+      "epoch": 0.4248099792704659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2879731894099061e-05,
+      "loss": 1.0932,
+      "step": 11681
+    },
+    {
+      "epoch": 0.4248463468742045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2878603908473579e-05,
+      "loss": 0.9219,
+      "step": 11682
+    },
+    {
+      "epoch": 0.42488271447794307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2877475882911553e-05,
+      "loss": 1.1064,
+      "step": 11683
+    },
+    {
+      "epoch": 0.42491908208168166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.287634781742863e-05,
+      "loss": 1.0158,
+      "step": 11684
+    },
+    {
+      "epoch": 0.4249554496854202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2875219712040461e-05,
+      "loss": 1.028,
+      "step": 11685
+    },
+    {
+      "epoch": 0.4249918172891588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.28740915667627e-05,
+      "loss": 1.126,
+      "step": 11686
+    },
+    {
+      "epoch": 0.4250281848928974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2872963381610995e-05,
+      "loss": 0.9953,
+      "step": 11687
+    },
+    {
+      "epoch": 0.425064552496636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2871835156601e-05,
+      "loss": 0.9873,
+      "step": 11688
+    },
+    {
+      "epoch": 0.4251009201003746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2870706891748364e-05,
+      "loss": 0.9781,
+      "step": 11689
+    },
+    {
+      "epoch": 0.4251372877041132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2869578587068747e-05,
+      "loss": 1.0901,
+      "step": 11690
+    },
+    {
+      "epoch": 0.4251736553078518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2868450242577797e-05,
+      "loss": 1.0053,
+      "step": 11691
+    },
+    {
+      "epoch": 0.4252100229115904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.286732185829117e-05,
+      "loss": 1.0265,
+      "step": 11692
+    },
+    {
+      "epoch": 0.4252463905153289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2866193434224518e-05,
+      "loss": 1.1267,
+      "step": 11693
+    },
+    {
+      "epoch": 0.4252827581190675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2865064970393502e-05,
+      "loss": 1.0706,
+      "step": 11694
+    },
+    {
+      "epoch": 0.4253191257228061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2863936466813771e-05,
+      "loss": 1.0356,
+      "step": 11695
+    },
+    {
+      "epoch": 0.4253554933265447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2862807923500989e-05,
+      "loss": 1.0739,
+      "step": 11696
+    },
+    {
+      "epoch": 0.4253918609302833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2861679340470806e-05,
+      "loss": 1.02,
+      "step": 11697
+    },
+    {
+      "epoch": 0.4254282285340219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2860550717738882e-05,
+      "loss": 1.0286,
+      "step": 11698
+    },
+    {
+      "epoch": 0.4254645961377605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2859422055320878e-05,
+      "loss": 1.0162,
+      "step": 11699
+    },
+    {
+      "epoch": 0.4255009637414991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2858293353232449e-05,
+      "loss": 1.0475,
+      "step": 11700
+    },
+    {
+      "epoch": 0.42553733134523763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2857164611489257e-05,
+      "loss": 1.1037,
+      "step": 11701
+    },
+    {
+      "epoch": 0.42557369894897623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2856035830106955e-05,
+      "loss": 1.0818,
+      "step": 11702
+    },
+    {
+      "epoch": 0.4256100665527148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2854907009101212e-05,
+      "loss": 1.1007,
+      "step": 11703
+    },
+    {
+      "epoch": 0.4256464341564534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2853778148487683e-05,
+      "loss": 1.0819,
+      "step": 11704
+    },
+    {
+      "epoch": 0.425682801760192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.285264924828203e-05,
+      "loss": 0.9561,
+      "step": 11705
+    },
+    {
+      "epoch": 0.4257191693639306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2851520308499918e-05,
+      "loss": 1.0471,
+      "step": 11706
+    },
+    {
+      "epoch": 0.4257555369676692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2850391329157009e-05,
+      "loss": 0.8964,
+      "step": 11707
+    },
+    {
+      "epoch": 0.4257919045714078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.284926231026896e-05,
+      "loss": 1.0061,
+      "step": 11708
+    },
+    {
+      "epoch": 0.4258282721751464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2848133251851442e-05,
+      "loss": 0.9596,
+      "step": 11709
+    },
+    {
+      "epoch": 0.42586463977888495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2847004153920115e-05,
+      "loss": 1.0092,
+      "step": 11710
+    },
+    {
+      "epoch": 0.42590100738262354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2845875016490646e-05,
+      "loss": 0.9476,
+      "step": 11711
+    },
+    {
+      "epoch": 0.42593737498636214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.28447458395787e-05,
+      "loss": 1.062,
+      "step": 11712
+    },
+    {
+      "epoch": 0.42597374259010073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2843616623199938e-05,
+      "loss": 1.2039,
+      "step": 11713
+    },
+    {
+      "epoch": 0.42601011019383933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2842487367370033e-05,
+      "loss": 1.0045,
+      "step": 11714
+    },
+    {
+      "epoch": 0.4260464777975779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2841358072104647e-05,
+      "loss": 1.0064,
+      "step": 11715
+    },
+    {
+      "epoch": 0.4260828454013165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.284022873741945e-05,
+      "loss": 0.9815,
+      "step": 11716
+    },
+    {
+      "epoch": 0.4261192130050551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2839099363330108e-05,
+      "loss": 0.9551,
+      "step": 11717
+    },
+    {
+      "epoch": 0.42615558060879366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2837969949852291e-05,
+      "loss": 1.1223,
+      "step": 11718
+    },
+    {
+      "epoch": 0.42619194821253226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.283684049700167e-05,
+      "loss": 1.0516,
+      "step": 11719
+    },
+    {
+      "epoch": 0.42622831581627085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.283571100479391e-05,
+      "loss": 1.0845,
+      "step": 11720
+    },
+    {
+      "epoch": 0.42626468342000945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2834581473244682e-05,
+      "loss": 1.0866,
+      "step": 11721
+    },
+    {
+      "epoch": 0.42630105102374805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.283345190236966e-05,
+      "loss": 1.0411,
+      "step": 11722
+    },
+    {
+      "epoch": 0.42633741862748664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.283232229218451e-05,
+      "loss": 1.0134,
+      "step": 11723
+    },
+    {
+      "epoch": 0.42637378623122524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2831192642704909e-05,
+      "loss": 1.1027,
+      "step": 11724
+    },
+    {
+      "epoch": 0.42641015383496383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2830062953946527e-05,
+      "loss": 1.1525,
+      "step": 11725
+    },
+    {
+      "epoch": 0.42644652143870243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2828933225925036e-05,
+      "loss": 1.0317,
+      "step": 11726
+    },
+    {
+      "epoch": 0.42648288904244097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2827803458656112e-05,
+      "loss": 1.0999,
+      "step": 11727
+    },
+    {
+      "epoch": 0.42651925664617957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2826673652155426e-05,
+      "loss": 1.0802,
+      "step": 11728
+    },
+    {
+      "epoch": 0.42655562424991816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2825543806438651e-05,
+      "loss": 1.0411,
+      "step": 11729
+    },
+    {
+      "epoch": 0.42659199185365676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2824413921521468e-05,
+      "loss": 1.1044,
+      "step": 11730
+    },
+    {
+      "epoch": 0.42662835945739536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2823283997419549e-05,
+      "loss": 0.9438,
+      "step": 11731
+    },
+    {
+      "epoch": 0.42666472706113395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2822154034148567e-05,
+      "loss": 0.942,
+      "step": 11732
+    },
+    {
+      "epoch": 0.42670109466487255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2821024031724205e-05,
+      "loss": 1.0546,
+      "step": 11733
+    },
+    {
+      "epoch": 0.42673746226861115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2819893990162138e-05,
+      "loss": 1.0499,
+      "step": 11734
+    },
+    {
+      "epoch": 0.4267738298723497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2818763909478041e-05,
+      "loss": 0.9163,
+      "step": 11735
+    },
+    {
+      "epoch": 0.4268101974760883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2817633789687593e-05,
+      "loss": 1.1082,
+      "step": 11736
+    },
+    {
+      "epoch": 0.4268465650798269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2816503630806474e-05,
+      "loss": 1.002,
+      "step": 11737
+    },
+    {
+      "epoch": 0.4268829326835655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2815373432850367e-05,
+      "loss": 1.0147,
+      "step": 11738
+    },
+    {
+      "epoch": 0.42691930028730407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2814243195834945e-05,
+      "loss": 1.1211,
+      "step": 11739
+    },
+    {
+      "epoch": 0.42695566789104267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2813112919775892e-05,
+      "loss": 1.0397,
+      "step": 11740
+    },
+    {
+      "epoch": 0.42699203549478126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2811982604688889e-05,
+      "loss": 1.0278,
+      "step": 11741
+    },
+    {
+      "epoch": 0.42702840309851986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2810852250589614e-05,
+      "loss": 1.0854,
+      "step": 11742
+    },
+    {
+      "epoch": 0.4270647707022584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2809721857493756e-05,
+      "loss": 1.0784,
+      "step": 11743
+    },
+    {
+      "epoch": 0.427101138305997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2808591425416992e-05,
+      "loss": 1.1686,
+      "step": 11744
+    },
+    {
+      "epoch": 0.4271375059097356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2807460954375006e-05,
+      "loss": 1.0668,
+      "step": 11745
+    },
+    {
+      "epoch": 0.4271738735134742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2806330444383486e-05,
+      "loss": 1.0728,
+      "step": 11746
+    },
+    {
+      "epoch": 0.4272102411172128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2805199895458108e-05,
+      "loss": 1.0633,
+      "step": 11747
+    },
+    {
+      "epoch": 0.4272466087209514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2804069307614564e-05,
+      "loss": 0.9308,
+      "step": 11748
+    },
+    {
+      "epoch": 0.42728297632469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2802938680868537e-05,
+      "loss": 0.9827,
+      "step": 11749
+    },
+    {
+      "epoch": 0.4273193439284286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2801808015235712e-05,
+      "loss": 0.9786,
+      "step": 11750
+    },
+    {
+      "epoch": 0.42735571153216717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2800677310731777e-05,
+      "loss": 1.0285,
+      "step": 11751
+    },
+    {
+      "epoch": 0.4273920791359057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2799546567372418e-05,
+      "loss": 0.9335,
+      "step": 11752
+    },
+    {
+      "epoch": 0.4274284467396443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.279841578517332e-05,
+      "loss": 1.0417,
+      "step": 11753
+    },
+    {
+      "epoch": 0.4274648143433829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2797284964150177e-05,
+      "loss": 0.9902,
+      "step": 11754
+    },
+    {
+      "epoch": 0.4275011819471215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2796154104318672e-05,
+      "loss": 0.9023,
+      "step": 11755
+    },
+    {
+      "epoch": 0.4275375495508601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2795023205694494e-05,
+      "loss": 1.0452,
+      "step": 11756
+    },
+    {
+      "epoch": 0.4275739171545987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.279389226829334e-05,
+      "loss": 1.128,
+      "step": 11757
+    },
+    {
+      "epoch": 0.4276102847583373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2792761292130891e-05,
+      "loss": 0.9762,
+      "step": 11758
+    },
+    {
+      "epoch": 0.4276466523620759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2791630277222842e-05,
+      "loss": 0.9353,
+      "step": 11759
+    },
+    {
+      "epoch": 0.4276830199658144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2790499223584885e-05,
+      "loss": 1.0653,
+      "step": 11760
+    },
+    {
+      "epoch": 0.427719387569553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2789368131232707e-05,
+      "loss": 1.0604,
+      "step": 11761
+    },
+    {
+      "epoch": 0.4277557551732916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2788237000182007e-05,
+      "loss": 0.8865,
+      "step": 11762
+    },
+    {
+      "epoch": 0.4277921227770302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2787105830448474e-05,
+      "loss": 0.9428,
+      "step": 11763
+    },
+    {
+      "epoch": 0.4278284903807688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2785974622047801e-05,
+      "loss": 0.8581,
+      "step": 11764
+    },
+    {
+      "epoch": 0.4278648579845074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2784843374995683e-05,
+      "loss": 1.0104,
+      "step": 11765
+    },
+    {
+      "epoch": 0.427901225588246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2783712089307815e-05,
+      "loss": 0.971,
+      "step": 11766
+    },
+    {
+      "epoch": 0.4279375931919846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2782580764999892e-05,
+      "loss": 1.128,
+      "step": 11767
+    },
+    {
+      "epoch": 0.4279739607957232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2781449402087609e-05,
+      "loss": 1.1238,
+      "step": 11768
+    },
+    {
+      "epoch": 0.42801032839946174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2780318000586658e-05,
+      "loss": 1.0076,
+      "step": 11769
+    },
+    {
+      "epoch": 0.42804669600320033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2779186560512745e-05,
+      "loss": 1.0777,
+      "step": 11770
+    },
+    {
+      "epoch": 0.42808306360693893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.277805508188156e-05,
+      "loss": 1.0309,
+      "step": 11771
+    },
+    {
+      "epoch": 0.4281194312106775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2776923564708798e-05,
+      "loss": 1.1317,
+      "step": 11772
+    },
+    {
+      "epoch": 0.4281557988144161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2775792009010166e-05,
+      "loss": 1.0801,
+      "step": 11773
+    },
+    {
+      "epoch": 0.4281921664181547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2774660414801359e-05,
+      "loss": 1.1816,
+      "step": 11774
+    },
+    {
+      "epoch": 0.4282285340218933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.277352878209807e-05,
+      "loss": 1.1274,
+      "step": 11775
+    },
+    {
+      "epoch": 0.4282649016256319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.277239711091601e-05,
+      "loss": 0.9132,
+      "step": 11776
+    },
+    {
+      "epoch": 0.42830126922937045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2771265401270871e-05,
+      "loss": 1.0022,
+      "step": 11777
+    },
+    {
+      "epoch": 0.42833763683310905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2770133653178358e-05,
+      "loss": 0.9574,
+      "step": 11778
+    },
+    {
+      "epoch": 0.42837400443684764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.276900186665417e-05,
+      "loss": 0.9733,
+      "step": 11779
+    },
+    {
+      "epoch": 0.42841037204058624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2767870041714008e-05,
+      "loss": 0.9855,
+      "step": 11780
+    },
+    {
+      "epoch": 0.42844673964432484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2766738178373578e-05,
+      "loss": 1.0027,
+      "step": 11781
+    },
+    {
+      "epoch": 0.42848310724806343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.276560627664858e-05,
+      "loss": 0.9838,
+      "step": 11782
+    },
+    {
+      "epoch": 0.42851947485180203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2764474336554721e-05,
+      "loss": 1.064,
+      "step": 11783
+    },
+    {
+      "epoch": 0.4285558424555406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2763342358107702e-05,
+      "loss": 0.9752,
+      "step": 11784
+    },
+    {
+      "epoch": 0.4285922100592792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2762210341323227e-05,
+      "loss": 1.0744,
+      "step": 11785
+    },
+    {
+      "epoch": 0.42862857766301776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2761078286217005e-05,
+      "loss": 1.0965,
+      "step": 11786
+    },
+    {
+      "epoch": 0.42866494526675636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.275994619280474e-05,
+      "loss": 1.0782,
+      "step": 11787
+    },
+    {
+      "epoch": 0.42870131287049495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2758814061102132e-05,
+      "loss": 0.939,
+      "step": 11788
+    },
+    {
+      "epoch": 0.42873768047423355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2757681891124898e-05,
+      "loss": 0.9634,
+      "step": 11789
+    },
+    {
+      "epoch": 0.42877404807797215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.275654968288874e-05,
+      "loss": 1.078,
+      "step": 11790
+    },
+    {
+      "epoch": 0.42881041568171074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2755417436409363e-05,
+      "loss": 1.1002,
+      "step": 11791
+    },
+    {
+      "epoch": 0.42884678328544934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2754285151702481e-05,
+      "loss": 0.9928,
+      "step": 11792
+    },
+    {
+      "epoch": 0.42888315088918794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2753152828783799e-05,
+      "loss": 0.9377,
+      "step": 11793
+    },
+    {
+      "epoch": 0.4289195184929265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.275202046766903e-05,
+      "loss": 1.0183,
+      "step": 11794
+    },
+    {
+      "epoch": 0.4289558860966651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2750888068373883e-05,
+      "loss": 0.951,
+      "step": 11795
+    },
+    {
+      "epoch": 0.42899225370040367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2749755630914063e-05,
+      "loss": 0.9986,
+      "step": 11796
+    },
+    {
+      "epoch": 0.42902862130414227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2748623155305286e-05,
+      "loss": 1.0669,
+      "step": 11797
+    },
+    {
+      "epoch": 0.42906498890788086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2747490641563263e-05,
+      "loss": 1.1035,
+      "step": 11798
+    },
+    {
+      "epoch": 0.42910135651161946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2746358089703705e-05,
+      "loss": 1.0578,
+      "step": 11799
+    },
+    {
+      "epoch": 0.42913772411535805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2745225499742327e-05,
+      "loss": 1.1228,
+      "step": 11800
+    },
+    {
+      "epoch": 0.42917409171909665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2744092871694837e-05,
+      "loss": 1.0464,
+      "step": 11801
+    },
+    {
+      "epoch": 0.4292104593228352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2742960205576953e-05,
+      "loss": 1.0019,
+      "step": 11802
+    },
+    {
+      "epoch": 0.4292468269265738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.274182750140439e-05,
+      "loss": 1.0727,
+      "step": 11803
+    },
+    {
+      "epoch": 0.4292831945303124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2740694759192858e-05,
+      "loss": 1.0933,
+      "step": 11804
+    },
+    {
+      "epoch": 0.429319562134051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2739561978958075e-05,
+      "loss": 0.9465,
+      "step": 11805
+    },
+    {
+      "epoch": 0.4293559297377896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2738429160715758e-05,
+      "loss": 1.0399,
+      "step": 11806
+    },
+    {
+      "epoch": 0.4293922973415282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.273729630448162e-05,
+      "loss": 1.0224,
+      "step": 11807
+    },
+    {
+      "epoch": 0.42942866494526677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2736163410271381e-05,
+      "loss": 1.1032,
+      "step": 11808
+    },
+    {
+      "epoch": 0.42946503254900537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2735030478100754e-05,
+      "loss": 1.0014,
+      "step": 11809
+    },
+    {
+      "epoch": 0.42950140015274396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2733897507985462e-05,
+      "loss": 1.0923,
+      "step": 11810
+    },
+    {
+      "epoch": 0.4295377677564825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2732764499941222e-05,
+      "loss": 1.0557,
+      "step": 11811
+    },
+    {
+      "epoch": 0.4295741353602211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.273163145398375e-05,
+      "loss": 0.8632,
+      "step": 11812
+    },
+    {
+      "epoch": 0.4296105029639597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2730498370128767e-05,
+      "loss": 0.9188,
+      "step": 11813
+    },
+    {
+      "epoch": 0.4296468705676983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2729365248391996e-05,
+      "loss": 0.9819,
+      "step": 11814
+    },
+    {
+      "epoch": 0.4296832381714369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2728232088789153e-05,
+      "loss": 0.9857,
+      "step": 11815
+    },
+    {
+      "epoch": 0.4297196057751755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.272709889133596e-05,
+      "loss": 1.076,
+      "step": 11816
+    },
+    {
+      "epoch": 0.4297559733789141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2725965656048138e-05,
+      "loss": 1.0398,
+      "step": 11817
+    },
+    {
+      "epoch": 0.4297923409826527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2724832382941412e-05,
+      "loss": 1.2073,
+      "step": 11818
+    },
+    {
+      "epoch": 0.4298287085863912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2723699072031503e-05,
+      "loss": 1.0288,
+      "step": 11819
+    },
+    {
+      "epoch": 0.4298650761901298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2722565723334133e-05,
+      "loss": 1.0813,
+      "step": 11820
+    },
+    {
+      "epoch": 0.4299014437938684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2721432336865028e-05,
+      "loss": 1.0103,
+      "step": 11821
+    },
+    {
+      "epoch": 0.429937811397607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2720298912639908e-05,
+      "loss": 1.002,
+      "step": 11822
+    },
+    {
+      "epoch": 0.4299741790013456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.27191654506745e-05,
+      "loss": 1.0575,
+      "step": 11823
+    },
+    {
+      "epoch": 0.4300105466050842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2718031950984531e-05,
+      "loss": 1.0413,
+      "step": 11824
+    },
+    {
+      "epoch": 0.4300469142088228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2716898413585724e-05,
+      "loss": 1.0013,
+      "step": 11825
+    },
+    {
+      "epoch": 0.4300832818125614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2715764838493808e-05,
+      "loss": 1.0054,
+      "step": 11826
+    },
+    {
+      "epoch": 0.4301196494163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2714631225724507e-05,
+      "loss": 0.996,
+      "step": 11827
+    },
+    {
+      "epoch": 0.43015601702003853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2713497575293547e-05,
+      "loss": 1.062,
+      "step": 11828
+    },
+    {
+      "epoch": 0.4301923846237771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.271236388721666e-05,
+      "loss": 1.097,
+      "step": 11829
+    },
+    {
+      "epoch": 0.4302287522275157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2711230161509576e-05,
+      "loss": 0.9177,
+      "step": 11830
+    },
+    {
+      "epoch": 0.4302651198312543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2710096398188015e-05,
+      "loss": 1.094,
+      "step": 11831
+    },
+    {
+      "epoch": 0.4303014874349929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2708962597267713e-05,
+      "loss": 1.0693,
+      "step": 11832
+    },
+    {
+      "epoch": 0.4303378550387315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.27078287587644e-05,
+      "loss": 0.9021,
+      "step": 11833
+    },
+    {
+      "epoch": 0.4303742226424701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2706694882693804e-05,
+      "loss": 0.9799,
+      "step": 11834
+    },
+    {
+      "epoch": 0.4304105902462087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.270556096907166e-05,
+      "loss": 0.9782,
+      "step": 11835
+    },
+    {
+      "epoch": 0.43044695784994724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.270442701791369e-05,
+      "loss": 1.124,
+      "step": 11836
+    },
+    {
+      "epoch": 0.43048332545368584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2703293029235635e-05,
+      "loss": 0.9155,
+      "step": 11837
+    },
+    {
+      "epoch": 0.43051969305742444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2702159003053226e-05,
+      "loss": 0.9621,
+      "step": 11838
+    },
+    {
+      "epoch": 0.43055606066116303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2701024939382191e-05,
+      "loss": 1.1163,
+      "step": 11839
+    },
+    {
+      "epoch": 0.4305924282649016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2699890838238271e-05,
+      "loss": 0.9796,
+      "step": 11840
+    },
+    {
+      "epoch": 0.4306287958686402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2698756699637196e-05,
+      "loss": 1.031,
+      "step": 11841
+    },
+    {
+      "epoch": 0.4306651634723788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2697622523594699e-05,
+      "loss": 1.0658,
+      "step": 11842
+    },
+    {
+      "epoch": 0.4307015310761174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2696488310126518e-05,
+      "loss": 0.8587,
+      "step": 11843
+    },
+    {
+      "epoch": 0.43073789867985596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2695354059248386e-05,
+      "loss": 1.1038,
+      "step": 11844
+    },
+    {
+      "epoch": 0.43077426628359455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2694219770976041e-05,
+      "loss": 1.0926,
+      "step": 11845
+    },
+    {
+      "epoch": 0.43081063388733315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2693085445325223e-05,
+      "loss": 1.0458,
+      "step": 11846
+    },
+    {
+      "epoch": 0.43084700149107175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2691951082311662e-05,
+      "loss": 1.0626,
+      "step": 11847
+    },
+    {
+      "epoch": 0.43088336909481034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2690816681951099e-05,
+      "loss": 0.9994,
+      "step": 11848
+    },
+    {
+      "epoch": 0.43091973669854894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2689682244259274e-05,
+      "loss": 0.989,
+      "step": 11849
+    },
+    {
+      "epoch": 0.43095610430228753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2688547769251922e-05,
+      "loss": 0.8791,
+      "step": 11850
+    },
+    {
+      "epoch": 0.43099247190602613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2687413256944785e-05,
+      "loss": 0.9238,
+      "step": 11851
+    },
+    {
+      "epoch": 0.4310288395097647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2686278707353603e-05,
+      "loss": 1.1995,
+      "step": 11852
+    },
+    {
+      "epoch": 0.43106520711350327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2685144120494117e-05,
+      "loss": 1.0644,
+      "step": 11853
+    },
+    {
+      "epoch": 0.43110157471724186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2684009496382062e-05,
+      "loss": 1.0876,
+      "step": 11854
+    },
+    {
+      "epoch": 0.43113794232098046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2682874835033185e-05,
+      "loss": 0.9959,
+      "step": 11855
+    },
+    {
+      "epoch": 0.43117430992471906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2681740136463227e-05,
+      "loss": 0.9927,
+      "step": 11856
+    },
+    {
+      "epoch": 0.43121067752845765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2680605400687927e-05,
+      "loss": 0.9671,
+      "step": 11857
+    },
+    {
+      "epoch": 0.43124704513219625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2679470627723035e-05,
+      "loss": 1.0217,
+      "step": 11858
+    },
+    {
+      "epoch": 0.43128341273593485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2678335817584286e-05,
+      "loss": 1.1211,
+      "step": 11859
+    },
+    {
+      "epoch": 0.43131978033967344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2677200970287427e-05,
+      "loss": 1.0606,
+      "step": 11860
+    },
+    {
+      "epoch": 0.431356147943412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2676066085848205e-05,
+      "loss": 1.0873,
+      "step": 11861
+    },
+    {
+      "epoch": 0.4313925155471506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2674931164282363e-05,
+      "loss": 1.0424,
+      "step": 11862
+    },
+    {
+      "epoch": 0.4314288831508892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2673796205605645e-05,
+      "loss": 1.0128,
+      "step": 11863
+    },
+    {
+      "epoch": 0.43146525075462777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2672661209833796e-05,
+      "loss": 0.9443,
+      "step": 11864
+    },
+    {
+      "epoch": 0.43150161835836637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2671526176982567e-05,
+      "loss": 1.0442,
+      "step": 11865
+    },
+    {
+      "epoch": 0.43153798596210496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2670391107067706e-05,
+      "loss": 0.9975,
+      "step": 11866
+    },
+    {
+      "epoch": 0.43157435356584356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2669256000104953e-05,
+      "loss": 0.9554,
+      "step": 11867
+    },
+    {
+      "epoch": 0.43161072116958216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2668120856110062e-05,
+      "loss": 1.0123,
+      "step": 11868
+    },
+    {
+      "epoch": 0.43164708877332075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2666985675098777e-05,
+      "loss": 1.0435,
+      "step": 11869
+    },
+    {
+      "epoch": 0.4316834563770593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2665850457086854e-05,
+      "loss": 1.0806,
+      "step": 11870
+    },
+    {
+      "epoch": 0.4317198239807979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2664715202090034e-05,
+      "loss": 1.0002,
+      "step": 11871
+    },
+    {
+      "epoch": 0.4317561915845365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2663579910124077e-05,
+      "loss": 0.9868,
+      "step": 11872
+    },
+    {
+      "epoch": 0.4317925591882751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2662444581204723e-05,
+      "loss": 1.0262,
+      "step": 11873
+    },
+    {
+      "epoch": 0.4318289267920137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.266130921534773e-05,
+      "loss": 1.0777,
+      "step": 11874
+    },
+    {
+      "epoch": 0.4318652943957523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2660173812568848e-05,
+      "loss": 1.0284,
+      "step": 11875
+    },
+    {
+      "epoch": 0.43190166199949087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2659038372883825e-05,
+      "loss": 0.9482,
+      "step": 11876
+    },
+    {
+      "epoch": 0.43193802960322947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.265790289630842e-05,
+      "loss": 1.0225,
+      "step": 11877
+    },
+    {
+      "epoch": 0.431974397206968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2656767382858386e-05,
+      "loss": 0.948,
+      "step": 11878
+    },
+    {
+      "epoch": 0.4320107648107066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.265563183254947e-05,
+      "loss": 0.99,
+      "step": 11879
+    },
+    {
+      "epoch": 0.4320471324144452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2654496245397432e-05,
+      "loss": 0.977,
+      "step": 11880
+    },
+    {
+      "epoch": 0.4320835000181838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2653360621418023e-05,
+      "loss": 1.0712,
+      "step": 11881
+    },
+    {
+      "epoch": 0.4321198676219224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2652224960627e-05,
+      "loss": 1.0204,
+      "step": 11882
+    },
+    {
+      "epoch": 0.432156235225661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2651089263040122e-05,
+      "loss": 0.9612,
+      "step": 11883
+    },
+    {
+      "epoch": 0.4321926028293996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2649953528673138e-05,
+      "loss": 1.105,
+      "step": 11884
+    },
+    {
+      "epoch": 0.4322289704331382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2648817757541812e-05,
+      "loss": 1.0972,
+      "step": 11885
+    },
+    {
+      "epoch": 0.4322653380368767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2647681949661896e-05,
+      "loss": 1.0606,
+      "step": 11886
+    },
+    {
+      "epoch": 0.4323017056406153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2646546105049149e-05,
+      "loss": 0.9629,
+      "step": 11887
+    },
+    {
+      "epoch": 0.4323380732443539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.264541022371933e-05,
+      "loss": 0.9805,
+      "step": 11888
+    },
+    {
+      "epoch": 0.4323744408480925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2644274305688198e-05,
+      "loss": 1.0151,
+      "step": 11889
+    },
+    {
+      "epoch": 0.4324108084518311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2643138350971512e-05,
+      "loss": 1.0085,
+      "step": 11890
+    },
+    {
+      "epoch": 0.4324471760555697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2642002359585033e-05,
+      "loss": 1.1237,
+      "step": 11891
+    },
+    {
+      "epoch": 0.4324835436593083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2640866331544516e-05,
+      "loss": 1.0599,
+      "step": 11892
+    },
+    {
+      "epoch": 0.4325199112630469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2639730266865728e-05,
+      "loss": 1.0198,
+      "step": 11893
+    },
+    {
+      "epoch": 0.4325562788667855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.263859416556443e-05,
+      "loss": 1.0031,
+      "step": 11894
+    },
+    {
+      "epoch": 0.43259264647052403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2637458027656377e-05,
+      "loss": 0.9765,
+      "step": 11895
+    },
+    {
+      "epoch": 0.43262901407426263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2636321853157338e-05,
+      "loss": 1.1364,
+      "step": 11896
+    },
+    {
+      "epoch": 0.4326653816780012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2635185642083073e-05,
+      "loss": 1.1347,
+      "step": 11897
+    },
+    {
+      "epoch": 0.4327017492817398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2634049394449346e-05,
+      "loss": 1.0414,
+      "step": 11898
+    },
+    {
+      "epoch": 0.4327381168854784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2632913110271922e-05,
+      "loss": 0.9451,
+      "step": 11899
+    },
+    {
+      "epoch": 0.432774484489217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2631776789566564e-05,
+      "loss": 0.977,
+      "step": 11900
+    },
+    {
+      "epoch": 0.4328108520929556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2630640432349037e-05,
+      "loss": 1.0506,
+      "step": 11901
+    },
+    {
+      "epoch": 0.4328472196966942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2629504038635108e-05,
+      "loss": 1.0887,
+      "step": 11902
+    },
+    {
+      "epoch": 0.43288358730043275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2628367608440538e-05,
+      "loss": 0.9599,
+      "step": 11903
+    },
+    {
+      "epoch": 0.43291995490417134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2627231141781098e-05,
+      "loss": 0.8564,
+      "step": 11904
+    },
+    {
+      "epoch": 0.43295632250790994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2626094638672552e-05,
+      "loss": 0.9801,
+      "step": 11905
+    },
+    {
+      "epoch": 0.43299269011164854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.262495809913067e-05,
+      "loss": 1.0836,
+      "step": 11906
+    },
+    {
+      "epoch": 0.43302905771538713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.262382152317122e-05,
+      "loss": 1.1049,
+      "step": 11907
+    },
+    {
+      "epoch": 0.43306542531912573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2622684910809966e-05,
+      "loss": 1.0079,
+      "step": 11908
+    },
+    {
+      "epoch": 0.4331017929228643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2621548262062684e-05,
+      "loss": 0.8982,
+      "step": 11909
+    },
+    {
+      "epoch": 0.4331381605266029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2620411576945139e-05,
+      "loss": 1.025,
+      "step": 11910
+    },
+    {
+      "epoch": 0.4331745281303415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2619274855473099e-05,
+      "loss": 1.0374,
+      "step": 11911
+    },
+    {
+      "epoch": 0.43321089573408006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2618138097662337e-05,
+      "loss": 1.0099,
+      "step": 11912
+    },
+    {
+      "epoch": 0.43324726333781866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2617001303528621e-05,
+      "loss": 1.0796,
+      "step": 11913
+    },
+    {
+      "epoch": 0.43328363094155725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.261586447308773e-05,
+      "loss": 0.9992,
+      "step": 11914
+    },
+    {
+      "epoch": 0.43331999854529585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.261472760635543e-05,
+      "loss": 1.0557,
+      "step": 11915
+    },
+    {
+      "epoch": 0.43335636614903444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.261359070334749e-05,
+      "loss": 0.9652,
+      "step": 11916
+    },
+    {
+      "epoch": 0.43339273375277304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2612453764079693e-05,
+      "loss": 1.0104,
+      "step": 11917
+    },
+    {
+      "epoch": 0.43342910135651164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2611316788567805e-05,
+      "loss": 0.9946,
+      "step": 11918
+    },
+    {
+      "epoch": 0.43346546896025023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2610179776827598e-05,
+      "loss": 1.0197,
+      "step": 11919
+    },
+    {
+      "epoch": 0.4335018365639888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2609042728874852e-05,
+      "loss": 0.9978,
+      "step": 11920
+    },
+    {
+      "epoch": 0.43353820416772737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2607905644725342e-05,
+      "loss": 0.9462,
+      "step": 11921
+    },
+    {
+      "epoch": 0.43357457177146597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2606768524394837e-05,
+      "loss": 1.0725,
+      "step": 11922
+    },
+    {
+      "epoch": 0.43361093937520456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2605631367899121e-05,
+      "loss": 1.0002,
+      "step": 11923
+    },
+    {
+      "epoch": 0.43364730697894316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2604494175253966e-05,
+      "loss": 1.1449,
+      "step": 11924
+    },
+    {
+      "epoch": 0.43368367458268176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2603356946475147e-05,
+      "loss": 1.0484,
+      "step": 11925
+    },
+    {
+      "epoch": 0.43372004218642035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2602219681578447e-05,
+      "loss": 0.9247,
+      "step": 11926
+    },
+    {
+      "epoch": 0.43375640979015895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.260108238057964e-05,
+      "loss": 0.8951,
+      "step": 11927
+    },
+    {
+      "epoch": 0.4337927773938975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2599945043494506e-05,
+      "loss": 1.0802,
+      "step": 11928
+    },
+    {
+      "epoch": 0.4338291449976361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2598807670338821e-05,
+      "loss": 1.0855,
+      "step": 11929
+    },
+    {
+      "epoch": 0.4338655126013747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2597670261128371e-05,
+      "loss": 1.0962,
+      "step": 11930
+    },
+    {
+      "epoch": 0.4339018802051133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.259653281587893e-05,
+      "loss": 0.949,
+      "step": 11931
+    },
+    {
+      "epoch": 0.4339382478088519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2595395334606277e-05,
+      "loss": 0.9659,
+      "step": 11932
+    },
+    {
+      "epoch": 0.43397461541259047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.25942578173262e-05,
+      "loss": 1.0494,
+      "step": 11933
+    },
+    {
+      "epoch": 0.43401098301632907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2593120264054477e-05,
+      "loss": 1.0557,
+      "step": 11934
+    },
+    {
+      "epoch": 0.43404735062006766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2591982674806886e-05,
+      "loss": 0.9656,
+      "step": 11935
+    },
+    {
+      "epoch": 0.43408371822380626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2590845049599217e-05,
+      "loss": 1.0197,
+      "step": 11936
+    },
+    {
+      "epoch": 0.4341200858275448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2589707388447243e-05,
+      "loss": 1.0797,
+      "step": 11937
+    },
+    {
+      "epoch": 0.4341564534312834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2588569691366757e-05,
+      "loss": 1.0886,
+      "step": 11938
+    },
+    {
+      "epoch": 0.434192821035022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.258743195837354e-05,
+      "loss": 1.1238,
+      "step": 11939
+    },
+    {
+      "epoch": 0.4342291886387606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2586294189483375e-05,
+      "loss": 0.997,
+      "step": 11940
+    },
+    {
+      "epoch": 0.4342655562424992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2585156384712045e-05,
+      "loss": 0.9735,
+      "step": 11941
+    },
+    {
+      "epoch": 0.4343019238462378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2584018544075343e-05,
+      "loss": 0.9397,
+      "step": 11942
+    },
+    {
+      "epoch": 0.4343382914499764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2582880667589046e-05,
+      "loss": 1.0619,
+      "step": 11943
+    },
+    {
+      "epoch": 0.434374659053715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2581742755268947e-05,
+      "loss": 0.9313,
+      "step": 11944
+    },
+    {
+      "epoch": 0.4344110266574535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2580604807130828e-05,
+      "loss": 1.0001,
+      "step": 11945
+    },
+    {
+      "epoch": 0.4344473942611921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2579466823190478e-05,
+      "loss": 1.0993,
+      "step": 11946
+    },
+    {
+      "epoch": 0.4344837618649307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2578328803463688e-05,
+      "loss": 1.1264,
+      "step": 11947
+    },
+    {
+      "epoch": 0.4345201294686693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2577190747966244e-05,
+      "loss": 1.0216,
+      "step": 11948
+    },
+    {
+      "epoch": 0.4345564970724079,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2576052656713934e-05,
+      "loss": 0.9928,
+      "step": 11949
+    },
+    {
+      "epoch": 0.4345928646761465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.257491452972255e-05,
+      "loss": 1.0176,
+      "step": 11950
+    },
+    {
+      "epoch": 0.4346292322798851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2573776367007878e-05,
+      "loss": 1.1141,
+      "step": 11951
+    },
+    {
+      "epoch": 0.4346655998836237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.257263816858571e-05,
+      "loss": 1.0318,
+      "step": 11952
+    },
+    {
+      "epoch": 0.4347019674873623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2571499934471839e-05,
+      "loss": 0.9652,
+      "step": 11953
+    },
+    {
+      "epoch": 0.4347383350911008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2570361664682055e-05,
+      "loss": 1.0378,
+      "step": 11954
+    },
+    {
+      "epoch": 0.4347747026948394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.256922335923215e-05,
+      "loss": 1.0712,
+      "step": 11955
+    },
+    {
+      "epoch": 0.434811070298578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2568085018137914e-05,
+      "loss": 1.0198,
+      "step": 11956
+    },
+    {
+      "epoch": 0.4348474379023166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2566946641415146e-05,
+      "loss": 1.1055,
+      "step": 11957
+    },
+    {
+      "epoch": 0.4348838055060552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2565808229079633e-05,
+      "loss": 1.0643,
+      "step": 11958
+    },
+    {
+      "epoch": 0.4349201731097938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2564669781147169e-05,
+      "loss": 1.0619,
+      "step": 11959
+    },
+    {
+      "epoch": 0.4349565407135324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2563531297633553e-05,
+      "loss": 1.137,
+      "step": 11960
+    },
+    {
+      "epoch": 0.434992908317271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2562392778554577e-05,
+      "loss": 1.0027,
+      "step": 11961
+    },
+    {
+      "epoch": 0.43502927592100954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2561254223926037e-05,
+      "loss": 0.9849,
+      "step": 11962
+    },
+    {
+      "epoch": 0.43506564352474814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.256011563376373e-05,
+      "loss": 1.0187,
+      "step": 11963
+    },
+    {
+      "epoch": 0.43510201112848673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2558977008083449e-05,
+      "loss": 1.0854,
+      "step": 11964
+    },
+    {
+      "epoch": 0.43513837873222533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2557838346900993e-05,
+      "loss": 1.0651,
+      "step": 11965
+    },
+    {
+      "epoch": 0.4351747463359639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.255669965023216e-05,
+      "loss": 1.1066,
+      "step": 11966
+    },
+    {
+      "epoch": 0.4352111139397025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2555560918092745e-05,
+      "loss": 1.0831,
+      "step": 11967
+    },
+    {
+      "epoch": 0.4352474815434411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.255442215049855e-05,
+      "loss": 1.0135,
+      "step": 11968
+    },
+    {
+      "epoch": 0.4352838491471797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.255328334746537e-05,
+      "loss": 0.9891,
+      "step": 11969
+    },
+    {
+      "epoch": 0.4353202167509183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2552144509009008e-05,
+      "loss": 1.0346,
+      "step": 11970
+    },
+    {
+      "epoch": 0.43535658435465685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2551005635145264e-05,
+      "loss": 1.0618,
+      "step": 11971
+    },
+    {
+      "epoch": 0.43539295195839545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2549866725889933e-05,
+      "loss": 1.0298,
+      "step": 11972
+    },
+    {
+      "epoch": 0.43542931956213404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2548727781258817e-05,
+      "loss": 0.9607,
+      "step": 11973
+    },
+    {
+      "epoch": 0.43546568716587264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2547588801267722e-05,
+      "loss": 1.1644,
+      "step": 11974
+    },
+    {
+      "epoch": 0.43550205476961124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2546449785932446e-05,
+      "loss": 1.0263,
+      "step": 11975
+    },
+    {
+      "epoch": 0.43553842237334983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2545310735268794e-05,
+      "loss": 0.9134,
+      "step": 11976
+    },
+    {
+      "epoch": 0.43557478997708843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2544171649292567e-05,
+      "loss": 0.9758,
+      "step": 11977
+    },
+    {
+      "epoch": 0.435611157580827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2543032528019566e-05,
+      "loss": 1.0475,
+      "step": 11978
+    },
+    {
+      "epoch": 0.43564752518456556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2541893371465596e-05,
+      "loss": 0.9691,
+      "step": 11979
+    },
+    {
+      "epoch": 0.43568389278830416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2540754179646463e-05,
+      "loss": 1.062,
+      "step": 11980
+    },
+    {
+      "epoch": 0.43572026039204276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2539614952577971e-05,
+      "loss": 0.8741,
+      "step": 11981
+    },
+    {
+      "epoch": 0.43575662799578135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2538475690275926e-05,
+      "loss": 0.9854,
+      "step": 11982
+    },
+    {
+      "epoch": 0.43579299559951995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.253733639275613e-05,
+      "loss": 0.8441,
+      "step": 11983
+    },
+    {
+      "epoch": 0.43582936320325855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2536197060034394e-05,
+      "loss": 1.0325,
+      "step": 11984
+    },
+    {
+      "epoch": 0.43586573080699714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.253505769212652e-05,
+      "loss": 1.013,
+      "step": 11985
+    },
+    {
+      "epoch": 0.43590209841073574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2533918289048317e-05,
+      "loss": 0.8964,
+      "step": 11986
+    },
+    {
+      "epoch": 0.4359384660144743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.25327788508156e-05,
+      "loss": 1.085,
+      "step": 11987
+    },
+    {
+      "epoch": 0.4359748336182129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.253163937744416e-05,
+      "loss": 1.0689,
+      "step": 11988
+    },
+    {
+      "epoch": 0.43601120122195147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2530499868949822e-05,
+      "loss": 1.0375,
+      "step": 11989
+    },
+    {
+      "epoch": 0.43604756882569007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2529360325348385e-05,
+      "loss": 1.1881,
+      "step": 11990
+    },
+    {
+      "epoch": 0.43608393642942866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2528220746655663e-05,
+      "loss": 0.9029,
+      "step": 11991
+    },
+    {
+      "epoch": 0.43612030403316726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2527081132887466e-05,
+      "loss": 1.165,
+      "step": 11992
+    },
+    {
+      "epoch": 0.43615667163690586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2525941484059606e-05,
+      "loss": 0.9649,
+      "step": 11993
+    },
+    {
+      "epoch": 0.43619303924064445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2524801800187887e-05,
+      "loss": 1.0662,
+      "step": 11994
+    },
+    {
+      "epoch": 0.43622940684438305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2523662081288128e-05,
+      "loss": 0.8231,
+      "step": 11995
+    },
+    {
+      "epoch": 0.4362657744481216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2522522327376136e-05,
+      "loss": 1.0959,
+      "step": 11996
+    },
+    {
+      "epoch": 0.4363021420518602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2521382538467727e-05,
+      "loss": 0.9857,
+      "step": 11997
+    },
+    {
+      "epoch": 0.4363385096555988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2520242714578715e-05,
+      "loss": 1.0755,
+      "step": 11998
+    },
+    {
+      "epoch": 0.4363748772593374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2519102855724907e-05,
+      "loss": 1.1226,
+      "step": 11999
+    },
+    {
+      "epoch": 0.436411244863076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2517962961922123e-05,
+      "loss": 1.0461,
+      "step": 12000
+    },
+    {
+      "epoch": 0.43644761246681457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2516823033186178e-05,
+      "loss": 1.0677,
+      "step": 12001
+    },
+    {
+      "epoch": 0.43648398007055317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.251568306953288e-05,
+      "loss": 1.0355,
+      "step": 12002
+    },
+    {
+      "epoch": 0.43652034767429176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2514543070978052e-05,
+      "loss": 1.1075,
+      "step": 12003
+    },
+    {
+      "epoch": 0.4365567152780303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2513403037537506e-05,
+      "loss": 0.8777,
+      "step": 12004
+    },
+    {
+      "epoch": 0.4365930828817689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2512262969227055e-05,
+      "loss": 0.9827,
+      "step": 12005
+    },
+    {
+      "epoch": 0.4366294504855075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2511122866062526e-05,
+      "loss": 1.022,
+      "step": 12006
+    },
+    {
+      "epoch": 0.4366658180892461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2509982728059726e-05,
+      "loss": 1.0941,
+      "step": 12007
+    },
+    {
+      "epoch": 0.4367021856929847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2508842555234479e-05,
+      "loss": 0.9284,
+      "step": 12008
+    },
+    {
+      "epoch": 0.4367385532967233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2507702347602597e-05,
+      "loss": 0.9991,
+      "step": 12009
+    },
+    {
+      "epoch": 0.4367749209004619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2506562105179907e-05,
+      "loss": 1.0105,
+      "step": 12010
+    },
+    {
+      "epoch": 0.4368112885042005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2505421827982222e-05,
+      "loss": 0.964,
+      "step": 12011
+    },
+    {
+      "epoch": 0.4368476561079391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2504281516025365e-05,
+      "loss": 0.9591,
+      "step": 12012
+    },
+    {
+      "epoch": 0.4368840237116776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2503141169325157e-05,
+      "loss": 0.9169,
+      "step": 12013
+    },
+    {
+      "epoch": 0.4369203913154162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2502000787897415e-05,
+      "loss": 1.0403,
+      "step": 12014
+    },
+    {
+      "epoch": 0.4369567589191548,
+      "grad_norm": 0.0,
+      "learning_rate": 1.250086037175796e-05,
+      "loss": 1.0173,
+      "step": 12015
+    },
+    {
+      "epoch": 0.4369931265228934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2499719920922616e-05,
+      "loss": 0.8288,
+      "step": 12016
+    },
+    {
+      "epoch": 0.437029494126632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2498579435407207e-05,
+      "loss": 1.0148,
+      "step": 12017
+    },
+    {
+      "epoch": 0.4370658617303706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2497438915227552e-05,
+      "loss": 0.998,
+      "step": 12018
+    },
+    {
+      "epoch": 0.4371022293341092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2496298360399475e-05,
+      "loss": 1.0272,
+      "step": 12019
+    },
+    {
+      "epoch": 0.4371385969378478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.24951577709388e-05,
+      "loss": 1.0974,
+      "step": 12020
+    },
+    {
+      "epoch": 0.43717496454158633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2494017146861353e-05,
+      "loss": 1.0048,
+      "step": 12021
+    },
+    {
+      "epoch": 0.4372113321453249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2492876488182955e-05,
+      "loss": 1.1645,
+      "step": 12022
+    },
+    {
+      "epoch": 0.4372476997490635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2491735794919433e-05,
+      "loss": 1.1213,
+      "step": 12023
+    },
+    {
+      "epoch": 0.4372840673528021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2490595067086612e-05,
+      "loss": 0.9059,
+      "step": 12024
+    },
+    {
+      "epoch": 0.4373204349565407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.248945430470032e-05,
+      "loss": 1.0699,
+      "step": 12025
+    },
+    {
+      "epoch": 0.4373568025602793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.248831350777638e-05,
+      "loss": 1.0953,
+      "step": 12026
+    },
+    {
+      "epoch": 0.4373931701640179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.248717267633062e-05,
+      "loss": 1.0456,
+      "step": 12027
+    },
+    {
+      "epoch": 0.4374295377677565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2486031810378867e-05,
+      "loss": 0.8851,
+      "step": 12028
+    },
+    {
+      "epoch": 0.43746590537149505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2484890909936954e-05,
+      "loss": 0.9074,
+      "step": 12029
+    },
+    {
+      "epoch": 0.43750227297523364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2483749975020702e-05,
+      "loss": 0.9791,
+      "step": 12030
+    },
+    {
+      "epoch": 0.43753864057897224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2482609005645943e-05,
+      "loss": 1.1424,
+      "step": 12031
+    },
+    {
+      "epoch": 0.43757500818271083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2481468001828507e-05,
+      "loss": 1.1018,
+      "step": 12032
+    },
+    {
+      "epoch": 0.43761137578644943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2480326963584224e-05,
+      "loss": 0.873,
+      "step": 12033
+    },
+    {
+      "epoch": 0.437647743390188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2479185890928922e-05,
+      "loss": 1.0266,
+      "step": 12034
+    },
+    {
+      "epoch": 0.4376841109939266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2478044783878433e-05,
+      "loss": 1.0837,
+      "step": 12035
+    },
+    {
+      "epoch": 0.4377204785976652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.247690364244859e-05,
+      "loss": 1.0542,
+      "step": 12036
+    },
+    {
+      "epoch": 0.4377568462014038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2475762466655224e-05,
+      "loss": 1.0672,
+      "step": 12037
+    },
+    {
+      "epoch": 0.43779321380514236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2474621256514165e-05,
+      "loss": 0.9631,
+      "step": 12038
+    },
+    {
+      "epoch": 0.43782958140888095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2473480012041246e-05,
+      "loss": 0.9982,
+      "step": 12039
+    },
+    {
+      "epoch": 0.43786594901261955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2472338733252302e-05,
+      "loss": 1.0741,
+      "step": 12040
+    },
+    {
+      "epoch": 0.43790231661635814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2471197420163169e-05,
+      "loss": 1.0525,
+      "step": 12041
+    },
+    {
+      "epoch": 0.43793868422009674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2470056072789674e-05,
+      "loss": 1.0965,
+      "step": 12042
+    },
+    {
+      "epoch": 0.43797505182383534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2468914691147655e-05,
+      "loss": 0.8809,
+      "step": 12043
+    },
+    {
+      "epoch": 0.43801141942757393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2467773275252952e-05,
+      "loss": 1.2034,
+      "step": 12044
+    },
+    {
+      "epoch": 0.43804778703131253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2466631825121394e-05,
+      "loss": 1.1585,
+      "step": 12045
+    },
+    {
+      "epoch": 0.43808415463505107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.246549034076882e-05,
+      "loss": 0.8919,
+      "step": 12046
+    },
+    {
+      "epoch": 0.43812052223878967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2464348822211063e-05,
+      "loss": 1.0717,
+      "step": 12047
+    },
+    {
+      "epoch": 0.43815688984252826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2463207269463964e-05,
+      "loss": 1.1874,
+      "step": 12048
+    },
+    {
+      "epoch": 0.43819325744626686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2462065682543358e-05,
+      "loss": 1.0914,
+      "step": 12049
+    },
+    {
+      "epoch": 0.43822962505000546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2460924061465085e-05,
+      "loss": 0.9891,
+      "step": 12050
+    },
+    {
+      "epoch": 0.43826599265374405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2459782406244982e-05,
+      "loss": 1.0319,
+      "step": 12051
+    },
+    {
+      "epoch": 0.43830236025748265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2458640716898887e-05,
+      "loss": 0.8978,
+      "step": 12052
+    },
+    {
+      "epoch": 0.43833872786122124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.245749899344264e-05,
+      "loss": 1.0228,
+      "step": 12053
+    },
+    {
+      "epoch": 0.43837509546495984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2456357235892085e-05,
+      "loss": 0.9906,
+      "step": 12054
+    },
+    {
+      "epoch": 0.4384114630686984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2455215444263057e-05,
+      "loss": 1.0105,
+      "step": 12055
+    },
+    {
+      "epoch": 0.438447830672437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2454073618571396e-05,
+      "loss": 0.9511,
+      "step": 12056
+    },
+    {
+      "epoch": 0.4384841982761756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2452931758832947e-05,
+      "loss": 1.0036,
+      "step": 12057
+    },
+    {
+      "epoch": 0.43852056587991417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2451789865063548e-05,
+      "loss": 1.0972,
+      "step": 12058
+    },
+    {
+      "epoch": 0.43855693348365277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2450647937279045e-05,
+      "loss": 0.9664,
+      "step": 12059
+    },
+    {
+      "epoch": 0.43859330108739136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2449505975495282e-05,
+      "loss": 1.0033,
+      "step": 12060
+    },
+    {
+      "epoch": 0.43862966869112996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2448363979728093e-05,
+      "loss": 1.0148,
+      "step": 12061
+    },
+    {
+      "epoch": 0.43866603629486856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2447221949993332e-05,
+      "loss": 1.0099,
+      "step": 12062
+    },
+    {
+      "epoch": 0.4387024038986071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2446079886306836e-05,
+      "loss": 1.0045,
+      "step": 12063
+    },
+    {
+      "epoch": 0.4387387715023457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2444937788684455e-05,
+      "loss": 1.0197,
+      "step": 12064
+    },
+    {
+      "epoch": 0.4387751391060843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.244379565714203e-05,
+      "loss": 1.0156,
+      "step": 12065
+    },
+    {
+      "epoch": 0.4388115067098229,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2442653491695404e-05,
+      "loss": 1.0631,
+      "step": 12066
+    },
+    {
+      "epoch": 0.4388478743135615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2441511292360431e-05,
+      "loss": 1.1326,
+      "step": 12067
+    },
+    {
+      "epoch": 0.4388842419173001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.244036905915295e-05,
+      "loss": 1.0762,
+      "step": 12068
+    },
+    {
+      "epoch": 0.4389206095210387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2439226792088812e-05,
+      "loss": 1.1096,
+      "step": 12069
+    },
+    {
+      "epoch": 0.43895697712477727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2438084491183866e-05,
+      "loss": 0.8658,
+      "step": 12070
+    },
+    {
+      "epoch": 0.4389933447285158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.243694215645395e-05,
+      "loss": 0.9804,
+      "step": 12071
+    },
+    {
+      "epoch": 0.4390297123322544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2435799787914924e-05,
+      "loss": 1.02,
+      "step": 12072
+    },
+    {
+      "epoch": 0.439066079935993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2434657385582634e-05,
+      "loss": 0.9587,
+      "step": 12073
+    },
+    {
+      "epoch": 0.4391024475397316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2433514949472923e-05,
+      "loss": 1.0857,
+      "step": 12074
+    },
+    {
+      "epoch": 0.4391388151434702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2432372479601644e-05,
+      "loss": 0.9626,
+      "step": 12075
+    },
+    {
+      "epoch": 0.4391751827472088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2431229975984652e-05,
+      "loss": 1.0219,
+      "step": 12076
+    },
+    {
+      "epoch": 0.4392115503509474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2430087438637788e-05,
+      "loss": 1.1771,
+      "step": 12077
+    },
+    {
+      "epoch": 0.439247917954686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2428944867576912e-05,
+      "loss": 1.0456,
+      "step": 12078
+    },
+    {
+      "epoch": 0.4392842855584246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2427802262817871e-05,
+      "loss": 1.0511,
+      "step": 12079
+    },
+    {
+      "epoch": 0.4393206531621631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.242665962437652e-05,
+      "loss": 0.9982,
+      "step": 12080
+    },
+    {
+      "epoch": 0.4393570207659017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.242551695226871e-05,
+      "loss": 0.8081,
+      "step": 12081
+    },
+    {
+      "epoch": 0.4393933883696403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.242437424651029e-05,
+      "loss": 1.0862,
+      "step": 12082
+    },
+    {
+      "epoch": 0.4394297559733789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2423231507117124e-05,
+      "loss": 1.0646,
+      "step": 12083
+    },
+    {
+      "epoch": 0.4394661235771175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2422088734105054e-05,
+      "loss": 1.0659,
+      "step": 12084
+    },
+    {
+      "epoch": 0.4395024911808561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.242094592748994e-05,
+      "loss": 1.0455,
+      "step": 12085
+    },
+    {
+      "epoch": 0.4395388587845947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2419803087287639e-05,
+      "loss": 1.0097,
+      "step": 12086
+    },
+    {
+      "epoch": 0.4395752263883333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2418660213513998e-05,
+      "loss": 1.1477,
+      "step": 12087
+    },
+    {
+      "epoch": 0.43961159399207184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2417517306184881e-05,
+      "loss": 1.0436,
+      "step": 12088
+    },
+    {
+      "epoch": 0.43964796159581043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2416374365316141e-05,
+      "loss": 1.0617,
+      "step": 12089
+    },
+    {
+      "epoch": 0.43968432919954903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2415231390923634e-05,
+      "loss": 1.1274,
+      "step": 12090
+    },
+    {
+      "epoch": 0.4397206968032876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2414088383023219e-05,
+      "loss": 0.9695,
+      "step": 12091
+    },
+    {
+      "epoch": 0.4397570644070262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2412945341630753e-05,
+      "loss": 0.9686,
+      "step": 12092
+    },
+    {
+      "epoch": 0.4397934320107648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2411802266762094e-05,
+      "loss": 1.1034,
+      "step": 12093
+    },
+    {
+      "epoch": 0.4398297996145034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2410659158433102e-05,
+      "loss": 1.0956,
+      "step": 12094
+    },
+    {
+      "epoch": 0.439866167218242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2409516016659631e-05,
+      "loss": 1.0326,
+      "step": 12095
+    },
+    {
+      "epoch": 0.4399025348219806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2408372841457547e-05,
+      "loss": 0.9407,
+      "step": 12096
+    },
+    {
+      "epoch": 0.43993890242571915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2407229632842706e-05,
+      "loss": 1.0309,
+      "step": 12097
+    },
+    {
+      "epoch": 0.43997527002945774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.240608639083097e-05,
+      "loss": 0.937,
+      "step": 12098
+    },
+    {
+      "epoch": 0.44001163763319634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2404943115438197e-05,
+      "loss": 0.9393,
+      "step": 12099
+    },
+    {
+      "epoch": 0.44004800523693494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2403799806680252e-05,
+      "loss": 1.0007,
+      "step": 12100
+    },
+    {
+      "epoch": 0.44008437284067353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2402656464572993e-05,
+      "loss": 1.0615,
+      "step": 12101
+    },
+    {
+      "epoch": 0.44012074044441213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2401513089132288e-05,
+      "loss": 1.0664,
+      "step": 12102
+    },
+    {
+      "epoch": 0.4401571080481507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2400369680373996e-05,
+      "loss": 0.9916,
+      "step": 12103
+    },
+    {
+      "epoch": 0.4401934756518893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.239922623831398e-05,
+      "loss": 1.0914,
+      "step": 12104
+    },
+    {
+      "epoch": 0.44022984325562786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2398082762968102e-05,
+      "loss": 1.0908,
+      "step": 12105
+    },
+    {
+      "epoch": 0.44026621085936646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2396939254352227e-05,
+      "loss": 1.0978,
+      "step": 12106
+    },
+    {
+      "epoch": 0.44030257846310505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2395795712482223e-05,
+      "loss": 1.0577,
+      "step": 12107
+    },
+    {
+      "epoch": 0.44033894606684365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2394652137373952e-05,
+      "loss": 1.1398,
+      "step": 12108
+    },
+    {
+      "epoch": 0.44037531367058225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2393508529043282e-05,
+      "loss": 1.1158,
+      "step": 12109
+    },
+    {
+      "epoch": 0.44041168127432084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2392364887506074e-05,
+      "loss": 0.9863,
+      "step": 12110
+    },
+    {
+      "epoch": 0.44044804887805944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.23912212127782e-05,
+      "loss": 1.0203,
+      "step": 12111
+    },
+    {
+      "epoch": 0.44048441648179804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2390077504875522e-05,
+      "loss": 1.1101,
+      "step": 12112
+    },
+    {
+      "epoch": 0.4405207840855366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2388933763813913e-05,
+      "loss": 0.9947,
+      "step": 12113
+    },
+    {
+      "epoch": 0.4405571516892752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2387789989609233e-05,
+      "loss": 1.0965,
+      "step": 12114
+    },
+    {
+      "epoch": 0.44059351929301377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.238664618227736e-05,
+      "loss": 1.0576,
+      "step": 12115
+    },
+    {
+      "epoch": 0.44062988689675237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2385502341834155e-05,
+      "loss": 1.0081,
+      "step": 12116
+    },
+    {
+      "epoch": 0.44066625450049096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2384358468295488e-05,
+      "loss": 1.0697,
+      "step": 12117
+    },
+    {
+      "epoch": 0.44070262210422956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2383214561677233e-05,
+      "loss": 1.1134,
+      "step": 12118
+    },
+    {
+      "epoch": 0.44073898970796815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2382070621995252e-05,
+      "loss": 1.0558,
+      "step": 12119
+    },
+    {
+      "epoch": 0.44077535731170675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2380926649265426e-05,
+      "loss": 0.9442,
+      "step": 12120
+    },
+    {
+      "epoch": 0.44081172491544535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2379782643503622e-05,
+      "loss": 1.0572,
+      "step": 12121
+    },
+    {
+      "epoch": 0.4408480925191839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2378638604725708e-05,
+      "loss": 0.9153,
+      "step": 12122
+    },
+    {
+      "epoch": 0.4408844601229225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2377494532947559e-05,
+      "loss": 0.9496,
+      "step": 12123
+    },
+    {
+      "epoch": 0.4409208277266611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2376350428185042e-05,
+      "loss": 0.9468,
+      "step": 12124
+    },
+    {
+      "epoch": 0.4409571953303997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.237520629045404e-05,
+      "loss": 1.0345,
+      "step": 12125
+    },
+    {
+      "epoch": 0.4409935629341383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2374062119770418e-05,
+      "loss": 0.9705,
+      "step": 12126
+    },
+    {
+      "epoch": 0.44102993053787687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2372917916150053e-05,
+      "loss": 1.0547,
+      "step": 12127
+    },
+    {
+      "epoch": 0.44106629814161546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2371773679608818e-05,
+      "loss": 1.0919,
+      "step": 12128
+    },
+    {
+      "epoch": 0.44110266574535406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2370629410162592e-05,
+      "loss": 1.0147,
+      "step": 12129
+    },
+    {
+      "epoch": 0.4411390333490926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2369485107827241e-05,
+      "loss": 1.0414,
+      "step": 12130
+    },
+    {
+      "epoch": 0.4411754009528312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.236834077261865e-05,
+      "loss": 1.0268,
+      "step": 12131
+    },
+    {
+      "epoch": 0.4412117685565698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.236719640455269e-05,
+      "loss": 1.0191,
+      "step": 12132
+    },
+    {
+      "epoch": 0.4412481361603084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2366052003645237e-05,
+      "loss": 1.0444,
+      "step": 12133
+    },
+    {
+      "epoch": 0.441284503764047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.236490756991217e-05,
+      "loss": 1.0088,
+      "step": 12134
+    },
+    {
+      "epoch": 0.4413208713677856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2363763103369365e-05,
+      "loss": 1.0756,
+      "step": 12135
+    },
+    {
+      "epoch": 0.4413572389715242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2362618604032703e-05,
+      "loss": 0.9727,
+      "step": 12136
+    },
+    {
+      "epoch": 0.4413936065752628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.236147407191806e-05,
+      "loss": 1.0429,
+      "step": 12137
+    },
+    {
+      "epoch": 0.44142997417900137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.236032950704131e-05,
+      "loss": 0.9982,
+      "step": 12138
+    },
+    {
+      "epoch": 0.4414663417827399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2359184909418343e-05,
+      "loss": 0.9264,
+      "step": 12139
+    },
+    {
+      "epoch": 0.4415027093864785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.235804027906503e-05,
+      "loss": 0.9993,
+      "step": 12140
+    },
+    {
+      "epoch": 0.4415390769902171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2356895615997255e-05,
+      "loss": 1.0078,
+      "step": 12141
+    },
+    {
+      "epoch": 0.4415754445939557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2355750920230898e-05,
+      "loss": 1.1277,
+      "step": 12142
+    },
+    {
+      "epoch": 0.4416118121976943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2354606191781837e-05,
+      "loss": 1.0555,
+      "step": 12143
+    },
+    {
+      "epoch": 0.4416481798014329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2353461430665956e-05,
+      "loss": 0.9103,
+      "step": 12144
+    },
+    {
+      "epoch": 0.4416845474051715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2352316636899139e-05,
+      "loss": 1.0116,
+      "step": 12145
+    },
+    {
+      "epoch": 0.4417209150089101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2351171810497262e-05,
+      "loss": 1.0555,
+      "step": 12146
+    },
+    {
+      "epoch": 0.4417572826126486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2350026951476214e-05,
+      "loss": 1.0311,
+      "step": 12147
+    },
+    {
+      "epoch": 0.4417936502163872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2348882059851879e-05,
+      "loss": 1.0844,
+      "step": 12148
+    },
+    {
+      "epoch": 0.4418300178201258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2347737135640134e-05,
+      "loss": 1.1033,
+      "step": 12149
+    },
+    {
+      "epoch": 0.4418663854238644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2346592178856871e-05,
+      "loss": 0.8808,
+      "step": 12150
+    },
+    {
+      "epoch": 0.441902753027603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2345447189517968e-05,
+      "loss": 1.0447,
+      "step": 12151
+    },
+    {
+      "epoch": 0.4419391206313416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2344302167639315e-05,
+      "loss": 0.9097,
+      "step": 12152
+    },
+    {
+      "epoch": 0.4419754882350802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2343157113236795e-05,
+      "loss": 1.0656,
+      "step": 12153
+    },
+    {
+      "epoch": 0.4420118558388188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2342012026326291e-05,
+      "loss": 0.9856,
+      "step": 12154
+    },
+    {
+      "epoch": 0.4420482234425574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2340866906923697e-05,
+      "loss": 1.0343,
+      "step": 12155
+    },
+    {
+      "epoch": 0.44208459104629594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2339721755044895e-05,
+      "loss": 1.0098,
+      "step": 12156
+    },
+    {
+      "epoch": 0.44212095865003453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2338576570705774e-05,
+      "loss": 1.0084,
+      "step": 12157
+    },
+    {
+      "epoch": 0.44215732625377313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.233743135392222e-05,
+      "loss": 0.9842,
+      "step": 12158
+    },
+    {
+      "epoch": 0.4421936938575117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2336286104710124e-05,
+      "loss": 1.0979,
+      "step": 12159
+    },
+    {
+      "epoch": 0.4422300614612503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2335140823085372e-05,
+      "loss": 0.9894,
+      "step": 12160
+    },
+    {
+      "epoch": 0.4422664290649889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2333995509063854e-05,
+      "loss": 0.9207,
+      "step": 12161
+    },
+    {
+      "epoch": 0.4423027966687275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2332850162661458e-05,
+      "loss": 1.039,
+      "step": 12162
+    },
+    {
+      "epoch": 0.4423391642724661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.233170478389408e-05,
+      "loss": 1.0516,
+      "step": 12163
+    },
+    {
+      "epoch": 0.44237553187620465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2330559372777602e-05,
+      "loss": 1.0608,
+      "step": 12164
+    },
+    {
+      "epoch": 0.44241189947994325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2329413929327922e-05,
+      "loss": 1.0168,
+      "step": 12165
+    },
+    {
+      "epoch": 0.44244826708368185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.232826845356093e-05,
+      "loss": 1.0042,
+      "step": 12166
+    },
+    {
+      "epoch": 0.44248463468742044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2327122945492514e-05,
+      "loss": 0.9891,
+      "step": 12167
+    },
+    {
+      "epoch": 0.44252100229115904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.232597740513857e-05,
+      "loss": 0.9175,
+      "step": 12168
+    },
+    {
+      "epoch": 0.44255736989489763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.232483183251499e-05,
+      "loss": 0.891,
+      "step": 12169
+    },
+    {
+      "epoch": 0.44259373749863623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2323686227637664e-05,
+      "loss": 0.9949,
+      "step": 12170
+    },
+    {
+      "epoch": 0.4426301051023748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.232254059052249e-05,
+      "loss": 1.0007,
+      "step": 12171
+    },
+    {
+      "epoch": 0.44266647270611337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2321394921185363e-05,
+      "loss": 0.9521,
+      "step": 12172
+    },
+    {
+      "epoch": 0.44270284030985196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.232024921964217e-05,
+      "loss": 1.079,
+      "step": 12173
+    },
+    {
+      "epoch": 0.44273920791359056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2319103485908816e-05,
+      "loss": 1.0439,
+      "step": 12174
+    },
+    {
+      "epoch": 0.44277557551732916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.231795772000119e-05,
+      "loss": 0.9796,
+      "step": 12175
+    },
+    {
+      "epoch": 0.44281194312106775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.231681192193519e-05,
+      "loss": 0.9753,
+      "step": 12176
+    },
+    {
+      "epoch": 0.44284831072480635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.231566609172671e-05,
+      "loss": 0.9783,
+      "step": 12177
+    },
+    {
+      "epoch": 0.44288467832854495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.231452022939165e-05,
+      "loss": 1.0009,
+      "step": 12178
+    },
+    {
+      "epoch": 0.44292104593228354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2313374334945906e-05,
+      "loss": 0.9138,
+      "step": 12179
+    },
+    {
+      "epoch": 0.44295741353602214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2312228408405374e-05,
+      "loss": 1.0604,
+      "step": 12180
+    },
+    {
+      "epoch": 0.4429937811397607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2311082449785954e-05,
+      "loss": 0.9388,
+      "step": 12181
+    },
+    {
+      "epoch": 0.4430301487434993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2309936459103544e-05,
+      "loss": 0.9389,
+      "step": 12182
+    },
+    {
+      "epoch": 0.44306651634723787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2308790436374043e-05,
+      "loss": 0.8969,
+      "step": 12183
+    },
+    {
+      "epoch": 0.44310288395097647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2307644381613351e-05,
+      "loss": 1.0884,
+      "step": 12184
+    },
+    {
+      "epoch": 0.44313925155471506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2306498294837366e-05,
+      "loss": 1.1008,
+      "step": 12185
+    },
+    {
+      "epoch": 0.44317561915845366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.230535217606199e-05,
+      "loss": 0.9797,
+      "step": 12186
+    },
+    {
+      "epoch": 0.44321198676219226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2304206025303125e-05,
+      "loss": 0.9762,
+      "step": 12187
+    },
+    {
+      "epoch": 0.44324835436593085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.230305984257667e-05,
+      "loss": 1.0477,
+      "step": 12188
+    },
+    {
+      "epoch": 0.4432847219696694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2301913627898524e-05,
+      "loss": 1.0106,
+      "step": 12189
+    },
+    {
+      "epoch": 0.443321089573408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2300767381284597e-05,
+      "loss": 0.9686,
+      "step": 12190
+    },
+    {
+      "epoch": 0.4433574571771466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2299621102750783e-05,
+      "loss": 1.0731,
+      "step": 12191
+    },
+    {
+      "epoch": 0.4433938247808852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.229847479231299e-05,
+      "loss": 0.9282,
+      "step": 12192
+    },
+    {
+      "epoch": 0.4434301923846238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2297328449987122e-05,
+      "loss": 0.9966,
+      "step": 12193
+    },
+    {
+      "epoch": 0.4434665599883624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2296182075789081e-05,
+      "loss": 0.9832,
+      "step": 12194
+    },
+    {
+      "epoch": 0.44350292759210097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2295035669734771e-05,
+      "loss": 1.2302,
+      "step": 12195
+    },
+    {
+      "epoch": 0.44353929519583957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.22938892318401e-05,
+      "loss": 0.979,
+      "step": 12196
+    },
+    {
+      "epoch": 0.44357566279957816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2292742762120966e-05,
+      "loss": 1.0241,
+      "step": 12197
+    },
+    {
+      "epoch": 0.4436120304033167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2291596260593284e-05,
+      "loss": 1.0106,
+      "step": 12198
+    },
+    {
+      "epoch": 0.4436483980070553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.229044972727295e-05,
+      "loss": 1.0134,
+      "step": 12199
+    },
+    {
+      "epoch": 0.4436847656107939,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2289303162175878e-05,
+      "loss": 0.9676,
+      "step": 12200
+    },
+    {
+      "epoch": 0.4437211332145325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2288156565317975e-05,
+      "loss": 1.0648,
+      "step": 12201
+    },
+    {
+      "epoch": 0.4437575008182711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2287009936715141e-05,
+      "loss": 1.0078,
+      "step": 12202
+    },
+    {
+      "epoch": 0.4437938684220097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2285863276383294e-05,
+      "loss": 1.147,
+      "step": 12203
+    },
+    {
+      "epoch": 0.4438302360257483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2284716584338335e-05,
+      "loss": 0.9734,
+      "step": 12204
+    },
+    {
+      "epoch": 0.4438666036294869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2283569860596175e-05,
+      "loss": 1.02,
+      "step": 12205
+    },
+    {
+      "epoch": 0.4439029712332254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2282423105172722e-05,
+      "loss": 0.9949,
+      "step": 12206
+    },
+    {
+      "epoch": 0.443939338836964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2281276318083887e-05,
+      "loss": 1.0312,
+      "step": 12207
+    },
+    {
+      "epoch": 0.4439757064407026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2280129499345581e-05,
+      "loss": 0.9559,
+      "step": 12208
+    },
+    {
+      "epoch": 0.4440120740444412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2278982648973715e-05,
+      "loss": 0.8884,
+      "step": 12209
+    },
+    {
+      "epoch": 0.4440484416481798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2277835766984194e-05,
+      "loss": 0.9508,
+      "step": 12210
+    },
+    {
+      "epoch": 0.4440848092519184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2276688853392938e-05,
+      "loss": 0.9299,
+      "step": 12211
+    },
+    {
+      "epoch": 0.444121176855657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2275541908215853e-05,
+      "loss": 1.046,
+      "step": 12212
+    },
+    {
+      "epoch": 0.4441575444593956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.227439493146885e-05,
+      "loss": 1.0532,
+      "step": 12213
+    },
+    {
+      "epoch": 0.44419391206313413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2273247923167846e-05,
+      "loss": 0.9762,
+      "step": 12214
+    },
+    {
+      "epoch": 0.44423027966687273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2272100883328752e-05,
+      "loss": 1.0237,
+      "step": 12215
+    },
+    {
+      "epoch": 0.4442666472706113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2270953811967484e-05,
+      "loss": 0.8947,
+      "step": 12216
+    },
+    {
+      "epoch": 0.4443030148743499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2269806709099954e-05,
+      "loss": 0.9536,
+      "step": 12217
+    },
+    {
+      "epoch": 0.4443393824780885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2268659574742073e-05,
+      "loss": 0.8884,
+      "step": 12218
+    },
+    {
+      "epoch": 0.4443757500818271,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2267512408909762e-05,
+      "loss": 0.8385,
+      "step": 12219
+    },
+    {
+      "epoch": 0.4444121176855657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2266365211618933e-05,
+      "loss": 0.9767,
+      "step": 12220
+    },
+    {
+      "epoch": 0.4444484852893043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.22652179828855e-05,
+      "loss": 1.0606,
+      "step": 12221
+    },
+    {
+      "epoch": 0.4444848528930429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2264070722725383e-05,
+      "loss": 1.0807,
+      "step": 12222
+    },
+    {
+      "epoch": 0.44452122049678144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2262923431154497e-05,
+      "loss": 1.0982,
+      "step": 12223
+    },
+    {
+      "epoch": 0.44455758810052004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2261776108188758e-05,
+      "loss": 1.0446,
+      "step": 12224
+    },
+    {
+      "epoch": 0.44459395570425864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2260628753844089e-05,
+      "loss": 0.9337,
+      "step": 12225
+    },
+    {
+      "epoch": 0.44463032330799723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2259481368136397e-05,
+      "loss": 1.0994,
+      "step": 12226
+    },
+    {
+      "epoch": 0.44466669091173583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2258333951081612e-05,
+      "loss": 0.9891,
+      "step": 12227
+    },
+    {
+      "epoch": 0.4447030585154744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2257186502695647e-05,
+      "loss": 0.975,
+      "step": 12228
+    },
+    {
+      "epoch": 0.444739426119213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2256039022994418e-05,
+      "loss": 0.9786,
+      "step": 12229
+    },
+    {
+      "epoch": 0.4447757937229516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.225489151199385e-05,
+      "loss": 0.9686,
+      "step": 12230
+    },
+    {
+      "epoch": 0.44481216132669016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2253743969709864e-05,
+      "loss": 1.0809,
+      "step": 12231
+    },
+    {
+      "epoch": 0.44484852893042875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2252596396158374e-05,
+      "loss": 1.0858,
+      "step": 12232
+    },
+    {
+      "epoch": 0.44488489653416735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2251448791355309e-05,
+      "loss": 0.9343,
+      "step": 12233
+    },
+    {
+      "epoch": 0.44492126413790595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2250301155316582e-05,
+      "loss": 0.9392,
+      "step": 12234
+    },
+    {
+      "epoch": 0.44495763174164454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2249153488058124e-05,
+      "loss": 1.0327,
+      "step": 12235
+    },
+    {
+      "epoch": 0.44499399934538314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2248005789595852e-05,
+      "loss": 1.1053,
+      "step": 12236
+    },
+    {
+      "epoch": 0.44503036694912174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2246858059945687e-05,
+      "loss": 0.9695,
+      "step": 12237
+    },
+    {
+      "epoch": 0.44506673455286033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2245710299123557e-05,
+      "loss": 0.8741,
+      "step": 12238
+    },
+    {
+      "epoch": 0.44510310215659893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2244562507145379e-05,
+      "loss": 1.1815,
+      "step": 12239
+    },
+    {
+      "epoch": 0.44513946976033747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2243414684027081e-05,
+      "loss": 1.1209,
+      "step": 12240
+    },
+    {
+      "epoch": 0.44517583736407607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.224226682978459e-05,
+      "loss": 1.0524,
+      "step": 12241
+    },
+    {
+      "epoch": 0.44521220496781466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2241118944433828e-05,
+      "loss": 0.9156,
+      "step": 12242
+    },
+    {
+      "epoch": 0.44524857257155326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2239971027990717e-05,
+      "loss": 0.9315,
+      "step": 12243
+    },
+    {
+      "epoch": 0.44528494017529185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2238823080471191e-05,
+      "loss": 1.0173,
+      "step": 12244
+    },
+    {
+      "epoch": 0.44532130777903045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2237675101891167e-05,
+      "loss": 1.057,
+      "step": 12245
+    },
+    {
+      "epoch": 0.44535767538276905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2236527092266578e-05,
+      "loss": 1.0051,
+      "step": 12246
+    },
+    {
+      "epoch": 0.44539404298650764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2235379051613347e-05,
+      "loss": 0.888,
+      "step": 12247
+    },
+    {
+      "epoch": 0.4454304105902462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2234230979947405e-05,
+      "loss": 1.0239,
+      "step": 12248
+    },
+    {
+      "epoch": 0.4454667781939848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2233082877284679e-05,
+      "loss": 1.0073,
+      "step": 12249
+    },
+    {
+      "epoch": 0.4455031457977234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2231934743641094e-05,
+      "loss": 1.0107,
+      "step": 12250
+    },
+    {
+      "epoch": 0.445539513401462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2230786579032583e-05,
+      "loss": 1.0058,
+      "step": 12251
+    },
+    {
+      "epoch": 0.44557588100520057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2229638383475073e-05,
+      "loss": 1.0152,
+      "step": 12252
+    },
+    {
+      "epoch": 0.44561224860893917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2228490156984491e-05,
+      "loss": 1.011,
+      "step": 12253
+    },
+    {
+      "epoch": 0.44564861621267776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2227341899576774e-05,
+      "loss": 0.9311,
+      "step": 12254
+    },
+    {
+      "epoch": 0.44568498381641636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2226193611267848e-05,
+      "loss": 0.9802,
+      "step": 12255
+    },
+    {
+      "epoch": 0.4457213514201549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2225045292073642e-05,
+      "loss": 1.029,
+      "step": 12256
+    },
+    {
+      "epoch": 0.4457577190238935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2223896942010092e-05,
+      "loss": 0.9485,
+      "step": 12257
+    },
+    {
+      "epoch": 0.4457940866276321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2222748561093122e-05,
+      "loss": 1.0622,
+      "step": 12258
+    },
+    {
+      "epoch": 0.4458304542313707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2221600149338675e-05,
+      "loss": 0.9908,
+      "step": 12259
+    },
+    {
+      "epoch": 0.4458668218351093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2220451706762677e-05,
+      "loss": 1.0997,
+      "step": 12260
+    },
+    {
+      "epoch": 0.4459031894388479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2219303233381058e-05,
+      "loss": 1.085,
+      "step": 12261
+    },
+    {
+      "epoch": 0.4459395570425865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2218154729209762e-05,
+      "loss": 1.0783,
+      "step": 12262
+    },
+    {
+      "epoch": 0.4459759246463251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2217006194264709e-05,
+      "loss": 1.0424,
+      "step": 12263
+    },
+    {
+      "epoch": 0.44601229225006367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2215857628561844e-05,
+      "loss": 0.9459,
+      "step": 12264
+    },
+    {
+      "epoch": 0.4460486598538022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.22147090321171e-05,
+      "loss": 1.0454,
+      "step": 12265
+    },
+    {
+      "epoch": 0.4460850274575408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2213560404946408e-05,
+      "loss": 1.0858,
+      "step": 12266
+    },
+    {
+      "epoch": 0.4461213950612794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2212411747065707e-05,
+      "loss": 0.8908,
+      "step": 12267
+    },
+    {
+      "epoch": 0.446157762665018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2211263058490932e-05,
+      "loss": 1.0025,
+      "step": 12268
+    },
+    {
+      "epoch": 0.4461941302687566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.221011433923802e-05,
+      "loss": 1.0498,
+      "step": 12269
+    },
+    {
+      "epoch": 0.4462304978724952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2208965589322906e-05,
+      "loss": 1.0716,
+      "step": 12270
+    },
+    {
+      "epoch": 0.4462668654762338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.220781680876153e-05,
+      "loss": 1.1223,
+      "step": 12271
+    },
+    {
+      "epoch": 0.4463032330799724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2206667997569826e-05,
+      "loss": 1.0397,
+      "step": 12272
+    },
+    {
+      "epoch": 0.4463396006837109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2205519155763737e-05,
+      "loss": 0.8915,
+      "step": 12273
+    },
+    {
+      "epoch": 0.4463759682874495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2204370283359196e-05,
+      "loss": 1.052,
+      "step": 12274
+    },
+    {
+      "epoch": 0.4464123358911881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.220322138037215e-05,
+      "loss": 1.1773,
+      "step": 12275
+    },
+    {
+      "epoch": 0.4464487034949267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.220207244681853e-05,
+      "loss": 1.0887,
+      "step": 12276
+    },
+    {
+      "epoch": 0.4464850710986653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2200923482714278e-05,
+      "loss": 1.0556,
+      "step": 12277
+    },
+    {
+      "epoch": 0.4465214387024039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2199774488075336e-05,
+      "loss": 1.1377,
+      "step": 12278
+    },
+    {
+      "epoch": 0.4465578063061425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2198625462917643e-05,
+      "loss": 0.9657,
+      "step": 12279
+    },
+    {
+      "epoch": 0.4465941739098811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2197476407257145e-05,
+      "loss": 1.0014,
+      "step": 12280
+    },
+    {
+      "epoch": 0.4466305415136197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2196327321109778e-05,
+      "loss": 1.0384,
+      "step": 12281
+    },
+    {
+      "epoch": 0.44666690911735824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2195178204491483e-05,
+      "loss": 1.1452,
+      "step": 12282
+    },
+    {
+      "epoch": 0.44670327672109683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2194029057418208e-05,
+      "loss": 0.9593,
+      "step": 12283
+    },
+    {
+      "epoch": 0.4467396443248354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2192879879905892e-05,
+      "loss": 1.0322,
+      "step": 12284
+    },
+    {
+      "epoch": 0.446776011928574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2191730671970476e-05,
+      "loss": 1.0472,
+      "step": 12285
+    },
+    {
+      "epoch": 0.4468123795323126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.219058143362791e-05,
+      "loss": 1.0518,
+      "step": 12286
+    },
+    {
+      "epoch": 0.4468487471360512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2189432164894135e-05,
+      "loss": 1.0246,
+      "step": 12287
+    },
+    {
+      "epoch": 0.4468851147397898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2188282865785093e-05,
+      "loss": 0.9313,
+      "step": 12288
+    },
+    {
+      "epoch": 0.4469214823435284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2187133536316733e-05,
+      "loss": 1.0393,
+      "step": 12289
+    },
+    {
+      "epoch": 0.44695784994726695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2185984176504994e-05,
+      "loss": 1.1283,
+      "step": 12290
+    },
+    {
+      "epoch": 0.44699421755100555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.218483478636583e-05,
+      "loss": 1.0749,
+      "step": 12291
+    },
+    {
+      "epoch": 0.44703058515474414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2183685365915183e-05,
+      "loss": 0.8475,
+      "step": 12292
+    },
+    {
+      "epoch": 0.44706695275848274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2182535915169002e-05,
+      "loss": 0.9998,
+      "step": 12293
+    },
+    {
+      "epoch": 0.44710332036222133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.218138643414323e-05,
+      "loss": 0.8134,
+      "step": 12294
+    },
+    {
+      "epoch": 0.44713968796595993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2180236922853812e-05,
+      "loss": 0.8954,
+      "step": 12295
+    },
+    {
+      "epoch": 0.4471760555696985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2179087381316705e-05,
+      "loss": 1.0944,
+      "step": 12296
+    },
+    {
+      "epoch": 0.4472124231734371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.217793780954785e-05,
+      "loss": 0.946,
+      "step": 12297
+    },
+    {
+      "epoch": 0.44724879077717566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.21767882075632e-05,
+      "loss": 1.0599,
+      "step": 12298
+    },
+    {
+      "epoch": 0.44728515838091426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2175638575378701e-05,
+      "loss": 0.8686,
+      "step": 12299
+    },
+    {
+      "epoch": 0.44732152598465286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2174488913010302e-05,
+      "loss": 1.0267,
+      "step": 12300
+    },
+    {
+      "epoch": 0.44735789358839145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2173339220473956e-05,
+      "loss": 0.8985,
+      "step": 12301
+    },
+    {
+      "epoch": 0.44739426119213005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2172189497785612e-05,
+      "loss": 1.063,
+      "step": 12302
+    },
+    {
+      "epoch": 0.44743062879586865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.217103974496122e-05,
+      "loss": 1.0151,
+      "step": 12303
+    },
+    {
+      "epoch": 0.44746699639960724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2169889962016731e-05,
+      "loss": 1.0359,
+      "step": 12304
+    },
+    {
+      "epoch": 0.44750336400334584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2168740148968101e-05,
+      "loss": 0.9392,
+      "step": 12305
+    },
+    {
+      "epoch": 0.44753973160708443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2167590305831274e-05,
+      "loss": 0.9869,
+      "step": 12306
+    },
+    {
+      "epoch": 0.447576099210823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.216644043262221e-05,
+      "loss": 0.9894,
+      "step": 12307
+    },
+    {
+      "epoch": 0.44761246681456157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2165290529356858e-05,
+      "loss": 1.08,
+      "step": 12308
+    },
+    {
+      "epoch": 0.44764883441830017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.216414059605117e-05,
+      "loss": 1.0457,
+      "step": 12309
+    },
+    {
+      "epoch": 0.44768520202203876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2162990632721106e-05,
+      "loss": 0.9847,
+      "step": 12310
+    },
+    {
+      "epoch": 0.44772156962577736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2161840639382614e-05,
+      "loss": 0.8294,
+      "step": 12311
+    },
+    {
+      "epoch": 0.44775793722951596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.216069061605165e-05,
+      "loss": 1.0653,
+      "step": 12312
+    },
+    {
+      "epoch": 0.44779430483325455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2159540562744169e-05,
+      "loss": 1.0546,
+      "step": 12313
+    },
+    {
+      "epoch": 0.44783067243699315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2158390479476125e-05,
+      "loss": 1.1242,
+      "step": 12314
+    },
+    {
+      "epoch": 0.4478670400407317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.215724036626348e-05,
+      "loss": 0.9127,
+      "step": 12315
+    },
+    {
+      "epoch": 0.4479034076444703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2156090223122183e-05,
+      "loss": 1.0102,
+      "step": 12316
+    },
+    {
+      "epoch": 0.4479397752482089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2154940050068191e-05,
+      "loss": 1.0971,
+      "step": 12317
+    },
+    {
+      "epoch": 0.4479761428519475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2153789847117468e-05,
+      "loss": 1.1298,
+      "step": 12318
+    },
+    {
+      "epoch": 0.4480125104556861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2152639614285962e-05,
+      "loss": 1.1113,
+      "step": 12319
+    },
+    {
+      "epoch": 0.44804887805942467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2151489351589638e-05,
+      "loss": 1.0256,
+      "step": 12320
+    },
+    {
+      "epoch": 0.44808524566316327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2150339059044452e-05,
+      "loss": 1.1272,
+      "step": 12321
+    },
+    {
+      "epoch": 0.44812161326690186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.214918873666636e-05,
+      "loss": 1.1294,
+      "step": 12322
+    },
+    {
+      "epoch": 0.44815798087064046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2148038384471328e-05,
+      "loss": 1.0869,
+      "step": 12323
+    },
+    {
+      "epoch": 0.448194348474379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2146888002475307e-05,
+      "loss": 1.03,
+      "step": 12324
+    },
+    {
+      "epoch": 0.4482307160781176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2145737590694263e-05,
+      "loss": 1.0307,
+      "step": 12325
+    },
+    {
+      "epoch": 0.4482670836818562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2144587149144154e-05,
+      "loss": 1.0789,
+      "step": 12326
+    },
+    {
+      "epoch": 0.4483034512855948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2143436677840943e-05,
+      "loss": 1.1459,
+      "step": 12327
+    },
+    {
+      "epoch": 0.4483398188893334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2142286176800586e-05,
+      "loss": 0.9827,
+      "step": 12328
+    },
+    {
+      "epoch": 0.448376186493072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2141135646039048e-05,
+      "loss": 0.8961,
+      "step": 12329
+    },
+    {
+      "epoch": 0.4484125540968106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2139985085572293e-05,
+      "loss": 1.0146,
+      "step": 12330
+    },
+    {
+      "epoch": 0.4484489217005492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2138834495416279e-05,
+      "loss": 0.9666,
+      "step": 12331
+    },
+    {
+      "epoch": 0.4484852893042877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2137683875586976e-05,
+      "loss": 1.0994,
+      "step": 12332
+    },
+    {
+      "epoch": 0.4485216569080263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2136533226100336e-05,
+      "loss": 0.9538,
+      "step": 12333
+    },
+    {
+      "epoch": 0.4485580245117649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2135382546972332e-05,
+      "loss": 1.0701,
+      "step": 12334
+    },
+    {
+      "epoch": 0.4485943921155035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2134231838218922e-05,
+      "loss": 0.9535,
+      "step": 12335
+    },
+    {
+      "epoch": 0.4486307597192421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2133081099856076e-05,
+      "loss": 1.1703,
+      "step": 12336
+    },
+    {
+      "epoch": 0.4486671273229807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2131930331899757e-05,
+      "loss": 1.1519,
+      "step": 12337
+    },
+    {
+      "epoch": 0.4487034949267193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2130779534365925e-05,
+      "loss": 1.0769,
+      "step": 12338
+    },
+    {
+      "epoch": 0.4487398625304579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2129628707270556e-05,
+      "loss": 0.9799,
+      "step": 12339
+    },
+    {
+      "epoch": 0.4487762301341965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2128477850629608e-05,
+      "loss": 0.9845,
+      "step": 12340
+    },
+    {
+      "epoch": 0.448812597737935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2127326964459047e-05,
+      "loss": 1.059,
+      "step": 12341
+    },
+    {
+      "epoch": 0.4488489653416736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2126176048774844e-05,
+      "loss": 1.043,
+      "step": 12342
+    },
+    {
+      "epoch": 0.4488853329454122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2125025103592966e-05,
+      "loss": 1.0299,
+      "step": 12343
+    },
+    {
+      "epoch": 0.4489217005491508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.212387412892938e-05,
+      "loss": 1.0292,
+      "step": 12344
+    },
+    {
+      "epoch": 0.4489580681528894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2122723124800054e-05,
+      "loss": 1.0888,
+      "step": 12345
+    },
+    {
+      "epoch": 0.448994435756628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2121572091220955e-05,
+      "loss": 1.0305,
+      "step": 12346
+    },
+    {
+      "epoch": 0.4490308033603666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2120421028208053e-05,
+      "loss": 1.0385,
+      "step": 12347
+    },
+    {
+      "epoch": 0.4490671709641052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.211926993577732e-05,
+      "loss": 1.0765,
+      "step": 12348
+    },
+    {
+      "epoch": 0.44910353856784374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2118118813944721e-05,
+      "loss": 0.9741,
+      "step": 12349
+    },
+    {
+      "epoch": 0.44913990617158234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.211696766272623e-05,
+      "loss": 0.9165,
+      "step": 12350
+    },
+    {
+      "epoch": 0.44917627377532093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2115816482137817e-05,
+      "loss": 1.0685,
+      "step": 12351
+    },
+    {
+      "epoch": 0.44921264137905953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2114665272195452e-05,
+      "loss": 1.0798,
+      "step": 12352
+    },
+    {
+      "epoch": 0.4492490089827981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2113514032915106e-05,
+      "loss": 1.0619,
+      "step": 12353
+    },
+    {
+      "epoch": 0.4492853765865367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.211236276431275e-05,
+      "loss": 1.0655,
+      "step": 12354
+    },
+    {
+      "epoch": 0.4493217441902753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2111211466404361e-05,
+      "loss": 1.0792,
+      "step": 12355
+    },
+    {
+      "epoch": 0.4493581117940139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2110060139205908e-05,
+      "loss": 0.8911,
+      "step": 12356
+    },
+    {
+      "epoch": 0.44939447939775246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.210890878273336e-05,
+      "loss": 1.1036,
+      "step": 12357
+    },
+    {
+      "epoch": 0.44943084700149105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.21077573970027e-05,
+      "loss": 0.9849,
+      "step": 12358
+    },
+    {
+      "epoch": 0.44946721460522965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2106605982029893e-05,
+      "loss": 0.976,
+      "step": 12359
+    },
+    {
+      "epoch": 0.44950358220896824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2105454537830919e-05,
+      "loss": 1.0511,
+      "step": 12360
+    },
+    {
+      "epoch": 0.44953994981270684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.210430306442175e-05,
+      "loss": 0.8807,
+      "step": 12361
+    },
+    {
+      "epoch": 0.44957631741644544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2103151561818359e-05,
+      "loss": 1.0695,
+      "step": 12362
+    },
+    {
+      "epoch": 0.44961268502018403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2102000030036727e-05,
+      "loss": 0.9514,
+      "step": 12363
+    },
+    {
+      "epoch": 0.44964905262392263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2100848469092827e-05,
+      "loss": 1.0457,
+      "step": 12364
+    },
+    {
+      "epoch": 0.4496854202276612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2099696879002633e-05,
+      "loss": 1.0144,
+      "step": 12365
+    },
+    {
+      "epoch": 0.44972178783139977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2098545259782126e-05,
+      "loss": 1.0425,
+      "step": 12366
+    },
+    {
+      "epoch": 0.44975815543513836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2097393611447282e-05,
+      "loss": 1.0431,
+      "step": 12367
+    },
+    {
+      "epoch": 0.44979452303887696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2096241934014074e-05,
+      "loss": 1.1278,
+      "step": 12368
+    },
+    {
+      "epoch": 0.44983089064261556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2095090227498486e-05,
+      "loss": 1.0597,
+      "step": 12369
+    },
+    {
+      "epoch": 0.44986725824635415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2093938491916493e-05,
+      "loss": 0.9031,
+      "step": 12370
+    },
+    {
+      "epoch": 0.44990362585009275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2092786727284077e-05,
+      "loss": 1.0041,
+      "step": 12371
+    },
+    {
+      "epoch": 0.44993999345383134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2091634933617212e-05,
+      "loss": 1.0982,
+      "step": 12372
+    },
+    {
+      "epoch": 0.44997636105756994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2090483110931877e-05,
+      "loss": 1.1431,
+      "step": 12373
+    },
+    {
+      "epoch": 0.4500127286613085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2089331259244057e-05,
+      "loss": 0.9721,
+      "step": 12374
+    },
+    {
+      "epoch": 0.4500490962650471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2088179378569729e-05,
+      "loss": 1.0367,
+      "step": 12375
+    },
+    {
+      "epoch": 0.4500854638687857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2087027468924877e-05,
+      "loss": 1.0276,
+      "step": 12376
+    },
+    {
+      "epoch": 0.45012183147252427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2085875530325482e-05,
+      "loss": 1.1341,
+      "step": 12377
+    },
+    {
+      "epoch": 0.45015819907626287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2084723562787519e-05,
+      "loss": 1.0416,
+      "step": 12378
+    },
+    {
+      "epoch": 0.45019456668000146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2083571566326977e-05,
+      "loss": 1.0401,
+      "step": 12379
+    },
+    {
+      "epoch": 0.45023093428374006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2082419540959835e-05,
+      "loss": 1.1405,
+      "step": 12380
+    },
+    {
+      "epoch": 0.45026730188747865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2081267486702076e-05,
+      "loss": 0.9724,
+      "step": 12381
+    },
+    {
+      "epoch": 0.45030366949121725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2080115403569684e-05,
+      "loss": 0.9424,
+      "step": 12382
+    },
+    {
+      "epoch": 0.4503400370949558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2078963291578642e-05,
+      "loss": 1.0528,
+      "step": 12383
+    },
+    {
+      "epoch": 0.4503764046986944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2077811150744934e-05,
+      "loss": 1.0242,
+      "step": 12384
+    },
+    {
+      "epoch": 0.450412772302433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2076658981084543e-05,
+      "loss": 0.98,
+      "step": 12385
+    },
+    {
+      "epoch": 0.4504491399061716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2075506782613457e-05,
+      "loss": 1.0194,
+      "step": 12386
+    },
+    {
+      "epoch": 0.4504855075099102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2074354555347658e-05,
+      "loss": 0.9418,
+      "step": 12387
+    },
+    {
+      "epoch": 0.4505218751136488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2073202299303134e-05,
+      "loss": 1.1556,
+      "step": 12388
+    },
+    {
+      "epoch": 0.45055824271738737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2072050014495868e-05,
+      "loss": 0.8632,
+      "step": 12389
+    },
+    {
+      "epoch": 0.45059461032112597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2070897700941851e-05,
+      "loss": 0.9371,
+      "step": 12390
+    },
+    {
+      "epoch": 0.4506309779248645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2069745358657064e-05,
+      "loss": 0.9463,
+      "step": 12391
+    },
+    {
+      "epoch": 0.4506673455286031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2068592987657497e-05,
+      "loss": 1.0476,
+      "step": 12392
+    },
+    {
+      "epoch": 0.4507037131323417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2067440587959138e-05,
+      "loss": 1.0598,
+      "step": 12393
+    },
+    {
+      "epoch": 0.4507400807360803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2066288159577972e-05,
+      "loss": 0.9674,
+      "step": 12394
+    },
+    {
+      "epoch": 0.4507764483398189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.206513570252999e-05,
+      "loss": 0.8846,
+      "step": 12395
+    },
+    {
+      "epoch": 0.4508128159435575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2063983216831183e-05,
+      "loss": 0.9198,
+      "step": 12396
+    },
+    {
+      "epoch": 0.4508491835472961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2062830702497535e-05,
+      "loss": 0.9311,
+      "step": 12397
+    },
+    {
+      "epoch": 0.4508855511510347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2061678159545039e-05,
+      "loss": 0.8403,
+      "step": 12398
+    },
+    {
+      "epoch": 0.4509219187547732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2060525587989681e-05,
+      "loss": 1.011,
+      "step": 12399
+    },
+    {
+      "epoch": 0.4509582863585118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2059372987847455e-05,
+      "loss": 0.9298,
+      "step": 12400
+    },
+    {
+      "epoch": 0.4509946539622504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2058220359134354e-05,
+      "loss": 1.0567,
+      "step": 12401
+    },
+    {
+      "epoch": 0.451031021565989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.205706770186636e-05,
+      "loss": 1.0398,
+      "step": 12402
+    },
+    {
+      "epoch": 0.4510673891697276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2055915016059474e-05,
+      "loss": 1.0522,
+      "step": 12403
+    },
+    {
+      "epoch": 0.4511037567734662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2054762301729685e-05,
+      "loss": 1.0445,
+      "step": 12404
+    },
+    {
+      "epoch": 0.4511401243772048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2053609558892983e-05,
+      "loss": 0.9689,
+      "step": 12405
+    },
+    {
+      "epoch": 0.4511764919809434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2052456787565361e-05,
+      "loss": 0.9474,
+      "step": 12406
+    },
+    {
+      "epoch": 0.451212859584682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2051303987762815e-05,
+      "loss": 0.96,
+      "step": 12407
+    },
+    {
+      "epoch": 0.45124922718842053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2050151159501335e-05,
+      "loss": 1.0296,
+      "step": 12408
+    },
+    {
+      "epoch": 0.45128559479215913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.204899830279692e-05,
+      "loss": 1.0768,
+      "step": 12409
+    },
+    {
+      "epoch": 0.4513219623958977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2047845417665556e-05,
+      "loss": 0.9657,
+      "step": 12410
+    },
+    {
+      "epoch": 0.4513583299996363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2046692504123245e-05,
+      "loss": 1.12,
+      "step": 12411
+    },
+    {
+      "epoch": 0.4513946976033749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2045539562185979e-05,
+      "loss": 1.1099,
+      "step": 12412
+    },
+    {
+      "epoch": 0.4514310652071135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2044386591869752e-05,
+      "loss": 1.0075,
+      "step": 12413
+    },
+    {
+      "epoch": 0.4514674328108521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2043233593190565e-05,
+      "loss": 1.0203,
+      "step": 12414
+    },
+    {
+      "epoch": 0.4515038004145907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2042080566164408e-05,
+      "loss": 0.9162,
+      "step": 12415
+    },
+    {
+      "epoch": 0.45154016801832925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.204092751080728e-05,
+      "loss": 1.0331,
+      "step": 12416
+    },
+    {
+      "epoch": 0.45157653562206784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.203977442713518e-05,
+      "loss": 0.9979,
+      "step": 12417
+    },
+    {
+      "epoch": 0.45161290322580644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2038621315164103e-05,
+      "loss": 1.0716,
+      "step": 12418
+    },
+    {
+      "epoch": 0.45164927082954504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2037468174910049e-05,
+      "loss": 1.0795,
+      "step": 12419
+    },
+    {
+      "epoch": 0.45168563843328363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2036315006389015e-05,
+      "loss": 1.1659,
+      "step": 12420
+    },
+    {
+      "epoch": 0.45172200603702223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2035161809616998e-05,
+      "loss": 1.0424,
+      "step": 12421
+    },
+    {
+      "epoch": 0.4517583736407608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.203400858461e-05,
+      "loss": 0.942,
+      "step": 12422
+    },
+    {
+      "epoch": 0.4517947412444994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2032855331384017e-05,
+      "loss": 1.0092,
+      "step": 12423
+    },
+    {
+      "epoch": 0.451831108848238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.203170204995505e-05,
+      "loss": 0.965,
+      "step": 12424
+    },
+    {
+      "epoch": 0.45186747645197656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2030548740339102e-05,
+      "loss": 1.0525,
+      "step": 12425
+    },
+    {
+      "epoch": 0.45190384405571515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2029395402552171e-05,
+      "loss": 0.977,
+      "step": 12426
+    },
+    {
+      "epoch": 0.45194021165945375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2028242036610257e-05,
+      "loss": 1.0349,
+      "step": 12427
+    },
+    {
+      "epoch": 0.45197657926319235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2027088642529366e-05,
+      "loss": 1.007,
+      "step": 12428
+    },
+    {
+      "epoch": 0.45201294686693094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.202593522032549e-05,
+      "loss": 0.9233,
+      "step": 12429
+    },
+    {
+      "epoch": 0.45204931447066954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2024781770014642e-05,
+      "loss": 1.0451,
+      "step": 12430
+    },
+    {
+      "epoch": 0.45208568207440814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.202362829161282e-05,
+      "loss": 0.9802,
+      "step": 12431
+    },
+    {
+      "epoch": 0.45212204967814673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.202247478513602e-05,
+      "loss": 0.9164,
+      "step": 12432
+    },
+    {
+      "epoch": 0.45215841728188527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2021321250600258e-05,
+      "loss": 1.0673,
+      "step": 12433
+    },
+    {
+      "epoch": 0.45219478488562387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.202016768802153e-05,
+      "loss": 0.9562,
+      "step": 12434
+    },
+    {
+      "epoch": 0.45223115248936246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2019014097415843e-05,
+      "loss": 0.9215,
+      "step": 12435
+    },
+    {
+      "epoch": 0.45226752009310106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2017860478799196e-05,
+      "loss": 0.9645,
+      "step": 12436
+    },
+    {
+      "epoch": 0.45230388769683966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2016706832187602e-05,
+      "loss": 1.0406,
+      "step": 12437
+    },
+    {
+      "epoch": 0.45234025530057825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.201555315759706e-05,
+      "loss": 0.9788,
+      "step": 12438
+    },
+    {
+      "epoch": 0.45237662290431685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2014399455043577e-05,
+      "loss": 1.1399,
+      "step": 12439
+    },
+    {
+      "epoch": 0.45241299050805545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.201324572454316e-05,
+      "loss": 1.0315,
+      "step": 12440
+    },
+    {
+      "epoch": 0.452449358111794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2012091966111816e-05,
+      "loss": 1.0334,
+      "step": 12441
+    },
+    {
+      "epoch": 0.4524857257155326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2010938179765553e-05,
+      "loss": 1.0657,
+      "step": 12442
+    },
+    {
+      "epoch": 0.4525220933192712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2009784365520371e-05,
+      "loss": 0.9065,
+      "step": 12443
+    },
+    {
+      "epoch": 0.4525584609230098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2008630523392286e-05,
+      "loss": 0.9609,
+      "step": 12444
+    },
+    {
+      "epoch": 0.45259482852674837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2007476653397298e-05,
+      "loss": 1.1146,
+      "step": 12445
+    },
+    {
+      "epoch": 0.45263119613048697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2006322755551425e-05,
+      "loss": 0.9931,
+      "step": 12446
+    },
+    {
+      "epoch": 0.45266756373422556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2005168829870668e-05,
+      "loss": 0.9465,
+      "step": 12447
+    },
+    {
+      "epoch": 0.45270393133796416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.200401487637104e-05,
+      "loss": 1.1149,
+      "step": 12448
+    },
+    {
+      "epoch": 0.45274029894170276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2002860895068549e-05,
+      "loss": 0.9983,
+      "step": 12449
+    },
+    {
+      "epoch": 0.4527766665454413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.20017068859792e-05,
+      "loss": 0.9881,
+      "step": 12450
+    },
+    {
+      "epoch": 0.4528130341491799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2000552849119012e-05,
+      "loss": 1.1039,
+      "step": 12451
+    },
+    {
+      "epoch": 0.4528494017529185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1999398784503993e-05,
+      "loss": 1.038,
+      "step": 12452
+    },
+    {
+      "epoch": 0.4528857693566571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1998244692150149e-05,
+      "loss": 1.0543,
+      "step": 12453
+    },
+    {
+      "epoch": 0.4529221369603957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1997090572073498e-05,
+      "loss": 1.0123,
+      "step": 12454
+    },
+    {
+      "epoch": 0.4529585045641343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1995936424290047e-05,
+      "loss": 0.951,
+      "step": 12455
+    },
+    {
+      "epoch": 0.4529948721678729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1994782248815808e-05,
+      "loss": 1.0177,
+      "step": 12456
+    },
+    {
+      "epoch": 0.45303123977161147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1993628045666799e-05,
+      "loss": 0.9598,
+      "step": 12457
+    },
+    {
+      "epoch": 0.45306760737535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.199247381485903e-05,
+      "loss": 1.0776,
+      "step": 12458
+    },
+    {
+      "epoch": 0.4531039749790886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.199131955640851e-05,
+      "loss": 0.9629,
+      "step": 12459
+    },
+    {
+      "epoch": 0.4531403425828272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.199016527033126e-05,
+      "loss": 1.0818,
+      "step": 12460
+    },
+    {
+      "epoch": 0.4531767101865658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1989010956643287e-05,
+      "loss": 0.9361,
+      "step": 12461
+    },
+    {
+      "epoch": 0.4532130777903044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.198785661536061e-05,
+      "loss": 1.0175,
+      "step": 12462
+    },
+    {
+      "epoch": 0.453249445394043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1986702246499246e-05,
+      "loss": 1.1177,
+      "step": 12463
+    },
+    {
+      "epoch": 0.4532858129977816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1985547850075206e-05,
+      "loss": 0.9725,
+      "step": 12464
+    },
+    {
+      "epoch": 0.4533221806015202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1984393426104506e-05,
+      "loss": 0.9038,
+      "step": 12465
+    },
+    {
+      "epoch": 0.4533585482052588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1983238974603161e-05,
+      "loss": 1.0415,
+      "step": 12466
+    },
+    {
+      "epoch": 0.4533949158089973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1982084495587195e-05,
+      "loss": 0.9917,
+      "step": 12467
+    },
+    {
+      "epoch": 0.4534312834127359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1980929989072614e-05,
+      "loss": 1.0735,
+      "step": 12468
+    },
+    {
+      "epoch": 0.4534676510164745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1979775455075437e-05,
+      "loss": 1.1387,
+      "step": 12469
+    },
+    {
+      "epoch": 0.4535040186202131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.197862089361169e-05,
+      "loss": 1.0815,
+      "step": 12470
+    },
+    {
+      "epoch": 0.4535403862239517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1977466304697382e-05,
+      "loss": 1.0016,
+      "step": 12471
+    },
+    {
+      "epoch": 0.4535767538276903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1976311688348537e-05,
+      "loss": 1.0518,
+      "step": 12472
+    },
+    {
+      "epoch": 0.4536131214314289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.197515704458117e-05,
+      "loss": 1.0803,
+      "step": 12473
+    },
+    {
+      "epoch": 0.4536494890351675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.19740023734113e-05,
+      "loss": 1.0011,
+      "step": 12474
+    },
+    {
+      "epoch": 0.45368585663890604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1972847674854951e-05,
+      "loss": 1.0089,
+      "step": 12475
+    },
+    {
+      "epoch": 0.45372222424264463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1971692948928137e-05,
+      "loss": 1.1326,
+      "step": 12476
+    },
+    {
+      "epoch": 0.45375859184638323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1970538195646881e-05,
+      "loss": 0.9486,
+      "step": 12477
+    },
+    {
+      "epoch": 0.4537949594501218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1969383415027207e-05,
+      "loss": 1.1461,
+      "step": 12478
+    },
+    {
+      "epoch": 0.4538313270538604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1968228607085129e-05,
+      "loss": 1.0724,
+      "step": 12479
+    },
+    {
+      "epoch": 0.453867694657599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.196707377183667e-05,
+      "loss": 1.0709,
+      "step": 12480
+    },
+    {
+      "epoch": 0.4539040622613376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1965918909297857e-05,
+      "loss": 0.7891,
+      "step": 12481
+    },
+    {
+      "epoch": 0.4539404298650762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1964764019484706e-05,
+      "loss": 0.9963,
+      "step": 12482
+    },
+    {
+      "epoch": 0.45397679746881475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1963609102413242e-05,
+      "loss": 0.9816,
+      "step": 12483
+    },
+    {
+      "epoch": 0.45401316507255335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1962454158099491e-05,
+      "loss": 0.9026,
+      "step": 12484
+    },
+    {
+      "epoch": 0.45404953267629194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.196129918655947e-05,
+      "loss": 1.0695,
+      "step": 12485
+    },
+    {
+      "epoch": 0.45408590028003054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1960144187809208e-05,
+      "loss": 1.0984,
+      "step": 12486
+    },
+    {
+      "epoch": 0.45412226788376914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1958989161864725e-05,
+      "loss": 1.1008,
+      "step": 12487
+    },
+    {
+      "epoch": 0.45415863548750773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1957834108742046e-05,
+      "loss": 0.9571,
+      "step": 12488
+    },
+    {
+      "epoch": 0.45419500309124633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1956679028457196e-05,
+      "loss": 1.0711,
+      "step": 12489
+    },
+    {
+      "epoch": 0.4542313706949849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.19555239210262e-05,
+      "loss": 0.993,
+      "step": 12490
+    },
+    {
+      "epoch": 0.4542677382987235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1954368786465087e-05,
+      "loss": 0.9886,
+      "step": 12491
+    },
+    {
+      "epoch": 0.45430410590246206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1953213624789879e-05,
+      "loss": 1.1096,
+      "step": 12492
+    },
+    {
+      "epoch": 0.45434047350620066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.19520584360166e-05,
+      "loss": 1.0331,
+      "step": 12493
+    },
+    {
+      "epoch": 0.45437684110993926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1950903220161286e-05,
+      "loss": 0.9906,
+      "step": 12494
+    },
+    {
+      "epoch": 0.45441320871367785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1949747977239953e-05,
+      "loss": 0.8717,
+      "step": 12495
+    },
+    {
+      "epoch": 0.45444957631741645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1948592707268634e-05,
+      "loss": 1.0927,
+      "step": 12496
+    },
+    {
+      "epoch": 0.45448594392115504,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1947437410263355e-05,
+      "loss": 0.9576,
+      "step": 12497
+    },
+    {
+      "epoch": 0.45452231152489364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1946282086240148e-05,
+      "loss": 1.0012,
+      "step": 12498
+    },
+    {
+      "epoch": 0.45455867912863224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1945126735215037e-05,
+      "loss": 1.0891,
+      "step": 12499
+    },
+    {
+      "epoch": 0.4545950467323708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1943971357204052e-05,
+      "loss": 1.0283,
+      "step": 12500
+    },
+    {
+      "epoch": 0.4546314143361094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.194281595222322e-05,
+      "loss": 1.0541,
+      "step": 12501
+    },
+    {
+      "epoch": 0.45466778193984797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1941660520288579e-05,
+      "loss": 1.0592,
+      "step": 12502
+    },
+    {
+      "epoch": 0.45470414954358657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.194050506141615e-05,
+      "loss": 1.0918,
+      "step": 12503
+    },
+    {
+      "epoch": 0.45474051714732516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1939349575621966e-05,
+      "loss": 1.067,
+      "step": 12504
+    },
+    {
+      "epoch": 0.45477688475106376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.193819406292206e-05,
+      "loss": 1.0276,
+      "step": 12505
+    },
+    {
+      "epoch": 0.45481325235480236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1937038523332457e-05,
+      "loss": 1.0446,
+      "step": 12506
+    },
+    {
+      "epoch": 0.45484961995854095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1935882956869198e-05,
+      "loss": 1.0383,
+      "step": 12507
+    },
+    {
+      "epoch": 0.45488598756227955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1934727363548308e-05,
+      "loss": 1.1129,
+      "step": 12508
+    },
+    {
+      "epoch": 0.4549223551660181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.193357174338582e-05,
+      "loss": 0.9878,
+      "step": 12509
+    },
+    {
+      "epoch": 0.4549587227697567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1932416096397767e-05,
+      "loss": 0.9168,
+      "step": 12510
+    },
+    {
+      "epoch": 0.4549950903734953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1931260422600183e-05,
+      "loss": 0.9615,
+      "step": 12511
+    },
+    {
+      "epoch": 0.4550314579772339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.19301047220091e-05,
+      "loss": 1.0219,
+      "step": 12512
+    },
+    {
+      "epoch": 0.4550678255809725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1928948994640554e-05,
+      "loss": 0.9957,
+      "step": 12513
+    },
+    {
+      "epoch": 0.45510419318471107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.192779324051058e-05,
+      "loss": 0.8726,
+      "step": 12514
+    },
+    {
+      "epoch": 0.45514056078844967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1926637459635203e-05,
+      "loss": 0.9675,
+      "step": 12515
+    },
+    {
+      "epoch": 0.45517692839218826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.192548165203047e-05,
+      "loss": 1.074,
+      "step": 12516
+    },
+    {
+      "epoch": 0.4552132959959268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1924325817712408e-05,
+      "loss": 1.063,
+      "step": 12517
+    },
+    {
+      "epoch": 0.4552496635996654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1923169956697058e-05,
+      "loss": 1.0547,
+      "step": 12518
+    },
+    {
+      "epoch": 0.455286031203404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1922014069000454e-05,
+      "loss": 1.0415,
+      "step": 12519
+    },
+    {
+      "epoch": 0.4553223988071426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1920858154638627e-05,
+      "loss": 0.9891,
+      "step": 12520
+    },
+    {
+      "epoch": 0.4553587664108812,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1919702213627624e-05,
+      "loss": 1.004,
+      "step": 12521
+    },
+    {
+      "epoch": 0.4553951340146198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1918546245983476e-05,
+      "loss": 1.0695,
+      "step": 12522
+    },
+    {
+      "epoch": 0.4554315016183584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1917390251722218e-05,
+      "loss": 1.1593,
+      "step": 12523
+    },
+    {
+      "epoch": 0.455467869222097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1916234230859893e-05,
+      "loss": 0.9918,
+      "step": 12524
+    },
+    {
+      "epoch": 0.4555042368258356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1915078183412534e-05,
+      "loss": 0.8587,
+      "step": 12525
+    },
+    {
+      "epoch": 0.4555406044295741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1913922109396185e-05,
+      "loss": 1.1434,
+      "step": 12526
+    },
+    {
+      "epoch": 0.4555769720333127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1912766008826882e-05,
+      "loss": 0.9894,
+      "step": 12527
+    },
+    {
+      "epoch": 0.4556133396370513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1911609881720663e-05,
+      "loss": 0.9975,
+      "step": 12528
+    },
+    {
+      "epoch": 0.4556497072407899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1910453728093569e-05,
+      "loss": 0.973,
+      "step": 12529
+    },
+    {
+      "epoch": 0.4556860748445285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1909297547961641e-05,
+      "loss": 1.0193,
+      "step": 12530
+    },
+    {
+      "epoch": 0.4557224424482671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1908141341340918e-05,
+      "loss": 0.9834,
+      "step": 12531
+    },
+    {
+      "epoch": 0.4557588100520057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1906985108247441e-05,
+      "loss": 0.9818,
+      "step": 12532
+    },
+    {
+      "epoch": 0.4557951776557443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1905828848697252e-05,
+      "loss": 1.0564,
+      "step": 12533
+    },
+    {
+      "epoch": 0.45583154525948283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1904672562706392e-05,
+      "loss": 1.0337,
+      "step": 12534
+    },
+    {
+      "epoch": 0.4558679128632214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1903516250290903e-05,
+      "loss": 1.0403,
+      "step": 12535
+    },
+    {
+      "epoch": 0.45590428046696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1902359911466826e-05,
+      "loss": 0.9597,
+      "step": 12536
+    },
+    {
+      "epoch": 0.4559406480706986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1901203546250205e-05,
+      "loss": 1.061,
+      "step": 12537
+    },
+    {
+      "epoch": 0.4559770156744372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1900047154657081e-05,
+      "loss": 0.982,
+      "step": 12538
+    },
+    {
+      "epoch": 0.4560133832781758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1898890736703499e-05,
+      "loss": 1.0887,
+      "step": 12539
+    },
+    {
+      "epoch": 0.4560497508819144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1897734292405503e-05,
+      "loss": 0.9949,
+      "step": 12540
+    },
+    {
+      "epoch": 0.456086118485653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1896577821779135e-05,
+      "loss": 0.9718,
+      "step": 12541
+    },
+    {
+      "epoch": 0.45612248608939154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.189542132484044e-05,
+      "loss": 1.0472,
+      "step": 12542
+    },
+    {
+      "epoch": 0.45615885369313014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1894264801605466e-05,
+      "loss": 1.0103,
+      "step": 12543
+    },
+    {
+      "epoch": 0.45619522129686874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1893108252090253e-05,
+      "loss": 1.0528,
+      "step": 12544
+    },
+    {
+      "epoch": 0.45623158890060733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1891951676310851e-05,
+      "loss": 1.0878,
+      "step": 12545
+    },
+    {
+      "epoch": 0.45626795650434593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1890795074283301e-05,
+      "loss": 1.0956,
+      "step": 12546
+    },
+    {
+      "epoch": 0.4563043241080845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1889638446023655e-05,
+      "loss": 1.0098,
+      "step": 12547
+    },
+    {
+      "epoch": 0.4563406917118231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1888481791547957e-05,
+      "loss": 1.063,
+      "step": 12548
+    },
+    {
+      "epoch": 0.4563770593155617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.188732511087225e-05,
+      "loss": 1.0744,
+      "step": 12549
+    },
+    {
+      "epoch": 0.4564134269193003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1886168404012589e-05,
+      "loss": 1.0776,
+      "step": 12550
+    },
+    {
+      "epoch": 0.45644979452303885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1885011670985014e-05,
+      "loss": 0.8961,
+      "step": 12551
+    },
+    {
+      "epoch": 0.45648616212677745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1883854911805576e-05,
+      "loss": 1.0431,
+      "step": 12552
+    },
+    {
+      "epoch": 0.45652252973051605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1882698126490327e-05,
+      "loss": 1.1008,
+      "step": 12553
+    },
+    {
+      "epoch": 0.45655889733425464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1881541315055314e-05,
+      "loss": 0.9399,
+      "step": 12554
+    },
+    {
+      "epoch": 0.45659526493799324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.188038447751658e-05,
+      "loss": 0.9816,
+      "step": 12555
+    },
+    {
+      "epoch": 0.45663163254173184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1879227613890181e-05,
+      "loss": 1.0902,
+      "step": 12556
+    },
+    {
+      "epoch": 0.45666800014547043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1878070724192165e-05,
+      "loss": 0.9537,
+      "step": 12557
+    },
+    {
+      "epoch": 0.45670436774920903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1876913808438584e-05,
+      "loss": 0.8654,
+      "step": 12558
+    },
+    {
+      "epoch": 0.45674073535294757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1875756866645487e-05,
+      "loss": 1.1716,
+      "step": 12559
+    },
+    {
+      "epoch": 0.45677710295668617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1874599898828923e-05,
+      "loss": 0.9454,
+      "step": 12560
+    },
+    {
+      "epoch": 0.45681347056042476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1873442905004946e-05,
+      "loss": 1.0055,
+      "step": 12561
+    },
+    {
+      "epoch": 0.45684983816416336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1872285885189607e-05,
+      "loss": 0.9362,
+      "step": 12562
+    },
+    {
+      "epoch": 0.45688620576790195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1871128839398958e-05,
+      "loss": 0.9596,
+      "step": 12563
+    },
+    {
+      "epoch": 0.45692257337164055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.186997176764905e-05,
+      "loss": 1.0211,
+      "step": 12564
+    },
+    {
+      "epoch": 0.45695894097537915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1868814669955936e-05,
+      "loss": 0.9536,
+      "step": 12565
+    },
+    {
+      "epoch": 0.45699530857911774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1867657546335674e-05,
+      "loss": 1.1066,
+      "step": 12566
+    },
+    {
+      "epoch": 0.45703167618285634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.186650039680431e-05,
+      "loss": 0.9939,
+      "step": 12567
+    },
+    {
+      "epoch": 0.4570680437865949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1865343221377903e-05,
+      "loss": 0.9881,
+      "step": 12568
+    },
+    {
+      "epoch": 0.4571044113903335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1864186020072503e-05,
+      "loss": 1.0305,
+      "step": 12569
+    },
+    {
+      "epoch": 0.4571407789940721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.186302879290417e-05,
+      "loss": 1.0673,
+      "step": 12570
+    },
+    {
+      "epoch": 0.45717714659781067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1861871539888951e-05,
+      "loss": 0.9514,
+      "step": 12571
+    },
+    {
+      "epoch": 0.45721351420154926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1860714261042911e-05,
+      "loss": 1.0847,
+      "step": 12572
+    },
+    {
+      "epoch": 0.45724988180528786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1859556956382097e-05,
+      "loss": 1.0584,
+      "step": 12573
+    },
+    {
+      "epoch": 0.45728624940902646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1858399625922573e-05,
+      "loss": 0.9741,
+      "step": 12574
+    },
+    {
+      "epoch": 0.45732261701276505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1857242269680386e-05,
+      "loss": 1.0244,
+      "step": 12575
+    },
+    {
+      "epoch": 0.4573589846165036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1856084887671599e-05,
+      "loss": 0.864,
+      "step": 12576
+    },
+    {
+      "epoch": 0.4573953522202422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.185492747991227e-05,
+      "loss": 1.0304,
+      "step": 12577
+    },
+    {
+      "epoch": 0.4574317198239808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1853770046418454e-05,
+      "loss": 0.9899,
+      "step": 12578
+    },
+    {
+      "epoch": 0.4574680874277194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1852612587206204e-05,
+      "loss": 0.9539,
+      "step": 12579
+    },
+    {
+      "epoch": 0.457504455031458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1851455102291588e-05,
+      "loss": 1.1104,
+      "step": 12580
+    },
+    {
+      "epoch": 0.4575408226351966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1850297591690658e-05,
+      "loss": 1.1514,
+      "step": 12581
+    },
+    {
+      "epoch": 0.45757719023893517,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1849140055419474e-05,
+      "loss": 0.886,
+      "step": 12582
+    },
+    {
+      "epoch": 0.45761355784267377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1847982493494097e-05,
+      "loss": 0.9957,
+      "step": 12583
+    },
+    {
+      "epoch": 0.4576499254464123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.184682490593058e-05,
+      "loss": 1.1172,
+      "step": 12584
+    },
+    {
+      "epoch": 0.4576862930501509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.184566729274499e-05,
+      "loss": 0.9923,
+      "step": 12585
+    },
+    {
+      "epoch": 0.4577226606538895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1844509653953389e-05,
+      "loss": 1.1219,
+      "step": 12586
+    },
+    {
+      "epoch": 0.4577590282576281,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1843351989571828e-05,
+      "loss": 1.1265,
+      "step": 12587
+    },
+    {
+      "epoch": 0.4577953958613667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1842194299616377e-05,
+      "loss": 1.1478,
+      "step": 12588
+    },
+    {
+      "epoch": 0.4578317634651053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.184103658410309e-05,
+      "loss": 1.0915,
+      "step": 12589
+    },
+    {
+      "epoch": 0.4578681310688439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1839878843048036e-05,
+      "loss": 1.0146,
+      "step": 12590
+    },
+    {
+      "epoch": 0.4579044986725825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1838721076467273e-05,
+      "loss": 1.0177,
+      "step": 12591
+    },
+    {
+      "epoch": 0.4579408662763211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.183756328437686e-05,
+      "loss": 1.0897,
+      "step": 12592
+    },
+    {
+      "epoch": 0.4579772338800596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1836405466792869e-05,
+      "loss": 1.0081,
+      "step": 12593
+    },
+    {
+      "epoch": 0.4580136014837982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1835247623731356e-05,
+      "loss": 1.0659,
+      "step": 12594
+    },
+    {
+      "epoch": 0.4580499690875368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1834089755208385e-05,
+      "loss": 1.0384,
+      "step": 12595
+    },
+    {
+      "epoch": 0.4580863366912754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1832931861240022e-05,
+      "loss": 1.0221,
+      "step": 12596
+    },
+    {
+      "epoch": 0.458122704295014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1831773941842331e-05,
+      "loss": 1.0327,
+      "step": 12597
+    },
+    {
+      "epoch": 0.4581590718987526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1830615997031372e-05,
+      "loss": 1.0909,
+      "step": 12598
+    },
+    {
+      "epoch": 0.4581954395024912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1829458026823215e-05,
+      "loss": 0.9914,
+      "step": 12599
+    },
+    {
+      "epoch": 0.4582318071062298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1828300031233926e-05,
+      "loss": 1.0697,
+      "step": 12600
+    },
+    {
+      "epoch": 0.45826817470996833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1827142010279567e-05,
+      "loss": 1.0539,
+      "step": 12601
+    },
+    {
+      "epoch": 0.45830454231370693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1825983963976202e-05,
+      "loss": 1.0682,
+      "step": 12602
+    },
+    {
+      "epoch": 0.4583409099174455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1824825892339902e-05,
+      "loss": 0.9906,
+      "step": 12603
+    },
+    {
+      "epoch": 0.4583772775211841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1823667795386733e-05,
+      "loss": 0.967,
+      "step": 12604
+    },
+    {
+      "epoch": 0.4584136451249227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1822509673132759e-05,
+      "loss": 1.0121,
+      "step": 12605
+    },
+    {
+      "epoch": 0.4584500127286613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.182135152559405e-05,
+      "loss": 1.1054,
+      "step": 12606
+    },
+    {
+      "epoch": 0.4584863803323999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1820193352786673e-05,
+      "loss": 1.0496,
+      "step": 12607
+    },
+    {
+      "epoch": 0.4585227479361385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1819035154726694e-05,
+      "loss": 1.1082,
+      "step": 12608
+    },
+    {
+      "epoch": 0.4585591155398771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1817876931430185e-05,
+      "loss": 1.0522,
+      "step": 12609
+    },
+    {
+      "epoch": 0.45859548314361565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1816718682913213e-05,
+      "loss": 1.0477,
+      "step": 12610
+    },
+    {
+      "epoch": 0.45863185074735424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1815560409191843e-05,
+      "loss": 0.9739,
+      "step": 12611
+    },
+    {
+      "epoch": 0.45866821835109284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1814402110282155e-05,
+      "loss": 0.9973,
+      "step": 12612
+    },
+    {
+      "epoch": 0.45870458595483143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1813243786200208e-05,
+      "loss": 0.9998,
+      "step": 12613
+    },
+    {
+      "epoch": 0.45874095355857003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1812085436962075e-05,
+      "loss": 1.1205,
+      "step": 12614
+    },
+    {
+      "epoch": 0.4587773211623086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.181092706258383e-05,
+      "loss": 1.1538,
+      "step": 12615
+    },
+    {
+      "epoch": 0.4588136887660472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.180976866308154e-05,
+      "loss": 0.9974,
+      "step": 12616
+    },
+    {
+      "epoch": 0.4588500563697858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1808610238471279e-05,
+      "loss": 1.0369,
+      "step": 12617
+    },
+    {
+      "epoch": 0.45888642397352436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1807451788769116e-05,
+      "loss": 1.0668,
+      "step": 12618
+    },
+    {
+      "epoch": 0.45892279157726296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1806293313991124e-05,
+      "loss": 0.9124,
+      "step": 12619
+    },
+    {
+      "epoch": 0.45895915918100155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1805134814153376e-05,
+      "loss": 0.9764,
+      "step": 12620
+    },
+    {
+      "epoch": 0.45899552678474015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1803976289271941e-05,
+      "loss": 0.975,
+      "step": 12621
+    },
+    {
+      "epoch": 0.45903189438847875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1802817739362898e-05,
+      "loss": 1.094,
+      "step": 12622
+    },
+    {
+      "epoch": 0.45906826199221734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1801659164442314e-05,
+      "loss": 1.1359,
+      "step": 12623
+    },
+    {
+      "epoch": 0.45910462959595594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1800500564526265e-05,
+      "loss": 1.0849,
+      "step": 12624
+    },
+    {
+      "epoch": 0.45914099719969453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1799341939630829e-05,
+      "loss": 1.0927,
+      "step": 12625
+    },
+    {
+      "epoch": 0.4591773648034331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1798183289772074e-05,
+      "loss": 1.0703,
+      "step": 12626
+    },
+    {
+      "epoch": 0.45921373240717167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1797024614966075e-05,
+      "loss": 0.9671,
+      "step": 12627
+    },
+    {
+      "epoch": 0.45925010001091027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1795865915228912e-05,
+      "loss": 0.8488,
+      "step": 12628
+    },
+    {
+      "epoch": 0.45928646761464886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1794707190576655e-05,
+      "loss": 1.1371,
+      "step": 12629
+    },
+    {
+      "epoch": 0.45932283521838746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1793548441025384e-05,
+      "loss": 1.0028,
+      "step": 12630
+    },
+    {
+      "epoch": 0.45935920282212606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1792389666591173e-05,
+      "loss": 0.9402,
+      "step": 12631
+    },
+    {
+      "epoch": 0.45939557042586465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1791230867290098e-05,
+      "loss": 1.0107,
+      "step": 12632
+    },
+    {
+      "epoch": 0.45943193802960325,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1790072043138237e-05,
+      "loss": 1.1523,
+      "step": 12633
+    },
+    {
+      "epoch": 0.45946830563334184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1788913194151665e-05,
+      "loss": 0.9447,
+      "step": 12634
+    },
+    {
+      "epoch": 0.4595046732370804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1787754320346461e-05,
+      "loss": 1.131,
+      "step": 12635
+    },
+    {
+      "epoch": 0.459541040840819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1786595421738702e-05,
+      "loss": 1.1344,
+      "step": 12636
+    },
+    {
+      "epoch": 0.4595774084445576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1785436498344466e-05,
+      "loss": 0.9793,
+      "step": 12637
+    },
+    {
+      "epoch": 0.4596137760482962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1784277550179834e-05,
+      "loss": 0.9619,
+      "step": 12638
+    },
+    {
+      "epoch": 0.45965014365203477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1783118577260883e-05,
+      "loss": 1.0286,
+      "step": 12639
+    },
+    {
+      "epoch": 0.45968651125577337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1781959579603688e-05,
+      "loss": 0.912,
+      "step": 12640
+    },
+    {
+      "epoch": 0.45972287885951196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1780800557224335e-05,
+      "loss": 0.9385,
+      "step": 12641
+    },
+    {
+      "epoch": 0.45975924646325056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.17796415101389e-05,
+      "loss": 0.9833,
+      "step": 12642
+    },
+    {
+      "epoch": 0.4597956140669891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1778482438363464e-05,
+      "loss": 1.0681,
+      "step": 12643
+    },
+    {
+      "epoch": 0.4598319816707277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1777323341914107e-05,
+      "loss": 1.035,
+      "step": 12644
+    },
+    {
+      "epoch": 0.4598683492744663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.177616422080691e-05,
+      "loss": 1.0208,
+      "step": 12645
+    },
+    {
+      "epoch": 0.4599047168782049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1775005075057955e-05,
+      "loss": 0.9062,
+      "step": 12646
+    },
+    {
+      "epoch": 0.4599410844819435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1773845904683323e-05,
+      "loss": 0.9703,
+      "step": 12647
+    },
+    {
+      "epoch": 0.4599774520856821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1772686709699094e-05,
+      "loss": 1.1325,
+      "step": 12648
+    },
+    {
+      "epoch": 0.4600138196894207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1771527490121356e-05,
+      "loss": 0.9881,
+      "step": 12649
+    },
+    {
+      "epoch": 0.4600501872931593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1770368245966187e-05,
+      "loss": 0.9451,
+      "step": 12650
+    },
+    {
+      "epoch": 0.46008655489689787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1769208977249666e-05,
+      "loss": 0.9448,
+      "step": 12651
+    },
+    {
+      "epoch": 0.4601229225006364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1768049683987884e-05,
+      "loss": 1.017,
+      "step": 12652
+    },
+    {
+      "epoch": 0.460159290104375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.176689036619692e-05,
+      "loss": 1.0837,
+      "step": 12653
+    },
+    {
+      "epoch": 0.4601956577081136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.176573102389286e-05,
+      "loss": 0.9576,
+      "step": 12654
+    },
+    {
+      "epoch": 0.4602320253118522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1764571657091787e-05,
+      "loss": 1.0187,
+      "step": 12655
+    },
+    {
+      "epoch": 0.4602683929155908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1763412265809784e-05,
+      "loss": 1.1212,
+      "step": 12656
+    },
+    {
+      "epoch": 0.4603047605193294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1762252850062943e-05,
+      "loss": 1.013,
+      "step": 12657
+    },
+    {
+      "epoch": 0.460341128123068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1761093409867341e-05,
+      "loss": 0.8993,
+      "step": 12658
+    },
+    {
+      "epoch": 0.4603774957268066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1759933945239066e-05,
+      "loss": 1.0623,
+      "step": 12659
+    },
+    {
+      "epoch": 0.4604138633305451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1758774456194204e-05,
+      "loss": 1.0492,
+      "step": 12660
+    },
+    {
+      "epoch": 0.4604502309342837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1757614942748843e-05,
+      "loss": 0.9901,
+      "step": 12661
+    },
+    {
+      "epoch": 0.4604865985380223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1756455404919068e-05,
+      "loss": 0.9509,
+      "step": 12662
+    },
+    {
+      "epoch": 0.4605229661417609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1755295842720966e-05,
+      "loss": 1.0763,
+      "step": 12663
+    },
+    {
+      "epoch": 0.4605593337454995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1754136256170626e-05,
+      "loss": 0.8538,
+      "step": 12664
+    },
+    {
+      "epoch": 0.4605957013492381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1752976645284133e-05,
+      "loss": 1.1569,
+      "step": 12665
+    },
+    {
+      "epoch": 0.4606320689529767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1751817010077578e-05,
+      "loss": 0.9,
+      "step": 12666
+    },
+    {
+      "epoch": 0.4606684365567153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1750657350567043e-05,
+      "loss": 0.8604,
+      "step": 12667
+    },
+    {
+      "epoch": 0.46070480416045384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1749497666768625e-05,
+      "loss": 1.1212,
+      "step": 12668
+    },
+    {
+      "epoch": 0.46074117176419244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1748337958698409e-05,
+      "loss": 0.9332,
+      "step": 12669
+    },
+    {
+      "epoch": 0.46077753936793103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1747178226372482e-05,
+      "loss": 1.0505,
+      "step": 12670
+    },
+    {
+      "epoch": 0.46081390697166963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1746018469806938e-05,
+      "loss": 1.1628,
+      "step": 12671
+    },
+    {
+      "epoch": 0.4608502745754082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1744858689017862e-05,
+      "loss": 1.0866,
+      "step": 12672
+    },
+    {
+      "epoch": 0.4608866421791468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.174369888402135e-05,
+      "loss": 1.0969,
+      "step": 12673
+    },
+    {
+      "epoch": 0.4609230097828854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.174253905483349e-05,
+      "loss": 1.1005,
+      "step": 12674
+    },
+    {
+      "epoch": 0.460959377386624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1741379201470373e-05,
+      "loss": 0.891,
+      "step": 12675
+    },
+    {
+      "epoch": 0.4609957449903626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.174021932394809e-05,
+      "loss": 1.0162,
+      "step": 12676
+    },
+    {
+      "epoch": 0.46103211259410115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1739059422282735e-05,
+      "loss": 0.9998,
+      "step": 12677
+    },
+    {
+      "epoch": 0.46106848019783975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1737899496490397e-05,
+      "loss": 0.9776,
+      "step": 12678
+    },
+    {
+      "epoch": 0.46110484780157834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1736739546587168e-05,
+      "loss": 1.0215,
+      "step": 12679
+    },
+    {
+      "epoch": 0.46114121540531694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.173557957258914e-05,
+      "loss": 0.9086,
+      "step": 12680
+    },
+    {
+      "epoch": 0.46117758300905554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.173441957451241e-05,
+      "loss": 1.0116,
+      "step": 12681
+    },
+    {
+      "epoch": 0.46121395061279413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1733259552373071e-05,
+      "loss": 1.067,
+      "step": 12682
+    },
+    {
+      "epoch": 0.46125031821653273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.173209950618721e-05,
+      "loss": 1.0572,
+      "step": 12683
+    },
+    {
+      "epoch": 0.4612866858202713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1730939435970932e-05,
+      "loss": 1.1023,
+      "step": 12684
+    },
+    {
+      "epoch": 0.46132305342400987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.172977934174032e-05,
+      "loss": 1.051,
+      "step": 12685
+    },
+    {
+      "epoch": 0.46135942102774846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1728619223511476e-05,
+      "loss": 0.9951,
+      "step": 12686
+    },
+    {
+      "epoch": 0.46139578863148706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1727459081300494e-05,
+      "loss": 0.9447,
+      "step": 12687
+    },
+    {
+      "epoch": 0.46143215623522565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1726298915123465e-05,
+      "loss": 0.9967,
+      "step": 12688
+    },
+    {
+      "epoch": 0.46146852383896425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.172513872499649e-05,
+      "loss": 0.9548,
+      "step": 12689
+    },
+    {
+      "epoch": 0.46150489144270285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1723978510935666e-05,
+      "loss": 1.0324,
+      "step": 12690
+    },
+    {
+      "epoch": 0.46154125904644144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1722818272957079e-05,
+      "loss": 1.1887,
+      "step": 12691
+    },
+    {
+      "epoch": 0.46157762665018004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1721658011076838e-05,
+      "loss": 0.9711,
+      "step": 12692
+    },
+    {
+      "epoch": 0.46161399425391864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1720497725311036e-05,
+      "loss": 1.1169,
+      "step": 12693
+    },
+    {
+      "epoch": 0.4616503618576572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1719337415675765e-05,
+      "loss": 0.9812,
+      "step": 12694
+    },
+    {
+      "epoch": 0.4616867294613958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1718177082187127e-05,
+      "loss": 0.9962,
+      "step": 12695
+    },
+    {
+      "epoch": 0.46172309706513437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1717016724861223e-05,
+      "loss": 0.9413,
+      "step": 12696
+    },
+    {
+      "epoch": 0.46175946466887297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1715856343714148e-05,
+      "loss": 0.936,
+      "step": 12697
+    },
+    {
+      "epoch": 0.46179583227261156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1714695938761997e-05,
+      "loss": 1.115,
+      "step": 12698
+    },
+    {
+      "epoch": 0.46183219987635016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1713535510020875e-05,
+      "loss": 1.0572,
+      "step": 12699
+    },
+    {
+      "epoch": 0.46186856748008875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.171237505750688e-05,
+      "loss": 1.0445,
+      "step": 12700
+    },
+    {
+      "epoch": 0.46190493508382735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1711214581236106e-05,
+      "loss": 1.0419,
+      "step": 12701
+    },
+    {
+      "epoch": 0.4619413026875659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1710054081224662e-05,
+      "loss": 1.1566,
+      "step": 12702
+    },
+    {
+      "epoch": 0.4619776702913045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1708893557488642e-05,
+      "loss": 1.0405,
+      "step": 12703
+    },
+    {
+      "epoch": 0.4620140378950431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1707733010044149e-05,
+      "loss": 1.2007,
+      "step": 12704
+    },
+    {
+      "epoch": 0.4620504054987817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1706572438907285e-05,
+      "loss": 1.0253,
+      "step": 12705
+    },
+    {
+      "epoch": 0.4620867731025203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.170541184409415e-05,
+      "loss": 1.1,
+      "step": 12706
+    },
+    {
+      "epoch": 0.4621231407062589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1704251225620842e-05,
+      "loss": 0.984,
+      "step": 12707
+    },
+    {
+      "epoch": 0.46215950830999747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1703090583503466e-05,
+      "loss": 0.9934,
+      "step": 12708
+    },
+    {
+      "epoch": 0.46219587591373607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1701929917758128e-05,
+      "loss": 0.9555,
+      "step": 12709
+    },
+    {
+      "epoch": 0.46223224351747466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1700769228400925e-05,
+      "loss": 1.0006,
+      "step": 12710
+    },
+    {
+      "epoch": 0.4622686111212132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1699608515447962e-05,
+      "loss": 0.9332,
+      "step": 12711
+    },
+    {
+      "epoch": 0.4623049787249518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1698447778915344e-05,
+      "loss": 1.0448,
+      "step": 12712
+    },
+    {
+      "epoch": 0.4623413463286904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1697287018819173e-05,
+      "loss": 1.0214,
+      "step": 12713
+    },
+    {
+      "epoch": 0.462377713932429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1696126235175551e-05,
+      "loss": 1.0356,
+      "step": 12714
+    },
+    {
+      "epoch": 0.4624140815361676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1694965428000585e-05,
+      "loss": 1.0478,
+      "step": 12715
+    },
+    {
+      "epoch": 0.4624504491399062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.169380459731038e-05,
+      "loss": 1.0005,
+      "step": 12716
+    },
+    {
+      "epoch": 0.4624868167436448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1692643743121036e-05,
+      "loss": 1.114,
+      "step": 12717
+    },
+    {
+      "epoch": 0.4625231843473834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1691482865448662e-05,
+      "loss": 0.9731,
+      "step": 12718
+    },
+    {
+      "epoch": 0.4625595519511219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1690321964309366e-05,
+      "loss": 1.0448,
+      "step": 12719
+    },
+    {
+      "epoch": 0.4625959195548605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1689161039719248e-05,
+      "loss": 1.0571,
+      "step": 12720
+    },
+    {
+      "epoch": 0.4626322871585991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1688000091694417e-05,
+      "loss": 0.9937,
+      "step": 12721
+    },
+    {
+      "epoch": 0.4626686547623377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1686839120250984e-05,
+      "loss": 1.0916,
+      "step": 12722
+    },
+    {
+      "epoch": 0.4627050223660763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1685678125405046e-05,
+      "loss": 1.0237,
+      "step": 12723
+    },
+    {
+      "epoch": 0.4627413899698149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1684517107172718e-05,
+      "loss": 1.1046,
+      "step": 12724
+    },
+    {
+      "epoch": 0.4627777575735535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1683356065570107e-05,
+      "loss": 1.0228,
+      "step": 12725
+    },
+    {
+      "epoch": 0.4628141251772921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1682195000613314e-05,
+      "loss": 1.0565,
+      "step": 12726
+    },
+    {
+      "epoch": 0.46285049278103063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1681033912318458e-05,
+      "loss": 0.9354,
+      "step": 12727
+    },
+    {
+      "epoch": 0.4628868603847692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1679872800701637e-05,
+      "loss": 0.9465,
+      "step": 12728
+    },
+    {
+      "epoch": 0.4629232279885078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1678711665778968e-05,
+      "loss": 1.0571,
+      "step": 12729
+    },
+    {
+      "epoch": 0.4629595955922464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1677550507566555e-05,
+      "loss": 1.0591,
+      "step": 12730
+    },
+    {
+      "epoch": 0.462995963195985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1676389326080506e-05,
+      "loss": 0.8579,
+      "step": 12731
+    },
+    {
+      "epoch": 0.4630323307997236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1675228121336937e-05,
+      "loss": 0.8993,
+      "step": 12732
+    },
+    {
+      "epoch": 0.4630686984034622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1674066893351954e-05,
+      "loss": 1.0364,
+      "step": 12733
+    },
+    {
+      "epoch": 0.4631050660072008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.167290564214167e-05,
+      "loss": 1.0357,
+      "step": 12734
+    },
+    {
+      "epoch": 0.4631414336109394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.167174436772219e-05,
+      "loss": 0.8433,
+      "step": 12735
+    },
+    {
+      "epoch": 0.46317780121467794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1670583070109632e-05,
+      "loss": 1.0144,
+      "step": 12736
+    },
+    {
+      "epoch": 0.46321416881841654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1669421749320102e-05,
+      "loss": 1.1401,
+      "step": 12737
+    },
+    {
+      "epoch": 0.46325053642215513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1668260405369716e-05,
+      "loss": 1.0995,
+      "step": 12738
+    },
+    {
+      "epoch": 0.46328690402589373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1667099038274579e-05,
+      "loss": 0.9426,
+      "step": 12739
+    },
+    {
+      "epoch": 0.4633232716296323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1665937648050814e-05,
+      "loss": 1.1389,
+      "step": 12740
+    },
+    {
+      "epoch": 0.4633596392333709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1664776234714525e-05,
+      "loss": 1.0971,
+      "step": 12741
+    },
+    {
+      "epoch": 0.4633960068371095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1663614798281827e-05,
+      "loss": 0.9943,
+      "step": 12742
+    },
+    {
+      "epoch": 0.4634323744408481,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1662453338768835e-05,
+      "loss": 1.1393,
+      "step": 12743
+    },
+    {
+      "epoch": 0.46346874204458666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1661291856191659e-05,
+      "loss": 1.0573,
+      "step": 12744
+    },
+    {
+      "epoch": 0.46350510964832525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1660130350566419e-05,
+      "loss": 1.0629,
+      "step": 12745
+    },
+    {
+      "epoch": 0.46354147725206385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1658968821909227e-05,
+      "loss": 1.122,
+      "step": 12746
+    },
+    {
+      "epoch": 0.46357784485580245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1657807270236192e-05,
+      "loss": 1.0307,
+      "step": 12747
+    },
+    {
+      "epoch": 0.46361421245954104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1656645695563435e-05,
+      "loss": 1.0334,
+      "step": 12748
+    },
+    {
+      "epoch": 0.46365058006327964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1655484097907073e-05,
+      "loss": 0.9105,
+      "step": 12749
+    },
+    {
+      "epoch": 0.46368694766701823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1654322477283213e-05,
+      "loss": 1.0039,
+      "step": 12750
+    },
+    {
+      "epoch": 0.46372331527075683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1653160833707977e-05,
+      "loss": 1.1164,
+      "step": 12751
+    },
+    {
+      "epoch": 0.4637596828744954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1651999167197484e-05,
+      "loss": 0.9431,
+      "step": 12752
+    },
+    {
+      "epoch": 0.46379605047823397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1650837477767843e-05,
+      "loss": 1.1729,
+      "step": 12753
+    },
+    {
+      "epoch": 0.46383241808197256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1649675765435173e-05,
+      "loss": 0.9716,
+      "step": 12754
+    },
+    {
+      "epoch": 0.46386878568571116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1648514030215593e-05,
+      "loss": 0.9346,
+      "step": 12755
+    },
+    {
+      "epoch": 0.46390515328944976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1647352272125221e-05,
+      "loss": 1.0392,
+      "step": 12756
+    },
+    {
+      "epoch": 0.46394152089318835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1646190491180172e-05,
+      "loss": 0.9767,
+      "step": 12757
+    },
+    {
+      "epoch": 0.46397788849692695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1645028687396567e-05,
+      "loss": 1.0054,
+      "step": 12758
+    },
+    {
+      "epoch": 0.46401425610066555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1643866860790523e-05,
+      "loss": 1.0515,
+      "step": 12759
+    },
+    {
+      "epoch": 0.46405062370440414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1642705011378156e-05,
+      "loss": 0.9329,
+      "step": 12760
+    },
+    {
+      "epoch": 0.4640869913081427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.164154313917559e-05,
+      "loss": 1.0644,
+      "step": 12761
+    },
+    {
+      "epoch": 0.4641233589118813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1640381244198944e-05,
+      "loss": 0.9176,
+      "step": 12762
+    },
+    {
+      "epoch": 0.4641597265156199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1639219326464332e-05,
+      "loss": 0.8533,
+      "step": 12763
+    },
+    {
+      "epoch": 0.46419609411935847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.163805738598788e-05,
+      "loss": 1.0713,
+      "step": 12764
+    },
+    {
+      "epoch": 0.46423246172309707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1636895422785706e-05,
+      "loss": 1.0761,
+      "step": 12765
+    },
+    {
+      "epoch": 0.46426882932683566,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1635733436873929e-05,
+      "loss": 0.8371,
+      "step": 12766
+    },
+    {
+      "epoch": 0.46430519693057426,
+      "grad_norm": 0.0,
+      "learning_rate": 1.163457142826867e-05,
+      "loss": 1.2035,
+      "step": 12767
+    },
+    {
+      "epoch": 0.46434156453431286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1633409396986052e-05,
+      "loss": 1.128,
+      "step": 12768
+    },
+    {
+      "epoch": 0.4643779321380514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1632247343042198e-05,
+      "loss": 1.1077,
+      "step": 12769
+    },
+    {
+      "epoch": 0.46441429974179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1631085266453228e-05,
+      "loss": 0.8854,
+      "step": 12770
+    },
+    {
+      "epoch": 0.4644506673455286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1629923167235263e-05,
+      "loss": 0.9845,
+      "step": 12771
+    },
+    {
+      "epoch": 0.4644870349492672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1628761045404432e-05,
+      "loss": 1.04,
+      "step": 12772
+    },
+    {
+      "epoch": 0.4645234025530058,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1627598900976843e-05,
+      "loss": 0.8905,
+      "step": 12773
+    },
+    {
+      "epoch": 0.4645597701567444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1626436733968636e-05,
+      "loss": 0.9591,
+      "step": 12774
+    },
+    {
+      "epoch": 0.464596137760483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1625274544395923e-05,
+      "loss": 0.9138,
+      "step": 12775
+    },
+    {
+      "epoch": 0.46463250536422157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1624112332274832e-05,
+      "loss": 1.0109,
+      "step": 12776
+    },
+    {
+      "epoch": 0.46466887296796017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1622950097621488e-05,
+      "loss": 1.1124,
+      "step": 12777
+    },
+    {
+      "epoch": 0.4647052405716987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1621787840452015e-05,
+      "loss": 1.1394,
+      "step": 12778
+    },
+    {
+      "epoch": 0.4647416081754373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1620625560782532e-05,
+      "loss": 1.065,
+      "step": 12779
+    },
+    {
+      "epoch": 0.4647779757791759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1619463258629174e-05,
+      "loss": 0.8761,
+      "step": 12780
+    },
+    {
+      "epoch": 0.4648143433829145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1618300934008059e-05,
+      "loss": 1.0253,
+      "step": 12781
+    },
+    {
+      "epoch": 0.4648507109866531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1617138586935313e-05,
+      "loss": 0.9537,
+      "step": 12782
+    },
+    {
+      "epoch": 0.4648870785903917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1615976217427065e-05,
+      "loss": 1.0338,
+      "step": 12783
+    },
+    {
+      "epoch": 0.4649234461941303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1614813825499438e-05,
+      "loss": 1.0929,
+      "step": 12784
+    },
+    {
+      "epoch": 0.4649598137978689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1613651411168562e-05,
+      "loss": 1.0272,
+      "step": 12785
+    },
+    {
+      "epoch": 0.4649961814016074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1612488974450565e-05,
+      "loss": 1.0817,
+      "step": 12786
+    },
+    {
+      "epoch": 0.465032549005346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1611326515361567e-05,
+      "loss": 1.0743,
+      "step": 12787
+    },
+    {
+      "epoch": 0.4650689166090846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.16101640339177e-05,
+      "loss": 0.8887,
+      "step": 12788
+    },
+    {
+      "epoch": 0.4651052842128232,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1609001530135095e-05,
+      "loss": 1.0101,
+      "step": 12789
+    },
+    {
+      "epoch": 0.4651416518165618,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1607839004029872e-05,
+      "loss": 0.9486,
+      "step": 12790
+    },
+    {
+      "epoch": 0.4651780194203004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1606676455618171e-05,
+      "loss": 1.013,
+      "step": 12791
+    },
+    {
+      "epoch": 0.465214387024039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1605513884916106e-05,
+      "loss": 1.0812,
+      "step": 12792
+    },
+    {
+      "epoch": 0.4652507546277776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.160435129193982e-05,
+      "loss": 1.0188,
+      "step": 12793
+    },
+    {
+      "epoch": 0.4652871222315162,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1603188676705433e-05,
+      "loss": 0.9751,
+      "step": 12794
+    },
+    {
+      "epoch": 0.46532348983525473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1602026039229077e-05,
+      "loss": 1.023,
+      "step": 12795
+    },
+    {
+      "epoch": 0.46535985743899333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1600863379526883e-05,
+      "loss": 1.0767,
+      "step": 12796
+    },
+    {
+      "epoch": 0.4653962250427319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1599700697614982e-05,
+      "loss": 0.9172,
+      "step": 12797
+    },
+    {
+      "epoch": 0.4654325926464705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1598537993509501e-05,
+      "loss": 1.0139,
+      "step": 12798
+    },
+    {
+      "epoch": 0.4654689602502091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1597375267226579e-05,
+      "loss": 0.8483,
+      "step": 12799
+    },
+    {
+      "epoch": 0.4655053278539477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1596212518782336e-05,
+      "loss": 0.9958,
+      "step": 12800
+    },
+    {
+      "epoch": 0.4655416954576863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.159504974819291e-05,
+      "loss": 1.192,
+      "step": 12801
+    },
+    {
+      "epoch": 0.4655780630614249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1593886955474434e-05,
+      "loss": 0.9854,
+      "step": 12802
+    },
+    {
+      "epoch": 0.46561443066516345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1592724140643035e-05,
+      "loss": 0.895,
+      "step": 12803
+    },
+    {
+      "epoch": 0.46565079826890204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1591561303714852e-05,
+      "loss": 1.1154,
+      "step": 12804
+    },
+    {
+      "epoch": 0.46568716587264064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.159039844470601e-05,
+      "loss": 1.0984,
+      "step": 12805
+    },
+    {
+      "epoch": 0.46572353347637924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1589235563632648e-05,
+      "loss": 1.1698,
+      "step": 12806
+    },
+    {
+      "epoch": 0.46575990108011783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1588072660510899e-05,
+      "loss": 1.0361,
+      "step": 12807
+    },
+    {
+      "epoch": 0.46579626868385643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1586909735356893e-05,
+      "loss": 1.1149,
+      "step": 12808
+    },
+    {
+      "epoch": 0.465832636287595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1585746788186765e-05,
+      "loss": 1.0197,
+      "step": 12809
+    },
+    {
+      "epoch": 0.4658690038913336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1584583819016653e-05,
+      "loss": 1.0236,
+      "step": 12810
+    },
+    {
+      "epoch": 0.46590537149507216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1583420827862689e-05,
+      "loss": 1.0177,
+      "step": 12811
+    },
+    {
+      "epoch": 0.46594173909881076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1582257814741005e-05,
+      "loss": 1.0439,
+      "step": 12812
+    },
+    {
+      "epoch": 0.46597810670254936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1581094779667741e-05,
+      "loss": 1.0602,
+      "step": 12813
+    },
+    {
+      "epoch": 0.46601447430628795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1579931722659028e-05,
+      "loss": 0.8849,
+      "step": 12814
+    },
+    {
+      "epoch": 0.46605084191002655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1578768643731007e-05,
+      "loss": 1.119,
+      "step": 12815
+    },
+    {
+      "epoch": 0.46608720951376514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1577605542899806e-05,
+      "loss": 0.9828,
+      "step": 12816
+    },
+    {
+      "epoch": 0.46612357711750374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.157644242018157e-05,
+      "loss": 0.8536,
+      "step": 12817
+    },
+    {
+      "epoch": 0.46615994472124234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1575279275592432e-05,
+      "loss": 1.002,
+      "step": 12818
+    },
+    {
+      "epoch": 0.46619631232498093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1574116109148527e-05,
+      "loss": 1.0311,
+      "step": 12819
+    },
+    {
+      "epoch": 0.4662326799287195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1572952920865997e-05,
+      "loss": 1.0398,
+      "step": 12820
+    },
+    {
+      "epoch": 0.46626904753245807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1571789710760977e-05,
+      "loss": 1.0979,
+      "step": 12821
+    },
+    {
+      "epoch": 0.46630541513619667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1570626478849602e-05,
+      "loss": 1.1859,
+      "step": 12822
+    },
+    {
+      "epoch": 0.46634178273993526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1569463225148014e-05,
+      "loss": 0.9981,
+      "step": 12823
+    },
+    {
+      "epoch": 0.46637815034367386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1568299949672352e-05,
+      "loss": 0.9361,
+      "step": 12824
+    },
+    {
+      "epoch": 0.46641451794741245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1567136652438752e-05,
+      "loss": 1.0161,
+      "step": 12825
+    },
+    {
+      "epoch": 0.46645088555115105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1565973333463356e-05,
+      "loss": 1.0014,
+      "step": 12826
+    },
+    {
+      "epoch": 0.46648725315488965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.15648099927623e-05,
+      "loss": 1.0273,
+      "step": 12827
+    },
+    {
+      "epoch": 0.4665236207586282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1563646630351726e-05,
+      "loss": 0.9785,
+      "step": 12828
+    },
+    {
+      "epoch": 0.4665599883623668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1562483246247775e-05,
+      "loss": 0.996,
+      "step": 12829
+    },
+    {
+      "epoch": 0.4665963559661054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1561319840466584e-05,
+      "loss": 1.0227,
+      "step": 12830
+    },
+    {
+      "epoch": 0.466632723569844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.15601564130243e-05,
+      "loss": 0.9884,
+      "step": 12831
+    },
+    {
+      "epoch": 0.4666690911735826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1558992963937052e-05,
+      "loss": 0.8337,
+      "step": 12832
+    },
+    {
+      "epoch": 0.46670545877732117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1557829493220994e-05,
+      "loss": 1.0892,
+      "step": 12833
+    },
+    {
+      "epoch": 0.46674182638105977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.155666600089226e-05,
+      "loss": 0.9834,
+      "step": 12834
+    },
+    {
+      "epoch": 0.46677819398479836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1555502486966992e-05,
+      "loss": 1.0891,
+      "step": 12835
+    },
+    {
+      "epoch": 0.46681456158853696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1554338951461337e-05,
+      "loss": 1.0461,
+      "step": 12836
+    },
+    {
+      "epoch": 0.4668509291922755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1553175394391433e-05,
+      "loss": 1.0125,
+      "step": 12837
+    },
+    {
+      "epoch": 0.4668872967960141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1552011815773426e-05,
+      "loss": 1.0755,
+      "step": 12838
+    },
+    {
+      "epoch": 0.4669236643997527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1550848215623454e-05,
+      "loss": 1.023,
+      "step": 12839
+    },
+    {
+      "epoch": 0.4669600320034913,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1549684593957665e-05,
+      "loss": 0.9372,
+      "step": 12840
+    },
+    {
+      "epoch": 0.4669963996072299,
+      "grad_norm": 0.0,
+      "learning_rate": 1.15485209507922e-05,
+      "loss": 0.9701,
+      "step": 12841
+    },
+    {
+      "epoch": 0.4670327672109685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1547357286143206e-05,
+      "loss": 1.0638,
+      "step": 12842
+    },
+    {
+      "epoch": 0.4670691348147071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1546193600026822e-05,
+      "loss": 1.0455,
+      "step": 12843
+    },
+    {
+      "epoch": 0.4671055024184457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1545029892459199e-05,
+      "loss": 1.0252,
+      "step": 12844
+    },
+    {
+      "epoch": 0.4671418700221842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1543866163456476e-05,
+      "loss": 1.0573,
+      "step": 12845
+    },
+    {
+      "epoch": 0.4671782376259228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1542702413034801e-05,
+      "loss": 1.0145,
+      "step": 12846
+    },
+    {
+      "epoch": 0.4672146052296614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.154153864121032e-05,
+      "loss": 1.0586,
+      "step": 12847
+    },
+    {
+      "epoch": 0.4672509728334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1540374847999176e-05,
+      "loss": 1.1591,
+      "step": 12848
+    },
+    {
+      "epoch": 0.4672873404371386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1539211033417516e-05,
+      "loss": 1.044,
+      "step": 12849
+    },
+    {
+      "epoch": 0.4673237080408772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.153804719748149e-05,
+      "loss": 0.8981,
+      "step": 12850
+    },
+    {
+      "epoch": 0.4673600756446158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1536883340207239e-05,
+      "loss": 0.9718,
+      "step": 12851
+    },
+    {
+      "epoch": 0.4673964432483544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1535719461610912e-05,
+      "loss": 1.1298,
+      "step": 12852
+    },
+    {
+      "epoch": 0.46743281085209293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1534555561708659e-05,
+      "loss": 0.9318,
+      "step": 12853
+    },
+    {
+      "epoch": 0.4674691784558315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.153339164051662e-05,
+      "loss": 1.0272,
+      "step": 12854
+    },
+    {
+      "epoch": 0.4675055460595701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1532227698050951e-05,
+      "loss": 1.0511,
+      "step": 12855
+    },
+    {
+      "epoch": 0.4675419136633087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1531063734327793e-05,
+      "loss": 0.965,
+      "step": 12856
+    },
+    {
+      "epoch": 0.4675782812670473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1529899749363303e-05,
+      "loss": 1.0574,
+      "step": 12857
+    },
+    {
+      "epoch": 0.4676146488707859,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1528735743173622e-05,
+      "loss": 1.0795,
+      "step": 12858
+    },
+    {
+      "epoch": 0.4676510164745245,
+      "grad_norm": 0.0,
+      "learning_rate": 1.15275717157749e-05,
+      "loss": 0.9544,
+      "step": 12859
+    },
+    {
+      "epoch": 0.4676873840782631,
+      "grad_norm": 0.0,
+      "learning_rate": 1.152640766718329e-05,
+      "loss": 0.9279,
+      "step": 12860
+    },
+    {
+      "epoch": 0.4677237516820017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1525243597414938e-05,
+      "loss": 0.9582,
+      "step": 12861
+    },
+    {
+      "epoch": 0.46776011928574024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1524079506485996e-05,
+      "loss": 1.0484,
+      "step": 12862
+    },
+    {
+      "epoch": 0.46779648688947884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1522915394412615e-05,
+      "loss": 1.0789,
+      "step": 12863
+    },
+    {
+      "epoch": 0.46783285449321743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1521751261210942e-05,
+      "loss": 1.0105,
+      "step": 12864
+    },
+    {
+      "epoch": 0.46786922209695603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1520587106897127e-05,
+      "loss": 1.0219,
+      "step": 12865
+    },
+    {
+      "epoch": 0.4679055897006946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1519422931487328e-05,
+      "loss": 1.0506,
+      "step": 12866
+    },
+    {
+      "epoch": 0.4679419573044332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1518258734997688e-05,
+      "loss": 0.9704,
+      "step": 12867
+    },
+    {
+      "epoch": 0.4679783249081718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1517094517444364e-05,
+      "loss": 1.1093,
+      "step": 12868
+    },
+    {
+      "epoch": 0.4680146925119104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1515930278843509e-05,
+      "loss": 0.9797,
+      "step": 12869
+    },
+    {
+      "epoch": 0.46805106011564895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1514766019211265e-05,
+      "loss": 0.9427,
+      "step": 12870
+    },
+    {
+      "epoch": 0.46808742771938755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1513601738563796e-05,
+      "loss": 0.988,
+      "step": 12871
+    },
+    {
+      "epoch": 0.46812379532312615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1512437436917248e-05,
+      "loss": 1.0013,
+      "step": 12872
+    },
+    {
+      "epoch": 0.46816016292686474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.151127311428778e-05,
+      "loss": 0.9564,
+      "step": 12873
+    },
+    {
+      "epoch": 0.46819653053060334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.151010877069154e-05,
+      "loss": 0.8924,
+      "step": 12874
+    },
+    {
+      "epoch": 0.46823289813434193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1508944406144681e-05,
+      "loss": 0.9656,
+      "step": 12875
+    },
+    {
+      "epoch": 0.46826926573808053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1507780020663364e-05,
+      "loss": 0.906,
+      "step": 12876
+    },
+    {
+      "epoch": 0.4683056333418191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1506615614263737e-05,
+      "loss": 0.9932,
+      "step": 12877
+    },
+    {
+      "epoch": 0.4683420009455577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1505451186961952e-05,
+      "loss": 0.9945,
+      "step": 12878
+    },
+    {
+      "epoch": 0.46837836854929626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.150428673877417e-05,
+      "loss": 1.1083,
+      "step": 12879
+    },
+    {
+      "epoch": 0.46841473615303486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1503122269716546e-05,
+      "loss": 0.9227,
+      "step": 12880
+    },
+    {
+      "epoch": 0.46845110375677346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1501957779805231e-05,
+      "loss": 1.1381,
+      "step": 12881
+    },
+    {
+      "epoch": 0.46848747136051205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1500793269056385e-05,
+      "loss": 1.1519,
+      "step": 12882
+    },
+    {
+      "epoch": 0.46852383896425065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1499628737486157e-05,
+      "loss": 1.0787,
+      "step": 12883
+    },
+    {
+      "epoch": 0.46856020656798925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.149846418511071e-05,
+      "loss": 0.9407,
+      "step": 12884
+    },
+    {
+      "epoch": 0.46859657417172784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1497299611946201e-05,
+      "loss": 1.0019,
+      "step": 12885
+    },
+    {
+      "epoch": 0.46863294177546644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1496135018008783e-05,
+      "loss": 1.1695,
+      "step": 12886
+    },
+    {
+      "epoch": 0.468669309379205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1494970403314618e-05,
+      "loss": 1.0761,
+      "step": 12887
+    },
+    {
+      "epoch": 0.4687056769829436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1493805767879853e-05,
+      "loss": 1.0068,
+      "step": 12888
+    },
+    {
+      "epoch": 0.46874204458668217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1492641111720657e-05,
+      "loss": 0.9409,
+      "step": 12889
+    },
+    {
+      "epoch": 0.46877841219042077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1491476434853182e-05,
+      "loss": 0.9692,
+      "step": 12890
+    },
+    {
+      "epoch": 0.46881477979415936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1490311737293587e-05,
+      "loss": 1.0777,
+      "step": 12891
+    },
+    {
+      "epoch": 0.46885114739789796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1489147019058031e-05,
+      "loss": 1.0487,
+      "step": 12892
+    },
+    {
+      "epoch": 0.46888751500163656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1487982280162674e-05,
+      "loss": 0.9748,
+      "step": 12893
+    },
+    {
+      "epoch": 0.46892388260537515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1486817520623673e-05,
+      "loss": 1.0348,
+      "step": 12894
+    },
+    {
+      "epoch": 0.4689602502091137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.148565274045719e-05,
+      "loss": 1.0934,
+      "step": 12895
+    },
+    {
+      "epoch": 0.4689966178128523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1484487939679383e-05,
+      "loss": 0.9353,
+      "step": 12896
+    },
+    {
+      "epoch": 0.4690329854165909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1483323118306408e-05,
+      "loss": 1.1232,
+      "step": 12897
+    },
+    {
+      "epoch": 0.4690693530203295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1482158276354433e-05,
+      "loss": 1.0298,
+      "step": 12898
+    },
+    {
+      "epoch": 0.4691057206240681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1480993413839613e-05,
+      "loss": 1.147,
+      "step": 12899
+    },
+    {
+      "epoch": 0.4691420882278067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1479828530778112e-05,
+      "loss": 1.0946,
+      "step": 12900
+    },
+    {
+      "epoch": 0.46917845583154527,
+      "grad_norm": 0.0,
+      "learning_rate": 1.147866362718609e-05,
+      "loss": 1.0414,
+      "step": 12901
+    },
+    {
+      "epoch": 0.46921482343528387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1477498703079706e-05,
+      "loss": 0.9624,
+      "step": 12902
+    },
+    {
+      "epoch": 0.46925119103902246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1476333758475127e-05,
+      "loss": 0.9569,
+      "step": 12903
+    },
+    {
+      "epoch": 0.469287558642761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1475168793388512e-05,
+      "loss": 0.9638,
+      "step": 12904
+    },
+    {
+      "epoch": 0.4693239262464996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1474003807836019e-05,
+      "loss": 0.9893,
+      "step": 12905
+    },
+    {
+      "epoch": 0.4693602938502382,
+      "grad_norm": 0.0,
+      "learning_rate": 1.147283880183382e-05,
+      "loss": 0.9683,
+      "step": 12906
+    },
+    {
+      "epoch": 0.4693966614539768,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1471673775398067e-05,
+      "loss": 0.9873,
+      "step": 12907
+    },
+    {
+      "epoch": 0.4694330290577154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1470508728544932e-05,
+      "loss": 0.8986,
+      "step": 12908
+    },
+    {
+      "epoch": 0.469469396661454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1469343661290573e-05,
+      "loss": 0.9693,
+      "step": 12909
+    },
+    {
+      "epoch": 0.4695057642651926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1468178573651154e-05,
+      "loss": 0.9685,
+      "step": 12910
+    },
+    {
+      "epoch": 0.4695421318689312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1467013465642842e-05,
+      "loss": 1.0161,
+      "step": 12911
+    },
+    {
+      "epoch": 0.4695784994726697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1465848337281797e-05,
+      "loss": 0.9905,
+      "step": 12912
+    },
+    {
+      "epoch": 0.4696148670764083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1464683188584189e-05,
+      "loss": 1.0877,
+      "step": 12913
+    },
+    {
+      "epoch": 0.4696512346801469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1463518019566182e-05,
+      "loss": 0.9294,
+      "step": 12914
+    },
+    {
+      "epoch": 0.4696876022838855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1462352830243935e-05,
+      "loss": 0.9958,
+      "step": 12915
+    },
+    {
+      "epoch": 0.4697239698876241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1461187620633618e-05,
+      "loss": 1.0605,
+      "step": 12916
+    },
+    {
+      "epoch": 0.4697603374913627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1460022390751397e-05,
+      "loss": 1.0386,
+      "step": 12917
+    },
+    {
+      "epoch": 0.4697967050951013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1458857140613434e-05,
+      "loss": 0.9761,
+      "step": 12918
+    },
+    {
+      "epoch": 0.4698330726988399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1457691870235903e-05,
+      "loss": 1.0974,
+      "step": 12919
+    },
+    {
+      "epoch": 0.4698694403025785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1456526579634963e-05,
+      "loss": 0.9219,
+      "step": 12920
+    },
+    {
+      "epoch": 0.46990580790631703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.145536126882678e-05,
+      "loss": 1.0531,
+      "step": 12921
+    },
+    {
+      "epoch": 0.4699421755100556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.145419593782753e-05,
+      "loss": 1.0684,
+      "step": 12922
+    },
+    {
+      "epoch": 0.4699785431137942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1453030586653372e-05,
+      "loss": 1.0903,
+      "step": 12923
+    },
+    {
+      "epoch": 0.4700149107175328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1451865215320477e-05,
+      "loss": 1.0455,
+      "step": 12924
+    },
+    {
+      "epoch": 0.4700512783212714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1450699823845011e-05,
+      "loss": 1.0534,
+      "step": 12925
+    },
+    {
+      "epoch": 0.47008764592501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1449534412243143e-05,
+      "loss": 0.9769,
+      "step": 12926
+    },
+    {
+      "epoch": 0.4701240135287486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1448368980531043e-05,
+      "loss": 0.9536,
+      "step": 12927
+    },
+    {
+      "epoch": 0.4701603811324872,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1447203528724877e-05,
+      "loss": 1.143,
+      "step": 12928
+    },
+    {
+      "epoch": 0.47019674873622574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1446038056840818e-05,
+      "loss": 1.1635,
+      "step": 12929
+    },
+    {
+      "epoch": 0.47023311633996434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.144487256489503e-05,
+      "loss": 1.0867,
+      "step": 12930
+    },
+    {
+      "epoch": 0.47026948394370294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1443707052903683e-05,
+      "loss": 0.9301,
+      "step": 12931
+    },
+    {
+      "epoch": 0.47030585154744153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1442541520882955e-05,
+      "loss": 1.1841,
+      "step": 12932
+    },
+    {
+      "epoch": 0.47034221915118013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1441375968849006e-05,
+      "loss": 1.0413,
+      "step": 12933
+    },
+    {
+      "epoch": 0.4703785867549187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.144021039681801e-05,
+      "loss": 1.0041,
+      "step": 12934
+    },
+    {
+      "epoch": 0.4704149543586573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1439044804806141e-05,
+      "loss": 0.9409,
+      "step": 12935
+    },
+    {
+      "epoch": 0.4704513219623959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1437879192829566e-05,
+      "loss": 1.1138,
+      "step": 12936
+    },
+    {
+      "epoch": 0.4704876895661345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1436713560904456e-05,
+      "loss": 0.9827,
+      "step": 12937
+    },
+    {
+      "epoch": 0.47052405716987306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1435547909046985e-05,
+      "loss": 1.0435,
+      "step": 12938
+    },
+    {
+      "epoch": 0.47056042477361165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.143438223727332e-05,
+      "loss": 0.9095,
+      "step": 12939
+    },
+    {
+      "epoch": 0.47059679237735025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1433216545599642e-05,
+      "loss": 1.0326,
+      "step": 12940
+    },
+    {
+      "epoch": 0.47063315998108884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1432050834042115e-05,
+      "loss": 1.0576,
+      "step": 12941
+    },
+    {
+      "epoch": 0.47066952758482744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1430885102616913e-05,
+      "loss": 1.0652,
+      "step": 12942
+    },
+    {
+      "epoch": 0.47070589518856604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1429719351340214e-05,
+      "loss": 0.9985,
+      "step": 12943
+    },
+    {
+      "epoch": 0.47074226279230463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1428553580228186e-05,
+      "loss": 1.0693,
+      "step": 12944
+    },
+    {
+      "epoch": 0.47077863039604323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1427387789297004e-05,
+      "loss": 1.0777,
+      "step": 12945
+    },
+    {
+      "epoch": 0.47081499799978177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1426221978562844e-05,
+      "loss": 0.9421,
+      "step": 12946
+    },
+    {
+      "epoch": 0.47085136560352037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1425056148041874e-05,
+      "loss": 1.1474,
+      "step": 12947
+    },
+    {
+      "epoch": 0.47088773320725896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.142389029775027e-05,
+      "loss": 0.9687,
+      "step": 12948
+    },
+    {
+      "epoch": 0.47092410081099756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1422724427704215e-05,
+      "loss": 1.0988,
+      "step": 12949
+    },
+    {
+      "epoch": 0.47096046841473616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.142155853791987e-05,
+      "loss": 1.0049,
+      "step": 12950
+    },
+    {
+      "epoch": 0.47099683601847475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1420392628413421e-05,
+      "loss": 1.0692,
+      "step": 12951
+    },
+    {
+      "epoch": 0.47103320362221335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1419226699201041e-05,
+      "loss": 1.0314,
+      "step": 12952
+    },
+    {
+      "epoch": 0.47106957122595194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.14180607502989e-05,
+      "loss": 0.991,
+      "step": 12953
+    },
+    {
+      "epoch": 0.4711059388296905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1416894781723183e-05,
+      "loss": 0.9898,
+      "step": 12954
+    },
+    {
+      "epoch": 0.4711423064334291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1415728793490055e-05,
+      "loss": 1.0535,
+      "step": 12955
+    },
+    {
+      "epoch": 0.4711786740371677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1414562785615703e-05,
+      "loss": 0.9414,
+      "step": 12956
+    },
+    {
+      "epoch": 0.4712150416409063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.14133967581163e-05,
+      "loss": 0.9578,
+      "step": 12957
+    },
+    {
+      "epoch": 0.47125140924464487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1412230711008019e-05,
+      "loss": 0.9021,
+      "step": 12958
+    },
+    {
+      "epoch": 0.47128777684838347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1411064644307044e-05,
+      "loss": 0.9605,
+      "step": 12959
+    },
+    {
+      "epoch": 0.47132414445212206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1409898558029548e-05,
+      "loss": 1.0209,
+      "step": 12960
+    },
+    {
+      "epoch": 0.47136051205586066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1408732452191709e-05,
+      "loss": 1.0247,
+      "step": 12961
+    },
+    {
+      "epoch": 0.47139687965959925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1407566326809707e-05,
+      "loss": 0.9986,
+      "step": 12962
+    },
+    {
+      "epoch": 0.4714332472633378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1406400181899719e-05,
+      "loss": 0.9324,
+      "step": 12963
+    },
+    {
+      "epoch": 0.4714696148670764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1405234017477924e-05,
+      "loss": 1.0356,
+      "step": 12964
+    },
+    {
+      "epoch": 0.471505982470815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1404067833560501e-05,
+      "loss": 0.9193,
+      "step": 12965
+    },
+    {
+      "epoch": 0.4715423500745536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1402901630163627e-05,
+      "loss": 0.9245,
+      "step": 12966
+    },
+    {
+      "epoch": 0.4715787176782922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1401735407303484e-05,
+      "loss": 1.1047,
+      "step": 12967
+    },
+    {
+      "epoch": 0.4716150852820308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1400569164996254e-05,
+      "loss": 1.086,
+      "step": 12968
+    },
+    {
+      "epoch": 0.4716514528857694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.139940290325811e-05,
+      "loss": 1.107,
+      "step": 12969
+    },
+    {
+      "epoch": 0.47168782048950797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.139823662210524e-05,
+      "loss": 1.1377,
+      "step": 12970
+    },
+    {
+      "epoch": 0.4717241880932465,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1397070321553816e-05,
+      "loss": 1.0269,
+      "step": 12971
+    },
+    {
+      "epoch": 0.4717605556969851,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1395904001620026e-05,
+      "loss": 1.1207,
+      "step": 12972
+    },
+    {
+      "epoch": 0.4717969233007237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.139473766232005e-05,
+      "loss": 0.9991,
+      "step": 12973
+    },
+    {
+      "epoch": 0.4718332909044623,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1393571303670064e-05,
+      "loss": 1.0433,
+      "step": 12974
+    },
+    {
+      "epoch": 0.4718696585082009,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1392404925686255e-05,
+      "loss": 1.0354,
+      "step": 12975
+    },
+    {
+      "epoch": 0.4719060261119395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1391238528384807e-05,
+      "loss": 0.8915,
+      "step": 12976
+    },
+    {
+      "epoch": 0.4719423937156781,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1390072111781894e-05,
+      "loss": 0.8919,
+      "step": 12977
+    },
+    {
+      "epoch": 0.4719787613194167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1388905675893705e-05,
+      "loss": 0.9467,
+      "step": 12978
+    },
+    {
+      "epoch": 0.4720151289231553,
+      "grad_norm": 0.0,
+      "learning_rate": 1.138773922073642e-05,
+      "loss": 1.0126,
+      "step": 12979
+    },
+    {
+      "epoch": 0.4720514965268938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1386572746326219e-05,
+      "loss": 1.0419,
+      "step": 12980
+    },
+    {
+      "epoch": 0.4720878641306324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1385406252679292e-05,
+      "loss": 1.0417,
+      "step": 12981
+    },
+    {
+      "epoch": 0.472124231734371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1384239739811815e-05,
+      "loss": 0.9824,
+      "step": 12982
+    },
+    {
+      "epoch": 0.4721605993381096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1383073207739983e-05,
+      "loss": 0.9228,
+      "step": 12983
+    },
+    {
+      "epoch": 0.4721969669418482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1381906656479967e-05,
+      "loss": 0.9197,
+      "step": 12984
+    },
+    {
+      "epoch": 0.4722333345455868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1380740086047958e-05,
+      "loss": 0.9528,
+      "step": 12985
+    },
+    {
+      "epoch": 0.4722697021493254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1379573496460139e-05,
+      "loss": 1.1165,
+      "step": 12986
+    },
+    {
+      "epoch": 0.472306069753064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1378406887732693e-05,
+      "loss": 1.142,
+      "step": 12987
+    },
+    {
+      "epoch": 0.47234243735680254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1377240259881812e-05,
+      "loss": 0.9489,
+      "step": 12988
+    },
+    {
+      "epoch": 0.47237880496054113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1376073612923675e-05,
+      "loss": 1.0018,
+      "step": 12989
+    },
+    {
+      "epoch": 0.47241517256427973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1374906946874467e-05,
+      "loss": 1.1307,
+      "step": 12990
+    },
+    {
+      "epoch": 0.4724515401680183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1373740261750377e-05,
+      "loss": 1.0932,
+      "step": 12991
+    },
+    {
+      "epoch": 0.4724879077717569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.137257355756759e-05,
+      "loss": 0.9796,
+      "step": 12992
+    },
+    {
+      "epoch": 0.4725242753754955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1371406834342289e-05,
+      "loss": 0.9571,
+      "step": 12993
+    },
+    {
+      "epoch": 0.4725606429792341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.137024009209067e-05,
+      "loss": 0.9919,
+      "step": 12994
+    },
+    {
+      "epoch": 0.4725970105829727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1369073330828908e-05,
+      "loss": 1.0195,
+      "step": 12995
+    },
+    {
+      "epoch": 0.47263337818671125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.13679065505732e-05,
+      "loss": 0.943,
+      "step": 12996
+    },
+    {
+      "epoch": 0.47266974579044985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1366739751339728e-05,
+      "loss": 0.9805,
+      "step": 12997
+    },
+    {
+      "epoch": 0.47270611339418844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.136557293314468e-05,
+      "loss": 1.0237,
+      "step": 12998
+    },
+    {
+      "epoch": 0.47274248099792704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1364406096004248e-05,
+      "loss": 1.0989,
+      "step": 12999
+    },
+    {
+      "epoch": 0.47277884860166564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1363239239934617e-05,
+      "loss": 1.0754,
+      "step": 13000
+    },
+    {
+      "epoch": 0.47281521620540423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1362072364951971e-05,
+      "loss": 0.9023,
+      "step": 13001
+    },
+    {
+      "epoch": 0.47285158380914283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.136090547107251e-05,
+      "loss": 0.8785,
+      "step": 13002
+    },
+    {
+      "epoch": 0.4728879514128814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1359738558312412e-05,
+      "loss": 1.0345,
+      "step": 13003
+    },
+    {
+      "epoch": 0.47292431901662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1358571626687874e-05,
+      "loss": 1.0289,
+      "step": 13004
+    },
+    {
+      "epoch": 0.47296068662035856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.135740467621508e-05,
+      "loss": 1.0452,
+      "step": 13005
+    },
+    {
+      "epoch": 0.47299705422409716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.135623770691022e-05,
+      "loss": 0.9002,
+      "step": 13006
+    },
+    {
+      "epoch": 0.47303342182783575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.135507071878949e-05,
+      "loss": 1.0262,
+      "step": 13007
+    },
+    {
+      "epoch": 0.47306978943157435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1353903711869074e-05,
+      "loss": 1.0526,
+      "step": 13008
+    },
+    {
+      "epoch": 0.47310615703531295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1352736686165164e-05,
+      "loss": 1.0876,
+      "step": 13009
+    },
+    {
+      "epoch": 0.47314252463905154,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1351569641693956e-05,
+      "loss": 1.0035,
+      "step": 13010
+    },
+    {
+      "epoch": 0.47317889224279014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1350402578471632e-05,
+      "loss": 0.9324,
+      "step": 13011
+    },
+    {
+      "epoch": 0.47321525984652874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.134923549651439e-05,
+      "loss": 0.9788,
+      "step": 13012
+    },
+    {
+      "epoch": 0.4732516274502673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1348068395838423e-05,
+      "loss": 0.881,
+      "step": 13013
+    },
+    {
+      "epoch": 0.47328799505400587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1346901276459913e-05,
+      "loss": 1.0309,
+      "step": 13014
+    },
+    {
+      "epoch": 0.47332436265774447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1345734138395063e-05,
+      "loss": 0.912,
+      "step": 13015
+    },
+    {
+      "epoch": 0.47336073026148306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1344566981660064e-05,
+      "loss": 0.9479,
+      "step": 13016
+    },
+    {
+      "epoch": 0.47339709786522166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.13433998062711e-05,
+      "loss": 0.8692,
+      "step": 13017
+    },
+    {
+      "epoch": 0.47343346546896026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1342232612244373e-05,
+      "loss": 1.0347,
+      "step": 13018
+    },
+    {
+      "epoch": 0.47346983307269885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1341065399596073e-05,
+      "loss": 1.0408,
+      "step": 13019
+    },
+    {
+      "epoch": 0.47350620067643745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1339898168342392e-05,
+      "loss": 0.9378,
+      "step": 13020
+    },
+    {
+      "epoch": 0.47354256828017605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1338730918499527e-05,
+      "loss": 1.0001,
+      "step": 13021
+    },
+    {
+      "epoch": 0.4735789358839146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1337563650083667e-05,
+      "loss": 1.0277,
+      "step": 13022
+    },
+    {
+      "epoch": 0.4736153034876532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1336396363111013e-05,
+      "loss": 0.9325,
+      "step": 13023
+    },
+    {
+      "epoch": 0.4736516710913918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1335229057597755e-05,
+      "loss": 1.0995,
+      "step": 13024
+    },
+    {
+      "epoch": 0.4736880386951304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1334061733560082e-05,
+      "loss": 0.9427,
+      "step": 13025
+    },
+    {
+      "epoch": 0.47372440629886897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1332894391014201e-05,
+      "loss": 0.9888,
+      "step": 13026
+    },
+    {
+      "epoch": 0.47376077390260757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.13317270299763e-05,
+      "loss": 1.0291,
+      "step": 13027
+    },
+    {
+      "epoch": 0.47379714150634616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1330559650462576e-05,
+      "loss": 0.9141,
+      "step": 13028
+    },
+    {
+      "epoch": 0.47383350911008476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1329392252489226e-05,
+      "loss": 0.9591,
+      "step": 13029
+    },
+    {
+      "epoch": 0.4738698767138233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1328224836072441e-05,
+      "loss": 1.0342,
+      "step": 13030
+    },
+    {
+      "epoch": 0.4739062443175619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1327057401228426e-05,
+      "loss": 0.9297,
+      "step": 13031
+    },
+    {
+      "epoch": 0.4739426119213005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1325889947973368e-05,
+      "loss": 1.1167,
+      "step": 13032
+    },
+    {
+      "epoch": 0.4739789795250391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1324722476323469e-05,
+      "loss": 1.2275,
+      "step": 13033
+    },
+    {
+      "epoch": 0.4740153471287777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1323554986294927e-05,
+      "loss": 0.9996,
+      "step": 13034
+    },
+    {
+      "epoch": 0.4740517147325163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1322387477903937e-05,
+      "loss": 1.0203,
+      "step": 13035
+    },
+    {
+      "epoch": 0.4740880823362549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1321219951166695e-05,
+      "loss": 1.0449,
+      "step": 13036
+    },
+    {
+      "epoch": 0.4741244499399935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1320052406099402e-05,
+      "loss": 0.9871,
+      "step": 13037
+    },
+    {
+      "epoch": 0.474160817543732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1318884842718253e-05,
+      "loss": 0.8687,
+      "step": 13038
+    },
+    {
+      "epoch": 0.4741971851474706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1317717261039452e-05,
+      "loss": 0.9774,
+      "step": 13039
+    },
+    {
+      "epoch": 0.4742335527512092,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1316549661079193e-05,
+      "loss": 1.0902,
+      "step": 13040
+    },
+    {
+      "epoch": 0.4742699203549478,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1315382042853673e-05,
+      "loss": 1.0888,
+      "step": 13041
+    },
+    {
+      "epoch": 0.4743062879586864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1314214406379096e-05,
+      "loss": 0.9677,
+      "step": 13042
+    },
+    {
+      "epoch": 0.474342655562425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1313046751671655e-05,
+      "loss": 0.9746,
+      "step": 13043
+    },
+    {
+      "epoch": 0.4743790231661636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1311879078747555e-05,
+      "loss": 1.1337,
+      "step": 13044
+    },
+    {
+      "epoch": 0.4744153907699022,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1310711387622997e-05,
+      "loss": 0.8512,
+      "step": 13045
+    },
+    {
+      "epoch": 0.4744517583736408,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1309543678314174e-05,
+      "loss": 1.0471,
+      "step": 13046
+    },
+    {
+      "epoch": 0.4744881259773793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1308375950837294e-05,
+      "loss": 1.0532,
+      "step": 13047
+    },
+    {
+      "epoch": 0.4745244935811179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1307208205208554e-05,
+      "loss": 0.9893,
+      "step": 13048
+    },
+    {
+      "epoch": 0.4745608611848565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.130604044144415e-05,
+      "loss": 1.0432,
+      "step": 13049
+    },
+    {
+      "epoch": 0.4745972287885951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1304872659560291e-05,
+      "loss": 1.1198,
+      "step": 13050
+    },
+    {
+      "epoch": 0.4746335963923337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1303704859573178e-05,
+      "loss": 1.0303,
+      "step": 13051
+    },
+    {
+      "epoch": 0.4746699639960723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1302537041499004e-05,
+      "loss": 1.0258,
+      "step": 13052
+    },
+    {
+      "epoch": 0.4747063315998109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.130136920535398e-05,
+      "loss": 0.9178,
+      "step": 13053
+    },
+    {
+      "epoch": 0.4747426992035495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1300201351154303e-05,
+      "loss": 1.0165,
+      "step": 13054
+    },
+    {
+      "epoch": 0.47477906680728804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1299033478916179e-05,
+      "loss": 1.0557,
+      "step": 13055
+    },
+    {
+      "epoch": 0.47481543441102664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1297865588655808e-05,
+      "loss": 1.1221,
+      "step": 13056
+    },
+    {
+      "epoch": 0.47485180201476523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1296697680389393e-05,
+      "loss": 0.9517,
+      "step": 13057
+    },
+    {
+      "epoch": 0.47488816961850383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1295529754133136e-05,
+      "loss": 0.9483,
+      "step": 13058
+    },
+    {
+      "epoch": 0.4749245372222424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1294361809903244e-05,
+      "loss": 1.0112,
+      "step": 13059
+    },
+    {
+      "epoch": 0.474960904825981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.129319384771592e-05,
+      "loss": 1.0225,
+      "step": 13060
+    },
+    {
+      "epoch": 0.4749972724297196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1292025867587364e-05,
+      "loss": 1.0195,
+      "step": 13061
+    },
+    {
+      "epoch": 0.4750336400334582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1290857869533783e-05,
+      "loss": 0.9924,
+      "step": 13062
+    },
+    {
+      "epoch": 0.4750700076371968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1289689853571379e-05,
+      "loss": 0.9462,
+      "step": 13063
+    },
+    {
+      "epoch": 0.47510637524093535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1288521819716358e-05,
+      "loss": 0.9347,
+      "step": 13064
+    },
+    {
+      "epoch": 0.47514274284467395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1287353767984925e-05,
+      "loss": 1.0055,
+      "step": 13065
+    },
+    {
+      "epoch": 0.47517911044841254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1286185698393286e-05,
+      "loss": 1.0661,
+      "step": 13066
+    },
+    {
+      "epoch": 0.47521547805215114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1285017610957644e-05,
+      "loss": 0.8934,
+      "step": 13067
+    },
+    {
+      "epoch": 0.47525184565588974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1283849505694208e-05,
+      "loss": 0.9437,
+      "step": 13068
+    },
+    {
+      "epoch": 0.47528821325962833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1282681382619183e-05,
+      "loss": 1.1105,
+      "step": 13069
+    },
+    {
+      "epoch": 0.47532458086336693,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1281513241748767e-05,
+      "loss": 1.0533,
+      "step": 13070
+    },
+    {
+      "epoch": 0.4753609484671055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.128034508309918e-05,
+      "loss": 1.0226,
+      "step": 13071
+    },
+    {
+      "epoch": 0.47539731607084407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.127917690668662e-05,
+      "loss": 1.1776,
+      "step": 13072
+    },
+    {
+      "epoch": 0.47543368367458266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1278008712527291e-05,
+      "loss": 1.0933,
+      "step": 13073
+    },
+    {
+      "epoch": 0.47547005127832126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1276840500637409e-05,
+      "loss": 0.8764,
+      "step": 13074
+    },
+    {
+      "epoch": 0.47550641888205986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1275672271033175e-05,
+      "loss": 1.0746,
+      "step": 13075
+    },
+    {
+      "epoch": 0.47554278648579845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1274504023730796e-05,
+      "loss": 1.1238,
+      "step": 13076
+    },
+    {
+      "epoch": 0.47557915408953705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1273335758746486e-05,
+      "loss": 1.0004,
+      "step": 13077
+    },
+    {
+      "epoch": 0.47561552169327564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1272167476096447e-05,
+      "loss": 1.0525,
+      "step": 13078
+    },
+    {
+      "epoch": 0.47565188929701424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.127099917579689e-05,
+      "loss": 0.8969,
+      "step": 13079
+    },
+    {
+      "epoch": 0.4756882569007528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1269830857864024e-05,
+      "loss": 1.0693,
+      "step": 13080
+    },
+    {
+      "epoch": 0.4757246245044914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1268662522314053e-05,
+      "loss": 1.0543,
+      "step": 13081
+    },
+    {
+      "epoch": 0.47576099210823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1267494169163193e-05,
+      "loss": 1.0068,
+      "step": 13082
+    },
+    {
+      "epoch": 0.47579735971196857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1266325798427646e-05,
+      "loss": 0.9652,
+      "step": 13083
+    },
+    {
+      "epoch": 0.47583372731570717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1265157410123628e-05,
+      "loss": 1.0432,
+      "step": 13084
+    },
+    {
+      "epoch": 0.47587009491944576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1263989004267346e-05,
+      "loss": 1.1075,
+      "step": 13085
+    },
+    {
+      "epoch": 0.47590646252318436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1262820580875007e-05,
+      "loss": 0.9774,
+      "step": 13086
+    },
+    {
+      "epoch": 0.47594283012692296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.126165213996283e-05,
+      "loss": 0.9134,
+      "step": 13087
+    },
+    {
+      "epoch": 0.47597919773066155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1260483681547014e-05,
+      "loss": 1.142,
+      "step": 13088
+    },
+    {
+      "epoch": 0.4760155653344001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1259315205643775e-05,
+      "loss": 0.9806,
+      "step": 13089
+    },
+    {
+      "epoch": 0.4760519329381387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1258146712269328e-05,
+      "loss": 0.9865,
+      "step": 13090
+    },
+    {
+      "epoch": 0.4760883005418773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1256978201439879e-05,
+      "loss": 0.8948,
+      "step": 13091
+    },
+    {
+      "epoch": 0.4761246681456159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.125580967317164e-05,
+      "loss": 1.0098,
+      "step": 13092
+    },
+    {
+      "epoch": 0.4761610357493545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1254641127480825e-05,
+      "loss": 0.9612,
+      "step": 13093
+    },
+    {
+      "epoch": 0.4761974033530931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1253472564383643e-05,
+      "loss": 1.1327,
+      "step": 13094
+    },
+    {
+      "epoch": 0.47623377095683167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.125230398389631e-05,
+      "loss": 0.9824,
+      "step": 13095
+    },
+    {
+      "epoch": 0.47627013856057027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1251135386035032e-05,
+      "loss": 1.0512,
+      "step": 13096
+    },
+    {
+      "epoch": 0.4763065061643088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1249966770816028e-05,
+      "loss": 1.0274,
+      "step": 13097
+    },
+    {
+      "epoch": 0.4763428737680474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.124879813825551e-05,
+      "loss": 1.0543,
+      "step": 13098
+    },
+    {
+      "epoch": 0.476379241371786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1247629488369687e-05,
+      "loss": 1.0793,
+      "step": 13099
+    },
+    {
+      "epoch": 0.4764156089755246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1246460821174778e-05,
+      "loss": 0.9795,
+      "step": 13100
+    },
+    {
+      "epoch": 0.4764519765792632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1245292136686991e-05,
+      "loss": 0.8685,
+      "step": 13101
+    },
+    {
+      "epoch": 0.4764883441830018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1244123434922542e-05,
+      "loss": 1.093,
+      "step": 13102
+    },
+    {
+      "epoch": 0.4765247117867404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1242954715897649e-05,
+      "loss": 1.0182,
+      "step": 13103
+    },
+    {
+      "epoch": 0.476561079390479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1241785979628522e-05,
+      "loss": 0.9963,
+      "step": 13104
+    },
+    {
+      "epoch": 0.4765974469942176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1240617226131371e-05,
+      "loss": 1.0004,
+      "step": 13105
+    },
+    {
+      "epoch": 0.4766338145979561,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1239448455422422e-05,
+      "loss": 1.0079,
+      "step": 13106
+    },
+    {
+      "epoch": 0.4766701822016947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1238279667517883e-05,
+      "loss": 0.9668,
+      "step": 13107
+    },
+    {
+      "epoch": 0.4767065498054333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1237110862433968e-05,
+      "loss": 0.9301,
+      "step": 13108
+    },
+    {
+      "epoch": 0.4767429174091719,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1235942040186897e-05,
+      "loss": 1.1064,
+      "step": 13109
+    },
+    {
+      "epoch": 0.4767792850129105,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1234773200792883e-05,
+      "loss": 0.9307,
+      "step": 13110
+    },
+    {
+      "epoch": 0.4768156526166491,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1233604344268145e-05,
+      "loss": 1.0572,
+      "step": 13111
+    },
+    {
+      "epoch": 0.4768520202203877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1232435470628895e-05,
+      "loss": 1.0027,
+      "step": 13112
+    },
+    {
+      "epoch": 0.4768883878241263,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1231266579891352e-05,
+      "loss": 1.031,
+      "step": 13113
+    },
+    {
+      "epoch": 0.47692475542786483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1230097672071732e-05,
+      "loss": 1.0124,
+      "step": 13114
+    },
+    {
+      "epoch": 0.47696112303160343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1228928747186254e-05,
+      "loss": 0.9911,
+      "step": 13115
+    },
+    {
+      "epoch": 0.476997490635342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1227759805251128e-05,
+      "loss": 1.0739,
+      "step": 13116
+    },
+    {
+      "epoch": 0.4770338582390806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1226590846282584e-05,
+      "loss": 1.0416,
+      "step": 13117
+    },
+    {
+      "epoch": 0.4770702258428192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1225421870296826e-05,
+      "loss": 0.9344,
+      "step": 13118
+    },
+    {
+      "epoch": 0.4771065934465578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1224252877310083e-05,
+      "loss": 1.0391,
+      "step": 13119
+    },
+    {
+      "epoch": 0.4771429610502964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1223083867338564e-05,
+      "loss": 1.0472,
+      "step": 13120
+    },
+    {
+      "epoch": 0.477179328654035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1221914840398494e-05,
+      "loss": 0.9143,
+      "step": 13121
+    },
+    {
+      "epoch": 0.4772156962577736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1220745796506089e-05,
+      "loss": 0.9937,
+      "step": 13122
+    },
+    {
+      "epoch": 0.47725206386151214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1219576735677569e-05,
+      "loss": 1.0569,
+      "step": 13123
+    },
+    {
+      "epoch": 0.47728843146525074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1218407657929149e-05,
+      "loss": 1.0422,
+      "step": 13124
+    },
+    {
+      "epoch": 0.47732479906898934,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1217238563277053e-05,
+      "loss": 0.9063,
+      "step": 13125
+    },
+    {
+      "epoch": 0.47736116667272793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1216069451737498e-05,
+      "loss": 0.9802,
+      "step": 13126
+    },
+    {
+      "epoch": 0.47739753427646653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1214900323326708e-05,
+      "loss": 1.0281,
+      "step": 13127
+    },
+    {
+      "epoch": 0.4774339018802051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1213731178060898e-05,
+      "loss": 1.005,
+      "step": 13128
+    },
+    {
+      "epoch": 0.4774702694839437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.121256201595629e-05,
+      "loss": 1.1096,
+      "step": 13129
+    },
+    {
+      "epoch": 0.4775066370876823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1211392837029104e-05,
+      "loss": 0.9028,
+      "step": 13130
+    },
+    {
+      "epoch": 0.47754300469142086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.121022364129556e-05,
+      "loss": 1.0472,
+      "step": 13131
+    },
+    {
+      "epoch": 0.47757937229515945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1209054428771881e-05,
+      "loss": 1.1332,
+      "step": 13132
+    },
+    {
+      "epoch": 0.47761573989889805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.120788519947429e-05,
+      "loss": 1.1061,
+      "step": 13133
+    },
+    {
+      "epoch": 0.47765210750263665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1206715953419003e-05,
+      "loss": 0.9351,
+      "step": 13134
+    },
+    {
+      "epoch": 0.47768847510637524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1205546690622244e-05,
+      "loss": 1.0588,
+      "step": 13135
+    },
+    {
+      "epoch": 0.47772484271011384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1204377411100237e-05,
+      "loss": 1.0754,
+      "step": 13136
+    },
+    {
+      "epoch": 0.47776121031385244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1203208114869199e-05,
+      "loss": 1.0599,
+      "step": 13137
+    },
+    {
+      "epoch": 0.47779757791759103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.120203880194536e-05,
+      "loss": 1.0408,
+      "step": 13138
+    },
+    {
+      "epoch": 0.4778339455213296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1200869472344933e-05,
+      "loss": 1.0621,
+      "step": 13139
+    },
+    {
+      "epoch": 0.47787031312506817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.119970012608415e-05,
+      "loss": 0.9745,
+      "step": 13140
+    },
+    {
+      "epoch": 0.47790668072880677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.119853076317923e-05,
+      "loss": 1.0721,
+      "step": 13141
+    },
+    {
+      "epoch": 0.47794304833254536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1197361383646393e-05,
+      "loss": 0.9523,
+      "step": 13142
+    },
+    {
+      "epoch": 0.47797941593628396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1196191987501866e-05,
+      "loss": 1.0674,
+      "step": 13143
+    },
+    {
+      "epoch": 0.47801578354002255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1195022574761874e-05,
+      "loss": 0.9033,
+      "step": 13144
+    },
+    {
+      "epoch": 0.47805215114376115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1193853145442638e-05,
+      "loss": 0.9667,
+      "step": 13145
+    },
+    {
+      "epoch": 0.47808851874749975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1192683699560385e-05,
+      "loss": 1.0861,
+      "step": 13146
+    },
+    {
+      "epoch": 0.47812488635123834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1191514237131339e-05,
+      "loss": 0.8284,
+      "step": 13147
+    },
+    {
+      "epoch": 0.4781612539549769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.119034475817172e-05,
+      "loss": 1.0228,
+      "step": 13148
+    },
+    {
+      "epoch": 0.4781976215587155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1189175262697758e-05,
+      "loss": 0.9604,
+      "step": 13149
+    },
+    {
+      "epoch": 0.4782339891624541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1188005750725674e-05,
+      "loss": 1.0246,
+      "step": 13150
+    },
+    {
+      "epoch": 0.4782703567661927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1186836222271699e-05,
+      "loss": 0.953,
+      "step": 13151
+    },
+    {
+      "epoch": 0.47830672436993127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1185666677352054e-05,
+      "loss": 1.1086,
+      "step": 13152
+    },
+    {
+      "epoch": 0.47834309197366986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1184497115982966e-05,
+      "loss": 0.9671,
+      "step": 13153
+    },
+    {
+      "epoch": 0.47837945957740846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1183327538180661e-05,
+      "loss": 1.0797,
+      "step": 13154
+    },
+    {
+      "epoch": 0.47841582718114706,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1182157943961367e-05,
+      "loss": 0.9973,
+      "step": 13155
+    },
+    {
+      "epoch": 0.4784521947848856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.118098833334131e-05,
+      "loss": 1.0049,
+      "step": 13156
+    },
+    {
+      "epoch": 0.4784885623886242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1179818706336714e-05,
+      "loss": 1.0058,
+      "step": 13157
+    },
+    {
+      "epoch": 0.4785249299923628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1178649062963805e-05,
+      "loss": 1.05,
+      "step": 13158
+    },
+    {
+      "epoch": 0.4785612975961014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1177479403238813e-05,
+      "loss": 0.91,
+      "step": 13159
+    },
+    {
+      "epoch": 0.47859766519984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1176309727177968e-05,
+      "loss": 0.9978,
+      "step": 13160
+    },
+    {
+      "epoch": 0.4786340328035786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1175140034797491e-05,
+      "loss": 1.1147,
+      "step": 13161
+    },
+    {
+      "epoch": 0.4786704004073172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1173970326113617e-05,
+      "loss": 0.9748,
+      "step": 13162
+    },
+    {
+      "epoch": 0.47870676801105577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1172800601142568e-05,
+      "loss": 1.0594,
+      "step": 13163
+    },
+    {
+      "epoch": 0.47874313561479437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1171630859900572e-05,
+      "loss": 1.0261,
+      "step": 13164
+    },
+    {
+      "epoch": 0.4787795032185329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1170461102403865e-05,
+      "loss": 1.034,
+      "step": 13165
+    },
+    {
+      "epoch": 0.4788158708222715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1169291328668667e-05,
+      "loss": 0.9633,
+      "step": 13166
+    },
+    {
+      "epoch": 0.4788522384260101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1168121538711214e-05,
+      "loss": 1.0527,
+      "step": 13167
+    },
+    {
+      "epoch": 0.4788886060297487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1166951732547731e-05,
+      "loss": 0.9056,
+      "step": 13168
+    },
+    {
+      "epoch": 0.4789249736334873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1165781910194446e-05,
+      "loss": 1.0274,
+      "step": 13169
+    },
+    {
+      "epoch": 0.4789613412372259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1164612071667596e-05,
+      "loss": 0.9113,
+      "step": 13170
+    },
+    {
+      "epoch": 0.4789977088409645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1163442216983402e-05,
+      "loss": 0.8844,
+      "step": 13171
+    },
+    {
+      "epoch": 0.4790340764447031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1162272346158098e-05,
+      "loss": 0.9809,
+      "step": 13172
+    },
+    {
+      "epoch": 0.4790704440484416,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1161102459207916e-05,
+      "loss": 1.0498,
+      "step": 13173
+    },
+    {
+      "epoch": 0.4791068116521802,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1159932556149084e-05,
+      "loss": 1.028,
+      "step": 13174
+    },
+    {
+      "epoch": 0.4791431792559188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1158762636997835e-05,
+      "loss": 1.0746,
+      "step": 13175
+    },
+    {
+      "epoch": 0.4791795468596574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1157592701770396e-05,
+      "loss": 1.0054,
+      "step": 13176
+    },
+    {
+      "epoch": 0.479215914463396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1156422750483003e-05,
+      "loss": 0.9102,
+      "step": 13177
+    },
+    {
+      "epoch": 0.4792522820671346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1155252783151884e-05,
+      "loss": 0.9569,
+      "step": 13178
+    },
+    {
+      "epoch": 0.4792886496708732,
+      "grad_norm": 0.0,
+      "learning_rate": 1.115408279979327e-05,
+      "loss": 1.0186,
+      "step": 13179
+    },
+    {
+      "epoch": 0.4793250172746118,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1152912800423396e-05,
+      "loss": 0.9637,
+      "step": 13180
+    },
+    {
+      "epoch": 0.47936138487835034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1151742785058493e-05,
+      "loss": 1.0984,
+      "step": 13181
+    },
+    {
+      "epoch": 0.47939775248208893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1150572753714792e-05,
+      "loss": 1.1105,
+      "step": 13182
+    },
+    {
+      "epoch": 0.47943412008582753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1149402706408528e-05,
+      "loss": 1.016,
+      "step": 13183
+    },
+    {
+      "epoch": 0.4794704876895661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1148232643155932e-05,
+      "loss": 1.0137,
+      "step": 13184
+    },
+    {
+      "epoch": 0.4795068552933047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1147062563973235e-05,
+      "loss": 0.9686,
+      "step": 13185
+    },
+    {
+      "epoch": 0.4795432228970433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1145892468876673e-05,
+      "loss": 0.9661,
+      "step": 13186
+    },
+    {
+      "epoch": 0.4795795905007819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1144722357882483e-05,
+      "loss": 0.9595,
+      "step": 13187
+    },
+    {
+      "epoch": 0.4796159581045205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.114355223100689e-05,
+      "loss": 1.0409,
+      "step": 13188
+    },
+    {
+      "epoch": 0.4796523257082591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1142382088266133e-05,
+      "loss": 0.9839,
+      "step": 13189
+    },
+    {
+      "epoch": 0.47968869331199765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1141211929676447e-05,
+      "loss": 0.9524,
+      "step": 13190
+    },
+    {
+      "epoch": 0.47972506091573625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.114004175525406e-05,
+      "loss": 0.8717,
+      "step": 13191
+    },
+    {
+      "epoch": 0.47976142851947484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1138871565015216e-05,
+      "loss": 1.0038,
+      "step": 13192
+    },
+    {
+      "epoch": 0.47979779612321344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1137701358976142e-05,
+      "loss": 1.0315,
+      "step": 13193
+    },
+    {
+      "epoch": 0.47983416372695203,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1136531137153082e-05,
+      "loss": 1.0376,
+      "step": 13194
+    },
+    {
+      "epoch": 0.47987053133069063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.113536089956226e-05,
+      "loss": 1.0458,
+      "step": 13195
+    },
+    {
+      "epoch": 0.4799068989344292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1134190646219916e-05,
+      "loss": 1.0812,
+      "step": 13196
+    },
+    {
+      "epoch": 0.4799432665381678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1133020377142287e-05,
+      "loss": 0.9479,
+      "step": 13197
+    },
+    {
+      "epoch": 0.47997963414190636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1131850092345606e-05,
+      "loss": 1.0174,
+      "step": 13198
+    },
+    {
+      "epoch": 0.48001600174564496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1130679791846112e-05,
+      "loss": 1.1261,
+      "step": 13199
+    },
+    {
+      "epoch": 0.48005236934938356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1129509475660042e-05,
+      "loss": 1.0679,
+      "step": 13200
+    },
+    {
+      "epoch": 0.48008873695312215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1128339143803628e-05,
+      "loss": 0.9742,
+      "step": 13201
+    },
+    {
+      "epoch": 0.48012510455686075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.112716879629311e-05,
+      "loss": 1.0641,
+      "step": 13202
+    },
+    {
+      "epoch": 0.48016147216059935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1125998433144727e-05,
+      "loss": 0.9961,
+      "step": 13203
+    },
+    {
+      "epoch": 0.48019783976433794,
+      "grad_norm": 0.0,
+      "learning_rate": 1.112482805437471e-05,
+      "loss": 0.9836,
+      "step": 13204
+    },
+    {
+      "epoch": 0.48023420736807654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1123657659999301e-05,
+      "loss": 0.9299,
+      "step": 13205
+    },
+    {
+      "epoch": 0.48027057497181513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1122487250034738e-05,
+      "loss": 1.0953,
+      "step": 13206
+    },
+    {
+      "epoch": 0.4803069425755537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1121316824497253e-05,
+      "loss": 0.9431,
+      "step": 13207
+    },
+    {
+      "epoch": 0.48034331017929227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.112014638340309e-05,
+      "loss": 1.0464,
+      "step": 13208
+    },
+    {
+      "epoch": 0.48037967778303087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1118975926768483e-05,
+      "loss": 1.1053,
+      "step": 13209
+    },
+    {
+      "epoch": 0.48041604538676946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1117805454609675e-05,
+      "loss": 0.8721,
+      "step": 13210
+    },
+    {
+      "epoch": 0.48045241299050806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1116634966942903e-05,
+      "loss": 0.9109,
+      "step": 13211
+    },
+    {
+      "epoch": 0.48048878059424666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1115464463784402e-05,
+      "loss": 1.1244,
+      "step": 13212
+    },
+    {
+      "epoch": 0.48052514819798525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.111429394515042e-05,
+      "loss": 1.0284,
+      "step": 13213
+    },
+    {
+      "epoch": 0.48056151580172385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1113123411057185e-05,
+      "loss": 1.0467,
+      "step": 13214
+    },
+    {
+      "epoch": 0.4805978834054624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1111952861520944e-05,
+      "loss": 1.0835,
+      "step": 13215
+    },
+    {
+      "epoch": 0.480634251009201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1110782296557935e-05,
+      "loss": 0.9493,
+      "step": 13216
+    },
+    {
+      "epoch": 0.4806706186129396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1109611716184397e-05,
+      "loss": 1.1244,
+      "step": 13217
+    },
+    {
+      "epoch": 0.4807069862166782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1108441120416571e-05,
+      "loss": 1.0705,
+      "step": 13218
+    },
+    {
+      "epoch": 0.4807433538204168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1107270509270698e-05,
+      "loss": 1.0321,
+      "step": 13219
+    },
+    {
+      "epoch": 0.48077972142415537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1106099882763015e-05,
+      "loss": 1.0572,
+      "step": 13220
+    },
+    {
+      "epoch": 0.48081608902789397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1104929240909767e-05,
+      "loss": 0.977,
+      "step": 13221
+    },
+    {
+      "epoch": 0.48085245663163256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1103758583727193e-05,
+      "loss": 1.0036,
+      "step": 13222
+    },
+    {
+      "epoch": 0.4808888242353711,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1102587911231535e-05,
+      "loss": 1.0078,
+      "step": 13223
+    },
+    {
+      "epoch": 0.4809251918391097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1101417223439035e-05,
+      "loss": 0.993,
+      "step": 13224
+    },
+    {
+      "epoch": 0.4809615594428483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1100246520365935e-05,
+      "loss": 1.0134,
+      "step": 13225
+    },
+    {
+      "epoch": 0.4809979270465869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1099075802028472e-05,
+      "loss": 1.0652,
+      "step": 13226
+    },
+    {
+      "epoch": 0.4810342946503255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1097905068442894e-05,
+      "loss": 1.0922,
+      "step": 13227
+    },
+    {
+      "epoch": 0.4810706622540641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.109673431962544e-05,
+      "loss": 0.8569,
+      "step": 13228
+    },
+    {
+      "epoch": 0.4811070298578027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1095563555592352e-05,
+      "loss": 1.0421,
+      "step": 13229
+    },
+    {
+      "epoch": 0.4811433974615413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1094392776359877e-05,
+      "loss": 1.0534,
+      "step": 13230
+    },
+    {
+      "epoch": 0.4811797650652799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1093221981944254e-05,
+      "loss": 0.8462,
+      "step": 13231
+    },
+    {
+      "epoch": 0.4812161326690184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.109205117236173e-05,
+      "loss": 0.9866,
+      "step": 13232
+    },
+    {
+      "epoch": 0.481252500272757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1090880347628538e-05,
+      "loss": 1.0298,
+      "step": 13233
+    },
+    {
+      "epoch": 0.4812888678764956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1089709507760935e-05,
+      "loss": 0.9826,
+      "step": 13234
+    },
+    {
+      "epoch": 0.4813252354802342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1088538652775158e-05,
+      "loss": 0.8943,
+      "step": 13235
+    },
+    {
+      "epoch": 0.4813616030839728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1087367782687449e-05,
+      "loss": 1.1442,
+      "step": 13236
+    },
+    {
+      "epoch": 0.4813979706877114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1086196897514057e-05,
+      "loss": 0.8805,
+      "step": 13237
+    },
+    {
+      "epoch": 0.48143433829145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1085025997271222e-05,
+      "loss": 1.1232,
+      "step": 13238
+    },
+    {
+      "epoch": 0.4814707058951886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1083855081975192e-05,
+      "loss": 1.1517,
+      "step": 13239
+    },
+    {
+      "epoch": 0.48150707349892713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.108268415164221e-05,
+      "loss": 1.0616,
+      "step": 13240
+    },
+    {
+      "epoch": 0.4815434411026657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1081513206288521e-05,
+      "loss": 1.0311,
+      "step": 13241
+    },
+    {
+      "epoch": 0.4815798087064043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1080342245930373e-05,
+      "loss": 1.0129,
+      "step": 13242
+    },
+    {
+      "epoch": 0.4816161763101429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1079171270584009e-05,
+      "loss": 0.9603,
+      "step": 13243
+    },
+    {
+      "epoch": 0.4816525439138815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1078000280265673e-05,
+      "loss": 1.0357,
+      "step": 13244
+    },
+    {
+      "epoch": 0.4816889115176201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1076829274991614e-05,
+      "loss": 1.0133,
+      "step": 13245
+    },
+    {
+      "epoch": 0.4817252791213587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1075658254778076e-05,
+      "loss": 0.9445,
+      "step": 13246
+    },
+    {
+      "epoch": 0.4817616467250973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1074487219641306e-05,
+      "loss": 1.0499,
+      "step": 13247
+    },
+    {
+      "epoch": 0.4817980143288359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.107331616959755e-05,
+      "loss": 0.9411,
+      "step": 13248
+    },
+    {
+      "epoch": 0.48183438193257444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1072145104663052e-05,
+      "loss": 0.9892,
+      "step": 13249
+    },
+    {
+      "epoch": 0.48187074953631304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1070974024854066e-05,
+      "loss": 1.0075,
+      "step": 13250
+    },
+    {
+      "epoch": 0.48190711714005163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1069802930186837e-05,
+      "loss": 1.034,
+      "step": 13251
+    },
+    {
+      "epoch": 0.48194348474379023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1068631820677606e-05,
+      "loss": 0.8313,
+      "step": 13252
+    },
+    {
+      "epoch": 0.4819798523475288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1067460696342627e-05,
+      "loss": 0.9298,
+      "step": 13253
+    },
+    {
+      "epoch": 0.4820162199512674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1066289557198142e-05,
+      "loss": 1.0432,
+      "step": 13254
+    },
+    {
+      "epoch": 0.482052587555006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1065118403260407e-05,
+      "loss": 0.9931,
+      "step": 13255
+    },
+    {
+      "epoch": 0.4820889551587446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1063947234545662e-05,
+      "loss": 1.0869,
+      "step": 13256
+    },
+    {
+      "epoch": 0.48212532276248315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1062776051070159e-05,
+      "loss": 1.022,
+      "step": 13257
+    },
+    {
+      "epoch": 0.48216169036622175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1061604852850148e-05,
+      "loss": 0.9898,
+      "step": 13258
+    },
+    {
+      "epoch": 0.48219805796996035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1060433639901874e-05,
+      "loss": 1.0981,
+      "step": 13259
+    },
+    {
+      "epoch": 0.48223442557369894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1059262412241586e-05,
+      "loss": 1.0069,
+      "step": 13260
+    },
+    {
+      "epoch": 0.48227079317743754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1058091169885538e-05,
+      "loss": 1.0042,
+      "step": 13261
+    },
+    {
+      "epoch": 0.48230716078117614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1056919912849977e-05,
+      "loss": 0.9322,
+      "step": 13262
+    },
+    {
+      "epoch": 0.48234352838491473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.105574864115115e-05,
+      "loss": 1.1671,
+      "step": 13263
+    },
+    {
+      "epoch": 0.48237989598865333,
+      "grad_norm": 0.0,
+      "learning_rate": 1.105457735480531e-05,
+      "loss": 0.9194,
+      "step": 13264
+    },
+    {
+      "epoch": 0.48241626359239187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1053406053828701e-05,
+      "loss": 0.9401,
+      "step": 13265
+    },
+    {
+      "epoch": 0.48245263119613047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1052234738237581e-05,
+      "loss": 1.04,
+      "step": 13266
+    },
+    {
+      "epoch": 0.48248899879986906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1051063408048197e-05,
+      "loss": 1.0137,
+      "step": 13267
+    },
+    {
+      "epoch": 0.48252536640360766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1049892063276799e-05,
+      "loss": 1.026,
+      "step": 13268
+    },
+    {
+      "epoch": 0.48256173400734625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1048720703939639e-05,
+      "loss": 0.996,
+      "step": 13269
+    },
+    {
+      "epoch": 0.48259810161108485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1047549330052967e-05,
+      "loss": 0.9299,
+      "step": 13270
+    },
+    {
+      "epoch": 0.48263446921482345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1046377941633034e-05,
+      "loss": 0.9955,
+      "step": 13271
+    },
+    {
+      "epoch": 0.48267083681856204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1045206538696094e-05,
+      "loss": 0.9731,
+      "step": 13272
+    },
+    {
+      "epoch": 0.48270720442230064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1044035121258391e-05,
+      "loss": 1.1122,
+      "step": 13273
+    },
+    {
+      "epoch": 0.4827435720260392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1042863689336188e-05,
+      "loss": 1.0041,
+      "step": 13274
+    },
+    {
+      "epoch": 0.4827799396297778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1041692242945727e-05,
+      "loss": 0.9677,
+      "step": 13275
+    },
+    {
+      "epoch": 0.4828163072335164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1040520782103264e-05,
+      "loss": 0.9293,
+      "step": 13276
+    },
+    {
+      "epoch": 0.48285267483725497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1039349306825053e-05,
+      "loss": 1.1025,
+      "step": 13277
+    },
+    {
+      "epoch": 0.48288904244099357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1038177817127345e-05,
+      "loss": 1.0049,
+      "step": 13278
+    },
+    {
+      "epoch": 0.48292541004473216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.103700631302639e-05,
+      "loss": 0.9404,
+      "step": 13279
+    },
+    {
+      "epoch": 0.48296177764847076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1035834794538446e-05,
+      "loss": 1.0131,
+      "step": 13280
+    },
+    {
+      "epoch": 0.48299814525220935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1034663261679763e-05,
+      "loss": 0.9661,
+      "step": 13281
+    },
+    {
+      "epoch": 0.4830345128559479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1033491714466598e-05,
+      "loss": 1.017,
+      "step": 13282
+    },
+    {
+      "epoch": 0.4830708804596865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.10323201529152e-05,
+      "loss": 0.9235,
+      "step": 13283
+    },
+    {
+      "epoch": 0.4831072480634251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1031148577041822e-05,
+      "loss": 0.9343,
+      "step": 13284
+    },
+    {
+      "epoch": 0.4831436156671637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1029976986862722e-05,
+      "loss": 0.968,
+      "step": 13285
+    },
+    {
+      "epoch": 0.4831799832709023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1028805382394156e-05,
+      "loss": 0.8635,
+      "step": 13286
+    },
+    {
+      "epoch": 0.4832163508746409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.102763376365237e-05,
+      "loss": 0.9387,
+      "step": 13287
+    },
+    {
+      "epoch": 0.4832527184783795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1026462130653627e-05,
+      "loss": 0.8588,
+      "step": 13288
+    },
+    {
+      "epoch": 0.48328908608211807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.102529048341418e-05,
+      "loss": 1.008,
+      "step": 13289
+    },
+    {
+      "epoch": 0.48332545368585667,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1024118821950278e-05,
+      "loss": 1.0341,
+      "step": 13290
+    },
+    {
+      "epoch": 0.4833618212895952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1022947146278185e-05,
+      "loss": 1.0304,
+      "step": 13291
+    },
+    {
+      "epoch": 0.4833981888933338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1021775456414146e-05,
+      "loss": 1.1338,
+      "step": 13292
+    },
+    {
+      "epoch": 0.4834345564970724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1020603752374426e-05,
+      "loss": 0.959,
+      "step": 13293
+    },
+    {
+      "epoch": 0.483470924100811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1019432034175275e-05,
+      "loss": 0.9792,
+      "step": 13294
+    },
+    {
+      "epoch": 0.4835072917045496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1018260301832952e-05,
+      "loss": 1.0042,
+      "step": 13295
+    },
+    {
+      "epoch": 0.4835436593082882,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1017088555363713e-05,
+      "loss": 0.9872,
+      "step": 13296
+    },
+    {
+      "epoch": 0.4835800269120268,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1015916794783809e-05,
+      "loss": 0.9031,
+      "step": 13297
+    },
+    {
+      "epoch": 0.4836163945157654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1014745020109505e-05,
+      "loss": 1.0286,
+      "step": 13298
+    },
+    {
+      "epoch": 0.4836527621195039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1013573231357053e-05,
+      "loss": 0.9074,
+      "step": 13299
+    },
+    {
+      "epoch": 0.4836891297232425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1012401428542709e-05,
+      "loss": 1.1096,
+      "step": 13300
+    },
+    {
+      "epoch": 0.4837254973269811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1011229611682734e-05,
+      "loss": 0.9485,
+      "step": 13301
+    },
+    {
+      "epoch": 0.4837618649307197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1010057780793382e-05,
+      "loss": 0.9499,
+      "step": 13302
+    },
+    {
+      "epoch": 0.4837982325344583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.100888593589091e-05,
+      "loss": 1.0266,
+      "step": 13303
+    },
+    {
+      "epoch": 0.4838346001381969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1007714076991578e-05,
+      "loss": 0.8916,
+      "step": 13304
+    },
+    {
+      "epoch": 0.4838709677419355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1006542204111644e-05,
+      "loss": 1.0474,
+      "step": 13305
+    },
+    {
+      "epoch": 0.4839073353456741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1005370317267364e-05,
+      "loss": 1.0393,
+      "step": 13306
+    },
+    {
+      "epoch": 0.4839437029494127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1004198416474999e-05,
+      "loss": 0.9211,
+      "step": 13307
+    },
+    {
+      "epoch": 0.48398007055315123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.10030265017508e-05,
+      "loss": 1.1109,
+      "step": 13308
+    },
+    {
+      "epoch": 0.4840164381568898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.100185457311104e-05,
+      "loss": 1.1639,
+      "step": 13309
+    },
+    {
+      "epoch": 0.4840528057606284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1000682630571962e-05,
+      "loss": 0.9377,
+      "step": 13310
+    },
+    {
+      "epoch": 0.484089173364367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0999510674149836e-05,
+      "loss": 0.9397,
+      "step": 13311
+    },
+    {
+      "epoch": 0.4841255409681056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0998338703860918e-05,
+      "loss": 1.0663,
+      "step": 13312
+    },
+    {
+      "epoch": 0.4841619085718442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0997166719721463e-05,
+      "loss": 1.0159,
+      "step": 13313
+    },
+    {
+      "epoch": 0.4841982761755828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0995994721747738e-05,
+      "loss": 0.9351,
+      "step": 13314
+    },
+    {
+      "epoch": 0.4842346437793214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0994822709956e-05,
+      "loss": 1.0082,
+      "step": 13315
+    },
+    {
+      "epoch": 0.48427101138305995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0993650684362505e-05,
+      "loss": 0.9589,
+      "step": 13316
+    },
+    {
+      "epoch": 0.48430737898679854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0992478644983517e-05,
+      "loss": 0.9409,
+      "step": 13317
+    },
+    {
+      "epoch": 0.48434374659053714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0991306591835297e-05,
+      "loss": 0.8989,
+      "step": 13318
+    },
+    {
+      "epoch": 0.48438011419427573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0990134524934107e-05,
+      "loss": 1.0268,
+      "step": 13319
+    },
+    {
+      "epoch": 0.48441648179801433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0988962444296203e-05,
+      "loss": 1.0758,
+      "step": 13320
+    },
+    {
+      "epoch": 0.4844528494017529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0987790349937846e-05,
+      "loss": 1.0876,
+      "step": 13321
+    },
+    {
+      "epoch": 0.4844892170054915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0986618241875301e-05,
+      "loss": 1.0046,
+      "step": 13322
+    },
+    {
+      "epoch": 0.4845255846092301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.098544612012483e-05,
+      "loss": 0.9917,
+      "step": 13323
+    },
+    {
+      "epoch": 0.48456195221296866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.098427398470269e-05,
+      "loss": 1.1074,
+      "step": 13324
+    },
+    {
+      "epoch": 0.48459831981670726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0983101835625146e-05,
+      "loss": 0.9712,
+      "step": 13325
+    },
+    {
+      "epoch": 0.48463468742044585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0981929672908457e-05,
+      "loss": 1.0793,
+      "step": 13326
+    },
+    {
+      "epoch": 0.48467105502418445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0980757496568888e-05,
+      "loss": 1.1654,
+      "step": 13327
+    },
+    {
+      "epoch": 0.48470742262792305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0979585306622704e-05,
+      "loss": 1.0918,
+      "step": 13328
+    },
+    {
+      "epoch": 0.48474379023166164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0978413103086158e-05,
+      "loss": 1.1425,
+      "step": 13329
+    },
+    {
+      "epoch": 0.48478015783540024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0977240885975522e-05,
+      "loss": 0.9246,
+      "step": 13330
+    },
+    {
+      "epoch": 0.48481652543913883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0976068655307056e-05,
+      "loss": 0.9995,
+      "step": 13331
+    },
+    {
+      "epoch": 0.48485289304287743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0974896411097019e-05,
+      "loss": 1.0053,
+      "step": 13332
+    },
+    {
+      "epoch": 0.48488926064661597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0973724153361678e-05,
+      "loss": 1.0403,
+      "step": 13333
+    },
+    {
+      "epoch": 0.48492562825035457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0972551882117297e-05,
+      "loss": 1.0326,
+      "step": 13334
+    },
+    {
+      "epoch": 0.48496199585409316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.097137959738014e-05,
+      "loss": 1.021,
+      "step": 13335
+    },
+    {
+      "epoch": 0.48499836345783176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0970207299166466e-05,
+      "loss": 1.0259,
+      "step": 13336
+    },
+    {
+      "epoch": 0.48503473106157036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0969034987492542e-05,
+      "loss": 1.1219,
+      "step": 13337
+    },
+    {
+      "epoch": 0.48507109866530895,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0967862662374635e-05,
+      "loss": 0.9209,
+      "step": 13338
+    },
+    {
+      "epoch": 0.48510746626904755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0966690323829007e-05,
+      "loss": 0.9671,
+      "step": 13339
+    },
+    {
+      "epoch": 0.48514383387278615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0965517971871919e-05,
+      "loss": 0.9943,
+      "step": 13340
+    },
+    {
+      "epoch": 0.4851802014765247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0964345606519642e-05,
+      "loss": 0.8968,
+      "step": 13341
+    },
+    {
+      "epoch": 0.4852165690802633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0963173227788437e-05,
+      "loss": 0.9944,
+      "step": 13342
+    },
+    {
+      "epoch": 0.4852529366840019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.096200083569457e-05,
+      "loss": 0.9514,
+      "step": 13343
+    },
+    {
+      "epoch": 0.4852893042877405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0960828430254307e-05,
+      "loss": 1.0033,
+      "step": 13344
+    },
+    {
+      "epoch": 0.48532567189147907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0959656011483913e-05,
+      "loss": 0.9162,
+      "step": 13345
+    },
+    {
+      "epoch": 0.48536203949521767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0958483579399652e-05,
+      "loss": 1.0032,
+      "step": 13346
+    },
+    {
+      "epoch": 0.48539840709895626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0957311134017795e-05,
+      "loss": 1.101,
+      "step": 13347
+    },
+    {
+      "epoch": 0.48543477470269486,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0956138675354599e-05,
+      "loss": 0.8851,
+      "step": 13348
+    },
+    {
+      "epoch": 0.48547114230643346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0954966203426338e-05,
+      "loss": 1.0822,
+      "step": 13349
+    },
+    {
+      "epoch": 0.485507509910172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0953793718249278e-05,
+      "loss": 0.8985,
+      "step": 13350
+    },
+    {
+      "epoch": 0.4855438775139106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0952621219839678e-05,
+      "loss": 0.9777,
+      "step": 13351
+    },
+    {
+      "epoch": 0.4855802451176492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0951448708213815e-05,
+      "loss": 1.0287,
+      "step": 13352
+    },
+    {
+      "epoch": 0.4856166127213878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0950276183387948e-05,
+      "loss": 1.0519,
+      "step": 13353
+    },
+    {
+      "epoch": 0.4856529803251264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0949103645378348e-05,
+      "loss": 1.1379,
+      "step": 13354
+    },
+    {
+      "epoch": 0.485689347928865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0947931094201282e-05,
+      "loss": 1.0851,
+      "step": 13355
+    },
+    {
+      "epoch": 0.4857257155326036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0946758529873015e-05,
+      "loss": 1.0489,
+      "step": 13356
+    },
+    {
+      "epoch": 0.48576208313634217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.094558595240982e-05,
+      "loss": 1.0931,
+      "step": 13357
+    },
+    {
+      "epoch": 0.4857984507400807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.094441336182796e-05,
+      "loss": 0.9288,
+      "step": 13358
+    },
+    {
+      "epoch": 0.4858348183438193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0943240758143705e-05,
+      "loss": 0.9759,
+      "step": 13359
+    },
+    {
+      "epoch": 0.4858711859475579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0942068141373323e-05,
+      "loss": 1.0675,
+      "step": 13360
+    },
+    {
+      "epoch": 0.4859075535512965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0940895511533084e-05,
+      "loss": 0.9943,
+      "step": 13361
+    },
+    {
+      "epoch": 0.4859439211550351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0939722868639248e-05,
+      "loss": 1.0367,
+      "step": 13362
+    },
+    {
+      "epoch": 0.4859802887587737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0938550212708097e-05,
+      "loss": 0.9942,
+      "step": 13363
+    },
+    {
+      "epoch": 0.4860166563625123,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0937377543755889e-05,
+      "loss": 1.1072,
+      "step": 13364
+    },
+    {
+      "epoch": 0.4860530239662509,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0936204861798903e-05,
+      "loss": 0.9548,
+      "step": 13365
+    },
+    {
+      "epoch": 0.4860893915699894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0935032166853399e-05,
+      "loss": 1.0883,
+      "step": 13366
+    },
+    {
+      "epoch": 0.486125759173728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0933859458935651e-05,
+      "loss": 0.9527,
+      "step": 13367
+    },
+    {
+      "epoch": 0.4861621267774666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0932686738061928e-05,
+      "loss": 1.0026,
+      "step": 13368
+    },
+    {
+      "epoch": 0.4861984943812052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.09315140042485e-05,
+      "loss": 1.0106,
+      "step": 13369
+    },
+    {
+      "epoch": 0.4862348619849438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0930341257511636e-05,
+      "loss": 0.9427,
+      "step": 13370
+    },
+    {
+      "epoch": 0.4862712295886824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.092916849786761e-05,
+      "loss": 1.016,
+      "step": 13371
+    },
+    {
+      "epoch": 0.486307597192421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0927995725332686e-05,
+      "loss": 0.942,
+      "step": 13372
+    },
+    {
+      "epoch": 0.4863439647961596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.092682293992314e-05,
+      "loss": 0.8641,
+      "step": 13373
+    },
+    {
+      "epoch": 0.4863803323998982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.092565014165524e-05,
+      "loss": 0.9738,
+      "step": 13374
+    },
+    {
+      "epoch": 0.48641670000363674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0924477330545255e-05,
+      "loss": 1.0781,
+      "step": 13375
+    },
+    {
+      "epoch": 0.48645306760737533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0923304506609462e-05,
+      "loss": 1.1302,
+      "step": 13376
+    },
+    {
+      "epoch": 0.48648943521111393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0922131669864127e-05,
+      "loss": 0.9429,
+      "step": 13377
+    },
+    {
+      "epoch": 0.4865258028148525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0920958820325525e-05,
+      "loss": 1.0371,
+      "step": 13378
+    },
+    {
+      "epoch": 0.4865621704185911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0919785958009926e-05,
+      "loss": 1.0955,
+      "step": 13379
+    },
+    {
+      "epoch": 0.4865985380223297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.09186130829336e-05,
+      "loss": 0.9506,
+      "step": 13380
+    },
+    {
+      "epoch": 0.4866349056260683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0917440195112822e-05,
+      "loss": 0.994,
+      "step": 13381
+    },
+    {
+      "epoch": 0.4866712732298069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0916267294563865e-05,
+      "loss": 1.0724,
+      "step": 13382
+    },
+    {
+      "epoch": 0.48670764083354545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0915094381302994e-05,
+      "loss": 0.9537,
+      "step": 13383
+    },
+    {
+      "epoch": 0.48674400843728405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0913921455346493e-05,
+      "loss": 1.0065,
+      "step": 13384
+    },
+    {
+      "epoch": 0.48678037604102264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0912748516710626e-05,
+      "loss": 1.0396,
+      "step": 13385
+    },
+    {
+      "epoch": 0.48681674364476124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0911575565411669e-05,
+      "loss": 0.9566,
+      "step": 13386
+    },
+    {
+      "epoch": 0.48685311124849984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0910402601465895e-05,
+      "loss": 1.0129,
+      "step": 13387
+    },
+    {
+      "epoch": 0.48688947885223843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0909229624889573e-05,
+      "loss": 0.9466,
+      "step": 13388
+    },
+    {
+      "epoch": 0.48692584645597703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0908056635698982e-05,
+      "loss": 0.8317,
+      "step": 13389
+    },
+    {
+      "epoch": 0.4869622140597156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0906883633910395e-05,
+      "loss": 1.0364,
+      "step": 13390
+    },
+    {
+      "epoch": 0.4869985816634542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.090571061954008e-05,
+      "loss": 1.0179,
+      "step": 13391
+    },
+    {
+      "epoch": 0.48703494926719276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0904537592604321e-05,
+      "loss": 1.0613,
+      "step": 13392
+    },
+    {
+      "epoch": 0.48707131687093136,
+      "grad_norm": 0.0,
+      "learning_rate": 1.090336455311938e-05,
+      "loss": 1.0153,
+      "step": 13393
+    },
+    {
+      "epoch": 0.48710768447466996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0902191501101541e-05,
+      "loss": 1.015,
+      "step": 13394
+    },
+    {
+      "epoch": 0.48714405207840855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0901018436567077e-05,
+      "loss": 0.9682,
+      "step": 13395
+    },
+    {
+      "epoch": 0.48718041968214715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0899845359532257e-05,
+      "loss": 0.9115,
+      "step": 13396
+    },
+    {
+      "epoch": 0.48721678728588574,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0898672270013361e-05,
+      "loss": 0.9593,
+      "step": 13397
+    },
+    {
+      "epoch": 0.48725315488962434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0897499168026662e-05,
+      "loss": 0.9813,
+      "step": 13398
+    },
+    {
+      "epoch": 0.48728952249336294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0896326053588436e-05,
+      "loss": 0.9079,
+      "step": 13399
+    },
+    {
+      "epoch": 0.4873258900971015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0895152926714955e-05,
+      "loss": 0.8667,
+      "step": 13400
+    },
+    {
+      "epoch": 0.4873622577008401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0893979787422502e-05,
+      "loss": 1.0465,
+      "step": 13401
+    },
+    {
+      "epoch": 0.48739862530457867,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0892806635727344e-05,
+      "loss": 1.0704,
+      "step": 13402
+    },
+    {
+      "epoch": 0.48743499290831727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0891633471645761e-05,
+      "loss": 0.9847,
+      "step": 13403
+    },
+    {
+      "epoch": 0.48747136051205586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.089046029519403e-05,
+      "loss": 0.9711,
+      "step": 13404
+    },
+    {
+      "epoch": 0.48750772811579446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0889287106388426e-05,
+      "loss": 0.9786,
+      "step": 13405
+    },
+    {
+      "epoch": 0.48754409571953305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0888113905245226e-05,
+      "loss": 0.9667,
+      "step": 13406
+    },
+    {
+      "epoch": 0.48758046332327165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0886940691780704e-05,
+      "loss": 1.0797,
+      "step": 13407
+    },
+    {
+      "epoch": 0.4876168309270102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0885767466011139e-05,
+      "loss": 1.0518,
+      "step": 13408
+    },
+    {
+      "epoch": 0.4876531985307488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0884594227952805e-05,
+      "loss": 1.0904,
+      "step": 13409
+    },
+    {
+      "epoch": 0.4876895661344874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0883420977621983e-05,
+      "loss": 1.0151,
+      "step": 13410
+    },
+    {
+      "epoch": 0.487725933738226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0882247715034949e-05,
+      "loss": 1.006,
+      "step": 13411
+    },
+    {
+      "epoch": 0.4877623013419646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0881074440207977e-05,
+      "loss": 0.9903,
+      "step": 13412
+    },
+    {
+      "epoch": 0.4877986689457032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0879901153157348e-05,
+      "loss": 1.0774,
+      "step": 13413
+    },
+    {
+      "epoch": 0.48783503654944177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0878727853899342e-05,
+      "loss": 0.9142,
+      "step": 13414
+    },
+    {
+      "epoch": 0.48787140415318037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.087755454245023e-05,
+      "loss": 0.9687,
+      "step": 13415
+    },
+    {
+      "epoch": 0.48790777175691896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0876381218826298e-05,
+      "loss": 1.0774,
+      "step": 13416
+    },
+    {
+      "epoch": 0.4879441393606575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0875207883043817e-05,
+      "loss": 0.8967,
+      "step": 13417
+    },
+    {
+      "epoch": 0.4879805069643961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0874034535119067e-05,
+      "loss": 0.9411,
+      "step": 13418
+    },
+    {
+      "epoch": 0.4880168745681347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.087286117506833e-05,
+      "loss": 0.9846,
+      "step": 13419
+    },
+    {
+      "epoch": 0.4880532421718733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0871687802907881e-05,
+      "loss": 1.0205,
+      "step": 13420
+    },
+    {
+      "epoch": 0.4880896097756119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0870514418654003e-05,
+      "loss": 1.0286,
+      "step": 13421
+    },
+    {
+      "epoch": 0.4881259773793505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0869341022322972e-05,
+      "loss": 0.9127,
+      "step": 13422
+    },
+    {
+      "epoch": 0.4881623449830891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0868167613931066e-05,
+      "loss": 1.058,
+      "step": 13423
+    },
+    {
+      "epoch": 0.4881987125868277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0866994193494571e-05,
+      "loss": 0.9569,
+      "step": 13424
+    },
+    {
+      "epoch": 0.4882350801905662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0865820761029755e-05,
+      "loss": 0.9462,
+      "step": 13425
+    },
+    {
+      "epoch": 0.4882714477943048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.086464731655291e-05,
+      "loss": 0.9054,
+      "step": 13426
+    },
+    {
+      "epoch": 0.4883078153980434,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0863473860080307e-05,
+      "loss": 1.0892,
+      "step": 13427
+    },
+    {
+      "epoch": 0.488344183001782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0862300391628229e-05,
+      "loss": 1.0287,
+      "step": 13428
+    },
+    {
+      "epoch": 0.4883805506055206,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0861126911212958e-05,
+      "loss": 0.969,
+      "step": 13429
+    },
+    {
+      "epoch": 0.4884169182092592,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0859953418850773e-05,
+      "loss": 0.9335,
+      "step": 13430
+    },
+    {
+      "epoch": 0.4884532858129978,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0858779914557952e-05,
+      "loss": 0.9687,
+      "step": 13431
+    },
+    {
+      "epoch": 0.4884896534167364,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0857606398350781e-05,
+      "loss": 1.0484,
+      "step": 13432
+    },
+    {
+      "epoch": 0.488526021020475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0856432870245537e-05,
+      "loss": 0.9901,
+      "step": 13433
+    },
+    {
+      "epoch": 0.48856238862421353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.08552593302585e-05,
+      "loss": 1.0603,
+      "step": 13434
+    },
+    {
+      "epoch": 0.4885987562279521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0854085778405956e-05,
+      "loss": 1.0188,
+      "step": 13435
+    },
+    {
+      "epoch": 0.4886351238316907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0852912214704184e-05,
+      "loss": 1.0732,
+      "step": 13436
+    },
+    {
+      "epoch": 0.4886714914354293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0851738639169462e-05,
+      "loss": 1.0759,
+      "step": 13437
+    },
+    {
+      "epoch": 0.4887078590391679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0850565051818079e-05,
+      "loss": 0.996,
+      "step": 13438
+    },
+    {
+      "epoch": 0.4887442266429065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.084939145266631e-05,
+      "loss": 0.9408,
+      "step": 13439
+    },
+    {
+      "epoch": 0.4887805942466451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.084821784173044e-05,
+      "loss": 0.965,
+      "step": 13440
+    },
+    {
+      "epoch": 0.4888169618503837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0847044219026752e-05,
+      "loss": 1.1035,
+      "step": 13441
+    },
+    {
+      "epoch": 0.48885332945412224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0845870584571525e-05,
+      "loss": 0.9298,
+      "step": 13442
+    },
+    {
+      "epoch": 0.48888969705786084,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0844696938381048e-05,
+      "loss": 0.9689,
+      "step": 13443
+    },
+    {
+      "epoch": 0.48892606466159944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0843523280471594e-05,
+      "loss": 1.1139,
+      "step": 13444
+    },
+    {
+      "epoch": 0.48896243226533803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0842349610859455e-05,
+      "loss": 0.8169,
+      "step": 13445
+    },
+    {
+      "epoch": 0.48899879986907663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.084117592956091e-05,
+      "loss": 0.8927,
+      "step": 13446
+    },
+    {
+      "epoch": 0.4890351674728152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.084000223659224e-05,
+      "loss": 0.9127,
+      "step": 13447
+    },
+    {
+      "epoch": 0.4890715350765538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0838828531969734e-05,
+      "loss": 0.8999,
+      "step": 13448
+    },
+    {
+      "epoch": 0.4891079026802924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0837654815709668e-05,
+      "loss": 1.2278,
+      "step": 13449
+    },
+    {
+      "epoch": 0.48914427028403096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0836481087828335e-05,
+      "loss": 0.9487,
+      "step": 13450
+    },
+    {
+      "epoch": 0.48918063788776955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.083530734834201e-05,
+      "loss": 0.9844,
+      "step": 13451
+    },
+    {
+      "epoch": 0.48921700549150815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.083413359726698e-05,
+      "loss": 0.9323,
+      "step": 13452
+    },
+    {
+      "epoch": 0.48925337309524675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0832959834619534e-05,
+      "loss": 1.0243,
+      "step": 13453
+    },
+    {
+      "epoch": 0.48928974069898534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.083178606041595e-05,
+      "loss": 0.9625,
+      "step": 13454
+    },
+    {
+      "epoch": 0.48932610830272394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0830612274672514e-05,
+      "loss": 0.8947,
+      "step": 13455
+    },
+    {
+      "epoch": 0.48936247590646254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0829438477405512e-05,
+      "loss": 0.9651,
+      "step": 13456
+    },
+    {
+      "epoch": 0.48939884351020113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0828264668631228e-05,
+      "loss": 1.0306,
+      "step": 13457
+    },
+    {
+      "epoch": 0.4894352111139397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0827090848365944e-05,
+      "loss": 1.1245,
+      "step": 13458
+    },
+    {
+      "epoch": 0.48947157871767827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0825917016625953e-05,
+      "loss": 0.968,
+      "step": 13459
+    },
+    {
+      "epoch": 0.48950794632141686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.082474317342753e-05,
+      "loss": 1.0607,
+      "step": 13460
+    },
+    {
+      "epoch": 0.48954431392515546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0823569318786971e-05,
+      "loss": 0.9858,
+      "step": 13461
+    },
+    {
+      "epoch": 0.48958068152889406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0822395452720555e-05,
+      "loss": 0.9996,
+      "step": 13462
+    },
+    {
+      "epoch": 0.48961704913263265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0821221575244567e-05,
+      "loss": 1.0922,
+      "step": 13463
+    },
+    {
+      "epoch": 0.48965341673637125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0820047686375295e-05,
+      "loss": 0.9999,
+      "step": 13464
+    },
+    {
+      "epoch": 0.48968978434010985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0818873786129022e-05,
+      "loss": 1.0283,
+      "step": 13465
+    },
+    {
+      "epoch": 0.48972615194384844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0817699874522043e-05,
+      "loss": 0.9565,
+      "step": 13466
+    },
+    {
+      "epoch": 0.489762519547587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0816525951570635e-05,
+      "loss": 0.9759,
+      "step": 13467
+    },
+    {
+      "epoch": 0.4897988871513256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0815352017291087e-05,
+      "loss": 1.0297,
+      "step": 13468
+    },
+    {
+      "epoch": 0.4898352547550642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0814178071699688e-05,
+      "loss": 1.1352,
+      "step": 13469
+    },
+    {
+      "epoch": 0.48987162235880277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0813004114812726e-05,
+      "loss": 0.8962,
+      "step": 13470
+    },
+    {
+      "epoch": 0.48990798996254137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.081183014664648e-05,
+      "loss": 0.9041,
+      "step": 13471
+    },
+    {
+      "epoch": 0.48994435756627996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0810656167217246e-05,
+      "loss": 1.018,
+      "step": 13472
+    },
+    {
+      "epoch": 0.48998072517001856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0809482176541308e-05,
+      "loss": 1.0137,
+      "step": 13473
+    },
+    {
+      "epoch": 0.49001709277375716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0808308174634951e-05,
+      "loss": 0.9821,
+      "step": 13474
+    },
+    {
+      "epoch": 0.49005346037749575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0807134161514468e-05,
+      "loss": 0.9261,
+      "step": 13475
+    },
+    {
+      "epoch": 0.4900898279812343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.080596013719614e-05,
+      "loss": 0.9236,
+      "step": 13476
+    },
+    {
+      "epoch": 0.4901261955849729,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0804786101696263e-05,
+      "loss": 0.9295,
+      "step": 13477
+    },
+    {
+      "epoch": 0.4901625631887115,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0803612055031118e-05,
+      "loss": 1.0917,
+      "step": 13478
+    },
+    {
+      "epoch": 0.4901989307924501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0802437997216994e-05,
+      "loss": 1.0083,
+      "step": 13479
+    },
+    {
+      "epoch": 0.4902352983961887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0801263928270185e-05,
+      "loss": 0.9666,
+      "step": 13480
+    },
+    {
+      "epoch": 0.4902716659999273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0800089848206976e-05,
+      "loss": 0.9483,
+      "step": 13481
+    },
+    {
+      "epoch": 0.49030803360366587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0798915757043656e-05,
+      "loss": 1.0203,
+      "step": 13482
+    },
+    {
+      "epoch": 0.49034440120740447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.079774165479651e-05,
+      "loss": 0.9577,
+      "step": 13483
+    },
+    {
+      "epoch": 0.490380768811143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0796567541481831e-05,
+      "loss": 1.0331,
+      "step": 13484
+    },
+    {
+      "epoch": 0.4904171364148816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0795393417115909e-05,
+      "loss": 1.0332,
+      "step": 13485
+    },
+    {
+      "epoch": 0.4904535040186202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0794219281715033e-05,
+      "loss": 1.0038,
+      "step": 13486
+    },
+    {
+      "epoch": 0.4904898716223588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0793045135295488e-05,
+      "loss": 0.9781,
+      "step": 13487
+    },
+    {
+      "epoch": 0.4905262392260974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0791870977873568e-05,
+      "loss": 0.9879,
+      "step": 13488
+    },
+    {
+      "epoch": 0.490562606829836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0790696809465565e-05,
+      "loss": 0.9206,
+      "step": 13489
+    },
+    {
+      "epoch": 0.4905989744335746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.078952263008776e-05,
+      "loss": 0.8979,
+      "step": 13490
+    },
+    {
+      "epoch": 0.4906353420373132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0788348439756451e-05,
+      "loss": 1.0598,
+      "step": 13491
+    },
+    {
+      "epoch": 0.4906717096410518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0787174238487925e-05,
+      "loss": 1.0341,
+      "step": 13492
+    },
+    {
+      "epoch": 0.4907080772447903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0786000026298475e-05,
+      "loss": 1.046,
+      "step": 13493
+    },
+    {
+      "epoch": 0.4907444448485289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.078482580320439e-05,
+      "loss": 0.9183,
+      "step": 13494
+    },
+    {
+      "epoch": 0.4907808124522675,
+      "grad_norm": 0.0,
+      "learning_rate": 1.078365156922196e-05,
+      "loss": 0.8964,
+      "step": 13495
+    },
+    {
+      "epoch": 0.4908171800560061,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0782477324367477e-05,
+      "loss": 1.0125,
+      "step": 13496
+    },
+    {
+      "epoch": 0.4908535476597447,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0781303068657232e-05,
+      "loss": 1.0084,
+      "step": 13497
+    },
+    {
+      "epoch": 0.4908899152634833,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0780128802107513e-05,
+      "loss": 0.9767,
+      "step": 13498
+    },
+    {
+      "epoch": 0.4909262828672219,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0778954524734616e-05,
+      "loss": 1.0617,
+      "step": 13499
+    },
+    {
+      "epoch": 0.4909626504709605,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0777780236554827e-05,
+      "loss": 0.9962,
+      "step": 13500
+    },
+    {
+      "epoch": 0.49099901807469903,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0776605937584444e-05,
+      "loss": 1.0067,
+      "step": 13501
+    },
+    {
+      "epoch": 0.49103538567843763,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0775431627839753e-05,
+      "loss": 1.1069,
+      "step": 13502
+    },
+    {
+      "epoch": 0.4910717532821762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.077425730733705e-05,
+      "loss": 0.9319,
+      "step": 13503
+    },
+    {
+      "epoch": 0.4911081208859148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0773082976092626e-05,
+      "loss": 0.9202,
+      "step": 13504
+    },
+    {
+      "epoch": 0.4911444884896534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0771908634122772e-05,
+      "loss": 1.1222,
+      "step": 13505
+    },
+    {
+      "epoch": 0.491180856093392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.077073428144378e-05,
+      "loss": 1.2379,
+      "step": 13506
+    },
+    {
+      "epoch": 0.4912172236971306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0769559918071944e-05,
+      "loss": 0.9768,
+      "step": 13507
+    },
+    {
+      "epoch": 0.4912535913008692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0768385544023554e-05,
+      "loss": 1.0631,
+      "step": 13508
+    },
+    {
+      "epoch": 0.49128995890460775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0767211159314908e-05,
+      "loss": 1.0646,
+      "step": 13509
+    },
+    {
+      "epoch": 0.49132632650834634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0766036763962293e-05,
+      "loss": 1.0058,
+      "step": 13510
+    },
+    {
+      "epoch": 0.49136269411208494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0764862357982007e-05,
+      "loss": 0.8786,
+      "step": 13511
+    },
+    {
+      "epoch": 0.49139906171582354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.076368794139034e-05,
+      "loss": 0.9881,
+      "step": 13512
+    },
+    {
+      "epoch": 0.49143542931956213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0762513514203588e-05,
+      "loss": 1.0739,
+      "step": 13513
+    },
+    {
+      "epoch": 0.49147179692330073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0761339076438039e-05,
+      "loss": 1.0073,
+      "step": 13514
+    },
+    {
+      "epoch": 0.4915081645270393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0760164628109993e-05,
+      "loss": 1.1465,
+      "step": 13515
+    },
+    {
+      "epoch": 0.4915445321307779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0758990169235742e-05,
+      "loss": 0.9831,
+      "step": 13516
+    },
+    {
+      "epoch": 0.4915808997345165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0757815699831575e-05,
+      "loss": 0.9899,
+      "step": 13517
+    },
+    {
+      "epoch": 0.49161726733825506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0756641219913795e-05,
+      "loss": 0.9564,
+      "step": 13518
+    },
+    {
+      "epoch": 0.49165363494199366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0755466729498689e-05,
+      "loss": 0.8602,
+      "step": 13519
+    },
+    {
+      "epoch": 0.49169000254573225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0754292228602559e-05,
+      "loss": 0.9959,
+      "step": 13520
+    },
+    {
+      "epoch": 0.49172637014947085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0753117717241688e-05,
+      "loss": 1.0026,
+      "step": 13521
+    },
+    {
+      "epoch": 0.49176273775320944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0751943195432382e-05,
+      "loss": 0.9971,
+      "step": 13522
+    },
+    {
+      "epoch": 0.49179910535694804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0750768663190929e-05,
+      "loss": 1.01,
+      "step": 13523
+    },
+    {
+      "epoch": 0.49183547296068664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0749594120533623e-05,
+      "loss": 1.0352,
+      "step": 13524
+    },
+    {
+      "epoch": 0.49187184056442523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0748419567476766e-05,
+      "loss": 1.0533,
+      "step": 13525
+    },
+    {
+      "epoch": 0.4919082081681638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.074724500403665e-05,
+      "loss": 0.8927,
+      "step": 13526
+    },
+    {
+      "epoch": 0.49194457577190237,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0746070430229567e-05,
+      "loss": 0.9659,
+      "step": 13527
+    },
+    {
+      "epoch": 0.49198094337564097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.074489584607182e-05,
+      "loss": 0.9931,
+      "step": 13528
+    },
+    {
+      "epoch": 0.49201731097937956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0743721251579694e-05,
+      "loss": 0.9769,
+      "step": 13529
+    },
+    {
+      "epoch": 0.49205367858311816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0742546646769493e-05,
+      "loss": 1.0941,
+      "step": 13530
+    },
+    {
+      "epoch": 0.49209004618685676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0741372031657513e-05,
+      "loss": 1.0088,
+      "step": 13531
+    },
+    {
+      "epoch": 0.49212641379059535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0740197406260045e-05,
+      "loss": 1.0803,
+      "step": 13532
+    },
+    {
+      "epoch": 0.49216278139433395,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0739022770593387e-05,
+      "loss": 0.8978,
+      "step": 13533
+    },
+    {
+      "epoch": 0.49219914899807254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0737848124673841e-05,
+      "loss": 1.1288,
+      "step": 13534
+    },
+    {
+      "epoch": 0.4922355166018111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0736673468517695e-05,
+      "loss": 0.981,
+      "step": 13535
+    },
+    {
+      "epoch": 0.4922718842055497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0735498802141253e-05,
+      "loss": 0.9321,
+      "step": 13536
+    },
+    {
+      "epoch": 0.4923082518092883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0734324125560809e-05,
+      "loss": 1.0023,
+      "step": 13537
+    },
+    {
+      "epoch": 0.4923446194130269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0733149438792657e-05,
+      "loss": 1.1133,
+      "step": 13538
+    },
+    {
+      "epoch": 0.49238098701676547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0731974741853102e-05,
+      "loss": 1.0672,
+      "step": 13539
+    },
+    {
+      "epoch": 0.49241735462050407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0730800034758428e-05,
+      "loss": 1.0981,
+      "step": 13540
+    },
+    {
+      "epoch": 0.49245372222424266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0729625317524945e-05,
+      "loss": 1.0098,
+      "step": 13541
+    },
+    {
+      "epoch": 0.49249008982798126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0728450590168948e-05,
+      "loss": 0.956,
+      "step": 13542
+    },
+    {
+      "epoch": 0.4925264574317198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.072727585270673e-05,
+      "loss": 1.0913,
+      "step": 13543
+    },
+    {
+      "epoch": 0.4925628250354584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.072610110515459e-05,
+      "loss": 0.9927,
+      "step": 13544
+    },
+    {
+      "epoch": 0.492599192639197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.072492634752883e-05,
+      "loss": 1.1237,
+      "step": 13545
+    },
+    {
+      "epoch": 0.4926355602429356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0723751579845743e-05,
+      "loss": 1.0369,
+      "step": 13546
+    },
+    {
+      "epoch": 0.4926719278466742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0722576802121631e-05,
+      "loss": 1.0076,
+      "step": 13547
+    },
+    {
+      "epoch": 0.4927082954504128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0721402014372788e-05,
+      "loss": 0.9935,
+      "step": 13548
+    },
+    {
+      "epoch": 0.4927446630541514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.072022721661552e-05,
+      "loss": 0.9701,
+      "step": 13549
+    },
+    {
+      "epoch": 0.49278103065789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.071905240886612e-05,
+      "loss": 0.8982,
+      "step": 13550
+    },
+    {
+      "epoch": 0.4928173982616285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0717877591140885e-05,
+      "loss": 0.9384,
+      "step": 13551
+    },
+    {
+      "epoch": 0.4928537658653671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0716702763456122e-05,
+      "loss": 1.0586,
+      "step": 13552
+    },
+    {
+      "epoch": 0.4928901334691057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0715527925828124e-05,
+      "loss": 1.0358,
+      "step": 13553
+    },
+    {
+      "epoch": 0.4929265010728443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0714353078273188e-05,
+      "loss": 0.9653,
+      "step": 13554
+    },
+    {
+      "epoch": 0.4929628686765829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.071317822080762e-05,
+      "loss": 1.0074,
+      "step": 13555
+    },
+    {
+      "epoch": 0.4929992362803215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0712003353447716e-05,
+      "loss": 1.1729,
+      "step": 13556
+    },
+    {
+      "epoch": 0.4930356038840601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.071082847620977e-05,
+      "loss": 1.0633,
+      "step": 13557
+    },
+    {
+      "epoch": 0.4930719714877987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0709653589110097e-05,
+      "loss": 0.9641,
+      "step": 13558
+    },
+    {
+      "epoch": 0.4931083390915373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.070847869216498e-05,
+      "loss": 1.0126,
+      "step": 13559
+    },
+    {
+      "epoch": 0.4931447066952758,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0707303785390727e-05,
+      "loss": 0.8721,
+      "step": 13560
+    },
+    {
+      "epoch": 0.4931810742990144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.070612886880364e-05,
+      "loss": 0.9089,
+      "step": 13561
+    },
+    {
+      "epoch": 0.493217441902753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0704953942420016e-05,
+      "loss": 1.0387,
+      "step": 13562
+    },
+    {
+      "epoch": 0.4932538095064916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0703779006256157e-05,
+      "loss": 0.9752,
+      "step": 13563
+    },
+    {
+      "epoch": 0.4932901771102302,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0702604060328359e-05,
+      "loss": 0.9904,
+      "step": 13564
+    },
+    {
+      "epoch": 0.4933265447139688,
+      "grad_norm": 0.0,
+      "learning_rate": 1.070142910465293e-05,
+      "loss": 1.0199,
+      "step": 13565
+    },
+    {
+      "epoch": 0.4933629123177074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0700254139246166e-05,
+      "loss": 0.9505,
+      "step": 13566
+    },
+    {
+      "epoch": 0.493399279921446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0699079164124368e-05,
+      "loss": 1.0256,
+      "step": 13567
+    },
+    {
+      "epoch": 0.49343564752518454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0697904179303841e-05,
+      "loss": 1.0405,
+      "step": 13568
+    },
+    {
+      "epoch": 0.49347201512892314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0696729184800884e-05,
+      "loss": 1.0308,
+      "step": 13569
+    },
+    {
+      "epoch": 0.49350838273266173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0695554180631798e-05,
+      "loss": 1.0183,
+      "step": 13570
+    },
+    {
+      "epoch": 0.49354475033640033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0694379166812884e-05,
+      "loss": 1.0349,
+      "step": 13571
+    },
+    {
+      "epoch": 0.4935811179401389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0693204143360445e-05,
+      "loss": 1.0173,
+      "step": 13572
+    },
+    {
+      "epoch": 0.4936174855438775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.069202911029078e-05,
+      "loss": 1.0286,
+      "step": 13573
+    },
+    {
+      "epoch": 0.4936538531476161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0690854067620194e-05,
+      "loss": 0.9694,
+      "step": 13574
+    },
+    {
+      "epoch": 0.4936902207513547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0689679015364986e-05,
+      "loss": 0.8405,
+      "step": 13575
+    },
+    {
+      "epoch": 0.4937265883550933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0688503953541465e-05,
+      "loss": 1.1028,
+      "step": 13576
+    },
+    {
+      "epoch": 0.49376295595883185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0687328882165928e-05,
+      "loss": 1.0881,
+      "step": 13577
+    },
+    {
+      "epoch": 0.49379932356257045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0686153801254673e-05,
+      "loss": 1.0872,
+      "step": 13578
+    },
+    {
+      "epoch": 0.49383569116630904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0684978710824012e-05,
+      "loss": 0.921,
+      "step": 13579
+    },
+    {
+      "epoch": 0.49387205877004764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.068380361089024e-05,
+      "loss": 1.0226,
+      "step": 13580
+    },
+    {
+      "epoch": 0.49390842637378624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0682628501469664e-05,
+      "loss": 1.0973,
+      "step": 13581
+    },
+    {
+      "epoch": 0.49394479397752483,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0681453382578588e-05,
+      "loss": 1.0448,
+      "step": 13582
+    },
+    {
+      "epoch": 0.49398116158126343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0680278254233312e-05,
+      "loss": 1.0095,
+      "step": 13583
+    },
+    {
+      "epoch": 0.494017529185002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.067910311645014e-05,
+      "loss": 1.1156,
+      "step": 13584
+    },
+    {
+      "epoch": 0.49405389678874057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0677927969245377e-05,
+      "loss": 0.9778,
+      "step": 13585
+    },
+    {
+      "epoch": 0.49409026439247916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0676752812635323e-05,
+      "loss": 0.929,
+      "step": 13586
+    },
+    {
+      "epoch": 0.49412663199621776,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0675577646636286e-05,
+      "loss": 1.0074,
+      "step": 13587
+    },
+    {
+      "epoch": 0.49416299959995635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0674402471264567e-05,
+      "loss": 1.097,
+      "step": 13588
+    },
+    {
+      "epoch": 0.49419936720369495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0673227286536469e-05,
+      "loss": 0.9774,
+      "step": 13589
+    },
+    {
+      "epoch": 0.49423573480743355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.06720520924683e-05,
+      "loss": 1.0667,
+      "step": 13590
+    },
+    {
+      "epoch": 0.49427210241117214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0670876889076358e-05,
+      "loss": 0.9085,
+      "step": 13591
+    },
+    {
+      "epoch": 0.49430847001491074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0669701676376955e-05,
+      "loss": 1.0056,
+      "step": 13592
+    },
+    {
+      "epoch": 0.4943448376186493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0668526454386392e-05,
+      "loss": 1.0377,
+      "step": 13593
+    },
+    {
+      "epoch": 0.4943812052223879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.066735122312097e-05,
+      "loss": 0.9712,
+      "step": 13594
+    },
+    {
+      "epoch": 0.4944175728261265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0666175982596997e-05,
+      "loss": 1.0321,
+      "step": 13595
+    },
+    {
+      "epoch": 0.49445394042986507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0665000732830778e-05,
+      "loss": 1.0176,
+      "step": 13596
+    },
+    {
+      "epoch": 0.49449030803360366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0663825473838621e-05,
+      "loss": 1.0575,
+      "step": 13597
+    },
+    {
+      "epoch": 0.49452667563734226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0662650205636825e-05,
+      "loss": 1.1153,
+      "step": 13598
+    },
+    {
+      "epoch": 0.49456304324108086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0661474928241692e-05,
+      "loss": 0.9229,
+      "step": 13599
+    },
+    {
+      "epoch": 0.49459941084481945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.066029964166954e-05,
+      "loss": 1.0394,
+      "step": 13600
+    },
+    {
+      "epoch": 0.49463577844855805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0659124345936665e-05,
+      "loss": 0.8385,
+      "step": 13601
+    },
+    {
+      "epoch": 0.4946721460522966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0657949041059374e-05,
+      "loss": 0.9227,
+      "step": 13602
+    },
+    {
+      "epoch": 0.4947085136560352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0656773727053975e-05,
+      "loss": 0.9481,
+      "step": 13603
+    },
+    {
+      "epoch": 0.4947448812597738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.065559840393677e-05,
+      "loss": 1.14,
+      "step": 13604
+    },
+    {
+      "epoch": 0.4947812488635124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0654423071724069e-05,
+      "loss": 1.0133,
+      "step": 13605
+    },
+    {
+      "epoch": 0.494817616467251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0653247730432177e-05,
+      "loss": 1.0655,
+      "step": 13606
+    },
+    {
+      "epoch": 0.49485398407098957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0652072380077398e-05,
+      "loss": 1.0225,
+      "step": 13607
+    },
+    {
+      "epoch": 0.49489035167472817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.065089702067604e-05,
+      "loss": 1.0573,
+      "step": 13608
+    },
+    {
+      "epoch": 0.49492671927846676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.064972165224441e-05,
+      "loss": 0.9139,
+      "step": 13609
+    },
+    {
+      "epoch": 0.4949630868822053,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0648546274798814e-05,
+      "loss": 0.9808,
+      "step": 13610
+    },
+    {
+      "epoch": 0.4949994544859439,
+      "grad_norm": 0.0,
+      "learning_rate": 1.064737088835556e-05,
+      "loss": 0.9957,
+      "step": 13611
+    },
+    {
+      "epoch": 0.4950358220896825,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0646195492930954e-05,
+      "loss": 0.9526,
+      "step": 13612
+    },
+    {
+      "epoch": 0.4950721896934211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0645020088541296e-05,
+      "loss": 0.9521,
+      "step": 13613
+    },
+    {
+      "epoch": 0.4951085572971597,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0643844675202905e-05,
+      "loss": 0.9521,
+      "step": 13614
+    },
+    {
+      "epoch": 0.4951449249008983,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0642669252932082e-05,
+      "loss": 1.0047,
+      "step": 13615
+    },
+    {
+      "epoch": 0.4951812925046369,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0641493821745135e-05,
+      "loss": 1.0953,
+      "step": 13616
+    },
+    {
+      "epoch": 0.4952176601083755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0640318381658372e-05,
+      "loss": 1.0153,
+      "step": 13617
+    },
+    {
+      "epoch": 0.4952540277121141,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0639142932688098e-05,
+      "loss": 1.0376,
+      "step": 13618
+    },
+    {
+      "epoch": 0.4952903953158526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0637967474850623e-05,
+      "loss": 1.0031,
+      "step": 13619
+    },
+    {
+      "epoch": 0.4953267629195912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0636792008162253e-05,
+      "loss": 1.0452,
+      "step": 13620
+    },
+    {
+      "epoch": 0.4953631305233298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0635616532639302e-05,
+      "loss": 0.9117,
+      "step": 13621
+    },
+    {
+      "epoch": 0.4953994981270684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.063444104829807e-05,
+      "loss": 1.0369,
+      "step": 13622
+    },
+    {
+      "epoch": 0.495435865730807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0633265555154868e-05,
+      "loss": 1.0863,
+      "step": 13623
+    },
+    {
+      "epoch": 0.4954722333345456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0632090053226007e-05,
+      "loss": 0.9644,
+      "step": 13624
+    },
+    {
+      "epoch": 0.4955086009382842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0630914542527793e-05,
+      "loss": 0.9836,
+      "step": 13625
+    },
+    {
+      "epoch": 0.4955449685420228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0629739023076533e-05,
+      "loss": 1.0145,
+      "step": 13626
+    },
+    {
+      "epoch": 0.49558133614576133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.062856349488854e-05,
+      "loss": 1.0246,
+      "step": 13627
+    },
+    {
+      "epoch": 0.4956177037494999,
+      "grad_norm": 0.0,
+      "learning_rate": 1.062738795798012e-05,
+      "loss": 1.0947,
+      "step": 13628
+    },
+    {
+      "epoch": 0.4956540713532385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0626212412367578e-05,
+      "loss": 0.9656,
+      "step": 13629
+    },
+    {
+      "epoch": 0.4956904389569771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0625036858067233e-05,
+      "loss": 1.0269,
+      "step": 13630
+    },
+    {
+      "epoch": 0.4957268065607157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0623861295095383e-05,
+      "loss": 0.9725,
+      "step": 13631
+    },
+    {
+      "epoch": 0.4957631741644543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0622685723468346e-05,
+      "loss": 0.9306,
+      "step": 13632
+    },
+    {
+      "epoch": 0.4957995417681929,
+      "grad_norm": 0.0,
+      "learning_rate": 1.062151014320243e-05,
+      "loss": 0.9019,
+      "step": 13633
+    },
+    {
+      "epoch": 0.4958359093719315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0620334554313938e-05,
+      "loss": 1.0307,
+      "step": 13634
+    },
+    {
+      "epoch": 0.49587227697567005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0619158956819191e-05,
+      "loss": 0.9342,
+      "step": 13635
+    },
+    {
+      "epoch": 0.49590864457940864,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0617983350734484e-05,
+      "loss": 1.0534,
+      "step": 13636
+    },
+    {
+      "epoch": 0.49594501218314724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.061680773607614e-05,
+      "loss": 0.9865,
+      "step": 13637
+    },
+    {
+      "epoch": 0.49598137978688583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0615632112860464e-05,
+      "loss": 1.1305,
+      "step": 13638
+    },
+    {
+      "epoch": 0.49601774739062443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0614456481103761e-05,
+      "loss": 1.0155,
+      "step": 13639
+    },
+    {
+      "epoch": 0.496054114994363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0613280840822352e-05,
+      "loss": 0.909,
+      "step": 13640
+    },
+    {
+      "epoch": 0.4960904825981016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0612105192032539e-05,
+      "loss": 1.0526,
+      "step": 13641
+    },
+    {
+      "epoch": 0.4961268502018402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.061092953475063e-05,
+      "loss": 0.8646,
+      "step": 13642
+    },
+    {
+      "epoch": 0.4961632178055788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0609753868992947e-05,
+      "loss": 0.9816,
+      "step": 13643
+    },
+    {
+      "epoch": 0.49619958540931736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0608578194775794e-05,
+      "loss": 0.8415,
+      "step": 13644
+    },
+    {
+      "epoch": 0.49623595301305595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.060740251211548e-05,
+      "loss": 1.0247,
+      "step": 13645
+    },
+    {
+      "epoch": 0.49627232061679455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0606226821028319e-05,
+      "loss": 0.9817,
+      "step": 13646
+    },
+    {
+      "epoch": 0.49630868822053315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0605051121530616e-05,
+      "loss": 0.9925,
+      "step": 13647
+    },
+    {
+      "epoch": 0.49634505582427174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0603875413638694e-05,
+      "loss": 0.9205,
+      "step": 13648
+    },
+    {
+      "epoch": 0.49638142342801034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0602699697368854e-05,
+      "loss": 1.0889,
+      "step": 13649
+    },
+    {
+      "epoch": 0.49641779103174893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.060152397273741e-05,
+      "loss": 0.9578,
+      "step": 13650
+    },
+    {
+      "epoch": 0.49645415863548753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0600348239760679e-05,
+      "loss": 0.9806,
+      "step": 13651
+    },
+    {
+      "epoch": 0.49649052623922607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0599172498454964e-05,
+      "loss": 1.0642,
+      "step": 13652
+    },
+    {
+      "epoch": 0.49652689384296467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0597996748836579e-05,
+      "loss": 0.9819,
+      "step": 13653
+    },
+    {
+      "epoch": 0.49656326144670326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0596820990921844e-05,
+      "loss": 0.9865,
+      "step": 13654
+    },
+    {
+      "epoch": 0.49659962905044186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0595645224727059e-05,
+      "loss": 0.9211,
+      "step": 13655
+    },
+    {
+      "epoch": 0.49663599665418046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0594469450268542e-05,
+      "loss": 1.0387,
+      "step": 13656
+    },
+    {
+      "epoch": 0.49667236425791905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.059329366756261e-05,
+      "loss": 0.9576,
+      "step": 13657
+    },
+    {
+      "epoch": 0.49670873186165765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0592117876625562e-05,
+      "loss": 1.0258,
+      "step": 13658
+    },
+    {
+      "epoch": 0.49674509946539624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0590942077473725e-05,
+      "loss": 0.9293,
+      "step": 13659
+    },
+    {
+      "epoch": 0.49678146706913484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0589766270123403e-05,
+      "loss": 0.9437,
+      "step": 13660
+    },
+    {
+      "epoch": 0.4968178346728734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.058859045459091e-05,
+      "loss": 1.1212,
+      "step": 13661
+    },
+    {
+      "epoch": 0.496854202276612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0587414630892561e-05,
+      "loss": 0.9887,
+      "step": 13662
+    },
+    {
+      "epoch": 0.4968905698803506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0586238799044666e-05,
+      "loss": 1.0214,
+      "step": 13663
+    },
+    {
+      "epoch": 0.49692693748408917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0585062959063538e-05,
+      "loss": 1.0162,
+      "step": 13664
+    },
+    {
+      "epoch": 0.49696330508782777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0583887110965497e-05,
+      "loss": 0.9438,
+      "step": 13665
+    },
+    {
+      "epoch": 0.49699967269156636,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0582711254766844e-05,
+      "loss": 1.0219,
+      "step": 13666
+    },
+    {
+      "epoch": 0.49703604029530496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0581535390483904e-05,
+      "loss": 0.9555,
+      "step": 13667
+    },
+    {
+      "epoch": 0.49707240789904356,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0580359518132985e-05,
+      "loss": 0.9926,
+      "step": 13668
+    },
+    {
+      "epoch": 0.4971087755027821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0579183637730398e-05,
+      "loss": 1.0717,
+      "step": 13669
+    },
+    {
+      "epoch": 0.4971451431065207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0578007749292462e-05,
+      "loss": 0.9642,
+      "step": 13670
+    },
+    {
+      "epoch": 0.4971815107102593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.057683185283549e-05,
+      "loss": 1.0819,
+      "step": 13671
+    },
+    {
+      "epoch": 0.4972178783139979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0575655948375792e-05,
+      "loss": 0.9426,
+      "step": 13672
+    },
+    {
+      "epoch": 0.4972542459177365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0574480035929688e-05,
+      "loss": 1.0801,
+      "step": 13673
+    },
+    {
+      "epoch": 0.4972906135214751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0573304115513483e-05,
+      "loss": 1.0767,
+      "step": 13674
+    },
+    {
+      "epoch": 0.4973269811252137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0572128187143502e-05,
+      "loss": 1.049,
+      "step": 13675
+    },
+    {
+      "epoch": 0.49736334872895227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.057095225083605e-05,
+      "loss": 1.0428,
+      "step": 13676
+    },
+    {
+      "epoch": 0.49739971633269087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0569776306607449e-05,
+      "loss": 1.0483,
+      "step": 13677
+    },
+    {
+      "epoch": 0.4974360839364294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0568600354474007e-05,
+      "loss": 0.9468,
+      "step": 13678
+    },
+    {
+      "epoch": 0.497472451540168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0567424394452045e-05,
+      "loss": 1.0742,
+      "step": 13679
+    },
+    {
+      "epoch": 0.4975088191439066,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0566248426557873e-05,
+      "loss": 0.9358,
+      "step": 13680
+    },
+    {
+      "epoch": 0.4975451867476452,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0565072450807807e-05,
+      "loss": 0.9854,
+      "step": 13681
+    },
+    {
+      "epoch": 0.4975815543513838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0563896467218162e-05,
+      "loss": 0.985,
+      "step": 13682
+    },
+    {
+      "epoch": 0.4976179219551224,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0562720475805255e-05,
+      "loss": 1.1066,
+      "step": 13683
+    },
+    {
+      "epoch": 0.497654289558861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0561544476585401e-05,
+      "loss": 0.9892,
+      "step": 13684
+    },
+    {
+      "epoch": 0.4976906571625996,
+      "grad_norm": 0.0,
+      "learning_rate": 1.056036846957491e-05,
+      "loss": 0.9937,
+      "step": 13685
+    },
+    {
+      "epoch": 0.4977270247663381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0559192454790106e-05,
+      "loss": 0.9218,
+      "step": 13686
+    },
+    {
+      "epoch": 0.4977633923700767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0558016432247296e-05,
+      "loss": 0.9312,
+      "step": 13687
+    },
+    {
+      "epoch": 0.4977997599738153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0556840401962802e-05,
+      "loss": 1.0709,
+      "step": 13688
+    },
+    {
+      "epoch": 0.4978361275775539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.055566436395294e-05,
+      "loss": 1.0842,
+      "step": 13689
+    },
+    {
+      "epoch": 0.4978724951812925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0554488318234018e-05,
+      "loss": 1.0502,
+      "step": 13690
+    },
+    {
+      "epoch": 0.4979088627850311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.055331226482236e-05,
+      "loss": 0.9925,
+      "step": 13691
+    },
+    {
+      "epoch": 0.4979452303887697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.055213620373428e-05,
+      "loss": 1.0588,
+      "step": 13692
+    },
+    {
+      "epoch": 0.4979815979925083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0550960134986094e-05,
+      "loss": 0.918,
+      "step": 13693
+    },
+    {
+      "epoch": 0.49801796559624684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0549784058594117e-05,
+      "loss": 0.998,
+      "step": 13694
+    },
+    {
+      "epoch": 0.49805433319998543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0548607974574666e-05,
+      "loss": 0.7725,
+      "step": 13695
+    },
+    {
+      "epoch": 0.49809070080372403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0547431882944057e-05,
+      "loss": 0.964,
+      "step": 13696
+    },
+    {
+      "epoch": 0.4981270684074626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0546255783718608e-05,
+      "loss": 1.0828,
+      "step": 13697
+    },
+    {
+      "epoch": 0.4981634360112012,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0545079676914636e-05,
+      "loss": 1.009,
+      "step": 13698
+    },
+    {
+      "epoch": 0.4981998036149398,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0543903562548457e-05,
+      "loss": 0.9053,
+      "step": 13699
+    },
+    {
+      "epoch": 0.4982361712186784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0542727440636387e-05,
+      "loss": 0.9437,
+      "step": 13700
+    },
+    {
+      "epoch": 0.498272538822417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0541551311194741e-05,
+      "loss": 1.0149,
+      "step": 13701
+    },
+    {
+      "epoch": 0.4983089064261556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0540375174239842e-05,
+      "loss": 0.8883,
+      "step": 13702
+    },
+    {
+      "epoch": 0.49834527402989415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0539199029788004e-05,
+      "loss": 0.8375,
+      "step": 13703
+    },
+    {
+      "epoch": 0.49838164163363274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0538022877855543e-05,
+      "loss": 1.0091,
+      "step": 13704
+    },
+    {
+      "epoch": 0.49841800923737134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.053684671845878e-05,
+      "loss": 1.0709,
+      "step": 13705
+    },
+    {
+      "epoch": 0.49845437684110994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0535670551614028e-05,
+      "loss": 0.9392,
+      "step": 13706
+    },
+    {
+      "epoch": 0.49849074444484853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.053449437733761e-05,
+      "loss": 1.073,
+      "step": 13707
+    },
+    {
+      "epoch": 0.49852711204858713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.053331819564584e-05,
+      "loss": 0.9427,
+      "step": 13708
+    },
+    {
+      "epoch": 0.4985634796523257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0532142006555032e-05,
+      "loss": 1.0371,
+      "step": 13709
+    },
+    {
+      "epoch": 0.4985998472560643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0530965810081515e-05,
+      "loss": 0.9148,
+      "step": 13710
+    },
+    {
+      "epoch": 0.49863621485980286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0529789606241598e-05,
+      "loss": 1.0782,
+      "step": 13711
+    },
+    {
+      "epoch": 0.49867258246354146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0528613395051603e-05,
+      "loss": 0.9787,
+      "step": 13712
+    },
+    {
+      "epoch": 0.49870895006728005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0527437176527848e-05,
+      "loss": 1.0612,
+      "step": 13713
+    },
+    {
+      "epoch": 0.49874531767101865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.052626095068665e-05,
+      "loss": 1.0394,
+      "step": 13714
+    },
+    {
+      "epoch": 0.49878168527475725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0525084717544325e-05,
+      "loss": 1.0369,
+      "step": 13715
+    },
+    {
+      "epoch": 0.49881805287849584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0523908477117197e-05,
+      "loss": 1.0045,
+      "step": 13716
+    },
+    {
+      "epoch": 0.49885442048223444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0522732229421582e-05,
+      "loss": 1.1103,
+      "step": 13717
+    },
+    {
+      "epoch": 0.49889078808597304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0521555974473799e-05,
+      "loss": 0.8637,
+      "step": 13718
+    },
+    {
+      "epoch": 0.49892715568971163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0520379712290166e-05,
+      "loss": 0.9603,
+      "step": 13719
+    },
+    {
+      "epoch": 0.4989635232934502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0519203442887003e-05,
+      "loss": 0.9679,
+      "step": 13720
+    },
+    {
+      "epoch": 0.49899989089718877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0518027166280632e-05,
+      "loss": 1.0469,
+      "step": 13721
+    },
+    {
+      "epoch": 0.49903625850092737,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0516850882487364e-05,
+      "loss": 1.002,
+      "step": 13722
+    },
+    {
+      "epoch": 0.49907262610466596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0515674591523526e-05,
+      "loss": 0.9784,
+      "step": 13723
+    },
+    {
+      "epoch": 0.49910899370840456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0514498293405435e-05,
+      "loss": 1.0857,
+      "step": 13724
+    },
+    {
+      "epoch": 0.49914536131214315,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0513321988149408e-05,
+      "loss": 1.0061,
+      "step": 13725
+    },
+    {
+      "epoch": 0.49918172891588175,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0512145675771768e-05,
+      "loss": 0.8947,
+      "step": 13726
+    },
+    {
+      "epoch": 0.49921809651962035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0510969356288836e-05,
+      "loss": 0.997,
+      "step": 13727
+    },
+    {
+      "epoch": 0.4992544641233589,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0509793029716924e-05,
+      "loss": 1.032,
+      "step": 13728
+    },
+    {
+      "epoch": 0.4992908317270975,
+      "grad_norm": 0.0,
+      "learning_rate": 1.050861669607236e-05,
+      "loss": 0.9283,
+      "step": 13729
+    },
+    {
+      "epoch": 0.4993271993308361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0507440355371463e-05,
+      "loss": 1.0396,
+      "step": 13730
+    },
+    {
+      "epoch": 0.4993635669345747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0506264007630548e-05,
+      "loss": 1.0586,
+      "step": 13731
+    },
+    {
+      "epoch": 0.4993999345383133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.050508765286594e-05,
+      "loss": 1.0399,
+      "step": 13732
+    },
+    {
+      "epoch": 0.49943630214205187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0503911291093956e-05,
+      "loss": 0.89,
+      "step": 13733
+    },
+    {
+      "epoch": 0.49947266974579047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.050273492233092e-05,
+      "loss": 0.8823,
+      "step": 13734
+    },
+    {
+      "epoch": 0.49950903734952906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0501558546593144e-05,
+      "loss": 0.9788,
+      "step": 13735
+    },
+    {
+      "epoch": 0.4995454049532676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.050038216389696e-05,
+      "loss": 1.0281,
+      "step": 13736
+    },
+    {
+      "epoch": 0.4995817725570062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0499205774258685e-05,
+      "loss": 1.04,
+      "step": 13737
+    },
+    {
+      "epoch": 0.4996181401607448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0498029377694633e-05,
+      "loss": 0.9803,
+      "step": 13738
+    },
+    {
+      "epoch": 0.4996545077644834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0496852974221134e-05,
+      "loss": 0.9428,
+      "step": 13739
+    },
+    {
+      "epoch": 0.499690875368222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0495676563854505e-05,
+      "loss": 0.9243,
+      "step": 13740
+    },
+    {
+      "epoch": 0.4997272429719606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0494500146611064e-05,
+      "loss": 1.1035,
+      "step": 13741
+    },
+    {
+      "epoch": 0.4997636105756992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0493323722507138e-05,
+      "loss": 0.9311,
+      "step": 13742
+    },
+    {
+      "epoch": 0.4997999781794378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0492147291559045e-05,
+      "loss": 0.9763,
+      "step": 13743
+    },
+    {
+      "epoch": 0.4998363457831764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0490970853783105e-05,
+      "loss": 0.8806,
+      "step": 13744
+    },
+    {
+      "epoch": 0.4998727133869149,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0489794409195641e-05,
+      "loss": 0.9768,
+      "step": 13745
+    },
+    {
+      "epoch": 0.4999090809906535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0488617957812974e-05,
+      "loss": 1.035,
+      "step": 13746
+    },
+    {
+      "epoch": 0.4999454485943921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0487441499651428e-05,
+      "loss": 0.9918,
+      "step": 13747
+    },
+    {
+      "epoch": 0.4999818161981307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0486265034727323e-05,
+      "loss": 1.0235,
+      "step": 13748
+    },
+    {
+      "epoch": 0.5000181838018692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0485088563056977e-05,
+      "loss": 0.9754,
+      "step": 13749
+    },
+    {
+      "epoch": 0.5000545514056078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0483912084656722e-05,
+      "loss": 0.9133,
+      "step": 13750
+    },
+    {
+      "epoch": 0.5000909190093464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0482735599542867e-05,
+      "loss": 0.9652,
+      "step": 13751
+    },
+    {
+      "epoch": 0.500127286613085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0481559107731745e-05,
+      "loss": 0.9509,
+      "step": 13752
+    },
+    {
+      "epoch": 0.5001636542168236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0480382609239672e-05,
+      "loss": 0.8794,
+      "step": 13753
+    },
+    {
+      "epoch": 0.5002000218205622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0479206104082969e-05,
+      "loss": 0.9833,
+      "step": 13754
+    },
+    {
+      "epoch": 0.5002363894243008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0478029592277966e-05,
+      "loss": 1.0259,
+      "step": 13755
+    },
+    {
+      "epoch": 0.5002727570280394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0476853073840977e-05,
+      "loss": 1.1091,
+      "step": 13756
+    },
+    {
+      "epoch": 0.500309124631778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.047567654878833e-05,
+      "loss": 1.0794,
+      "step": 13757
+    },
+    {
+      "epoch": 0.5003454922355166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0474500017136345e-05,
+      "loss": 1.0476,
+      "step": 13758
+    },
+    {
+      "epoch": 0.5003818598392552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0473323478901346e-05,
+      "loss": 0.9686,
+      "step": 13759
+    },
+    {
+      "epoch": 0.5004182274429938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0472146934099654e-05,
+      "loss": 1.0264,
+      "step": 13760
+    },
+    {
+      "epoch": 0.5004545950467324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0470970382747594e-05,
+      "loss": 1.1205,
+      "step": 13761
+    },
+    {
+      "epoch": 0.500490962650471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0469793824861487e-05,
+      "loss": 0.9836,
+      "step": 13762
+    },
+    {
+      "epoch": 0.5005273302542096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0468617260457659e-05,
+      "loss": 1.0721,
+      "step": 13763
+    },
+    {
+      "epoch": 0.5005636978579482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0467440689552433e-05,
+      "loss": 0.8856,
+      "step": 13764
+    },
+    {
+      "epoch": 0.5006000654616868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0466264112162126e-05,
+      "loss": 1.0075,
+      "step": 13765
+    },
+    {
+      "epoch": 0.5006364330654253,
+      "grad_norm": 0.0,
+      "learning_rate": 1.046508752830307e-05,
+      "loss": 1.0488,
+      "step": 13766
+    },
+    {
+      "epoch": 0.5006728006691639,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0463910937991584e-05,
+      "loss": 0.9536,
+      "step": 13767
+    },
+    {
+      "epoch": 0.5007091682729025,
+      "grad_norm": 0.0,
+      "learning_rate": 1.046273434124399e-05,
+      "loss": 1.0295,
+      "step": 13768
+    },
+    {
+      "epoch": 0.5007455358766411,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0461557738076615e-05,
+      "loss": 0.9807,
+      "step": 13769
+    },
+    {
+      "epoch": 0.5007819034803797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0460381128505781e-05,
+      "loss": 1.0154,
+      "step": 13770
+    },
+    {
+      "epoch": 0.5008182710841182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0459204512547811e-05,
+      "loss": 1.0687,
+      "step": 13771
+    },
+    {
+      "epoch": 0.5008546386878568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0458027890219032e-05,
+      "loss": 0.952,
+      "step": 13772
+    },
+    {
+      "epoch": 0.5008910062915954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0456851261535763e-05,
+      "loss": 0.9465,
+      "step": 13773
+    },
+    {
+      "epoch": 0.500927373895334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0455674626514334e-05,
+      "loss": 0.9047,
+      "step": 13774
+    },
+    {
+      "epoch": 0.5009637414990726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0454497985171064e-05,
+      "loss": 1.0227,
+      "step": 13775
+    },
+    {
+      "epoch": 0.5010001091028112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.045332133752228e-05,
+      "loss": 1.1011,
+      "step": 13776
+    },
+    {
+      "epoch": 0.5010364767065498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0452144683584306e-05,
+      "loss": 1.0306,
+      "step": 13777
+    },
+    {
+      "epoch": 0.5010728443102884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0450968023373464e-05,
+      "loss": 0.8622,
+      "step": 13778
+    },
+    {
+      "epoch": 0.501109211914027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0449791356906082e-05,
+      "loss": 1.0158,
+      "step": 13779
+    },
+    {
+      "epoch": 0.5011455795177656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0448614684198483e-05,
+      "loss": 1.0065,
+      "step": 13780
+    },
+    {
+      "epoch": 0.5011819471215042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0447438005266991e-05,
+      "loss": 1.0751,
+      "step": 13781
+    },
+    {
+      "epoch": 0.5012183147252428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0446261320127932e-05,
+      "loss": 1.0975,
+      "step": 13782
+    },
+    {
+      "epoch": 0.5012546823289813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0445084628797633e-05,
+      "loss": 1.0275,
+      "step": 13783
+    },
+    {
+      "epoch": 0.5012910499327199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0443907931292411e-05,
+      "loss": 1.0644,
+      "step": 13784
+    },
+    {
+      "epoch": 0.5013274175364585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0442731227628602e-05,
+      "loss": 0.8602,
+      "step": 13785
+    },
+    {
+      "epoch": 0.5013637851401971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0441554517822521e-05,
+      "loss": 0.9618,
+      "step": 13786
+    },
+    {
+      "epoch": 0.5014001527439357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.04403778018905e-05,
+      "loss": 0.9756,
+      "step": 13787
+    },
+    {
+      "epoch": 0.5014365203476743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0439201079848864e-05,
+      "loss": 0.9285,
+      "step": 13788
+    },
+    {
+      "epoch": 0.5014728879514129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.043802435171393e-05,
+      "loss": 1.0429,
+      "step": 13789
+    },
+    {
+      "epoch": 0.5015092555551515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0436847617502035e-05,
+      "loss": 0.8906,
+      "step": 13790
+    },
+    {
+      "epoch": 0.5015456231588901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0435670877229496e-05,
+      "loss": 1.0652,
+      "step": 13791
+    },
+    {
+      "epoch": 0.5015819907626287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0434494130912642e-05,
+      "loss": 1.0172,
+      "step": 13792
+    },
+    {
+      "epoch": 0.5016183583663673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.04333173785678e-05,
+      "loss": 1.048,
+      "step": 13793
+    },
+    {
+      "epoch": 0.5016547259701059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0432140620211292e-05,
+      "loss": 0.9622,
+      "step": 13794
+    },
+    {
+      "epoch": 0.5016910935738444,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0430963855859448e-05,
+      "loss": 1.061,
+      "step": 13795
+    },
+    {
+      "epoch": 0.501727461177583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.042978708552859e-05,
+      "loss": 0.9026,
+      "step": 13796
+    },
+    {
+      "epoch": 0.5017638287813216,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0428610309235047e-05,
+      "loss": 1.0885,
+      "step": 13797
+    },
+    {
+      "epoch": 0.5018001963850602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0427433526995146e-05,
+      "loss": 1.049,
+      "step": 13798
+    },
+    {
+      "epoch": 0.5018365639887987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0426256738825208e-05,
+      "loss": 0.9927,
+      "step": 13799
+    },
+    {
+      "epoch": 0.5018729315925373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0425079944741563e-05,
+      "loss": 0.882,
+      "step": 13800
+    },
+    {
+      "epoch": 0.5019092991962759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0423903144760538e-05,
+      "loss": 1.0779,
+      "step": 13801
+    },
+    {
+      "epoch": 0.5019456668000145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0422726338898459e-05,
+      "loss": 1.2419,
+      "step": 13802
+    },
+    {
+      "epoch": 0.5019820344037531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.042154952717165e-05,
+      "loss": 0.9846,
+      "step": 13803
+    },
+    {
+      "epoch": 0.5020184020074917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0420372709596442e-05,
+      "loss": 0.9675,
+      "step": 13804
+    },
+    {
+      "epoch": 0.5020547696112303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0419195886189156e-05,
+      "loss": 0.8315,
+      "step": 13805
+    },
+    {
+      "epoch": 0.5020911372149689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0418019056966125e-05,
+      "loss": 1.0938,
+      "step": 13806
+    },
+    {
+      "epoch": 0.5021275048187075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.041684222194367e-05,
+      "loss": 0.9996,
+      "step": 13807
+    },
+    {
+      "epoch": 0.5021638724224461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0415665381138123e-05,
+      "loss": 1.1447,
+      "step": 13808
+    },
+    {
+      "epoch": 0.5022002400261847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0414488534565807e-05,
+      "loss": 1.1377,
+      "step": 13809
+    },
+    {
+      "epoch": 0.5022366076299233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.041331168224305e-05,
+      "loss": 1.0496,
+      "step": 13810
+    },
+    {
+      "epoch": 0.5022729752336619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0412134824186181e-05,
+      "loss": 1.032,
+      "step": 13811
+    },
+    {
+      "epoch": 0.5023093428374005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0410957960411527e-05,
+      "loss": 0.9009,
+      "step": 13812
+    },
+    {
+      "epoch": 0.5023457104411391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0409781090935412e-05,
+      "loss": 0.9471,
+      "step": 13813
+    },
+    {
+      "epoch": 0.5023820780448777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0408604215774167e-05,
+      "loss": 1.0475,
+      "step": 13814
+    },
+    {
+      "epoch": 0.5024184456486163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.040742733494412e-05,
+      "loss": 0.9725,
+      "step": 13815
+    },
+    {
+      "epoch": 0.5024548132523547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0406250448461593e-05,
+      "loss": 0.9496,
+      "step": 13816
+    },
+    {
+      "epoch": 0.5024911808560933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.040507355634292e-05,
+      "loss": 1.0216,
+      "step": 13817
+    },
+    {
+      "epoch": 0.5025275484598319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0403896658604425e-05,
+      "loss": 0.9351,
+      "step": 13818
+    },
+    {
+      "epoch": 0.5025639160635705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0402719755262436e-05,
+      "loss": 0.9752,
+      "step": 13819
+    },
+    {
+      "epoch": 0.5026002836673091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0401542846333286e-05,
+      "loss": 0.953,
+      "step": 13820
+    },
+    {
+      "epoch": 0.5026366512710477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0400365931833294e-05,
+      "loss": 1.097,
+      "step": 13821
+    },
+    {
+      "epoch": 0.5026730188747863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0399189011778795e-05,
+      "loss": 0.9093,
+      "step": 13822
+    },
+    {
+      "epoch": 0.5027093864785249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0398012086186116e-05,
+      "loss": 0.9744,
+      "step": 13823
+    },
+    {
+      "epoch": 0.5027457540822635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0396835155071581e-05,
+      "loss": 0.9999,
+      "step": 13824
+    },
+    {
+      "epoch": 0.5027821216860021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0395658218451521e-05,
+      "loss": 0.9024,
+      "step": 13825
+    },
+    {
+      "epoch": 0.5028184892897407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0394481276342268e-05,
+      "loss": 1.0868,
+      "step": 13826
+    },
+    {
+      "epoch": 0.5028548568934793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0393304328760145e-05,
+      "loss": 0.9687,
+      "step": 13827
+    },
+    {
+      "epoch": 0.5028912244972179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0392127375721483e-05,
+      "loss": 1.1074,
+      "step": 13828
+    },
+    {
+      "epoch": 0.5029275921009565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0390950417242608e-05,
+      "loss": 1.0134,
+      "step": 13829
+    },
+    {
+      "epoch": 0.5029639597046951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0389773453339856e-05,
+      "loss": 0.8572,
+      "step": 13830
+    },
+    {
+      "epoch": 0.5030003273084337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0388596484029546e-05,
+      "loss": 1.0021,
+      "step": 13831
+    },
+    {
+      "epoch": 0.5030366949121723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0387419509328011e-05,
+      "loss": 1.0688,
+      "step": 13832
+    },
+    {
+      "epoch": 0.5030730625159108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0386242529251582e-05,
+      "loss": 1.0038,
+      "step": 13833
+    },
+    {
+      "epoch": 0.5031094301196494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0385065543816583e-05,
+      "loss": 0.995,
+      "step": 13834
+    },
+    {
+      "epoch": 0.503145797723388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0383888553039347e-05,
+      "loss": 0.9788,
+      "step": 13835
+    },
+    {
+      "epoch": 0.5031821653271266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0382711556936204e-05,
+      "loss": 1.0611,
+      "step": 13836
+    },
+    {
+      "epoch": 0.5032185329308652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0381534555523479e-05,
+      "loss": 1.0678,
+      "step": 13837
+    },
+    {
+      "epoch": 0.5032549005346038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0380357548817506e-05,
+      "loss": 1.142,
+      "step": 13838
+    },
+    {
+      "epoch": 0.5032912681383424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.037918053683461e-05,
+      "loss": 1.0637,
+      "step": 13839
+    },
+    {
+      "epoch": 0.503327635742081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.037800351959112e-05,
+      "loss": 0.8866,
+      "step": 13840
+    },
+    {
+      "epoch": 0.5033640033458195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0376826497103369e-05,
+      "loss": 1.0185,
+      "step": 13841
+    },
+    {
+      "epoch": 0.5034003709495581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0375649469387685e-05,
+      "loss": 1.0106,
+      "step": 13842
+    },
+    {
+      "epoch": 0.5034367385532967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.03744724364604e-05,
+      "loss": 0.8906,
+      "step": 13843
+    },
+    {
+      "epoch": 0.5034731061570353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0373295398337839e-05,
+      "loss": 0.9252,
+      "step": 13844
+    },
+    {
+      "epoch": 0.5035094737607739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0372118355036334e-05,
+      "loss": 0.9734,
+      "step": 13845
+    },
+    {
+      "epoch": 0.5035458413645125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0370941306572215e-05,
+      "loss": 1.0386,
+      "step": 13846
+    },
+    {
+      "epoch": 0.5035822089682511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.036976425296181e-05,
+      "loss": 0.9941,
+      "step": 13847
+    },
+    {
+      "epoch": 0.5036185765719897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0368587194221455e-05,
+      "loss": 1.0069,
+      "step": 13848
+    },
+    {
+      "epoch": 0.5036549441757283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0367410130367472e-05,
+      "loss": 0.885,
+      "step": 13849
+    },
+    {
+      "epoch": 0.5036913117794668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0366233061416195e-05,
+      "loss": 1.0176,
+      "step": 13850
+    },
+    {
+      "epoch": 0.5037276793832054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0365055987383956e-05,
+      "loss": 1.0066,
+      "step": 13851
+    },
+    {
+      "epoch": 0.503764046986944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0363878908287082e-05,
+      "loss": 1.0747,
+      "step": 13852
+    },
+    {
+      "epoch": 0.5038004145906826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0362701824141903e-05,
+      "loss": 1.0444,
+      "step": 13853
+    },
+    {
+      "epoch": 0.5038367821944212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0361524734964752e-05,
+      "loss": 0.9345,
+      "step": 13854
+    },
+    {
+      "epoch": 0.5038731497981598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.036034764077196e-05,
+      "loss": 1.0018,
+      "step": 13855
+    },
+    {
+      "epoch": 0.5039095174018984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0359170541579853e-05,
+      "loss": 0.9335,
+      "step": 13856
+    },
+    {
+      "epoch": 0.503945885005637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0357993437404765e-05,
+      "loss": 0.9812,
+      "step": 13857
+    },
+    {
+      "epoch": 0.5039822526093756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0356816328263024e-05,
+      "loss": 0.946,
+      "step": 13858
+    },
+    {
+      "epoch": 0.5040186202131142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0355639214170966e-05,
+      "loss": 1.1154,
+      "step": 13859
+    },
+    {
+      "epoch": 0.5040549878168528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0354462095144917e-05,
+      "loss": 0.941,
+      "step": 13860
+    },
+    {
+      "epoch": 0.5040913554205914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.035328497120121e-05,
+      "loss": 0.9412,
+      "step": 13861
+    },
+    {
+      "epoch": 0.50412772302433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0352107842356176e-05,
+      "loss": 0.9407,
+      "step": 13862
+    },
+    {
+      "epoch": 0.5041640906280686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0350930708626144e-05,
+      "loss": 1.0672,
+      "step": 13863
+    },
+    {
+      "epoch": 0.5042004582318071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0349753570027446e-05,
+      "loss": 0.9986,
+      "step": 13864
+    },
+    {
+      "epoch": 0.5042368258355457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0348576426576416e-05,
+      "loss": 1.0503,
+      "step": 13865
+    },
+    {
+      "epoch": 0.5042731934392843,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0347399278289379e-05,
+      "loss": 1.1135,
+      "step": 13866
+    },
+    {
+      "epoch": 0.5043095610430228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0346222125182672e-05,
+      "loss": 0.9547,
+      "step": 13867
+    },
+    {
+      "epoch": 0.5043459286467614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0345044967272625e-05,
+      "loss": 0.8577,
+      "step": 13868
+    },
+    {
+      "epoch": 0.5043822962505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0343867804575567e-05,
+      "loss": 1.0045,
+      "step": 13869
+    },
+    {
+      "epoch": 0.5044186638542386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0342690637107831e-05,
+      "loss": 0.9169,
+      "step": 13870
+    },
+    {
+      "epoch": 0.5044550314579772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.034151346488575e-05,
+      "loss": 0.9796,
+      "step": 13871
+    },
+    {
+      "epoch": 0.5044913990617158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0340336287925652e-05,
+      "loss": 1.056,
+      "step": 13872
+    },
+    {
+      "epoch": 0.5045277666654544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0339159106243875e-05,
+      "loss": 1.0771,
+      "step": 13873
+    },
+    {
+      "epoch": 0.504564134269193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0337981919856742e-05,
+      "loss": 0.9693,
+      "step": 13874
+    },
+    {
+      "epoch": 0.5046005018729316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0336804728780592e-05,
+      "loss": 1.1277,
+      "step": 13875
+    },
+    {
+      "epoch": 0.5046368694766702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0335627533031754e-05,
+      "loss": 1.123,
+      "step": 13876
+    },
+    {
+      "epoch": 0.5046732370804088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0334450332626558e-05,
+      "loss": 0.9038,
+      "step": 13877
+    },
+    {
+      "epoch": 0.5047096046841474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.033327312758134e-05,
+      "loss": 1.0774,
+      "step": 13878
+    },
+    {
+      "epoch": 0.504745972287886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.033209591791243e-05,
+      "loss": 1.0446,
+      "step": 13879
+    },
+    {
+      "epoch": 0.5047823398916246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0330918703636158e-05,
+      "loss": 1.0821,
+      "step": 13880
+    },
+    {
+      "epoch": 0.5048187074953632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0329741484768862e-05,
+      "loss": 1.0014,
+      "step": 13881
+    },
+    {
+      "epoch": 0.5048550750991018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0328564261326872e-05,
+      "loss": 0.8656,
+      "step": 13882
+    },
+    {
+      "epoch": 0.5048914427028404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0327387033326515e-05,
+      "loss": 0.9969,
+      "step": 13883
+    },
+    {
+      "epoch": 0.5049278103065789,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0326209800784133e-05,
+      "loss": 0.972,
+      "step": 13884
+    },
+    {
+      "epoch": 0.5049641779103174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0325032563716047e-05,
+      "loss": 0.8485,
+      "step": 13885
+    },
+    {
+      "epoch": 0.505000545514056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0323855322138598e-05,
+      "loss": 1.0126,
+      "step": 13886
+    },
+    {
+      "epoch": 0.5050369131177946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0322678076068116e-05,
+      "loss": 1.0052,
+      "step": 13887
+    },
+    {
+      "epoch": 0.5050732807215332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0321500825520932e-05,
+      "loss": 0.9493,
+      "step": 13888
+    },
+    {
+      "epoch": 0.5051096483252718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.032032357051338e-05,
+      "loss": 1.0004,
+      "step": 13889
+    },
+    {
+      "epoch": 0.5051460159290104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0319146311061791e-05,
+      "loss": 1.0861,
+      "step": 13890
+    },
+    {
+      "epoch": 0.505182383532749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0317969047182505e-05,
+      "loss": 0.9111,
+      "step": 13891
+    },
+    {
+      "epoch": 0.5052187511364876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0316791778891848e-05,
+      "loss": 1.0031,
+      "step": 13892
+    },
+    {
+      "epoch": 0.5052551187402262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0315614506206152e-05,
+      "loss": 1.0078,
+      "step": 13893
+    },
+    {
+      "epoch": 0.5052914863439648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0314437229141755e-05,
+      "loss": 0.988,
+      "step": 13894
+    },
+    {
+      "epoch": 0.5053278539477034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0313259947714988e-05,
+      "loss": 1.0516,
+      "step": 13895
+    },
+    {
+      "epoch": 0.505364221551442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.031208266194218e-05,
+      "loss": 0.9438,
+      "step": 13896
+    },
+    {
+      "epoch": 0.5054005891551806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0310905371839673e-05,
+      "loss": 1.0204,
+      "step": 13897
+    },
+    {
+      "epoch": 0.5054369567589192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.030972807742379e-05,
+      "loss": 1.0062,
+      "step": 13898
+    },
+    {
+      "epoch": 0.5054733243626578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0308550778710873e-05,
+      "loss": 0.9766,
+      "step": 13899
+    },
+    {
+      "epoch": 0.5055096919663963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.030737347571725e-05,
+      "loss": 0.9957,
+      "step": 13900
+    },
+    {
+      "epoch": 0.5055460595701349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0306196168459255e-05,
+      "loss": 1.0782,
+      "step": 13901
+    },
+    {
+      "epoch": 0.5055824271738735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0305018856953226e-05,
+      "loss": 1.0537,
+      "step": 13902
+    },
+    {
+      "epoch": 0.5056187947776121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0303841541215491e-05,
+      "loss": 1.1575,
+      "step": 13903
+    },
+    {
+      "epoch": 0.5056551623813507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0302664221262389e-05,
+      "loss": 1.0318,
+      "step": 13904
+    },
+    {
+      "epoch": 0.5056915299850893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0301486897110246e-05,
+      "loss": 0.9216,
+      "step": 13905
+    },
+    {
+      "epoch": 0.5057278975888279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0300309568775402e-05,
+      "loss": 1.134,
+      "step": 13906
+    },
+    {
+      "epoch": 0.5057642651925665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0299132236274188e-05,
+      "loss": 1.0221,
+      "step": 13907
+    },
+    {
+      "epoch": 0.505800632796305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0297954899622939e-05,
+      "loss": 0.9788,
+      "step": 13908
+    },
+    {
+      "epoch": 0.5058370004000436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0296777558837988e-05,
+      "loss": 1.0667,
+      "step": 13909
+    },
+    {
+      "epoch": 0.5058733680037822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.029560021393567e-05,
+      "loss": 1.1089,
+      "step": 13910
+    },
+    {
+      "epoch": 0.5059097356075208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0294422864932318e-05,
+      "loss": 0.9645,
+      "step": 13911
+    },
+    {
+      "epoch": 0.5059461032112594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0293245511844268e-05,
+      "loss": 1.0179,
+      "step": 13912
+    },
+    {
+      "epoch": 0.505982470814998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.029206815468785e-05,
+      "loss": 0.9588,
+      "step": 13913
+    },
+    {
+      "epoch": 0.5060188384187366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0290890793479402e-05,
+      "loss": 0.9731,
+      "step": 13914
+    },
+    {
+      "epoch": 0.5060552060224752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0289713428235257e-05,
+      "loss": 0.8929,
+      "step": 13915
+    },
+    {
+      "epoch": 0.5060915736262138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.028853605897175e-05,
+      "loss": 0.8958,
+      "step": 13916
+    },
+    {
+      "epoch": 0.5061279412299523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0287358685705213e-05,
+      "loss": 0.9812,
+      "step": 13917
+    },
+    {
+      "epoch": 0.5061643088336909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0286181308451984e-05,
+      "loss": 1.0614,
+      "step": 13918
+    },
+    {
+      "epoch": 0.5062006764374295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0285003927228394e-05,
+      "loss": 0.9849,
+      "step": 13919
+    },
+    {
+      "epoch": 0.5062370440411681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0283826542050778e-05,
+      "loss": 0.8756,
+      "step": 13920
+    },
+    {
+      "epoch": 0.5062734116449067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0282649152935473e-05,
+      "loss": 0.9983,
+      "step": 13921
+    },
+    {
+      "epoch": 0.5063097792486453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0281471759898814e-05,
+      "loss": 0.8995,
+      "step": 13922
+    },
+    {
+      "epoch": 0.5063461468523839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.028029436295713e-05,
+      "loss": 0.9966,
+      "step": 13923
+    },
+    {
+      "epoch": 0.5063825144561225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0279116962126763e-05,
+      "loss": 1.0858,
+      "step": 13924
+    },
+    {
+      "epoch": 0.5064188820598611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.027793955742404e-05,
+      "loss": 0.9839,
+      "step": 13925
+    },
+    {
+      "epoch": 0.5064552496635997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0276762148865303e-05,
+      "loss": 1.0475,
+      "step": 13926
+    },
+    {
+      "epoch": 0.5064916172673383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0275584736466882e-05,
+      "loss": 0.9043,
+      "step": 13927
+    },
+    {
+      "epoch": 0.5065279848710769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0274407320245112e-05,
+      "loss": 0.9065,
+      "step": 13928
+    },
+    {
+      "epoch": 0.5065643524748155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0273229900216335e-05,
+      "loss": 0.9847,
+      "step": 13929
+    },
+    {
+      "epoch": 0.5066007200785541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0272052476396876e-05,
+      "loss": 0.8848,
+      "step": 13930
+    },
+    {
+      "epoch": 0.5066370876822927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0270875048803077e-05,
+      "loss": 0.9443,
+      "step": 13931
+    },
+    {
+      "epoch": 0.5066734552860312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0269697617451271e-05,
+      "loss": 0.9765,
+      "step": 13932
+    },
+    {
+      "epoch": 0.5067098228897698,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0268520182357792e-05,
+      "loss": 1.0237,
+      "step": 13933
+    },
+    {
+      "epoch": 0.5067461904935083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.026734274353898e-05,
+      "loss": 1.0065,
+      "step": 13934
+    },
+    {
+      "epoch": 0.5067825580972469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0266165301011163e-05,
+      "loss": 1.1095,
+      "step": 13935
+    },
+    {
+      "epoch": 0.5068189257009855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.026498785479068e-05,
+      "loss": 1.0222,
+      "step": 13936
+    },
+    {
+      "epoch": 0.5068552933047241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.026381040489387e-05,
+      "loss": 0.9499,
+      "step": 13937
+    },
+    {
+      "epoch": 0.5068916609084627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0262632951337064e-05,
+      "loss": 0.97,
+      "step": 13938
+    },
+    {
+      "epoch": 0.5069280285122013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0261455494136594e-05,
+      "loss": 0.9404,
+      "step": 13939
+    },
+    {
+      "epoch": 0.5069643961159399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0260278033308804e-05,
+      "loss": 0.8693,
+      "step": 13940
+    },
+    {
+      "epoch": 0.5070007637196785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0259100568870026e-05,
+      "loss": 0.9447,
+      "step": 13941
+    },
+    {
+      "epoch": 0.5070371313234171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0257923100836596e-05,
+      "loss": 1.0396,
+      "step": 13942
+    },
+    {
+      "epoch": 0.5070734989271557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0256745629224846e-05,
+      "loss": 1.0026,
+      "step": 13943
+    },
+    {
+      "epoch": 0.5071098665308943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0255568154051115e-05,
+      "loss": 0.9628,
+      "step": 13944
+    },
+    {
+      "epoch": 0.5071462341346329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.025439067533174e-05,
+      "loss": 1.0056,
+      "step": 13945
+    },
+    {
+      "epoch": 0.5071826017383715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0253213193083052e-05,
+      "loss": 0.9797,
+      "step": 13946
+    },
+    {
+      "epoch": 0.5072189693421101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0252035707321393e-05,
+      "loss": 0.963,
+      "step": 13947
+    },
+    {
+      "epoch": 0.5072553369458487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0250858218063098e-05,
+      "loss": 1.0278,
+      "step": 13948
+    },
+    {
+      "epoch": 0.5072917045495873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0249680725324496e-05,
+      "loss": 1.015,
+      "step": 13949
+    },
+    {
+      "epoch": 0.5073280721533259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0248503229121933e-05,
+      "loss": 0.9447,
+      "step": 13950
+    },
+    {
+      "epoch": 0.5073644397570644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0247325729471738e-05,
+      "loss": 0.9771,
+      "step": 13951
+    },
+    {
+      "epoch": 0.507400807360803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.024614822639025e-05,
+      "loss": 0.9819,
+      "step": 13952
+    },
+    {
+      "epoch": 0.5074371749645415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0244970719893804e-05,
+      "loss": 1.0408,
+      "step": 13953
+    },
+    {
+      "epoch": 0.5074735425682801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0243793209998739e-05,
+      "loss": 1.0258,
+      "step": 13954
+    },
+    {
+      "epoch": 0.5075099101720187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0242615696721386e-05,
+      "loss": 1.0599,
+      "step": 13955
+    },
+    {
+      "epoch": 0.5075462777757573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0241438180078085e-05,
+      "loss": 0.9964,
+      "step": 13956
+    },
+    {
+      "epoch": 0.5075826453794959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0240260660085171e-05,
+      "loss": 0.9655,
+      "step": 13957
+    },
+    {
+      "epoch": 0.5076190129832345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0239083136758982e-05,
+      "loss": 1.0598,
+      "step": 13958
+    },
+    {
+      "epoch": 0.5076553805869731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0237905610115858e-05,
+      "loss": 1.0185,
+      "step": 13959
+    },
+    {
+      "epoch": 0.5076917481907117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0236728080172127e-05,
+      "loss": 1.0687,
+      "step": 13960
+    },
+    {
+      "epoch": 0.5077281157944503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0235550546944132e-05,
+      "loss": 0.9065,
+      "step": 13961
+    },
+    {
+      "epoch": 0.5077644833981889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0234373010448205e-05,
+      "loss": 1.1592,
+      "step": 13962
+    },
+    {
+      "epoch": 0.5078008510019275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0233195470700685e-05,
+      "loss": 1.0111,
+      "step": 13963
+    },
+    {
+      "epoch": 0.5078372186056661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.023201792771791e-05,
+      "loss": 0.9488,
+      "step": 13964
+    },
+    {
+      "epoch": 0.5078735862094047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0230840381516214e-05,
+      "loss": 0.8725,
+      "step": 13965
+    },
+    {
+      "epoch": 0.5079099538131433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0229662832111937e-05,
+      "loss": 1.0246,
+      "step": 13966
+    },
+    {
+      "epoch": 0.5079463214168819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0228485279521414e-05,
+      "loss": 1.0844,
+      "step": 13967
+    },
+    {
+      "epoch": 0.5079826890206204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.022730772376098e-05,
+      "loss": 1.1093,
+      "step": 13968
+    },
+    {
+      "epoch": 0.508019056624359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0226130164846974e-05,
+      "loss": 1.0881,
+      "step": 13969
+    },
+    {
+      "epoch": 0.5080554242280976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0224952602795737e-05,
+      "loss": 1.1746,
+      "step": 13970
+    },
+    {
+      "epoch": 0.5080917918318362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0223775037623596e-05,
+      "loss": 0.9664,
+      "step": 13971
+    },
+    {
+      "epoch": 0.5081281594355748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0222597469346898e-05,
+      "loss": 1.0783,
+      "step": 13972
+    },
+    {
+      "epoch": 0.5081645270393134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0221419897981973e-05,
+      "loss": 0.9095,
+      "step": 13973
+    },
+    {
+      "epoch": 0.508200894643052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0220242323545162e-05,
+      "loss": 1.0591,
+      "step": 13974
+    },
+    {
+      "epoch": 0.5082372622467906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0219064746052803e-05,
+      "loss": 1.0275,
+      "step": 13975
+    },
+    {
+      "epoch": 0.5082736298505292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0217887165521228e-05,
+      "loss": 1.0992,
+      "step": 13976
+    },
+    {
+      "epoch": 0.5083099974542677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0216709581966781e-05,
+      "loss": 1.202,
+      "step": 13977
+    },
+    {
+      "epoch": 0.5083463650580063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0215531995405797e-05,
+      "loss": 0.8468,
+      "step": 13978
+    },
+    {
+      "epoch": 0.5083827326617449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.021435440585461e-05,
+      "loss": 1.0489,
+      "step": 13979
+    },
+    {
+      "epoch": 0.5084191002654835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0213176813329559e-05,
+      "loss": 0.9382,
+      "step": 13980
+    },
+    {
+      "epoch": 0.5084554678692221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0211999217846984e-05,
+      "loss": 0.99,
+      "step": 13981
+    },
+    {
+      "epoch": 0.5084918354729607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0210821619423219e-05,
+      "loss": 1.029,
+      "step": 13982
+    },
+    {
+      "epoch": 0.5085282030766993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0209644018074604e-05,
+      "loss": 1.0314,
+      "step": 13983
+    },
+    {
+      "epoch": 0.5085645706804378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0208466413817475e-05,
+      "loss": 1.0639,
+      "step": 13984
+    },
+    {
+      "epoch": 0.5086009382841764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0207288806668172e-05,
+      "loss": 0.9838,
+      "step": 13985
+    },
+    {
+      "epoch": 0.508637305887915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.020611119664303e-05,
+      "loss": 0.9837,
+      "step": 13986
+    },
+    {
+      "epoch": 0.5086736734916536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0204933583758387e-05,
+      "loss": 1.0569,
+      "step": 13987
+    },
+    {
+      "epoch": 0.5087100410953922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0203755968030581e-05,
+      "loss": 0.9699,
+      "step": 13988
+    },
+    {
+      "epoch": 0.5087464086991308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0202578349475951e-05,
+      "loss": 0.9567,
+      "step": 13989
+    },
+    {
+      "epoch": 0.5087827763028694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0201400728110833e-05,
+      "loss": 0.9667,
+      "step": 13990
+    },
+    {
+      "epoch": 0.508819143906608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0200223103951567e-05,
+      "loss": 1.0046,
+      "step": 13991
+    },
+    {
+      "epoch": 0.5088555115103466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0199045477014488e-05,
+      "loss": 1.037,
+      "step": 13992
+    },
+    {
+      "epoch": 0.5088918791140852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0197867847315937e-05,
+      "loss": 1.1121,
+      "step": 13993
+    },
+    {
+      "epoch": 0.5089282467178238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.019669021487225e-05,
+      "loss": 0.9318,
+      "step": 13994
+    },
+    {
+      "epoch": 0.5089646143215624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0195512579699764e-05,
+      "loss": 1.0658,
+      "step": 13995
+    },
+    {
+      "epoch": 0.509000981925301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0194334941814819e-05,
+      "loss": 1.0955,
+      "step": 13996
+    },
+    {
+      "epoch": 0.5090373495290396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.019315730123375e-05,
+      "loss": 0.9702,
+      "step": 13997
+    },
+    {
+      "epoch": 0.5090737171327782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.01919796579729e-05,
+      "loss": 1.0369,
+      "step": 13998
+    },
+    {
+      "epoch": 0.5091100847365168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0190802012048611e-05,
+      "loss": 1.0578,
+      "step": 13999
+    },
+    {
+      "epoch": 0.5091464523402554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0189624363477207e-05,
+      "loss": 1.0428,
+      "step": 14000
+    },
+    {
+      "epoch": 0.5091828199439938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0188446712275035e-05,
+      "loss": 0.942,
+      "step": 14001
+    },
+    {
+      "epoch": 0.5092191875477324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0187269058458431e-05,
+      "loss": 0.9578,
+      "step": 14002
+    },
+    {
+      "epoch": 0.509255555151471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0186091402043739e-05,
+      "loss": 0.9184,
+      "step": 14003
+    },
+    {
+      "epoch": 0.5092919227552096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0184913743047288e-05,
+      "loss": 0.9357,
+      "step": 14004
+    },
+    {
+      "epoch": 0.5093282903589482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0183736081485422e-05,
+      "loss": 0.8695,
+      "step": 14005
+    },
+    {
+      "epoch": 0.5093646579626868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0182558417374482e-05,
+      "loss": 1.0099,
+      "step": 14006
+    },
+    {
+      "epoch": 0.5094010255664254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0181380750730801e-05,
+      "loss": 1.0354,
+      "step": 14007
+    },
+    {
+      "epoch": 0.509437393170164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0180203081570719e-05,
+      "loss": 0.9866,
+      "step": 14008
+    },
+    {
+      "epoch": 0.5094737607739026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0179025409910575e-05,
+      "loss": 1.0242,
+      "step": 14009
+    },
+    {
+      "epoch": 0.5095101283776412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0177847735766709e-05,
+      "loss": 1.0038,
+      "step": 14010
+    },
+    {
+      "epoch": 0.5095464959813798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0176670059155456e-05,
+      "loss": 0.8946,
+      "step": 14011
+    },
+    {
+      "epoch": 0.5095828635851184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0175492380093158e-05,
+      "loss": 0.9625,
+      "step": 14012
+    },
+    {
+      "epoch": 0.509619231188857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0174314698596148e-05,
+      "loss": 0.9437,
+      "step": 14013
+    },
+    {
+      "epoch": 0.5096555987925956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0173137014680774e-05,
+      "loss": 0.9885,
+      "step": 14014
+    },
+    {
+      "epoch": 0.5096919663963342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0171959328363367e-05,
+      "loss": 1.0606,
+      "step": 14015
+    },
+    {
+      "epoch": 0.5097283340000728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.017078163966027e-05,
+      "loss": 1.0099,
+      "step": 14016
+    },
+    {
+      "epoch": 0.5097647016038114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0169603948587818e-05,
+      "loss": 0.9625,
+      "step": 14017
+    },
+    {
+      "epoch": 0.5098010692075499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0168426255162353e-05,
+      "loss": 1.0033,
+      "step": 14018
+    },
+    {
+      "epoch": 0.5098374368112885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0167248559400216e-05,
+      "loss": 1.1155,
+      "step": 14019
+    },
+    {
+      "epoch": 0.509873804415027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0166070861317738e-05,
+      "loss": 1.0042,
+      "step": 14020
+    },
+    {
+      "epoch": 0.5099101720187657,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0164893160931263e-05,
+      "loss": 0.9245,
+      "step": 14021
+    },
+    {
+      "epoch": 0.5099465396225042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.016371545825713e-05,
+      "loss": 0.9837,
+      "step": 14022
+    },
+    {
+      "epoch": 0.5099829072262428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0162537753311678e-05,
+      "loss": 1.0143,
+      "step": 14023
+    },
+    {
+      "epoch": 0.5100192748299814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0161360046111243e-05,
+      "loss": 1.0925,
+      "step": 14024
+    },
+    {
+      "epoch": 0.51005564243372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0160182336672171e-05,
+      "loss": 0.9875,
+      "step": 14025
+    },
+    {
+      "epoch": 0.5100920100374586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0159004625010794e-05,
+      "loss": 0.9142,
+      "step": 14026
+    },
+    {
+      "epoch": 0.5101283776411972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0157826911143449e-05,
+      "loss": 0.9596,
+      "step": 14027
+    },
+    {
+      "epoch": 0.5101647452449358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0156649195086485e-05,
+      "loss": 1.0573,
+      "step": 14028
+    },
+    {
+      "epoch": 0.5102011128486744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0155471476856234e-05,
+      "loss": 0.8982,
+      "step": 14029
+    },
+    {
+      "epoch": 0.510237480452413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0154293756469036e-05,
+      "loss": 0.9979,
+      "step": 14030
+    },
+    {
+      "epoch": 0.5102738480561516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0153116033941233e-05,
+      "loss": 0.9723,
+      "step": 14031
+    },
+    {
+      "epoch": 0.5103102156598902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.015193830928916e-05,
+      "loss": 0.9719,
+      "step": 14032
+    },
+    {
+      "epoch": 0.5103465832636288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.015076058252916e-05,
+      "loss": 1.0957,
+      "step": 14033
+    },
+    {
+      "epoch": 0.5103829508673674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0149582853677571e-05,
+      "loss": 0.8281,
+      "step": 14034
+    },
+    {
+      "epoch": 0.5104193184711059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.014840512275073e-05,
+      "loss": 0.8934,
+      "step": 14035
+    },
+    {
+      "epoch": 0.5104556860748445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.014722738976498e-05,
+      "loss": 0.9295,
+      "step": 14036
+    },
+    {
+      "epoch": 0.5104920536785831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0146049654736661e-05,
+      "loss": 1.0471,
+      "step": 14037
+    },
+    {
+      "epoch": 0.5105284212823217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.014487191768211e-05,
+      "loss": 0.955,
+      "step": 14038
+    },
+    {
+      "epoch": 0.5105647888860603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0143694178617664e-05,
+      "loss": 1.0877,
+      "step": 14039
+    },
+    {
+      "epoch": 0.5106011564897989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0142516437559664e-05,
+      "loss": 1.0023,
+      "step": 14040
+    },
+    {
+      "epoch": 0.5106375240935375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0141338694524455e-05,
+      "loss": 1.037,
+      "step": 14041
+    },
+    {
+      "epoch": 0.5106738916972761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0140160949528366e-05,
+      "loss": 1.0324,
+      "step": 14042
+    },
+    {
+      "epoch": 0.5107102593010147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0138983202587749e-05,
+      "loss": 0.9987,
+      "step": 14043
+    },
+    {
+      "epoch": 0.5107466269047533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0137805453718934e-05,
+      "loss": 0.9962,
+      "step": 14044
+    },
+    {
+      "epoch": 0.5107829945084918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0136627702938265e-05,
+      "loss": 1.0674,
+      "step": 14045
+    },
+    {
+      "epoch": 0.5108193621122304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.013544995026208e-05,
+      "loss": 1.0372,
+      "step": 14046
+    },
+    {
+      "epoch": 0.510855729715969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0134272195706717e-05,
+      "loss": 1.0617,
+      "step": 14047
+    },
+    {
+      "epoch": 0.5108920973197076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.013309443928852e-05,
+      "loss": 1.0319,
+      "step": 14048
+    },
+    {
+      "epoch": 0.5109284649234462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0131916681023828e-05,
+      "loss": 1.1133,
+      "step": 14049
+    },
+    {
+      "epoch": 0.5109648325271848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0130738920928977e-05,
+      "loss": 0.9605,
+      "step": 14050
+    },
+    {
+      "epoch": 0.5110012001309234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0129561159020309e-05,
+      "loss": 1.0372,
+      "step": 14051
+    },
+    {
+      "epoch": 0.5110375677346619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0128383395314166e-05,
+      "loss": 1.0378,
+      "step": 14052
+    },
+    {
+      "epoch": 0.5110739353384005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0127205629826885e-05,
+      "loss": 1.069,
+      "step": 14053
+    },
+    {
+      "epoch": 0.5111103029421391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0126027862574804e-05,
+      "loss": 1.1227,
+      "step": 14054
+    },
+    {
+      "epoch": 0.5111466705458777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0124850093574266e-05,
+      "loss": 0.9286,
+      "step": 14055
+    },
+    {
+      "epoch": 0.5111830381496163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0123672322841611e-05,
+      "loss": 0.9892,
+      "step": 14056
+    },
+    {
+      "epoch": 0.5112194057533549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0122494550393178e-05,
+      "loss": 0.9684,
+      "step": 14057
+    },
+    {
+      "epoch": 0.5112557733570935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0121316776245304e-05,
+      "loss": 1.0073,
+      "step": 14058
+    },
+    {
+      "epoch": 0.5112921409608321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0120139000414338e-05,
+      "loss": 0.9821,
+      "step": 14059
+    },
+    {
+      "epoch": 0.5113285085645707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.011896122291661e-05,
+      "loss": 1.041,
+      "step": 14060
+    },
+    {
+      "epoch": 0.5113648761683093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0117783443768462e-05,
+      "loss": 1.045,
+      "step": 14061
+    },
+    {
+      "epoch": 0.5114012437720479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0116605662986238e-05,
+      "loss": 1.0125,
+      "step": 14062
+    },
+    {
+      "epoch": 0.5114376113757865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0115427880586277e-05,
+      "loss": 0.8695,
+      "step": 14063
+    },
+    {
+      "epoch": 0.5114739789795251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0114250096584914e-05,
+      "loss": 0.9651,
+      "step": 14064
+    },
+    {
+      "epoch": 0.5115103465832637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0113072310998497e-05,
+      "loss": 0.9091,
+      "step": 14065
+    },
+    {
+      "epoch": 0.5115467141870023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0111894523843361e-05,
+      "loss": 1.0377,
+      "step": 14066
+    },
+    {
+      "epoch": 0.5115830817907409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0110716735135846e-05,
+      "loss": 0.9607,
+      "step": 14067
+    },
+    {
+      "epoch": 0.5116194493944795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0109538944892294e-05,
+      "loss": 1.0441,
+      "step": 14068
+    },
+    {
+      "epoch": 0.5116558169982179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0108361153129045e-05,
+      "loss": 1.0746,
+      "step": 14069
+    },
+    {
+      "epoch": 0.5116921846019565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0107183359862438e-05,
+      "loss": 0.8661,
+      "step": 14070
+    },
+    {
+      "epoch": 0.5117285522056951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0106005565108817e-05,
+      "loss": 1.0295,
+      "step": 14071
+    },
+    {
+      "epoch": 0.5117649198094337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0104827768884516e-05,
+      "loss": 1.0649,
+      "step": 14072
+    },
+    {
+      "epoch": 0.5118012874131723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0103649971205878e-05,
+      "loss": 1.1656,
+      "step": 14073
+    },
+    {
+      "epoch": 0.5118376550169109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0102472172089246e-05,
+      "loss": 1.1289,
+      "step": 14074
+    },
+    {
+      "epoch": 0.5118740226206495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0101294371550958e-05,
+      "loss": 1.0338,
+      "step": 14075
+    },
+    {
+      "epoch": 0.5119103902243881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0100116569607355e-05,
+      "loss": 0.8056,
+      "step": 14076
+    },
+    {
+      "epoch": 0.5119467578281267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0098938766274773e-05,
+      "loss": 0.9576,
+      "step": 14077
+    },
+    {
+      "epoch": 0.5119831254318653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0097760961569558e-05,
+      "loss": 1.023,
+      "step": 14078
+    },
+    {
+      "epoch": 0.5120194930356039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0096583155508049e-05,
+      "loss": 1.031,
+      "step": 14079
+    },
+    {
+      "epoch": 0.5120558606393425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0095405348106584e-05,
+      "loss": 0.9831,
+      "step": 14080
+    },
+    {
+      "epoch": 0.5120922282430811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0094227539381507e-05,
+      "loss": 1.0524,
+      "step": 14081
+    },
+    {
+      "epoch": 0.5121285958468197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0093049729349156e-05,
+      "loss": 1.0437,
+      "step": 14082
+    },
+    {
+      "epoch": 0.5121649634505583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.009187191802587e-05,
+      "loss": 0.9952,
+      "step": 14083
+    },
+    {
+      "epoch": 0.5122013310542969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0090694105427994e-05,
+      "loss": 0.9646,
+      "step": 14084
+    },
+    {
+      "epoch": 0.5122376986580354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0089516291571864e-05,
+      "loss": 1.1096,
+      "step": 14085
+    },
+    {
+      "epoch": 0.512274066261774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0088338476473824e-05,
+      "loss": 1.0121,
+      "step": 14086
+    },
+    {
+      "epoch": 0.5123104338655126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0087160660150213e-05,
+      "loss": 1.1053,
+      "step": 14087
+    },
+    {
+      "epoch": 0.5123468014692512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.008598284261737e-05,
+      "loss": 0.9337,
+      "step": 14088
+    },
+    {
+      "epoch": 0.5123831690729898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0084805023891639e-05,
+      "loss": 1.1045,
+      "step": 14089
+    },
+    {
+      "epoch": 0.5124195366767283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0083627203989359e-05,
+      "loss": 0.9967,
+      "step": 14090
+    },
+    {
+      "epoch": 0.512455904280467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0082449382926866e-05,
+      "loss": 1.1053,
+      "step": 14091
+    },
+    {
+      "epoch": 0.5124922718842055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0081271560720509e-05,
+      "loss": 1.0663,
+      "step": 14092
+    },
+    {
+      "epoch": 0.5125286394879441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0080093737386626e-05,
+      "loss": 0.9083,
+      "step": 14093
+    },
+    {
+      "epoch": 0.5125650070916827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0078915912941554e-05,
+      "loss": 1.0251,
+      "step": 14094
+    },
+    {
+      "epoch": 0.5126013746954213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0077738087401634e-05,
+      "loss": 1.0501,
+      "step": 14095
+    },
+    {
+      "epoch": 0.5126377422991599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0076560260783208e-05,
+      "loss": 1.0105,
+      "step": 14096
+    },
+    {
+      "epoch": 0.5126741099028985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0075382433102619e-05,
+      "loss": 1.0062,
+      "step": 14097
+    },
+    {
+      "epoch": 0.5127104775066371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0074204604376206e-05,
+      "loss": 0.9287,
+      "step": 14098
+    },
+    {
+      "epoch": 0.5127468451103757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0073026774620307e-05,
+      "loss": 1.1308,
+      "step": 14099
+    },
+    {
+      "epoch": 0.5127832127141143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0071848943851268e-05,
+      "loss": 1.06,
+      "step": 14100
+    },
+    {
+      "epoch": 0.5128195803178529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0070671112085421e-05,
+      "loss": 0.9032,
+      "step": 14101
+    },
+    {
+      "epoch": 0.5128559479215914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.006949327933912e-05,
+      "loss": 0.9884,
+      "step": 14102
+    },
+    {
+      "epoch": 0.51289231552533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0068315445628696e-05,
+      "loss": 0.8609,
+      "step": 14103
+    },
+    {
+      "epoch": 0.5129286831290686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.006713761097049e-05,
+      "loss": 0.9723,
+      "step": 14104
+    },
+    {
+      "epoch": 0.5129650507328072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0065959775380847e-05,
+      "loss": 0.9039,
+      "step": 14105
+    },
+    {
+      "epoch": 0.5130014183365458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0064781938876105e-05,
+      "loss": 1.0484,
+      "step": 14106
+    },
+    {
+      "epoch": 0.5130377859402844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0063604101472605e-05,
+      "loss": 0.8983,
+      "step": 14107
+    },
+    {
+      "epoch": 0.513074153544023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0062426263186689e-05,
+      "loss": 0.9735,
+      "step": 14108
+    },
+    {
+      "epoch": 0.5131105211477616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0061248424034698e-05,
+      "loss": 0.92,
+      "step": 14109
+    },
+    {
+      "epoch": 0.5131468887515002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0060070584032968e-05,
+      "loss": 0.9868,
+      "step": 14110
+    },
+    {
+      "epoch": 0.5131832563552388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0058892743197847e-05,
+      "loss": 1.0117,
+      "step": 14111
+    },
+    {
+      "epoch": 0.5132196239589774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0057714901545672e-05,
+      "loss": 0.9554,
+      "step": 14112
+    },
+    {
+      "epoch": 0.513255991562716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0056537059092787e-05,
+      "loss": 1.0648,
+      "step": 14113
+    },
+    {
+      "epoch": 0.5132923591664545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.005535921585553e-05,
+      "loss": 0.9486,
+      "step": 14114
+    },
+    {
+      "epoch": 0.5133287267701931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.005418137185024e-05,
+      "loss": 0.8914,
+      "step": 14115
+    },
+    {
+      "epoch": 0.5133650943739317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0053003527093263e-05,
+      "loss": 0.9669,
+      "step": 14116
+    },
+    {
+      "epoch": 0.5134014619776703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0051825681600936e-05,
+      "loss": 1.0876,
+      "step": 14117
+    },
+    {
+      "epoch": 0.5134378295814089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.00506478353896e-05,
+      "loss": 1.0379,
+      "step": 14118
+    },
+    {
+      "epoch": 0.5134741971851474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0049469988475599e-05,
+      "loss": 0.9914,
+      "step": 14119
+    },
+    {
+      "epoch": 0.513510564788886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.004829214087527e-05,
+      "loss": 0.9373,
+      "step": 14120
+    },
+    {
+      "epoch": 0.5135469323926246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0047114292604957e-05,
+      "loss": 1.0552,
+      "step": 14121
+    },
+    {
+      "epoch": 0.5135832999963632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0045936443681001e-05,
+      "loss": 0.8485,
+      "step": 14122
+    },
+    {
+      "epoch": 0.5136196676001018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0044758594119742e-05,
+      "loss": 0.9537,
+      "step": 14123
+    },
+    {
+      "epoch": 0.5136560352038404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.004358074393752e-05,
+      "loss": 1.0229,
+      "step": 14124
+    },
+    {
+      "epoch": 0.513692402807579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0042402893150679e-05,
+      "loss": 1.0067,
+      "step": 14125
+    },
+    {
+      "epoch": 0.5137287704113176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0041225041775553e-05,
+      "loss": 0.9573,
+      "step": 14126
+    },
+    {
+      "epoch": 0.5137651380150562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0040047189828492e-05,
+      "loss": 0.9604,
+      "step": 14127
+    },
+    {
+      "epoch": 0.5138015056187948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0038869337325831e-05,
+      "loss": 1.0688,
+      "step": 14128
+    },
+    {
+      "epoch": 0.5138378732225334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0037691484283916e-05,
+      "loss": 0.8325,
+      "step": 14129
+    },
+    {
+      "epoch": 0.513874240826272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0036513630719084e-05,
+      "loss": 1.1523,
+      "step": 14130
+    },
+    {
+      "epoch": 0.5139106084300106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0035335776647677e-05,
+      "loss": 1.0181,
+      "step": 14131
+    },
+    {
+      "epoch": 0.5139469760337492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0034157922086037e-05,
+      "loss": 0.9036,
+      "step": 14132
+    },
+    {
+      "epoch": 0.5139833436374878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0032980067050504e-05,
+      "loss": 0.9239,
+      "step": 14133
+    },
+    {
+      "epoch": 0.5140197112412264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.003180221155742e-05,
+      "loss": 0.9496,
+      "step": 14134
+    },
+    {
+      "epoch": 0.514056078844965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0030624355623125e-05,
+      "loss": 0.9653,
+      "step": 14135
+    },
+    {
+      "epoch": 0.5140924464487034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0029446499263957e-05,
+      "loss": 0.9818,
+      "step": 14136
+    },
+    {
+      "epoch": 0.514128814052442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0028268642496264e-05,
+      "loss": 0.9657,
+      "step": 14137
+    },
+    {
+      "epoch": 0.5141651816561806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0027090785336384e-05,
+      "loss": 0.8931,
+      "step": 14138
+    },
+    {
+      "epoch": 0.5142015492599192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0025912927800655e-05,
+      "loss": 1.0317,
+      "step": 14139
+    },
+    {
+      "epoch": 0.5142379168636578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0024735069905424e-05,
+      "loss": 0.9272,
+      "step": 14140
+    },
+    {
+      "epoch": 0.5142742844673964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0023557211667026e-05,
+      "loss": 1.0949,
+      "step": 14141
+    },
+    {
+      "epoch": 0.514310652071135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0022379353101808e-05,
+      "loss": 0.9543,
+      "step": 14142
+    },
+    {
+      "epoch": 0.5143470196748736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0021201494226107e-05,
+      "loss": 1.0414,
+      "step": 14143
+    },
+    {
+      "epoch": 0.5143833872786122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0020023635056264e-05,
+      "loss": 1.0537,
+      "step": 14144
+    },
+    {
+      "epoch": 0.5144197548823508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0018845775608623e-05,
+      "loss": 1.075,
+      "step": 14145
+    },
+    {
+      "epoch": 0.5144561224860894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0017667915899525e-05,
+      "loss": 1.0291,
+      "step": 14146
+    },
+    {
+      "epoch": 0.514492490089828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0016490055945307e-05,
+      "loss": 0.9394,
+      "step": 14147
+    },
+    {
+      "epoch": 0.5145288576935666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0015312195762314e-05,
+      "loss": 1.0633,
+      "step": 14148
+    },
+    {
+      "epoch": 0.5145652252973052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0014134335366887e-05,
+      "loss": 1.0319,
+      "step": 14149
+    },
+    {
+      "epoch": 0.5146015929010438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0012956474775364e-05,
+      "loss": 0.9216,
+      "step": 14150
+    },
+    {
+      "epoch": 0.5146379605047824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0011778614004091e-05,
+      "loss": 0.9295,
+      "step": 14151
+    },
+    {
+      "epoch": 0.514674328108521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0010600753069406e-05,
+      "loss": 1.0043,
+      "step": 14152
+    },
+    {
+      "epoch": 0.5147106957122595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.000942289198765e-05,
+      "loss": 1.0237,
+      "step": 14153
+    },
+    {
+      "epoch": 0.5147470633159981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0008245030775165e-05,
+      "loss": 1.0416,
+      "step": 14154
+    },
+    {
+      "epoch": 0.5147834309197367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.000706716944829e-05,
+      "loss": 0.9159,
+      "step": 14155
+    },
+    {
+      "epoch": 0.5148197985234753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0005889308023369e-05,
+      "loss": 0.8636,
+      "step": 14156
+    },
+    {
+      "epoch": 0.5148561661272139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0004711446516741e-05,
+      "loss": 1.0031,
+      "step": 14157
+    },
+    {
+      "epoch": 0.5148925337309525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0003533584944752e-05,
+      "loss": 1.0707,
+      "step": 14158
+    },
+    {
+      "epoch": 0.514928901334691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0002355723323737e-05,
+      "loss": 1.1076,
+      "step": 14159
+    },
+    {
+      "epoch": 0.5149652689384296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.000117786167004e-05,
+      "loss": 0.9274,
+      "step": 14160
+    },
+    {
+      "epoch": 0.5150016365421682,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.9341,
+      "step": 14161
+    },
+    {
+      "epoch": 0.5150380041459068,
+      "grad_norm": 0.0,
+      "learning_rate": 9.998822138329964e-06,
+      "loss": 1.0238,
+      "step": 14162
+    },
+    {
+      "epoch": 0.5150743717496454,
+      "grad_norm": 0.0,
+      "learning_rate": 9.997644276676266e-06,
+      "loss": 0.9412,
+      "step": 14163
+    },
+    {
+      "epoch": 0.515110739353384,
+      "grad_norm": 0.0,
+      "learning_rate": 9.996466415055253e-06,
+      "loss": 0.8205,
+      "step": 14164
+    },
+    {
+      "epoch": 0.5151471069571226,
+      "grad_norm": 0.0,
+      "learning_rate": 9.99528855348326e-06,
+      "loss": 0.9094,
+      "step": 14165
+    },
+    {
+      "epoch": 0.5151834745608612,
+      "grad_norm": 0.0,
+      "learning_rate": 9.994110691976635e-06,
+      "loss": 0.8462,
+      "step": 14166
+    },
+    {
+      "epoch": 0.5152198421645998,
+      "grad_norm": 0.0,
+      "learning_rate": 9.992932830551714e-06,
+      "loss": 0.9055,
+      "step": 14167
+    },
+    {
+      "epoch": 0.5152562097683384,
+      "grad_norm": 0.0,
+      "learning_rate": 9.99175496922484e-06,
+      "loss": 0.9301,
+      "step": 14168
+    },
+    {
+      "epoch": 0.5152925773720769,
+      "grad_norm": 0.0,
+      "learning_rate": 9.990577108012353e-06,
+      "loss": 1.0264,
+      "step": 14169
+    },
+    {
+      "epoch": 0.5153289449758155,
+      "grad_norm": 0.0,
+      "learning_rate": 9.9893992469306e-06,
+      "loss": 0.9499,
+      "step": 14170
+    },
+    {
+      "epoch": 0.5153653125795541,
+      "grad_norm": 0.0,
+      "learning_rate": 9.98822138599591e-06,
+      "loss": 1.0149,
+      "step": 14171
+    },
+    {
+      "epoch": 0.5154016801832927,
+      "grad_norm": 0.0,
+      "learning_rate": 9.98704352522464e-06,
+      "loss": 1.1114,
+      "step": 14172
+    },
+    {
+      "epoch": 0.5154380477870313,
+      "grad_norm": 0.0,
+      "learning_rate": 9.985865664633116e-06,
+      "loss": 0.8945,
+      "step": 14173
+    },
+    {
+      "epoch": 0.5154744153907699,
+      "grad_norm": 0.0,
+      "learning_rate": 9.984687804237689e-06,
+      "loss": 1.1028,
+      "step": 14174
+    },
+    {
+      "epoch": 0.5155107829945085,
+      "grad_norm": 0.0,
+      "learning_rate": 9.983509944054693e-06,
+      "loss": 1.0659,
+      "step": 14175
+    },
+    {
+      "epoch": 0.5155471505982471,
+      "grad_norm": 0.0,
+      "learning_rate": 9.982332084100478e-06,
+      "loss": 1.0113,
+      "step": 14176
+    },
+    {
+      "epoch": 0.5155835182019857,
+      "grad_norm": 0.0,
+      "learning_rate": 9.981154224391377e-06,
+      "loss": 1.0659,
+      "step": 14177
+    },
+    {
+      "epoch": 0.5156198858057243,
+      "grad_norm": 0.0,
+      "learning_rate": 9.979976364943739e-06,
+      "loss": 1.0272,
+      "step": 14178
+    },
+    {
+      "epoch": 0.5156562534094629,
+      "grad_norm": 0.0,
+      "learning_rate": 9.978798505773894e-06,
+      "loss": 1.1552,
+      "step": 14179
+    },
+    {
+      "epoch": 0.5156926210132015,
+      "grad_norm": 0.0,
+      "learning_rate": 9.977620646898195e-06,
+      "loss": 0.9169,
+      "step": 14180
+    },
+    {
+      "epoch": 0.51572898861694,
+      "grad_norm": 0.0,
+      "learning_rate": 9.976442788332976e-06,
+      "loss": 1.042,
+      "step": 14181
+    },
+    {
+      "epoch": 0.5157653562206787,
+      "grad_norm": 0.0,
+      "learning_rate": 9.97526493009458e-06,
+      "loss": 0.902,
+      "step": 14182
+    },
+    {
+      "epoch": 0.5158017238244172,
+      "grad_norm": 0.0,
+      "learning_rate": 9.97408707219935e-06,
+      "loss": 1.0291,
+      "step": 14183
+    },
+    {
+      "epoch": 0.5158380914281558,
+      "grad_norm": 0.0,
+      "learning_rate": 9.97290921466362e-06,
+      "loss": 0.9415,
+      "step": 14184
+    },
+    {
+      "epoch": 0.5158744590318944,
+      "grad_norm": 0.0,
+      "learning_rate": 9.971731357503741e-06,
+      "loss": 1.028,
+      "step": 14185
+    },
+    {
+      "epoch": 0.5159108266356329,
+      "grad_norm": 0.0,
+      "learning_rate": 9.970553500736045e-06,
+      "loss": 1.1049,
+      "step": 14186
+    },
+    {
+      "epoch": 0.5159471942393715,
+      "grad_norm": 0.0,
+      "learning_rate": 9.96937564437688e-06,
+      "loss": 1.0628,
+      "step": 14187
+    },
+    {
+      "epoch": 0.5159835618431101,
+      "grad_norm": 0.0,
+      "learning_rate": 9.968197788442585e-06,
+      "loss": 1.1213,
+      "step": 14188
+    },
+    {
+      "epoch": 0.5160199294468487,
+      "grad_norm": 0.0,
+      "learning_rate": 9.9670199329495e-06,
+      "loss": 1.1037,
+      "step": 14189
+    },
+    {
+      "epoch": 0.5160562970505873,
+      "grad_norm": 0.0,
+      "learning_rate": 9.965842077913963e-06,
+      "loss": 0.9926,
+      "step": 14190
+    },
+    {
+      "epoch": 0.5160926646543259,
+      "grad_norm": 0.0,
+      "learning_rate": 9.964664223352327e-06,
+      "loss": 0.943,
+      "step": 14191
+    },
+    {
+      "epoch": 0.5161290322580645,
+      "grad_norm": 0.0,
+      "learning_rate": 9.963486369280919e-06,
+      "loss": 0.988,
+      "step": 14192
+    },
+    {
+      "epoch": 0.5161653998618031,
+      "grad_norm": 0.0,
+      "learning_rate": 9.962308515716086e-06,
+      "loss": 1.017,
+      "step": 14193
+    },
+    {
+      "epoch": 0.5162017674655417,
+      "grad_norm": 0.0,
+      "learning_rate": 9.961130662674169e-06,
+      "loss": 1.0408,
+      "step": 14194
+    },
+    {
+      "epoch": 0.5162381350692803,
+      "grad_norm": 0.0,
+      "learning_rate": 9.95995281017151e-06,
+      "loss": 1.033,
+      "step": 14195
+    },
+    {
+      "epoch": 0.5162745026730189,
+      "grad_norm": 0.0,
+      "learning_rate": 9.958774958224447e-06,
+      "loss": 1.0524,
+      "step": 14196
+    },
+    {
+      "epoch": 0.5163108702767575,
+      "grad_norm": 0.0,
+      "learning_rate": 9.957597106849325e-06,
+      "loss": 1.088,
+      "step": 14197
+    },
+    {
+      "epoch": 0.5163472378804961,
+      "grad_norm": 0.0,
+      "learning_rate": 9.95641925606248e-06,
+      "loss": 1.1225,
+      "step": 14198
+    },
+    {
+      "epoch": 0.5163836054842347,
+      "grad_norm": 0.0,
+      "learning_rate": 9.955241405880261e-06,
+      "loss": 1.0289,
+      "step": 14199
+    },
+    {
+      "epoch": 0.5164199730879733,
+      "grad_norm": 0.0,
+      "learning_rate": 9.954063556319004e-06,
+      "loss": 0.8837,
+      "step": 14200
+    },
+    {
+      "epoch": 0.5164563406917119,
+      "grad_norm": 0.0,
+      "learning_rate": 9.952885707395044e-06,
+      "loss": 0.9804,
+      "step": 14201
+    },
+    {
+      "epoch": 0.5164927082954505,
+      "grad_norm": 0.0,
+      "learning_rate": 9.951707859124733e-06,
+      "loss": 0.9848,
+      "step": 14202
+    },
+    {
+      "epoch": 0.516529075899189,
+      "grad_norm": 0.0,
+      "learning_rate": 9.950530011524404e-06,
+      "loss": 1.0242,
+      "step": 14203
+    },
+    {
+      "epoch": 0.5165654435029275,
+      "grad_norm": 0.0,
+      "learning_rate": 9.949352164610405e-06,
+      "loss": 0.9499,
+      "step": 14204
+    },
+    {
+      "epoch": 0.5166018111066661,
+      "grad_norm": 0.0,
+      "learning_rate": 9.948174318399068e-06,
+      "loss": 1.0139,
+      "step": 14205
+    },
+    {
+      "epoch": 0.5166381787104047,
+      "grad_norm": 0.0,
+      "learning_rate": 9.946996472906742e-06,
+      "loss": 1.0551,
+      "step": 14206
+    },
+    {
+      "epoch": 0.5166745463141433,
+      "grad_norm": 0.0,
+      "learning_rate": 9.945818628149762e-06,
+      "loss": 0.9401,
+      "step": 14207
+    },
+    {
+      "epoch": 0.5167109139178819,
+      "grad_norm": 0.0,
+      "learning_rate": 9.944640784144474e-06,
+      "loss": 1.0721,
+      "step": 14208
+    },
+    {
+      "epoch": 0.5167472815216205,
+      "grad_norm": 0.0,
+      "learning_rate": 9.943462940907213e-06,
+      "loss": 0.9569,
+      "step": 14209
+    },
+    {
+      "epoch": 0.5167836491253591,
+      "grad_norm": 0.0,
+      "learning_rate": 9.942285098454331e-06,
+      "loss": 1.0399,
+      "step": 14210
+    },
+    {
+      "epoch": 0.5168200167290977,
+      "grad_norm": 0.0,
+      "learning_rate": 9.941107256802153e-06,
+      "loss": 1.2013,
+      "step": 14211
+    },
+    {
+      "epoch": 0.5168563843328363,
+      "grad_norm": 0.0,
+      "learning_rate": 9.939929415967034e-06,
+      "loss": 1.053,
+      "step": 14212
+    },
+    {
+      "epoch": 0.5168927519365749,
+      "grad_norm": 0.0,
+      "learning_rate": 9.938751575965306e-06,
+      "loss": 0.9785,
+      "step": 14213
+    },
+    {
+      "epoch": 0.5169291195403135,
+      "grad_norm": 0.0,
+      "learning_rate": 9.937573736813314e-06,
+      "loss": 1.157,
+      "step": 14214
+    },
+    {
+      "epoch": 0.5169654871440521,
+      "grad_norm": 0.0,
+      "learning_rate": 9.936395898527397e-06,
+      "loss": 0.9807,
+      "step": 14215
+    },
+    {
+      "epoch": 0.5170018547477907,
+      "grad_norm": 0.0,
+      "learning_rate": 9.935218061123898e-06,
+      "loss": 1.0088,
+      "step": 14216
+    },
+    {
+      "epoch": 0.5170382223515293,
+      "grad_norm": 0.0,
+      "learning_rate": 9.934040224619154e-06,
+      "loss": 1.0519,
+      "step": 14217
+    },
+    {
+      "epoch": 0.5170745899552679,
+      "grad_norm": 0.0,
+      "learning_rate": 9.932862389029514e-06,
+      "loss": 1.0886,
+      "step": 14218
+    },
+    {
+      "epoch": 0.5171109575590065,
+      "grad_norm": 0.0,
+      "learning_rate": 9.931684554371309e-06,
+      "loss": 0.9958,
+      "step": 14219
+    },
+    {
+      "epoch": 0.517147325162745,
+      "grad_norm": 0.0,
+      "learning_rate": 9.930506720660884e-06,
+      "loss": 0.9681,
+      "step": 14220
+    },
+    {
+      "epoch": 0.5171836927664836,
+      "grad_norm": 0.0,
+      "learning_rate": 9.92932888791458e-06,
+      "loss": 1.0138,
+      "step": 14221
+    },
+    {
+      "epoch": 0.5172200603702222,
+      "grad_norm": 0.0,
+      "learning_rate": 9.928151056148735e-06,
+      "loss": 0.9732,
+      "step": 14222
+    },
+    {
+      "epoch": 0.5172564279739608,
+      "grad_norm": 0.0,
+      "learning_rate": 9.926973225379698e-06,
+      "loss": 0.9214,
+      "step": 14223
+    },
+    {
+      "epoch": 0.5172927955776994,
+      "grad_norm": 0.0,
+      "learning_rate": 9.925795395623798e-06,
+      "loss": 0.9976,
+      "step": 14224
+    },
+    {
+      "epoch": 0.517329163181438,
+      "grad_norm": 0.0,
+      "learning_rate": 9.924617566897386e-06,
+      "loss": 1.065,
+      "step": 14225
+    },
+    {
+      "epoch": 0.5173655307851766,
+      "grad_norm": 0.0,
+      "learning_rate": 9.923439739216796e-06,
+      "loss": 1.0204,
+      "step": 14226
+    },
+    {
+      "epoch": 0.5174018983889151,
+      "grad_norm": 0.0,
+      "learning_rate": 9.922261912598371e-06,
+      "loss": 1.0261,
+      "step": 14227
+    },
+    {
+      "epoch": 0.5174382659926537,
+      "grad_norm": 0.0,
+      "learning_rate": 9.92108408705845e-06,
+      "loss": 0.9169,
+      "step": 14228
+    },
+    {
+      "epoch": 0.5174746335963923,
+      "grad_norm": 0.0,
+      "learning_rate": 9.919906262613378e-06,
+      "loss": 1.0539,
+      "step": 14229
+    },
+    {
+      "epoch": 0.5175110012001309,
+      "grad_norm": 0.0,
+      "learning_rate": 9.91872843927949e-06,
+      "loss": 0.8867,
+      "step": 14230
+    },
+    {
+      "epoch": 0.5175473688038695,
+      "grad_norm": 0.0,
+      "learning_rate": 9.917550617073136e-06,
+      "loss": 1.112,
+      "step": 14231
+    },
+    {
+      "epoch": 0.5175837364076081,
+      "grad_norm": 0.0,
+      "learning_rate": 9.916372796010644e-06,
+      "loss": 0.9867,
+      "step": 14232
+    },
+    {
+      "epoch": 0.5176201040113467,
+      "grad_norm": 0.0,
+      "learning_rate": 9.915194976108363e-06,
+      "loss": 0.8925,
+      "step": 14233
+    },
+    {
+      "epoch": 0.5176564716150853,
+      "grad_norm": 0.0,
+      "learning_rate": 9.91401715738263e-06,
+      "loss": 0.9503,
+      "step": 14234
+    },
+    {
+      "epoch": 0.5176928392188239,
+      "grad_norm": 0.0,
+      "learning_rate": 9.912839339849789e-06,
+      "loss": 0.9648,
+      "step": 14235
+    },
+    {
+      "epoch": 0.5177292068225625,
+      "grad_norm": 0.0,
+      "learning_rate": 9.911661523526176e-06,
+      "loss": 1.0893,
+      "step": 14236
+    },
+    {
+      "epoch": 0.517765574426301,
+      "grad_norm": 0.0,
+      "learning_rate": 9.91048370842814e-06,
+      "loss": 1.1383,
+      "step": 14237
+    },
+    {
+      "epoch": 0.5178019420300396,
+      "grad_norm": 0.0,
+      "learning_rate": 9.909305894572011e-06,
+      "loss": 1.0843,
+      "step": 14238
+    },
+    {
+      "epoch": 0.5178383096337782,
+      "grad_norm": 0.0,
+      "learning_rate": 9.908128081974133e-06,
+      "loss": 0.9371,
+      "step": 14239
+    },
+    {
+      "epoch": 0.5178746772375168,
+      "grad_norm": 0.0,
+      "learning_rate": 9.906950270650849e-06,
+      "loss": 0.9135,
+      "step": 14240
+    },
+    {
+      "epoch": 0.5179110448412554,
+      "grad_norm": 0.0,
+      "learning_rate": 9.905772460618496e-06,
+      "loss": 1.0034,
+      "step": 14241
+    },
+    {
+      "epoch": 0.517947412444994,
+      "grad_norm": 0.0,
+      "learning_rate": 9.904594651893423e-06,
+      "loss": 1.1074,
+      "step": 14242
+    },
+    {
+      "epoch": 0.5179837800487326,
+      "grad_norm": 0.0,
+      "learning_rate": 9.903416844491955e-06,
+      "loss": 0.8694,
+      "step": 14243
+    },
+    {
+      "epoch": 0.5180201476524712,
+      "grad_norm": 0.0,
+      "learning_rate": 9.902239038430447e-06,
+      "loss": 1.1138,
+      "step": 14244
+    },
+    {
+      "epoch": 0.5180565152562098,
+      "grad_norm": 0.0,
+      "learning_rate": 9.90106123372523e-06,
+      "loss": 1.01,
+      "step": 14245
+    },
+    {
+      "epoch": 0.5180928828599484,
+      "grad_norm": 0.0,
+      "learning_rate": 9.899883430392652e-06,
+      "loss": 0.9651,
+      "step": 14246
+    },
+    {
+      "epoch": 0.518129250463687,
+      "grad_norm": 0.0,
+      "learning_rate": 9.898705628449044e-06,
+      "loss": 1.0921,
+      "step": 14247
+    },
+    {
+      "epoch": 0.5181656180674256,
+      "grad_norm": 0.0,
+      "learning_rate": 9.897527827910757e-06,
+      "loss": 0.9444,
+      "step": 14248
+    },
+    {
+      "epoch": 0.5182019856711642,
+      "grad_norm": 0.0,
+      "learning_rate": 9.896350028794122e-06,
+      "loss": 1.046,
+      "step": 14249
+    },
+    {
+      "epoch": 0.5182383532749028,
+      "grad_norm": 0.0,
+      "learning_rate": 9.895172231115489e-06,
+      "loss": 1.0818,
+      "step": 14250
+    },
+    {
+      "epoch": 0.5182747208786413,
+      "grad_norm": 0.0,
+      "learning_rate": 9.893994434891185e-06,
+      "loss": 0.9058,
+      "step": 14251
+    },
+    {
+      "epoch": 0.51831108848238,
+      "grad_norm": 0.0,
+      "learning_rate": 9.892816640137564e-06,
+      "loss": 0.9604,
+      "step": 14252
+    },
+    {
+      "epoch": 0.5183474560861185,
+      "grad_norm": 0.0,
+      "learning_rate": 9.891638846870955e-06,
+      "loss": 0.9364,
+      "step": 14253
+    },
+    {
+      "epoch": 0.518383823689857,
+      "grad_norm": 0.0,
+      "learning_rate": 9.890461055107707e-06,
+      "loss": 1.0745,
+      "step": 14254
+    },
+    {
+      "epoch": 0.5184201912935956,
+      "grad_norm": 0.0,
+      "learning_rate": 9.889283264864154e-06,
+      "loss": 0.9782,
+      "step": 14255
+    },
+    {
+      "epoch": 0.5184565588973342,
+      "grad_norm": 0.0,
+      "learning_rate": 9.88810547615664e-06,
+      "loss": 1.0592,
+      "step": 14256
+    },
+    {
+      "epoch": 0.5184929265010728,
+      "grad_norm": 0.0,
+      "learning_rate": 9.886927689001508e-06,
+      "loss": 1.0896,
+      "step": 14257
+    },
+    {
+      "epoch": 0.5185292941048114,
+      "grad_norm": 0.0,
+      "learning_rate": 9.885749903415089e-06,
+      "loss": 1.0311,
+      "step": 14258
+    },
+    {
+      "epoch": 0.51856566170855,
+      "grad_norm": 0.0,
+      "learning_rate": 9.884572119413728e-06,
+      "loss": 0.9684,
+      "step": 14259
+    },
+    {
+      "epoch": 0.5186020293122886,
+      "grad_norm": 0.0,
+      "learning_rate": 9.883394337013764e-06,
+      "loss": 1.0257,
+      "step": 14260
+    },
+    {
+      "epoch": 0.5186383969160272,
+      "grad_norm": 0.0,
+      "learning_rate": 9.882216556231541e-06,
+      "loss": 1.045,
+      "step": 14261
+    },
+    {
+      "epoch": 0.5186747645197658,
+      "grad_norm": 0.0,
+      "learning_rate": 9.881038777083394e-06,
+      "loss": 1.0995,
+      "step": 14262
+    },
+    {
+      "epoch": 0.5187111321235044,
+      "grad_norm": 0.0,
+      "learning_rate": 9.879860999585669e-06,
+      "loss": 1.1011,
+      "step": 14263
+    },
+    {
+      "epoch": 0.518747499727243,
+      "grad_norm": 0.0,
+      "learning_rate": 9.878683223754698e-06,
+      "loss": 0.8818,
+      "step": 14264
+    },
+    {
+      "epoch": 0.5187838673309816,
+      "grad_norm": 0.0,
+      "learning_rate": 9.877505449606827e-06,
+      "loss": 0.9172,
+      "step": 14265
+    },
+    {
+      "epoch": 0.5188202349347202,
+      "grad_norm": 0.0,
+      "learning_rate": 9.876327677158392e-06,
+      "loss": 1.0035,
+      "step": 14266
+    },
+    {
+      "epoch": 0.5188566025384588,
+      "grad_norm": 0.0,
+      "learning_rate": 9.875149906425737e-06,
+      "loss": 1.0679,
+      "step": 14267
+    },
+    {
+      "epoch": 0.5188929701421974,
+      "grad_norm": 0.0,
+      "learning_rate": 9.873972137425198e-06,
+      "loss": 1.0108,
+      "step": 14268
+    },
+    {
+      "epoch": 0.518929337745936,
+      "grad_norm": 0.0,
+      "learning_rate": 9.872794370173119e-06,
+      "loss": 1.0831,
+      "step": 14269
+    },
+    {
+      "epoch": 0.5189657053496745,
+      "grad_norm": 0.0,
+      "learning_rate": 9.871616604685836e-06,
+      "loss": 0.9238,
+      "step": 14270
+    },
+    {
+      "epoch": 0.519002072953413,
+      "grad_norm": 0.0,
+      "learning_rate": 9.870438840979694e-06,
+      "loss": 1.023,
+      "step": 14271
+    },
+    {
+      "epoch": 0.5190384405571516,
+      "grad_norm": 0.0,
+      "learning_rate": 9.869261079071026e-06,
+      "loss": 0.9605,
+      "step": 14272
+    },
+    {
+      "epoch": 0.5190748081608902,
+      "grad_norm": 0.0,
+      "learning_rate": 9.868083318976175e-06,
+      "loss": 0.8996,
+      "step": 14273
+    },
+    {
+      "epoch": 0.5191111757646288,
+      "grad_norm": 0.0,
+      "learning_rate": 9.86690556071148e-06,
+      "loss": 1.0446,
+      "step": 14274
+    },
+    {
+      "epoch": 0.5191475433683674,
+      "grad_norm": 0.0,
+      "learning_rate": 9.865727804293284e-06,
+      "loss": 1.1711,
+      "step": 14275
+    },
+    {
+      "epoch": 0.519183910972106,
+      "grad_norm": 0.0,
+      "learning_rate": 9.864550049737926e-06,
+      "loss": 0.9132,
+      "step": 14276
+    },
+    {
+      "epoch": 0.5192202785758446,
+      "grad_norm": 0.0,
+      "learning_rate": 9.86337229706174e-06,
+      "loss": 0.9424,
+      "step": 14277
+    },
+    {
+      "epoch": 0.5192566461795832,
+      "grad_norm": 0.0,
+      "learning_rate": 9.86219454628107e-06,
+      "loss": 1.0399,
+      "step": 14278
+    },
+    {
+      "epoch": 0.5192930137833218,
+      "grad_norm": 0.0,
+      "learning_rate": 9.861016797412254e-06,
+      "loss": 0.9391,
+      "step": 14279
+    },
+    {
+      "epoch": 0.5193293813870604,
+      "grad_norm": 0.0,
+      "learning_rate": 9.859839050471636e-06,
+      "loss": 0.9464,
+      "step": 14280
+    },
+    {
+      "epoch": 0.519365748990799,
+      "grad_norm": 0.0,
+      "learning_rate": 9.85866130547555e-06,
+      "loss": 0.9355,
+      "step": 14281
+    },
+    {
+      "epoch": 0.5194021165945376,
+      "grad_norm": 0.0,
+      "learning_rate": 9.857483562440341e-06,
+      "loss": 0.9292,
+      "step": 14282
+    },
+    {
+      "epoch": 0.5194384841982762,
+      "grad_norm": 0.0,
+      "learning_rate": 9.85630582138234e-06,
+      "loss": 1.0602,
+      "step": 14283
+    },
+    {
+      "epoch": 0.5194748518020148,
+      "grad_norm": 0.0,
+      "learning_rate": 9.855128082317897e-06,
+      "loss": 1.0776,
+      "step": 14284
+    },
+    {
+      "epoch": 0.5195112194057534,
+      "grad_norm": 0.0,
+      "learning_rate": 9.853950345263342e-06,
+      "loss": 0.9587,
+      "step": 14285
+    },
+    {
+      "epoch": 0.519547587009492,
+      "grad_norm": 0.0,
+      "learning_rate": 9.852772610235022e-06,
+      "loss": 0.9094,
+      "step": 14286
+    },
+    {
+      "epoch": 0.5195839546132305,
+      "grad_norm": 0.0,
+      "learning_rate": 9.851594877249271e-06,
+      "loss": 1.1232,
+      "step": 14287
+    },
+    {
+      "epoch": 0.5196203222169691,
+      "grad_norm": 0.0,
+      "learning_rate": 9.850417146322432e-06,
+      "loss": 1.0444,
+      "step": 14288
+    },
+    {
+      "epoch": 0.5196566898207077,
+      "grad_norm": 0.0,
+      "learning_rate": 9.849239417470841e-06,
+      "loss": 1.0754,
+      "step": 14289
+    },
+    {
+      "epoch": 0.5196930574244463,
+      "grad_norm": 0.0,
+      "learning_rate": 9.848061690710843e-06,
+      "loss": 1.0952,
+      "step": 14290
+    },
+    {
+      "epoch": 0.5197294250281849,
+      "grad_norm": 0.0,
+      "learning_rate": 9.84688396605877e-06,
+      "loss": 0.9619,
+      "step": 14291
+    },
+    {
+      "epoch": 0.5197657926319235,
+      "grad_norm": 0.0,
+      "learning_rate": 9.845706243530967e-06,
+      "loss": 0.9148,
+      "step": 14292
+    },
+    {
+      "epoch": 0.5198021602356621,
+      "grad_norm": 0.0,
+      "learning_rate": 9.844528523143767e-06,
+      "loss": 0.9861,
+      "step": 14293
+    },
+    {
+      "epoch": 0.5198385278394007,
+      "grad_norm": 0.0,
+      "learning_rate": 9.843350804913517e-06,
+      "loss": 1.005,
+      "step": 14294
+    },
+    {
+      "epoch": 0.5198748954431393,
+      "grad_norm": 0.0,
+      "learning_rate": 9.84217308885655e-06,
+      "loss": 1.0498,
+      "step": 14295
+    },
+    {
+      "epoch": 0.5199112630468778,
+      "grad_norm": 0.0,
+      "learning_rate": 9.84099537498921e-06,
+      "loss": 1.0168,
+      "step": 14296
+    },
+    {
+      "epoch": 0.5199476306506164,
+      "grad_norm": 0.0,
+      "learning_rate": 9.839817663327835e-06,
+      "loss": 0.9862,
+      "step": 14297
+    },
+    {
+      "epoch": 0.519983998254355,
+      "grad_norm": 0.0,
+      "learning_rate": 9.838639953888758e-06,
+      "loss": 0.9946,
+      "step": 14298
+    },
+    {
+      "epoch": 0.5200203658580936,
+      "grad_norm": 0.0,
+      "learning_rate": 9.837462246688327e-06,
+      "loss": 0.9378,
+      "step": 14299
+    },
+    {
+      "epoch": 0.5200567334618322,
+      "grad_norm": 0.0,
+      "learning_rate": 9.836284541742871e-06,
+      "loss": 1.0977,
+      "step": 14300
+    },
+    {
+      "epoch": 0.5200931010655708,
+      "grad_norm": 0.0,
+      "learning_rate": 9.83510683906874e-06,
+      "loss": 1.0344,
+      "step": 14301
+    },
+    {
+      "epoch": 0.5201294686693094,
+      "grad_norm": 0.0,
+      "learning_rate": 9.833929138682263e-06,
+      "loss": 0.91,
+      "step": 14302
+    },
+    {
+      "epoch": 0.520165836273048,
+      "grad_norm": 0.0,
+      "learning_rate": 9.832751440599791e-06,
+      "loss": 0.9178,
+      "step": 14303
+    },
+    {
+      "epoch": 0.5202022038767865,
+      "grad_norm": 0.0,
+      "learning_rate": 9.831573744837648e-06,
+      "loss": 0.9703,
+      "step": 14304
+    },
+    {
+      "epoch": 0.5202385714805251,
+      "grad_norm": 0.0,
+      "learning_rate": 9.830396051412185e-06,
+      "loss": 1.0836,
+      "step": 14305
+    },
+    {
+      "epoch": 0.5202749390842637,
+      "grad_norm": 0.0,
+      "learning_rate": 9.829218360339732e-06,
+      "loss": 1.0294,
+      "step": 14306
+    },
+    {
+      "epoch": 0.5203113066880023,
+      "grad_norm": 0.0,
+      "learning_rate": 9.828040671636635e-06,
+      "loss": 0.962,
+      "step": 14307
+    },
+    {
+      "epoch": 0.5203476742917409,
+      "grad_norm": 0.0,
+      "learning_rate": 9.826862985319228e-06,
+      "loss": 0.9996,
+      "step": 14308
+    },
+    {
+      "epoch": 0.5203840418954795,
+      "grad_norm": 0.0,
+      "learning_rate": 9.825685301403854e-06,
+      "loss": 0.9292,
+      "step": 14309
+    },
+    {
+      "epoch": 0.5204204094992181,
+      "grad_norm": 0.0,
+      "learning_rate": 9.824507619906844e-06,
+      "loss": 1.0714,
+      "step": 14310
+    },
+    {
+      "epoch": 0.5204567771029567,
+      "grad_norm": 0.0,
+      "learning_rate": 9.823329940844549e-06,
+      "loss": 1.0336,
+      "step": 14311
+    },
+    {
+      "epoch": 0.5204931447066953,
+      "grad_norm": 0.0,
+      "learning_rate": 9.822152264233294e-06,
+      "loss": 1.0476,
+      "step": 14312
+    },
+    {
+      "epoch": 0.5205295123104339,
+      "grad_norm": 0.0,
+      "learning_rate": 9.820974590089428e-06,
+      "loss": 0.968,
+      "step": 14313
+    },
+    {
+      "epoch": 0.5205658799141725,
+      "grad_norm": 0.0,
+      "learning_rate": 9.819796918429283e-06,
+      "loss": 1.0167,
+      "step": 14314
+    },
+    {
+      "epoch": 0.5206022475179111,
+      "grad_norm": 0.0,
+      "learning_rate": 9.8186192492692e-06,
+      "loss": 0.943,
+      "step": 14315
+    },
+    {
+      "epoch": 0.5206386151216497,
+      "grad_norm": 0.0,
+      "learning_rate": 9.817441582625523e-06,
+      "loss": 0.8644,
+      "step": 14316
+    },
+    {
+      "epoch": 0.5206749827253883,
+      "grad_norm": 0.0,
+      "learning_rate": 9.81626391851458e-06,
+      "loss": 0.9524,
+      "step": 14317
+    },
+    {
+      "epoch": 0.5207113503291269,
+      "grad_norm": 0.0,
+      "learning_rate": 9.815086256952715e-06,
+      "loss": 1.0267,
+      "step": 14318
+    },
+    {
+      "epoch": 0.5207477179328655,
+      "grad_norm": 0.0,
+      "learning_rate": 9.813908597956265e-06,
+      "loss": 0.9977,
+      "step": 14319
+    },
+    {
+      "epoch": 0.520784085536604,
+      "grad_norm": 0.0,
+      "learning_rate": 9.812730941541572e-06,
+      "loss": 1.0003,
+      "step": 14320
+    },
+    {
+      "epoch": 0.5208204531403425,
+      "grad_norm": 0.0,
+      "learning_rate": 9.811553287724967e-06,
+      "loss": 1.0096,
+      "step": 14321
+    },
+    {
+      "epoch": 0.5208568207440811,
+      "grad_norm": 0.0,
+      "learning_rate": 9.810375636522798e-06,
+      "loss": 1.0415,
+      "step": 14322
+    },
+    {
+      "epoch": 0.5208931883478197,
+      "grad_norm": 0.0,
+      "learning_rate": 9.809197987951394e-06,
+      "loss": 1.0991,
+      "step": 14323
+    },
+    {
+      "epoch": 0.5209295559515583,
+      "grad_norm": 0.0,
+      "learning_rate": 9.808020342027101e-06,
+      "loss": 0.9854,
+      "step": 14324
+    },
+    {
+      "epoch": 0.5209659235552969,
+      "grad_norm": 0.0,
+      "learning_rate": 9.80684269876625e-06,
+      "loss": 1.1151,
+      "step": 14325
+    },
+    {
+      "epoch": 0.5210022911590355,
+      "grad_norm": 0.0,
+      "learning_rate": 9.805665058185183e-06,
+      "loss": 0.9048,
+      "step": 14326
+    },
+    {
+      "epoch": 0.5210386587627741,
+      "grad_norm": 0.0,
+      "learning_rate": 9.804487420300237e-06,
+      "loss": 0.8985,
+      "step": 14327
+    },
+    {
+      "epoch": 0.5210750263665127,
+      "grad_norm": 0.0,
+      "learning_rate": 9.803309785127752e-06,
+      "loss": 0.9836,
+      "step": 14328
+    },
+    {
+      "epoch": 0.5211113939702513,
+      "grad_norm": 0.0,
+      "learning_rate": 9.802132152684063e-06,
+      "loss": 0.9565,
+      "step": 14329
+    },
+    {
+      "epoch": 0.5211477615739899,
+      "grad_norm": 0.0,
+      "learning_rate": 9.800954522985515e-06,
+      "loss": 0.9676,
+      "step": 14330
+    },
+    {
+      "epoch": 0.5211841291777285,
+      "grad_norm": 0.0,
+      "learning_rate": 9.799776896048435e-06,
+      "loss": 0.9219,
+      "step": 14331
+    },
+    {
+      "epoch": 0.5212204967814671,
+      "grad_norm": 0.0,
+      "learning_rate": 9.798599271889169e-06,
+      "loss": 0.978,
+      "step": 14332
+    },
+    {
+      "epoch": 0.5212568643852057,
+      "grad_norm": 0.0,
+      "learning_rate": 9.79742165052405e-06,
+      "loss": 0.9801,
+      "step": 14333
+    },
+    {
+      "epoch": 0.5212932319889443,
+      "grad_norm": 0.0,
+      "learning_rate": 9.79624403196942e-06,
+      "loss": 0.9384,
+      "step": 14334
+    },
+    {
+      "epoch": 0.5213295995926829,
+      "grad_norm": 0.0,
+      "learning_rate": 9.795066416241618e-06,
+      "loss": 1.1178,
+      "step": 14335
+    },
+    {
+      "epoch": 0.5213659671964215,
+      "grad_norm": 0.0,
+      "learning_rate": 9.793888803356974e-06,
+      "loss": 1.0908,
+      "step": 14336
+    },
+    {
+      "epoch": 0.5214023348001601,
+      "grad_norm": 0.0,
+      "learning_rate": 9.792711193331833e-06,
+      "loss": 0.8733,
+      "step": 14337
+    },
+    {
+      "epoch": 0.5214387024038986,
+      "grad_norm": 0.0,
+      "learning_rate": 9.791533586182527e-06,
+      "loss": 0.8948,
+      "step": 14338
+    },
+    {
+      "epoch": 0.5214750700076372,
+      "grad_norm": 0.0,
+      "learning_rate": 9.7903559819254e-06,
+      "loss": 1.0665,
+      "step": 14339
+    },
+    {
+      "epoch": 0.5215114376113757,
+      "grad_norm": 0.0,
+      "learning_rate": 9.789178380576783e-06,
+      "loss": 1.0554,
+      "step": 14340
+    },
+    {
+      "epoch": 0.5215478052151143,
+      "grad_norm": 0.0,
+      "learning_rate": 9.78800078215302e-06,
+      "loss": 0.9758,
+      "step": 14341
+    },
+    {
+      "epoch": 0.5215841728188529,
+      "grad_norm": 0.0,
+      "learning_rate": 9.786823186670443e-06,
+      "loss": 1.0057,
+      "step": 14342
+    },
+    {
+      "epoch": 0.5216205404225915,
+      "grad_norm": 0.0,
+      "learning_rate": 9.785645594145396e-06,
+      "loss": 1.0077,
+      "step": 14343
+    },
+    {
+      "epoch": 0.5216569080263301,
+      "grad_norm": 0.0,
+      "learning_rate": 9.784468004594207e-06,
+      "loss": 0.9641,
+      "step": 14344
+    },
+    {
+      "epoch": 0.5216932756300687,
+      "grad_norm": 0.0,
+      "learning_rate": 9.783290418033222e-06,
+      "loss": 1.0171,
+      "step": 14345
+    },
+    {
+      "epoch": 0.5217296432338073,
+      "grad_norm": 0.0,
+      "learning_rate": 9.782112834478772e-06,
+      "loss": 0.9858,
+      "step": 14346
+    },
+    {
+      "epoch": 0.5217660108375459,
+      "grad_norm": 0.0,
+      "learning_rate": 9.7809352539472e-06,
+      "loss": 0.9305,
+      "step": 14347
+    },
+    {
+      "epoch": 0.5218023784412845,
+      "grad_norm": 0.0,
+      "learning_rate": 9.779757676454838e-06,
+      "loss": 0.9552,
+      "step": 14348
+    },
+    {
+      "epoch": 0.5218387460450231,
+      "grad_norm": 0.0,
+      "learning_rate": 9.77858010201803e-06,
+      "loss": 1.1198,
+      "step": 14349
+    },
+    {
+      "epoch": 0.5218751136487617,
+      "grad_norm": 0.0,
+      "learning_rate": 9.777402530653103e-06,
+      "loss": 1.0195,
+      "step": 14350
+    },
+    {
+      "epoch": 0.5219114812525003,
+      "grad_norm": 0.0,
+      "learning_rate": 9.776224962376407e-06,
+      "loss": 1.073,
+      "step": 14351
+    },
+    {
+      "epoch": 0.5219478488562389,
+      "grad_norm": 0.0,
+      "learning_rate": 9.775047397204267e-06,
+      "loss": 1.0427,
+      "step": 14352
+    },
+    {
+      "epoch": 0.5219842164599775,
+      "grad_norm": 0.0,
+      "learning_rate": 9.773869835153029e-06,
+      "loss": 0.9622,
+      "step": 14353
+    },
+    {
+      "epoch": 0.522020584063716,
+      "grad_norm": 0.0,
+      "learning_rate": 9.772692276239025e-06,
+      "loss": 1.0528,
+      "step": 14354
+    },
+    {
+      "epoch": 0.5220569516674546,
+      "grad_norm": 0.0,
+      "learning_rate": 9.77151472047859e-06,
+      "loss": 0.9723,
+      "step": 14355
+    },
+    {
+      "epoch": 0.5220933192711932,
+      "grad_norm": 0.0,
+      "learning_rate": 9.77033716788807e-06,
+      "loss": 1.1179,
+      "step": 14356
+    },
+    {
+      "epoch": 0.5221296868749318,
+      "grad_norm": 0.0,
+      "learning_rate": 9.769159618483789e-06,
+      "loss": 0.9699,
+      "step": 14357
+    },
+    {
+      "epoch": 0.5221660544786704,
+      "grad_norm": 0.0,
+      "learning_rate": 9.767982072282095e-06,
+      "loss": 0.99,
+      "step": 14358
+    },
+    {
+      "epoch": 0.522202422082409,
+      "grad_norm": 0.0,
+      "learning_rate": 9.766804529299317e-06,
+      "loss": 0.9613,
+      "step": 14359
+    },
+    {
+      "epoch": 0.5222387896861476,
+      "grad_norm": 0.0,
+      "learning_rate": 9.7656269895518e-06,
+      "loss": 1.0491,
+      "step": 14360
+    },
+    {
+      "epoch": 0.5222751572898862,
+      "grad_norm": 0.0,
+      "learning_rate": 9.76444945305587e-06,
+      "loss": 1.0409,
+      "step": 14361
+    },
+    {
+      "epoch": 0.5223115248936248,
+      "grad_norm": 0.0,
+      "learning_rate": 9.763271919827878e-06,
+      "loss": 0.9568,
+      "step": 14362
+    },
+    {
+      "epoch": 0.5223478924973634,
+      "grad_norm": 0.0,
+      "learning_rate": 9.762094389884146e-06,
+      "loss": 1.0206,
+      "step": 14363
+    },
+    {
+      "epoch": 0.522384260101102,
+      "grad_norm": 0.0,
+      "learning_rate": 9.76091686324102e-06,
+      "loss": 1.0498,
+      "step": 14364
+    },
+    {
+      "epoch": 0.5224206277048405,
+      "grad_norm": 0.0,
+      "learning_rate": 9.75973933991483e-06,
+      "loss": 0.8192,
+      "step": 14365
+    },
+    {
+      "epoch": 0.5224569953085791,
+      "grad_norm": 0.0,
+      "learning_rate": 9.758561819921918e-06,
+      "loss": 0.9614,
+      "step": 14366
+    },
+    {
+      "epoch": 0.5224933629123177,
+      "grad_norm": 0.0,
+      "learning_rate": 9.757384303278616e-06,
+      "loss": 0.9472,
+      "step": 14367
+    },
+    {
+      "epoch": 0.5225297305160563,
+      "grad_norm": 0.0,
+      "learning_rate": 9.756206790001265e-06,
+      "loss": 0.9796,
+      "step": 14368
+    },
+    {
+      "epoch": 0.5225660981197949,
+      "grad_norm": 0.0,
+      "learning_rate": 9.755029280106196e-06,
+      "loss": 1.042,
+      "step": 14369
+    },
+    {
+      "epoch": 0.5226024657235335,
+      "grad_norm": 0.0,
+      "learning_rate": 9.753851773609754e-06,
+      "loss": 1.0427,
+      "step": 14370
+    },
+    {
+      "epoch": 0.522638833327272,
+      "grad_norm": 0.0,
+      "learning_rate": 9.752674270528264e-06,
+      "loss": 1.0445,
+      "step": 14371
+    },
+    {
+      "epoch": 0.5226752009310106,
+      "grad_norm": 0.0,
+      "learning_rate": 9.75149677087807e-06,
+      "loss": 1.0516,
+      "step": 14372
+    },
+    {
+      "epoch": 0.5227115685347492,
+      "grad_norm": 0.0,
+      "learning_rate": 9.750319274675506e-06,
+      "loss": 0.9169,
+      "step": 14373
+    },
+    {
+      "epoch": 0.5227479361384878,
+      "grad_norm": 0.0,
+      "learning_rate": 9.749141781936907e-06,
+      "loss": 1.0173,
+      "step": 14374
+    },
+    {
+      "epoch": 0.5227843037422264,
+      "grad_norm": 0.0,
+      "learning_rate": 9.747964292678612e-06,
+      "loss": 0.9938,
+      "step": 14375
+    },
+    {
+      "epoch": 0.522820671345965,
+      "grad_norm": 0.0,
+      "learning_rate": 9.74678680691695e-06,
+      "loss": 1.1173,
+      "step": 14376
+    },
+    {
+      "epoch": 0.5228570389497036,
+      "grad_norm": 0.0,
+      "learning_rate": 9.745609324668266e-06,
+      "loss": 0.9318,
+      "step": 14377
+    },
+    {
+      "epoch": 0.5228934065534422,
+      "grad_norm": 0.0,
+      "learning_rate": 9.744431845948888e-06,
+      "loss": 1.0616,
+      "step": 14378
+    },
+    {
+      "epoch": 0.5229297741571808,
+      "grad_norm": 0.0,
+      "learning_rate": 9.743254370775159e-06,
+      "loss": 1.0135,
+      "step": 14379
+    },
+    {
+      "epoch": 0.5229661417609194,
+      "grad_norm": 0.0,
+      "learning_rate": 9.74207689916341e-06,
+      "loss": 1.0462,
+      "step": 14380
+    },
+    {
+      "epoch": 0.523002509364658,
+      "grad_norm": 0.0,
+      "learning_rate": 9.74089943112998e-06,
+      "loss": 0.8969,
+      "step": 14381
+    },
+    {
+      "epoch": 0.5230388769683966,
+      "grad_norm": 0.0,
+      "learning_rate": 9.739721966691196e-06,
+      "loss": 0.9784,
+      "step": 14382
+    },
+    {
+      "epoch": 0.5230752445721352,
+      "grad_norm": 0.0,
+      "learning_rate": 9.73854450586341e-06,
+      "loss": 0.7796,
+      "step": 14383
+    },
+    {
+      "epoch": 0.5231116121758738,
+      "grad_norm": 0.0,
+      "learning_rate": 9.73736704866294e-06,
+      "loss": 1.0106,
+      "step": 14384
+    },
+    {
+      "epoch": 0.5231479797796124,
+      "grad_norm": 0.0,
+      "learning_rate": 9.736189595106133e-06,
+      "loss": 1.0992,
+      "step": 14385
+    },
+    {
+      "epoch": 0.523184347383351,
+      "grad_norm": 0.0,
+      "learning_rate": 9.73501214520932e-06,
+      "loss": 0.9801,
+      "step": 14386
+    },
+    {
+      "epoch": 0.5232207149870896,
+      "grad_norm": 0.0,
+      "learning_rate": 9.733834698988838e-06,
+      "loss": 1.065,
+      "step": 14387
+    },
+    {
+      "epoch": 0.523257082590828,
+      "grad_norm": 0.0,
+      "learning_rate": 9.732657256461022e-06,
+      "loss": 0.9096,
+      "step": 14388
+    },
+    {
+      "epoch": 0.5232934501945666,
+      "grad_norm": 0.0,
+      "learning_rate": 9.731479817642211e-06,
+      "loss": 0.9859,
+      "step": 14389
+    },
+    {
+      "epoch": 0.5233298177983052,
+      "grad_norm": 0.0,
+      "learning_rate": 9.730302382548728e-06,
+      "loss": 0.886,
+      "step": 14390
+    },
+    {
+      "epoch": 0.5233661854020438,
+      "grad_norm": 0.0,
+      "learning_rate": 9.729124951196926e-06,
+      "loss": 1.1462,
+      "step": 14391
+    },
+    {
+      "epoch": 0.5234025530057824,
+      "grad_norm": 0.0,
+      "learning_rate": 9.727947523603127e-06,
+      "loss": 0.8492,
+      "step": 14392
+    },
+    {
+      "epoch": 0.523438920609521,
+      "grad_norm": 0.0,
+      "learning_rate": 9.726770099783669e-06,
+      "loss": 1.0303,
+      "step": 14393
+    },
+    {
+      "epoch": 0.5234752882132596,
+      "grad_norm": 0.0,
+      "learning_rate": 9.725592679754891e-06,
+      "loss": 1.0679,
+      "step": 14394
+    },
+    {
+      "epoch": 0.5235116558169982,
+      "grad_norm": 0.0,
+      "learning_rate": 9.724415263533121e-06,
+      "loss": 1.0427,
+      "step": 14395
+    },
+    {
+      "epoch": 0.5235480234207368,
+      "grad_norm": 0.0,
+      "learning_rate": 9.723237851134704e-06,
+      "loss": 1.0084,
+      "step": 14396
+    },
+    {
+      "epoch": 0.5235843910244754,
+      "grad_norm": 0.0,
+      "learning_rate": 9.722060442575964e-06,
+      "loss": 1.0591,
+      "step": 14397
+    },
+    {
+      "epoch": 0.523620758628214,
+      "grad_norm": 0.0,
+      "learning_rate": 9.720883037873244e-06,
+      "loss": 0.993,
+      "step": 14398
+    },
+    {
+      "epoch": 0.5236571262319526,
+      "grad_norm": 0.0,
+      "learning_rate": 9.719705637042873e-06,
+      "loss": 0.9628,
+      "step": 14399
+    },
+    {
+      "epoch": 0.5236934938356912,
+      "grad_norm": 0.0,
+      "learning_rate": 9.718528240101191e-06,
+      "loss": 0.9856,
+      "step": 14400
+    },
+    {
+      "epoch": 0.5237298614394298,
+      "grad_norm": 0.0,
+      "learning_rate": 9.717350847064527e-06,
+      "loss": 1.0226,
+      "step": 14401
+    },
+    {
+      "epoch": 0.5237662290431684,
+      "grad_norm": 0.0,
+      "learning_rate": 9.716173457949225e-06,
+      "loss": 1.0048,
+      "step": 14402
+    },
+    {
+      "epoch": 0.523802596646907,
+      "grad_norm": 0.0,
+      "learning_rate": 9.71499607277161e-06,
+      "loss": 0.983,
+      "step": 14403
+    },
+    {
+      "epoch": 0.5238389642506456,
+      "grad_norm": 0.0,
+      "learning_rate": 9.71381869154802e-06,
+      "loss": 0.9571,
+      "step": 14404
+    },
+    {
+      "epoch": 0.5238753318543841,
+      "grad_norm": 0.0,
+      "learning_rate": 9.712641314294788e-06,
+      "loss": 1.0043,
+      "step": 14405
+    },
+    {
+      "epoch": 0.5239116994581227,
+      "grad_norm": 0.0,
+      "learning_rate": 9.711463941028253e-06,
+      "loss": 0.9966,
+      "step": 14406
+    },
+    {
+      "epoch": 0.5239480670618613,
+      "grad_norm": 0.0,
+      "learning_rate": 9.710286571764743e-06,
+      "loss": 0.9718,
+      "step": 14407
+    },
+    {
+      "epoch": 0.5239844346655999,
+      "grad_norm": 0.0,
+      "learning_rate": 9.709109206520601e-06,
+      "loss": 1.0411,
+      "step": 14408
+    },
+    {
+      "epoch": 0.5240208022693384,
+      "grad_norm": 0.0,
+      "learning_rate": 9.70793184531215e-06,
+      "loss": 1.0411,
+      "step": 14409
+    },
+    {
+      "epoch": 0.524057169873077,
+      "grad_norm": 0.0,
+      "learning_rate": 9.706754488155736e-06,
+      "loss": 1.0141,
+      "step": 14410
+    },
+    {
+      "epoch": 0.5240935374768156,
+      "grad_norm": 0.0,
+      "learning_rate": 9.705577135067687e-06,
+      "loss": 1.0043,
+      "step": 14411
+    },
+    {
+      "epoch": 0.5241299050805542,
+      "grad_norm": 0.0,
+      "learning_rate": 9.704399786064333e-06,
+      "loss": 1.0011,
+      "step": 14412
+    },
+    {
+      "epoch": 0.5241662726842928,
+      "grad_norm": 0.0,
+      "learning_rate": 9.703222441162015e-06,
+      "loss": 1.0475,
+      "step": 14413
+    },
+    {
+      "epoch": 0.5242026402880314,
+      "grad_norm": 0.0,
+      "learning_rate": 9.702045100377063e-06,
+      "loss": 1.0387,
+      "step": 14414
+    },
+    {
+      "epoch": 0.52423900789177,
+      "grad_norm": 0.0,
+      "learning_rate": 9.700867763725817e-06,
+      "loss": 1.0646,
+      "step": 14415
+    },
+    {
+      "epoch": 0.5242753754955086,
+      "grad_norm": 0.0,
+      "learning_rate": 9.699690431224603e-06,
+      "loss": 0.9922,
+      "step": 14416
+    },
+    {
+      "epoch": 0.5243117430992472,
+      "grad_norm": 0.0,
+      "learning_rate": 9.698513102889759e-06,
+      "loss": 1.066,
+      "step": 14417
+    },
+    {
+      "epoch": 0.5243481107029858,
+      "grad_norm": 0.0,
+      "learning_rate": 9.697335778737616e-06,
+      "loss": 0.9611,
+      "step": 14418
+    },
+    {
+      "epoch": 0.5243844783067244,
+      "grad_norm": 0.0,
+      "learning_rate": 9.696158458784512e-06,
+      "loss": 0.9246,
+      "step": 14419
+    },
+    {
+      "epoch": 0.524420845910463,
+      "grad_norm": 0.0,
+      "learning_rate": 9.694981143046776e-06,
+      "loss": 0.95,
+      "step": 14420
+    },
+    {
+      "epoch": 0.5244572135142016,
+      "grad_norm": 0.0,
+      "learning_rate": 9.693803831540747e-06,
+      "loss": 1.1162,
+      "step": 14421
+    },
+    {
+      "epoch": 0.5244935811179401,
+      "grad_norm": 0.0,
+      "learning_rate": 9.69262652428275e-06,
+      "loss": 1.096,
+      "step": 14422
+    },
+    {
+      "epoch": 0.5245299487216787,
+      "grad_norm": 0.0,
+      "learning_rate": 9.69144922128913e-06,
+      "loss": 1.1221,
+      "step": 14423
+    },
+    {
+      "epoch": 0.5245663163254173,
+      "grad_norm": 0.0,
+      "learning_rate": 9.690271922576211e-06,
+      "loss": 0.9734,
+      "step": 14424
+    },
+    {
+      "epoch": 0.5246026839291559,
+      "grad_norm": 0.0,
+      "learning_rate": 9.689094628160332e-06,
+      "loss": 0.9463,
+      "step": 14425
+    },
+    {
+      "epoch": 0.5246390515328945,
+      "grad_norm": 0.0,
+      "learning_rate": 9.687917338057821e-06,
+      "loss": 1.0501,
+      "step": 14426
+    },
+    {
+      "epoch": 0.5246754191366331,
+      "grad_norm": 0.0,
+      "learning_rate": 9.686740052285016e-06,
+      "loss": 1.0731,
+      "step": 14427
+    },
+    {
+      "epoch": 0.5247117867403717,
+      "grad_norm": 0.0,
+      "learning_rate": 9.685562770858246e-06,
+      "loss": 1.0228,
+      "step": 14428
+    },
+    {
+      "epoch": 0.5247481543441103,
+      "grad_norm": 0.0,
+      "learning_rate": 9.684385493793851e-06,
+      "loss": 0.946,
+      "step": 14429
+    },
+    {
+      "epoch": 0.5247845219478489,
+      "grad_norm": 0.0,
+      "learning_rate": 9.683208221108157e-06,
+      "loss": 0.9506,
+      "step": 14430
+    },
+    {
+      "epoch": 0.5248208895515875,
+      "grad_norm": 0.0,
+      "learning_rate": 9.682030952817498e-06,
+      "loss": 0.9552,
+      "step": 14431
+    },
+    {
+      "epoch": 0.524857257155326,
+      "grad_norm": 0.0,
+      "learning_rate": 9.68085368893821e-06,
+      "loss": 1.0643,
+      "step": 14432
+    },
+    {
+      "epoch": 0.5248936247590646,
+      "grad_norm": 0.0,
+      "learning_rate": 9.679676429486624e-06,
+      "loss": 1.1356,
+      "step": 14433
+    },
+    {
+      "epoch": 0.5249299923628032,
+      "grad_norm": 0.0,
+      "learning_rate": 9.678499174479075e-06,
+      "loss": 0.9868,
+      "step": 14434
+    },
+    {
+      "epoch": 0.5249663599665418,
+      "grad_norm": 0.0,
+      "learning_rate": 9.677321923931888e-06,
+      "loss": 0.9978,
+      "step": 14435
+    },
+    {
+      "epoch": 0.5250027275702804,
+      "grad_norm": 0.0,
+      "learning_rate": 9.676144677861407e-06,
+      "loss": 0.8907,
+      "step": 14436
+    },
+    {
+      "epoch": 0.525039095174019,
+      "grad_norm": 0.0,
+      "learning_rate": 9.674967436283957e-06,
+      "loss": 0.9549,
+      "step": 14437
+    },
+    {
+      "epoch": 0.5250754627777576,
+      "grad_norm": 0.0,
+      "learning_rate": 9.673790199215874e-06,
+      "loss": 1.0258,
+      "step": 14438
+    },
+    {
+      "epoch": 0.5251118303814961,
+      "grad_norm": 0.0,
+      "learning_rate": 9.672612966673485e-06,
+      "loss": 0.9636,
+      "step": 14439
+    },
+    {
+      "epoch": 0.5251481979852347,
+      "grad_norm": 0.0,
+      "learning_rate": 9.671435738673131e-06,
+      "loss": 0.8727,
+      "step": 14440
+    },
+    {
+      "epoch": 0.5251845655889733,
+      "grad_norm": 0.0,
+      "learning_rate": 9.670258515231137e-06,
+      "loss": 1.0187,
+      "step": 14441
+    },
+    {
+      "epoch": 0.5252209331927119,
+      "grad_norm": 0.0,
+      "learning_rate": 9.669081296363843e-06,
+      "loss": 0.924,
+      "step": 14442
+    },
+    {
+      "epoch": 0.5252573007964505,
+      "grad_norm": 0.0,
+      "learning_rate": 9.667904082087573e-06,
+      "loss": 1.0269,
+      "step": 14443
+    },
+    {
+      "epoch": 0.5252936684001891,
+      "grad_norm": 0.0,
+      "learning_rate": 9.666726872418663e-06,
+      "loss": 0.9825,
+      "step": 14444
+    },
+    {
+      "epoch": 0.5253300360039277,
+      "grad_norm": 0.0,
+      "learning_rate": 9.665549667373442e-06,
+      "loss": 1.0703,
+      "step": 14445
+    },
+    {
+      "epoch": 0.5253664036076663,
+      "grad_norm": 0.0,
+      "learning_rate": 9.66437246696825e-06,
+      "loss": 0.9129,
+      "step": 14446
+    },
+    {
+      "epoch": 0.5254027712114049,
+      "grad_norm": 0.0,
+      "learning_rate": 9.66319527121941e-06,
+      "loss": 0.9613,
+      "step": 14447
+    },
+    {
+      "epoch": 0.5254391388151435,
+      "grad_norm": 0.0,
+      "learning_rate": 9.662018080143262e-06,
+      "loss": 1.0362,
+      "step": 14448
+    },
+    {
+      "epoch": 0.5254755064188821,
+      "grad_norm": 0.0,
+      "learning_rate": 9.660840893756132e-06,
+      "loss": 1.1229,
+      "step": 14449
+    },
+    {
+      "epoch": 0.5255118740226207,
+      "grad_norm": 0.0,
+      "learning_rate": 9.659663712074351e-06,
+      "loss": 1.0315,
+      "step": 14450
+    },
+    {
+      "epoch": 0.5255482416263593,
+      "grad_norm": 0.0,
+      "learning_rate": 9.658486535114254e-06,
+      "loss": 0.8741,
+      "step": 14451
+    },
+    {
+      "epoch": 0.5255846092300979,
+      "grad_norm": 0.0,
+      "learning_rate": 9.65730936289217e-06,
+      "loss": 1.0301,
+      "step": 14452
+    },
+    {
+      "epoch": 0.5256209768338365,
+      "grad_norm": 0.0,
+      "learning_rate": 9.65613219542444e-06,
+      "loss": 0.9406,
+      "step": 14453
+    },
+    {
+      "epoch": 0.5256573444375751,
+      "grad_norm": 0.0,
+      "learning_rate": 9.654955032727378e-06,
+      "loss": 1.0037,
+      "step": 14454
+    },
+    {
+      "epoch": 0.5256937120413135,
+      "grad_norm": 0.0,
+      "learning_rate": 9.653777874817333e-06,
+      "loss": 0.9116,
+      "step": 14455
+    },
+    {
+      "epoch": 0.5257300796450521,
+      "grad_norm": 0.0,
+      "learning_rate": 9.652600721710623e-06,
+      "loss": 0.9556,
+      "step": 14456
+    },
+    {
+      "epoch": 0.5257664472487907,
+      "grad_norm": 0.0,
+      "learning_rate": 9.651423573423589e-06,
+      "loss": 0.9617,
+      "step": 14457
+    },
+    {
+      "epoch": 0.5258028148525293,
+      "grad_norm": 0.0,
+      "learning_rate": 9.650246429972556e-06,
+      "loss": 0.933,
+      "step": 14458
+    },
+    {
+      "epoch": 0.5258391824562679,
+      "grad_norm": 0.0,
+      "learning_rate": 9.64906929137386e-06,
+      "loss": 0.9872,
+      "step": 14459
+    },
+    {
+      "epoch": 0.5258755500600065,
+      "grad_norm": 0.0,
+      "learning_rate": 9.647892157643825e-06,
+      "loss": 1.0939,
+      "step": 14460
+    },
+    {
+      "epoch": 0.5259119176637451,
+      "grad_norm": 0.0,
+      "learning_rate": 9.646715028798793e-06,
+      "loss": 0.8953,
+      "step": 14461
+    },
+    {
+      "epoch": 0.5259482852674837,
+      "grad_norm": 0.0,
+      "learning_rate": 9.645537904855083e-06,
+      "loss": 0.9371,
+      "step": 14462
+    },
+    {
+      "epoch": 0.5259846528712223,
+      "grad_norm": 0.0,
+      "learning_rate": 9.644360785829036e-06,
+      "loss": 0.8648,
+      "step": 14463
+    },
+    {
+      "epoch": 0.5260210204749609,
+      "grad_norm": 0.0,
+      "learning_rate": 9.643183671736976e-06,
+      "loss": 0.9245,
+      "step": 14464
+    },
+    {
+      "epoch": 0.5260573880786995,
+      "grad_norm": 0.0,
+      "learning_rate": 9.642006562595238e-06,
+      "loss": 0.9917,
+      "step": 14465
+    },
+    {
+      "epoch": 0.5260937556824381,
+      "grad_norm": 0.0,
+      "learning_rate": 9.640829458420149e-06,
+      "loss": 0.8847,
+      "step": 14466
+    },
+    {
+      "epoch": 0.5261301232861767,
+      "grad_norm": 0.0,
+      "learning_rate": 9.639652359228044e-06,
+      "loss": 0.9249,
+      "step": 14467
+    },
+    {
+      "epoch": 0.5261664908899153,
+      "grad_norm": 0.0,
+      "learning_rate": 9.638475265035253e-06,
+      "loss": 0.9602,
+      "step": 14468
+    },
+    {
+      "epoch": 0.5262028584936539,
+      "grad_norm": 0.0,
+      "learning_rate": 9.6372981758581e-06,
+      "loss": 0.9642,
+      "step": 14469
+    },
+    {
+      "epoch": 0.5262392260973925,
+      "grad_norm": 0.0,
+      "learning_rate": 9.636121091712923e-06,
+      "loss": 1.0966,
+      "step": 14470
+    },
+    {
+      "epoch": 0.5262755937011311,
+      "grad_norm": 0.0,
+      "learning_rate": 9.634944012616048e-06,
+      "loss": 0.9197,
+      "step": 14471
+    },
+    {
+      "epoch": 0.5263119613048696,
+      "grad_norm": 0.0,
+      "learning_rate": 9.633766938583807e-06,
+      "loss": 0.9241,
+      "step": 14472
+    },
+    {
+      "epoch": 0.5263483289086082,
+      "grad_norm": 0.0,
+      "learning_rate": 9.63258986963253e-06,
+      "loss": 0.9457,
+      "step": 14473
+    },
+    {
+      "epoch": 0.5263846965123468,
+      "grad_norm": 0.0,
+      "learning_rate": 9.63141280577855e-06,
+      "loss": 1.0852,
+      "step": 14474
+    },
+    {
+      "epoch": 0.5264210641160854,
+      "grad_norm": 0.0,
+      "learning_rate": 9.630235747038191e-06,
+      "loss": 0.9285,
+      "step": 14475
+    },
+    {
+      "epoch": 0.526457431719824,
+      "grad_norm": 0.0,
+      "learning_rate": 9.629058693427789e-06,
+      "loss": 1.0135,
+      "step": 14476
+    },
+    {
+      "epoch": 0.5264937993235626,
+      "grad_norm": 0.0,
+      "learning_rate": 9.627881644963667e-06,
+      "loss": 0.9548,
+      "step": 14477
+    },
+    {
+      "epoch": 0.5265301669273011,
+      "grad_norm": 0.0,
+      "learning_rate": 9.626704601662164e-06,
+      "loss": 1.0647,
+      "step": 14478
+    },
+    {
+      "epoch": 0.5265665345310397,
+      "grad_norm": 0.0,
+      "learning_rate": 9.625527563539602e-06,
+      "loss": 0.9686,
+      "step": 14479
+    },
+    {
+      "epoch": 0.5266029021347783,
+      "grad_norm": 0.0,
+      "learning_rate": 9.624350530612318e-06,
+      "loss": 0.9698,
+      "step": 14480
+    },
+    {
+      "epoch": 0.5266392697385169,
+      "grad_norm": 0.0,
+      "learning_rate": 9.62317350289663e-06,
+      "loss": 0.9442,
+      "step": 14481
+    },
+    {
+      "epoch": 0.5266756373422555,
+      "grad_norm": 0.0,
+      "learning_rate": 9.621996480408883e-06,
+      "loss": 0.9943,
+      "step": 14482
+    },
+    {
+      "epoch": 0.5267120049459941,
+      "grad_norm": 0.0,
+      "learning_rate": 9.620819463165393e-06,
+      "loss": 0.9002,
+      "step": 14483
+    },
+    {
+      "epoch": 0.5267483725497327,
+      "grad_norm": 0.0,
+      "learning_rate": 9.619642451182498e-06,
+      "loss": 1.0294,
+      "step": 14484
+    },
+    {
+      "epoch": 0.5267847401534713,
+      "grad_norm": 0.0,
+      "learning_rate": 9.618465444476521e-06,
+      "loss": 0.9682,
+      "step": 14485
+    },
+    {
+      "epoch": 0.5268211077572099,
+      "grad_norm": 0.0,
+      "learning_rate": 9.617288443063798e-06,
+      "loss": 1.0358,
+      "step": 14486
+    },
+    {
+      "epoch": 0.5268574753609485,
+      "grad_norm": 0.0,
+      "learning_rate": 9.616111446960656e-06,
+      "loss": 1.0945,
+      "step": 14487
+    },
+    {
+      "epoch": 0.5268938429646871,
+      "grad_norm": 0.0,
+      "learning_rate": 9.61493445618342e-06,
+      "loss": 1.0064,
+      "step": 14488
+    },
+    {
+      "epoch": 0.5269302105684256,
+      "grad_norm": 0.0,
+      "learning_rate": 9.613757470748423e-06,
+      "loss": 1.0019,
+      "step": 14489
+    },
+    {
+      "epoch": 0.5269665781721642,
+      "grad_norm": 0.0,
+      "learning_rate": 9.612580490671992e-06,
+      "loss": 1.007,
+      "step": 14490
+    },
+    {
+      "epoch": 0.5270029457759028,
+      "grad_norm": 0.0,
+      "learning_rate": 9.61140351597046e-06,
+      "loss": 1.0851,
+      "step": 14491
+    },
+    {
+      "epoch": 0.5270393133796414,
+      "grad_norm": 0.0,
+      "learning_rate": 9.610226546660147e-06,
+      "loss": 0.9506,
+      "step": 14492
+    },
+    {
+      "epoch": 0.52707568098338,
+      "grad_norm": 0.0,
+      "learning_rate": 9.609049582757395e-06,
+      "loss": 0.9845,
+      "step": 14493
+    },
+    {
+      "epoch": 0.5271120485871186,
+      "grad_norm": 0.0,
+      "learning_rate": 9.607872624278519e-06,
+      "loss": 0.9097,
+      "step": 14494
+    },
+    {
+      "epoch": 0.5271484161908572,
+      "grad_norm": 0.0,
+      "learning_rate": 9.606695671239858e-06,
+      "loss": 0.87,
+      "step": 14495
+    },
+    {
+      "epoch": 0.5271847837945958,
+      "grad_norm": 0.0,
+      "learning_rate": 9.605518723657736e-06,
+      "loss": 0.998,
+      "step": 14496
+    },
+    {
+      "epoch": 0.5272211513983344,
+      "grad_norm": 0.0,
+      "learning_rate": 9.60434178154848e-06,
+      "loss": 1.009,
+      "step": 14497
+    },
+    {
+      "epoch": 0.527257519002073,
+      "grad_norm": 0.0,
+      "learning_rate": 9.60316484492842e-06,
+      "loss": 0.9454,
+      "step": 14498
+    },
+    {
+      "epoch": 0.5272938866058116,
+      "grad_norm": 0.0,
+      "learning_rate": 9.601987913813889e-06,
+      "loss": 1.0337,
+      "step": 14499
+    },
+    {
+      "epoch": 0.5273302542095502,
+      "grad_norm": 0.0,
+      "learning_rate": 9.600810988221207e-06,
+      "loss": 1.0523,
+      "step": 14500
+    },
+    {
+      "epoch": 0.5273666218132887,
+      "grad_norm": 0.0,
+      "learning_rate": 9.59963406816671e-06,
+      "loss": 0.9861,
+      "step": 14501
+    },
+    {
+      "epoch": 0.5274029894170273,
+      "grad_norm": 0.0,
+      "learning_rate": 9.598457153666719e-06,
+      "loss": 1.0565,
+      "step": 14502
+    },
+    {
+      "epoch": 0.5274393570207659,
+      "grad_norm": 0.0,
+      "learning_rate": 9.597280244737565e-06,
+      "loss": 1.0018,
+      "step": 14503
+    },
+    {
+      "epoch": 0.5274757246245045,
+      "grad_norm": 0.0,
+      "learning_rate": 9.596103341395577e-06,
+      "loss": 0.9971,
+      "step": 14504
+    },
+    {
+      "epoch": 0.5275120922282431,
+      "grad_norm": 0.0,
+      "learning_rate": 9.594926443657084e-06,
+      "loss": 0.8963,
+      "step": 14505
+    },
+    {
+      "epoch": 0.5275484598319816,
+      "grad_norm": 0.0,
+      "learning_rate": 9.593749551538412e-06,
+      "loss": 1.0454,
+      "step": 14506
+    },
+    {
+      "epoch": 0.5275848274357202,
+      "grad_norm": 0.0,
+      "learning_rate": 9.592572665055884e-06,
+      "loss": 0.9107,
+      "step": 14507
+    },
+    {
+      "epoch": 0.5276211950394588,
+      "grad_norm": 0.0,
+      "learning_rate": 9.591395784225838e-06,
+      "loss": 1.0417,
+      "step": 14508
+    },
+    {
+      "epoch": 0.5276575626431974,
+      "grad_norm": 0.0,
+      "learning_rate": 9.590218909064591e-06,
+      "loss": 0.9048,
+      "step": 14509
+    },
+    {
+      "epoch": 0.527693930246936,
+      "grad_norm": 0.0,
+      "learning_rate": 9.589042039588478e-06,
+      "loss": 1.041,
+      "step": 14510
+    },
+    {
+      "epoch": 0.5277302978506746,
+      "grad_norm": 0.0,
+      "learning_rate": 9.58786517581382e-06,
+      "loss": 0.9222,
+      "step": 14511
+    },
+    {
+      "epoch": 0.5277666654544132,
+      "grad_norm": 0.0,
+      "learning_rate": 9.586688317756953e-06,
+      "loss": 1.1511,
+      "step": 14512
+    },
+    {
+      "epoch": 0.5278030330581518,
+      "grad_norm": 0.0,
+      "learning_rate": 9.585511465434195e-06,
+      "loss": 1.0743,
+      "step": 14513
+    },
+    {
+      "epoch": 0.5278394006618904,
+      "grad_norm": 0.0,
+      "learning_rate": 9.584334618861882e-06,
+      "loss": 0.9142,
+      "step": 14514
+    },
+    {
+      "epoch": 0.527875768265629,
+      "grad_norm": 0.0,
+      "learning_rate": 9.583157778056333e-06,
+      "loss": 1.0335,
+      "step": 14515
+    },
+    {
+      "epoch": 0.5279121358693676,
+      "grad_norm": 0.0,
+      "learning_rate": 9.581980943033879e-06,
+      "loss": 0.945,
+      "step": 14516
+    },
+    {
+      "epoch": 0.5279485034731062,
+      "grad_norm": 0.0,
+      "learning_rate": 9.580804113810844e-06,
+      "loss": 0.8948,
+      "step": 14517
+    },
+    {
+      "epoch": 0.5279848710768448,
+      "grad_norm": 0.0,
+      "learning_rate": 9.57962729040356e-06,
+      "loss": 0.9948,
+      "step": 14518
+    },
+    {
+      "epoch": 0.5280212386805834,
+      "grad_norm": 0.0,
+      "learning_rate": 9.57845047282835e-06,
+      "loss": 0.9518,
+      "step": 14519
+    },
+    {
+      "epoch": 0.528057606284322,
+      "grad_norm": 0.0,
+      "learning_rate": 9.577273661101545e-06,
+      "loss": 0.9965,
+      "step": 14520
+    },
+    {
+      "epoch": 0.5280939738880606,
+      "grad_norm": 0.0,
+      "learning_rate": 9.57609685523946e-06,
+      "loss": 0.9825,
+      "step": 14521
+    },
+    {
+      "epoch": 0.5281303414917992,
+      "grad_norm": 0.0,
+      "learning_rate": 9.574920055258439e-06,
+      "loss": 1.0081,
+      "step": 14522
+    },
+    {
+      "epoch": 0.5281667090955376,
+      "grad_norm": 0.0,
+      "learning_rate": 9.573743261174793e-06,
+      "loss": 1.0281,
+      "step": 14523
+    },
+    {
+      "epoch": 0.5282030766992762,
+      "grad_norm": 0.0,
+      "learning_rate": 9.572566473004857e-06,
+      "loss": 1.0334,
+      "step": 14524
+    },
+    {
+      "epoch": 0.5282394443030148,
+      "grad_norm": 0.0,
+      "learning_rate": 9.571389690764953e-06,
+      "loss": 0.9581,
+      "step": 14525
+    },
+    {
+      "epoch": 0.5282758119067534,
+      "grad_norm": 0.0,
+      "learning_rate": 9.570212914471411e-06,
+      "loss": 1.0038,
+      "step": 14526
+    },
+    {
+      "epoch": 0.528312179510492,
+      "grad_norm": 0.0,
+      "learning_rate": 9.569036144140558e-06,
+      "loss": 0.8506,
+      "step": 14527
+    },
+    {
+      "epoch": 0.5283485471142306,
+      "grad_norm": 0.0,
+      "learning_rate": 9.567859379788711e-06,
+      "loss": 0.9789,
+      "step": 14528
+    },
+    {
+      "epoch": 0.5283849147179692,
+      "grad_norm": 0.0,
+      "learning_rate": 9.566682621432205e-06,
+      "loss": 0.882,
+      "step": 14529
+    },
+    {
+      "epoch": 0.5284212823217078,
+      "grad_norm": 0.0,
+      "learning_rate": 9.56550586908736e-06,
+      "loss": 1.0429,
+      "step": 14530
+    },
+    {
+      "epoch": 0.5284576499254464,
+      "grad_norm": 0.0,
+      "learning_rate": 9.564329122770507e-06,
+      "loss": 1.1246,
+      "step": 14531
+    },
+    {
+      "epoch": 0.528494017529185,
+      "grad_norm": 0.0,
+      "learning_rate": 9.563152382497967e-06,
+      "loss": 0.9879,
+      "step": 14532
+    },
+    {
+      "epoch": 0.5285303851329236,
+      "grad_norm": 0.0,
+      "learning_rate": 9.561975648286074e-06,
+      "loss": 0.8798,
+      "step": 14533
+    },
+    {
+      "epoch": 0.5285667527366622,
+      "grad_norm": 0.0,
+      "learning_rate": 9.56079892015114e-06,
+      "loss": 1.1099,
+      "step": 14534
+    },
+    {
+      "epoch": 0.5286031203404008,
+      "grad_norm": 0.0,
+      "learning_rate": 9.559622198109503e-06,
+      "loss": 0.941,
+      "step": 14535
+    },
+    {
+      "epoch": 0.5286394879441394,
+      "grad_norm": 0.0,
+      "learning_rate": 9.558445482177479e-06,
+      "loss": 0.88,
+      "step": 14536
+    },
+    {
+      "epoch": 0.528675855547878,
+      "grad_norm": 0.0,
+      "learning_rate": 9.5572687723714e-06,
+      "loss": 0.8917,
+      "step": 14537
+    },
+    {
+      "epoch": 0.5287122231516166,
+      "grad_norm": 0.0,
+      "learning_rate": 9.556092068707587e-06,
+      "loss": 1.0333,
+      "step": 14538
+    },
+    {
+      "epoch": 0.5287485907553551,
+      "grad_norm": 0.0,
+      "learning_rate": 9.55491537120237e-06,
+      "loss": 1.0128,
+      "step": 14539
+    },
+    {
+      "epoch": 0.5287849583590937,
+      "grad_norm": 0.0,
+      "learning_rate": 9.553738679872066e-06,
+      "loss": 1.0325,
+      "step": 14540
+    },
+    {
+      "epoch": 0.5288213259628323,
+      "grad_norm": 0.0,
+      "learning_rate": 9.552561994733012e-06,
+      "loss": 1.012,
+      "step": 14541
+    },
+    {
+      "epoch": 0.5288576935665709,
+      "grad_norm": 0.0,
+      "learning_rate": 9.551385315801518e-06,
+      "loss": 0.9368,
+      "step": 14542
+    },
+    {
+      "epoch": 0.5288940611703095,
+      "grad_norm": 0.0,
+      "learning_rate": 9.550208643093921e-06,
+      "loss": 0.9812,
+      "step": 14543
+    },
+    {
+      "epoch": 0.5289304287740481,
+      "grad_norm": 0.0,
+      "learning_rate": 9.549031976626538e-06,
+      "loss": 1.0226,
+      "step": 14544
+    },
+    {
+      "epoch": 0.5289667963777867,
+      "grad_norm": 0.0,
+      "learning_rate": 9.547855316415697e-06,
+      "loss": 1.0452,
+      "step": 14545
+    },
+    {
+      "epoch": 0.5290031639815252,
+      "grad_norm": 0.0,
+      "learning_rate": 9.546678662477727e-06,
+      "loss": 0.9706,
+      "step": 14546
+    },
+    {
+      "epoch": 0.5290395315852638,
+      "grad_norm": 0.0,
+      "learning_rate": 9.54550201482894e-06,
+      "loss": 0.9645,
+      "step": 14547
+    },
+    {
+      "epoch": 0.5290758991890024,
+      "grad_norm": 0.0,
+      "learning_rate": 9.54432537348567e-06,
+      "loss": 0.9406,
+      "step": 14548
+    },
+    {
+      "epoch": 0.529112266792741,
+      "grad_norm": 0.0,
+      "learning_rate": 9.54314873846424e-06,
+      "loss": 1.0753,
+      "step": 14549
+    },
+    {
+      "epoch": 0.5291486343964796,
+      "grad_norm": 0.0,
+      "learning_rate": 9.541972109780972e-06,
+      "loss": 1.0307,
+      "step": 14550
+    },
+    {
+      "epoch": 0.5291850020002182,
+      "grad_norm": 0.0,
+      "learning_rate": 9.54079548745219e-06,
+      "loss": 0.8779,
+      "step": 14551
+    },
+    {
+      "epoch": 0.5292213696039568,
+      "grad_norm": 0.0,
+      "learning_rate": 9.539618871494226e-06,
+      "loss": 0.9947,
+      "step": 14552
+    },
+    {
+      "epoch": 0.5292577372076954,
+      "grad_norm": 0.0,
+      "learning_rate": 9.538442261923387e-06,
+      "loss": 0.9788,
+      "step": 14553
+    },
+    {
+      "epoch": 0.529294104811434,
+      "grad_norm": 0.0,
+      "learning_rate": 9.537265658756014e-06,
+      "loss": 0.997,
+      "step": 14554
+    },
+    {
+      "epoch": 0.5293304724151726,
+      "grad_norm": 0.0,
+      "learning_rate": 9.53608906200842e-06,
+      "loss": 0.969,
+      "step": 14555
+    },
+    {
+      "epoch": 0.5293668400189111,
+      "grad_norm": 0.0,
+      "learning_rate": 9.534912471696933e-06,
+      "loss": 0.8941,
+      "step": 14556
+    },
+    {
+      "epoch": 0.5294032076226497,
+      "grad_norm": 0.0,
+      "learning_rate": 9.533735887837874e-06,
+      "loss": 1.1546,
+      "step": 14557
+    },
+    {
+      "epoch": 0.5294395752263883,
+      "grad_norm": 0.0,
+      "learning_rate": 9.53255931044757e-06,
+      "loss": 0.9704,
+      "step": 14558
+    },
+    {
+      "epoch": 0.5294759428301269,
+      "grad_norm": 0.0,
+      "learning_rate": 9.531382739542341e-06,
+      "loss": 1.049,
+      "step": 14559
+    },
+    {
+      "epoch": 0.5295123104338655,
+      "grad_norm": 0.0,
+      "learning_rate": 9.530206175138516e-06,
+      "loss": 1.0096,
+      "step": 14560
+    },
+    {
+      "epoch": 0.5295486780376041,
+      "grad_norm": 0.0,
+      "learning_rate": 9.529029617252406e-06,
+      "loss": 0.9817,
+      "step": 14561
+    },
+    {
+      "epoch": 0.5295850456413427,
+      "grad_norm": 0.0,
+      "learning_rate": 9.52785306590035e-06,
+      "loss": 1.0292,
+      "step": 14562
+    },
+    {
+      "epoch": 0.5296214132450813,
+      "grad_norm": 0.0,
+      "learning_rate": 9.526676521098656e-06,
+      "loss": 0.9834,
+      "step": 14563
+    },
+    {
+      "epoch": 0.5296577808488199,
+      "grad_norm": 0.0,
+      "learning_rate": 9.525499982863658e-06,
+      "loss": 0.8984,
+      "step": 14564
+    },
+    {
+      "epoch": 0.5296941484525585,
+      "grad_norm": 0.0,
+      "learning_rate": 9.524323451211677e-06,
+      "loss": 0.9116,
+      "step": 14565
+    },
+    {
+      "epoch": 0.5297305160562971,
+      "grad_norm": 0.0,
+      "learning_rate": 9.523146926159024e-06,
+      "loss": 1.0179,
+      "step": 14566
+    },
+    {
+      "epoch": 0.5297668836600357,
+      "grad_norm": 0.0,
+      "learning_rate": 9.52197040772204e-06,
+      "loss": 0.8815,
+      "step": 14567
+    },
+    {
+      "epoch": 0.5298032512637743,
+      "grad_norm": 0.0,
+      "learning_rate": 9.520793895917033e-06,
+      "loss": 0.988,
+      "step": 14568
+    },
+    {
+      "epoch": 0.5298396188675129,
+      "grad_norm": 0.0,
+      "learning_rate": 9.519617390760333e-06,
+      "loss": 0.9268,
+      "step": 14569
+    },
+    {
+      "epoch": 0.5298759864712514,
+      "grad_norm": 0.0,
+      "learning_rate": 9.518440892268259e-06,
+      "loss": 0.9153,
+      "step": 14570
+    },
+    {
+      "epoch": 0.52991235407499,
+      "grad_norm": 0.0,
+      "learning_rate": 9.517264400457136e-06,
+      "loss": 0.9792,
+      "step": 14571
+    },
+    {
+      "epoch": 0.5299487216787286,
+      "grad_norm": 0.0,
+      "learning_rate": 9.516087915343281e-06,
+      "loss": 0.897,
+      "step": 14572
+    },
+    {
+      "epoch": 0.5299850892824671,
+      "grad_norm": 0.0,
+      "learning_rate": 9.514911436943024e-06,
+      "loss": 0.9108,
+      "step": 14573
+    },
+    {
+      "epoch": 0.5300214568862057,
+      "grad_norm": 0.0,
+      "learning_rate": 9.51373496527268e-06,
+      "loss": 0.8163,
+      "step": 14574
+    },
+    {
+      "epoch": 0.5300578244899443,
+      "grad_norm": 0.0,
+      "learning_rate": 9.512558500348575e-06,
+      "loss": 0.9023,
+      "step": 14575
+    },
+    {
+      "epoch": 0.5300941920936829,
+      "grad_norm": 0.0,
+      "learning_rate": 9.511382042187026e-06,
+      "loss": 0.9374,
+      "step": 14576
+    },
+    {
+      "epoch": 0.5301305596974215,
+      "grad_norm": 0.0,
+      "learning_rate": 9.51020559080436e-06,
+      "loss": 1.0097,
+      "step": 14577
+    },
+    {
+      "epoch": 0.5301669273011601,
+      "grad_norm": 0.0,
+      "learning_rate": 9.509029146216897e-06,
+      "loss": 1.0097,
+      "step": 14578
+    },
+    {
+      "epoch": 0.5302032949048987,
+      "grad_norm": 0.0,
+      "learning_rate": 9.507852708440957e-06,
+      "loss": 1.0326,
+      "step": 14579
+    },
+    {
+      "epoch": 0.5302396625086373,
+      "grad_norm": 0.0,
+      "learning_rate": 9.506676277492862e-06,
+      "loss": 1.0997,
+      "step": 14580
+    },
+    {
+      "epoch": 0.5302760301123759,
+      "grad_norm": 0.0,
+      "learning_rate": 9.505499853388938e-06,
+      "loss": 0.9709,
+      "step": 14581
+    },
+    {
+      "epoch": 0.5303123977161145,
+      "grad_norm": 0.0,
+      "learning_rate": 9.504323436145497e-06,
+      "loss": 1.0303,
+      "step": 14582
+    },
+    {
+      "epoch": 0.5303487653198531,
+      "grad_norm": 0.0,
+      "learning_rate": 9.503147025778867e-06,
+      "loss": 1.0008,
+      "step": 14583
+    },
+    {
+      "epoch": 0.5303851329235917,
+      "grad_norm": 0.0,
+      "learning_rate": 9.501970622305368e-06,
+      "loss": 1.0084,
+      "step": 14584
+    },
+    {
+      "epoch": 0.5304215005273303,
+      "grad_norm": 0.0,
+      "learning_rate": 9.500794225741318e-06,
+      "loss": 0.9854,
+      "step": 14585
+    },
+    {
+      "epoch": 0.5304578681310689,
+      "grad_norm": 0.0,
+      "learning_rate": 9.499617836103044e-06,
+      "loss": 1.0166,
+      "step": 14586
+    },
+    {
+      "epoch": 0.5304942357348075,
+      "grad_norm": 0.0,
+      "learning_rate": 9.498441453406858e-06,
+      "loss": 1.0113,
+      "step": 14587
+    },
+    {
+      "epoch": 0.5305306033385461,
+      "grad_norm": 0.0,
+      "learning_rate": 9.497265077669087e-06,
+      "loss": 1.1447,
+      "step": 14588
+    },
+    {
+      "epoch": 0.5305669709422847,
+      "grad_norm": 0.0,
+      "learning_rate": 9.496088708906047e-06,
+      "loss": 0.967,
+      "step": 14589
+    },
+    {
+      "epoch": 0.5306033385460232,
+      "grad_norm": 0.0,
+      "learning_rate": 9.494912347134064e-06,
+      "loss": 0.9562,
+      "step": 14590
+    },
+    {
+      "epoch": 0.5306397061497617,
+      "grad_norm": 0.0,
+      "learning_rate": 9.493735992369453e-06,
+      "loss": 0.9326,
+      "step": 14591
+    },
+    {
+      "epoch": 0.5306760737535003,
+      "grad_norm": 0.0,
+      "learning_rate": 9.492559644628543e-06,
+      "loss": 1.0645,
+      "step": 14592
+    },
+    {
+      "epoch": 0.5307124413572389,
+      "grad_norm": 0.0,
+      "learning_rate": 9.49138330392764e-06,
+      "loss": 1.1083,
+      "step": 14593
+    },
+    {
+      "epoch": 0.5307488089609775,
+      "grad_norm": 0.0,
+      "learning_rate": 9.490206970283079e-06,
+      "loss": 0.97,
+      "step": 14594
+    },
+    {
+      "epoch": 0.5307851765647161,
+      "grad_norm": 0.0,
+      "learning_rate": 9.489030643711168e-06,
+      "loss": 0.8839,
+      "step": 14595
+    },
+    {
+      "epoch": 0.5308215441684547,
+      "grad_norm": 0.0,
+      "learning_rate": 9.487854324228234e-06,
+      "loss": 0.9703,
+      "step": 14596
+    },
+    {
+      "epoch": 0.5308579117721933,
+      "grad_norm": 0.0,
+      "learning_rate": 9.486678011850592e-06,
+      "loss": 0.9995,
+      "step": 14597
+    },
+    {
+      "epoch": 0.5308942793759319,
+      "grad_norm": 0.0,
+      "learning_rate": 9.485501706594568e-06,
+      "loss": 0.9266,
+      "step": 14598
+    },
+    {
+      "epoch": 0.5309306469796705,
+      "grad_norm": 0.0,
+      "learning_rate": 9.484325408476475e-06,
+      "loss": 1.078,
+      "step": 14599
+    },
+    {
+      "epoch": 0.5309670145834091,
+      "grad_norm": 0.0,
+      "learning_rate": 9.48314911751264e-06,
+      "loss": 1.0698,
+      "step": 14600
+    },
+    {
+      "epoch": 0.5310033821871477,
+      "grad_norm": 0.0,
+      "learning_rate": 9.481972833719373e-06,
+      "loss": 1.0029,
+      "step": 14601
+    },
+    {
+      "epoch": 0.5310397497908863,
+      "grad_norm": 0.0,
+      "learning_rate": 9.480796557112998e-06,
+      "loss": 1.0547,
+      "step": 14602
+    },
+    {
+      "epoch": 0.5310761173946249,
+      "grad_norm": 0.0,
+      "learning_rate": 9.479620287709838e-06,
+      "loss": 1.1356,
+      "step": 14603
+    },
+    {
+      "epoch": 0.5311124849983635,
+      "grad_norm": 0.0,
+      "learning_rate": 9.478444025526203e-06,
+      "loss": 0.8891,
+      "step": 14604
+    },
+    {
+      "epoch": 0.5311488526021021,
+      "grad_norm": 0.0,
+      "learning_rate": 9.477267770578423e-06,
+      "loss": 1.1433,
+      "step": 14605
+    },
+    {
+      "epoch": 0.5311852202058407,
+      "grad_norm": 0.0,
+      "learning_rate": 9.476091522882804e-06,
+      "loss": 1.001,
+      "step": 14606
+    },
+    {
+      "epoch": 0.5312215878095792,
+      "grad_norm": 0.0,
+      "learning_rate": 9.474915282455678e-06,
+      "loss": 0.9222,
+      "step": 14607
+    },
+    {
+      "epoch": 0.5312579554133178,
+      "grad_norm": 0.0,
+      "learning_rate": 9.473739049313354e-06,
+      "loss": 0.9849,
+      "step": 14608
+    },
+    {
+      "epoch": 0.5312943230170564,
+      "grad_norm": 0.0,
+      "learning_rate": 9.472562823472157e-06,
+      "loss": 1.0128,
+      "step": 14609
+    },
+    {
+      "epoch": 0.531330690620795,
+      "grad_norm": 0.0,
+      "learning_rate": 9.471386604948399e-06,
+      "loss": 0.8936,
+      "step": 14610
+    },
+    {
+      "epoch": 0.5313670582245336,
+      "grad_norm": 0.0,
+      "learning_rate": 9.470210393758404e-06,
+      "loss": 0.937,
+      "step": 14611
+    },
+    {
+      "epoch": 0.5314034258282722,
+      "grad_norm": 0.0,
+      "learning_rate": 9.469034189918485e-06,
+      "loss": 0.9668,
+      "step": 14612
+    },
+    {
+      "epoch": 0.5314397934320108,
+      "grad_norm": 0.0,
+      "learning_rate": 9.46785799344497e-06,
+      "loss": 0.909,
+      "step": 14613
+    },
+    {
+      "epoch": 0.5314761610357494,
+      "grad_norm": 0.0,
+      "learning_rate": 9.466681804354164e-06,
+      "loss": 0.8702,
+      "step": 14614
+    },
+    {
+      "epoch": 0.531512528639488,
+      "grad_norm": 0.0,
+      "learning_rate": 9.465505622662395e-06,
+      "loss": 0.921,
+      "step": 14615
+    },
+    {
+      "epoch": 0.5315488962432265,
+      "grad_norm": 0.0,
+      "learning_rate": 9.464329448385972e-06,
+      "loss": 0.9695,
+      "step": 14616
+    },
+    {
+      "epoch": 0.5315852638469651,
+      "grad_norm": 0.0,
+      "learning_rate": 9.463153281541223e-06,
+      "loss": 0.8942,
+      "step": 14617
+    },
+    {
+      "epoch": 0.5316216314507037,
+      "grad_norm": 0.0,
+      "learning_rate": 9.461977122144457e-06,
+      "loss": 0.9532,
+      "step": 14618
+    },
+    {
+      "epoch": 0.5316579990544423,
+      "grad_norm": 0.0,
+      "learning_rate": 9.460800970212001e-06,
+      "loss": 0.9678,
+      "step": 14619
+    },
+    {
+      "epoch": 0.5316943666581809,
+      "grad_norm": 0.0,
+      "learning_rate": 9.45962482576016e-06,
+      "loss": 1.086,
+      "step": 14620
+    },
+    {
+      "epoch": 0.5317307342619195,
+      "grad_norm": 0.0,
+      "learning_rate": 9.458448688805262e-06,
+      "loss": 0.9525,
+      "step": 14621
+    },
+    {
+      "epoch": 0.5317671018656581,
+      "grad_norm": 0.0,
+      "learning_rate": 9.45727255936362e-06,
+      "loss": 1.0355,
+      "step": 14622
+    },
+    {
+      "epoch": 0.5318034694693967,
+      "grad_norm": 0.0,
+      "learning_rate": 9.456096437451548e-06,
+      "loss": 0.9247,
+      "step": 14623
+    },
+    {
+      "epoch": 0.5318398370731352,
+      "grad_norm": 0.0,
+      "learning_rate": 9.45492032308537e-06,
+      "loss": 1.1244,
+      "step": 14624
+    },
+    {
+      "epoch": 0.5318762046768738,
+      "grad_norm": 0.0,
+      "learning_rate": 9.453744216281394e-06,
+      "loss": 0.9571,
+      "step": 14625
+    },
+    {
+      "epoch": 0.5319125722806124,
+      "grad_norm": 0.0,
+      "learning_rate": 9.452568117055948e-06,
+      "loss": 1.0541,
+      "step": 14626
+    },
+    {
+      "epoch": 0.531948939884351,
+      "grad_norm": 0.0,
+      "learning_rate": 9.451392025425338e-06,
+      "loss": 1.0119,
+      "step": 14627
+    },
+    {
+      "epoch": 0.5319853074880896,
+      "grad_norm": 0.0,
+      "learning_rate": 9.450215941405886e-06,
+      "loss": 0.9548,
+      "step": 14628
+    },
+    {
+      "epoch": 0.5320216750918282,
+      "grad_norm": 0.0,
+      "learning_rate": 9.449039865013909e-06,
+      "loss": 1.0117,
+      "step": 14629
+    },
+    {
+      "epoch": 0.5320580426955668,
+      "grad_norm": 0.0,
+      "learning_rate": 9.447863796265721e-06,
+      "loss": 1.0451,
+      "step": 14630
+    },
+    {
+      "epoch": 0.5320944102993054,
+      "grad_norm": 0.0,
+      "learning_rate": 9.44668773517764e-06,
+      "loss": 1.0367,
+      "step": 14631
+    },
+    {
+      "epoch": 0.532130777903044,
+      "grad_norm": 0.0,
+      "learning_rate": 9.445511681765985e-06,
+      "loss": 1.0132,
+      "step": 14632
+    },
+    {
+      "epoch": 0.5321671455067826,
+      "grad_norm": 0.0,
+      "learning_rate": 9.444335636047062e-06,
+      "loss": 0.9521,
+      "step": 14633
+    },
+    {
+      "epoch": 0.5322035131105212,
+      "grad_norm": 0.0,
+      "learning_rate": 9.4431595980372e-06,
+      "loss": 1.1053,
+      "step": 14634
+    },
+    {
+      "epoch": 0.5322398807142598,
+      "grad_norm": 0.0,
+      "learning_rate": 9.441983567752705e-06,
+      "loss": 0.9304,
+      "step": 14635
+    },
+    {
+      "epoch": 0.5322762483179984,
+      "grad_norm": 0.0,
+      "learning_rate": 9.440807545209897e-06,
+      "loss": 0.9654,
+      "step": 14636
+    },
+    {
+      "epoch": 0.532312615921737,
+      "grad_norm": 0.0,
+      "learning_rate": 9.43963153042509e-06,
+      "loss": 0.9829,
+      "step": 14637
+    },
+    {
+      "epoch": 0.5323489835254755,
+      "grad_norm": 0.0,
+      "learning_rate": 9.438455523414602e-06,
+      "loss": 0.9768,
+      "step": 14638
+    },
+    {
+      "epoch": 0.5323853511292141,
+      "grad_norm": 0.0,
+      "learning_rate": 9.437279524194745e-06,
+      "loss": 1.1053,
+      "step": 14639
+    },
+    {
+      "epoch": 0.5324217187329526,
+      "grad_norm": 0.0,
+      "learning_rate": 9.436103532781841e-06,
+      "loss": 1.0674,
+      "step": 14640
+    },
+    {
+      "epoch": 0.5324580863366912,
+      "grad_norm": 0.0,
+      "learning_rate": 9.434927549192198e-06,
+      "loss": 0.9721,
+      "step": 14641
+    },
+    {
+      "epoch": 0.5324944539404298,
+      "grad_norm": 0.0,
+      "learning_rate": 9.43375157344213e-06,
+      "loss": 1.0073,
+      "step": 14642
+    },
+    {
+      "epoch": 0.5325308215441684,
+      "grad_norm": 0.0,
+      "learning_rate": 9.432575605547959e-06,
+      "loss": 0.9794,
+      "step": 14643
+    },
+    {
+      "epoch": 0.532567189147907,
+      "grad_norm": 0.0,
+      "learning_rate": 9.431399645525994e-06,
+      "loss": 1.0474,
+      "step": 14644
+    },
+    {
+      "epoch": 0.5326035567516456,
+      "grad_norm": 0.0,
+      "learning_rate": 9.430223693392558e-06,
+      "loss": 1.0232,
+      "step": 14645
+    },
+    {
+      "epoch": 0.5326399243553842,
+      "grad_norm": 0.0,
+      "learning_rate": 9.429047749163953e-06,
+      "loss": 0.9422,
+      "step": 14646
+    },
+    {
+      "epoch": 0.5326762919591228,
+      "grad_norm": 0.0,
+      "learning_rate": 9.427871812856503e-06,
+      "loss": 1.0279,
+      "step": 14647
+    },
+    {
+      "epoch": 0.5327126595628614,
+      "grad_norm": 0.0,
+      "learning_rate": 9.426695884486519e-06,
+      "loss": 1.0409,
+      "step": 14648
+    },
+    {
+      "epoch": 0.5327490271666,
+      "grad_norm": 0.0,
+      "learning_rate": 9.425519964070317e-06,
+      "loss": 1.0126,
+      "step": 14649
+    },
+    {
+      "epoch": 0.5327853947703386,
+      "grad_norm": 0.0,
+      "learning_rate": 9.42434405162421e-06,
+      "loss": 0.9362,
+      "step": 14650
+    },
+    {
+      "epoch": 0.5328217623740772,
+      "grad_norm": 0.0,
+      "learning_rate": 9.423168147164513e-06,
+      "loss": 1.0385,
+      "step": 14651
+    },
+    {
+      "epoch": 0.5328581299778158,
+      "grad_norm": 0.0,
+      "learning_rate": 9.421992250707538e-06,
+      "loss": 1.0864,
+      "step": 14652
+    },
+    {
+      "epoch": 0.5328944975815544,
+      "grad_norm": 0.0,
+      "learning_rate": 9.420816362269605e-06,
+      "loss": 1.048,
+      "step": 14653
+    },
+    {
+      "epoch": 0.532930865185293,
+      "grad_norm": 0.0,
+      "learning_rate": 9.419640481867018e-06,
+      "loss": 0.984,
+      "step": 14654
+    },
+    {
+      "epoch": 0.5329672327890316,
+      "grad_norm": 0.0,
+      "learning_rate": 9.4184646095161e-06,
+      "loss": 1.0035,
+      "step": 14655
+    },
+    {
+      "epoch": 0.5330036003927702,
+      "grad_norm": 0.0,
+      "learning_rate": 9.417288745233156e-06,
+      "loss": 0.9946,
+      "step": 14656
+    },
+    {
+      "epoch": 0.5330399679965087,
+      "grad_norm": 0.0,
+      "learning_rate": 9.416112889034506e-06,
+      "loss": 0.9458,
+      "step": 14657
+    },
+    {
+      "epoch": 0.5330763356002473,
+      "grad_norm": 0.0,
+      "learning_rate": 9.414937040936462e-06,
+      "loss": 1.07,
+      "step": 14658
+    },
+    {
+      "epoch": 0.5331127032039858,
+      "grad_norm": 0.0,
+      "learning_rate": 9.413761200955337e-06,
+      "loss": 1.0155,
+      "step": 14659
+    },
+    {
+      "epoch": 0.5331490708077244,
+      "grad_norm": 0.0,
+      "learning_rate": 9.412585369107444e-06,
+      "loss": 1.0007,
+      "step": 14660
+    },
+    {
+      "epoch": 0.533185438411463,
+      "grad_norm": 0.0,
+      "learning_rate": 9.411409545409094e-06,
+      "loss": 0.9572,
+      "step": 14661
+    },
+    {
+      "epoch": 0.5332218060152016,
+      "grad_norm": 0.0,
+      "learning_rate": 9.4102337298766e-06,
+      "loss": 0.9313,
+      "step": 14662
+    },
+    {
+      "epoch": 0.5332581736189402,
+      "grad_norm": 0.0,
+      "learning_rate": 9.409057922526278e-06,
+      "loss": 0.8771,
+      "step": 14663
+    },
+    {
+      "epoch": 0.5332945412226788,
+      "grad_norm": 0.0,
+      "learning_rate": 9.407882123374441e-06,
+      "loss": 0.9638,
+      "step": 14664
+    },
+    {
+      "epoch": 0.5333309088264174,
+      "grad_norm": 0.0,
+      "learning_rate": 9.406706332437395e-06,
+      "loss": 0.9855,
+      "step": 14665
+    },
+    {
+      "epoch": 0.533367276430156,
+      "grad_norm": 0.0,
+      "learning_rate": 9.405530549731461e-06,
+      "loss": 0.9847,
+      "step": 14666
+    },
+    {
+      "epoch": 0.5334036440338946,
+      "grad_norm": 0.0,
+      "learning_rate": 9.404354775272943e-06,
+      "loss": 1.0259,
+      "step": 14667
+    },
+    {
+      "epoch": 0.5334400116376332,
+      "grad_norm": 0.0,
+      "learning_rate": 9.403179009078161e-06,
+      "loss": 0.9856,
+      "step": 14668
+    },
+    {
+      "epoch": 0.5334763792413718,
+      "grad_norm": 0.0,
+      "learning_rate": 9.402003251163421e-06,
+      "loss": 0.925,
+      "step": 14669
+    },
+    {
+      "epoch": 0.5335127468451104,
+      "grad_norm": 0.0,
+      "learning_rate": 9.400827501545039e-06,
+      "loss": 0.9768,
+      "step": 14670
+    },
+    {
+      "epoch": 0.533549114448849,
+      "grad_norm": 0.0,
+      "learning_rate": 9.399651760239323e-06,
+      "loss": 1.0553,
+      "step": 14671
+    },
+    {
+      "epoch": 0.5335854820525876,
+      "grad_norm": 0.0,
+      "learning_rate": 9.398476027262591e-06,
+      "loss": 0.939,
+      "step": 14672
+    },
+    {
+      "epoch": 0.5336218496563262,
+      "grad_norm": 0.0,
+      "learning_rate": 9.397300302631149e-06,
+      "loss": 1.1064,
+      "step": 14673
+    },
+    {
+      "epoch": 0.5336582172600647,
+      "grad_norm": 0.0,
+      "learning_rate": 9.39612458636131e-06,
+      "loss": 0.9473,
+      "step": 14674
+    },
+    {
+      "epoch": 0.5336945848638033,
+      "grad_norm": 0.0,
+      "learning_rate": 9.394948878469383e-06,
+      "loss": 1.0429,
+      "step": 14675
+    },
+    {
+      "epoch": 0.5337309524675419,
+      "grad_norm": 0.0,
+      "learning_rate": 9.393773178971686e-06,
+      "loss": 1.0332,
+      "step": 14676
+    },
+    {
+      "epoch": 0.5337673200712805,
+      "grad_norm": 0.0,
+      "learning_rate": 9.392597487884522e-06,
+      "loss": 1.0734,
+      "step": 14677
+    },
+    {
+      "epoch": 0.5338036876750191,
+      "grad_norm": 0.0,
+      "learning_rate": 9.39142180522421e-06,
+      "loss": 1.1015,
+      "step": 14678
+    },
+    {
+      "epoch": 0.5338400552787577,
+      "grad_norm": 0.0,
+      "learning_rate": 9.390246131007057e-06,
+      "loss": 0.9451,
+      "step": 14679
+    },
+    {
+      "epoch": 0.5338764228824963,
+      "grad_norm": 0.0,
+      "learning_rate": 9.389070465249371e-06,
+      "loss": 0.9504,
+      "step": 14680
+    },
+    {
+      "epoch": 0.5339127904862349,
+      "grad_norm": 0.0,
+      "learning_rate": 9.387894807967466e-06,
+      "loss": 0.9712,
+      "step": 14681
+    },
+    {
+      "epoch": 0.5339491580899735,
+      "grad_norm": 0.0,
+      "learning_rate": 9.386719159177652e-06,
+      "loss": 1.0849,
+      "step": 14682
+    },
+    {
+      "epoch": 0.533985525693712,
+      "grad_norm": 0.0,
+      "learning_rate": 9.38554351889624e-06,
+      "loss": 1.0385,
+      "step": 14683
+    },
+    {
+      "epoch": 0.5340218932974506,
+      "grad_norm": 0.0,
+      "learning_rate": 9.38436788713954e-06,
+      "loss": 0.9667,
+      "step": 14684
+    },
+    {
+      "epoch": 0.5340582609011892,
+      "grad_norm": 0.0,
+      "learning_rate": 9.383192263923865e-06,
+      "loss": 1.0704,
+      "step": 14685
+    },
+    {
+      "epoch": 0.5340946285049278,
+      "grad_norm": 0.0,
+      "learning_rate": 9.382016649265518e-06,
+      "loss": 0.9277,
+      "step": 14686
+    },
+    {
+      "epoch": 0.5341309961086664,
+      "grad_norm": 0.0,
+      "learning_rate": 9.380841043180815e-06,
+      "loss": 1.0208,
+      "step": 14687
+    },
+    {
+      "epoch": 0.534167363712405,
+      "grad_norm": 0.0,
+      "learning_rate": 9.379665445686062e-06,
+      "loss": 0.9317,
+      "step": 14688
+    },
+    {
+      "epoch": 0.5342037313161436,
+      "grad_norm": 0.0,
+      "learning_rate": 9.378489856797574e-06,
+      "loss": 1.0556,
+      "step": 14689
+    },
+    {
+      "epoch": 0.5342400989198822,
+      "grad_norm": 0.0,
+      "learning_rate": 9.377314276531653e-06,
+      "loss": 0.9203,
+      "step": 14690
+    },
+    {
+      "epoch": 0.5342764665236207,
+      "grad_norm": 0.0,
+      "learning_rate": 9.376138704904619e-06,
+      "loss": 1.0477,
+      "step": 14691
+    },
+    {
+      "epoch": 0.5343128341273593,
+      "grad_norm": 0.0,
+      "learning_rate": 9.374963141932769e-06,
+      "loss": 0.9044,
+      "step": 14692
+    },
+    {
+      "epoch": 0.5343492017310979,
+      "grad_norm": 0.0,
+      "learning_rate": 9.373787587632424e-06,
+      "loss": 0.934,
+      "step": 14693
+    },
+    {
+      "epoch": 0.5343855693348365,
+      "grad_norm": 0.0,
+      "learning_rate": 9.372612042019883e-06,
+      "loss": 0.9208,
+      "step": 14694
+    },
+    {
+      "epoch": 0.5344219369385751,
+      "grad_norm": 0.0,
+      "learning_rate": 9.371436505111464e-06,
+      "loss": 0.8873,
+      "step": 14695
+    },
+    {
+      "epoch": 0.5344583045423137,
+      "grad_norm": 0.0,
+      "learning_rate": 9.370260976923469e-06,
+      "loss": 1.0731,
+      "step": 14696
+    },
+    {
+      "epoch": 0.5344946721460523,
+      "grad_norm": 0.0,
+      "learning_rate": 9.36908545747221e-06,
+      "loss": 1.0373,
+      "step": 14697
+    },
+    {
+      "epoch": 0.5345310397497909,
+      "grad_norm": 0.0,
+      "learning_rate": 9.367909946774e-06,
+      "loss": 0.9817,
+      "step": 14698
+    },
+    {
+      "epoch": 0.5345674073535295,
+      "grad_norm": 0.0,
+      "learning_rate": 9.366734444845135e-06,
+      "loss": 0.971,
+      "step": 14699
+    },
+    {
+      "epoch": 0.5346037749572681,
+      "grad_norm": 0.0,
+      "learning_rate": 9.365558951701936e-06,
+      "loss": 0.9339,
+      "step": 14700
+    },
+    {
+      "epoch": 0.5346401425610067,
+      "grad_norm": 0.0,
+      "learning_rate": 9.364383467360703e-06,
+      "loss": 0.9786,
+      "step": 14701
+    },
+    {
+      "epoch": 0.5346765101647453,
+      "grad_norm": 0.0,
+      "learning_rate": 9.363207991837749e-06,
+      "loss": 0.9926,
+      "step": 14702
+    },
+    {
+      "epoch": 0.5347128777684839,
+      "grad_norm": 0.0,
+      "learning_rate": 9.362032525149379e-06,
+      "loss": 0.9414,
+      "step": 14703
+    },
+    {
+      "epoch": 0.5347492453722225,
+      "grad_norm": 0.0,
+      "learning_rate": 9.360857067311909e-06,
+      "loss": 0.9357,
+      "step": 14704
+    },
+    {
+      "epoch": 0.534785612975961,
+      "grad_norm": 0.0,
+      "learning_rate": 9.35968161834163e-06,
+      "loss": 0.8946,
+      "step": 14705
+    },
+    {
+      "epoch": 0.5348219805796997,
+      "grad_norm": 0.0,
+      "learning_rate": 9.358506178254868e-06,
+      "loss": 0.9426,
+      "step": 14706
+    },
+    {
+      "epoch": 0.5348583481834382,
+      "grad_norm": 0.0,
+      "learning_rate": 9.357330747067921e-06,
+      "loss": 1.0506,
+      "step": 14707
+    },
+    {
+      "epoch": 0.5348947157871767,
+      "grad_norm": 0.0,
+      "learning_rate": 9.356155324797097e-06,
+      "loss": 1.0634,
+      "step": 14708
+    },
+    {
+      "epoch": 0.5349310833909153,
+      "grad_norm": 0.0,
+      "learning_rate": 9.354979911458704e-06,
+      "loss": 0.915,
+      "step": 14709
+    },
+    {
+      "epoch": 0.5349674509946539,
+      "grad_norm": 0.0,
+      "learning_rate": 9.353804507069051e-06,
+      "loss": 1.0436,
+      "step": 14710
+    },
+    {
+      "epoch": 0.5350038185983925,
+      "grad_norm": 0.0,
+      "learning_rate": 9.352629111644442e-06,
+      "loss": 1.0604,
+      "step": 14711
+    },
+    {
+      "epoch": 0.5350401862021311,
+      "grad_norm": 0.0,
+      "learning_rate": 9.35145372520119e-06,
+      "loss": 1.0627,
+      "step": 14712
+    },
+    {
+      "epoch": 0.5350765538058697,
+      "grad_norm": 0.0,
+      "learning_rate": 9.350278347755591e-06,
+      "loss": 1.0352,
+      "step": 14713
+    },
+    {
+      "epoch": 0.5351129214096083,
+      "grad_norm": 0.0,
+      "learning_rate": 9.349102979323961e-06,
+      "loss": 0.9844,
+      "step": 14714
+    },
+    {
+      "epoch": 0.5351492890133469,
+      "grad_norm": 0.0,
+      "learning_rate": 9.347927619922604e-06,
+      "loss": 0.9878,
+      "step": 14715
+    },
+    {
+      "epoch": 0.5351856566170855,
+      "grad_norm": 0.0,
+      "learning_rate": 9.346752269567826e-06,
+      "loss": 0.9068,
+      "step": 14716
+    },
+    {
+      "epoch": 0.5352220242208241,
+      "grad_norm": 0.0,
+      "learning_rate": 9.345576928275936e-06,
+      "loss": 0.9059,
+      "step": 14717
+    },
+    {
+      "epoch": 0.5352583918245627,
+      "grad_norm": 0.0,
+      "learning_rate": 9.344401596063234e-06,
+      "loss": 0.8648,
+      "step": 14718
+    },
+    {
+      "epoch": 0.5352947594283013,
+      "grad_norm": 0.0,
+      "learning_rate": 9.343226272946032e-06,
+      "loss": 1.032,
+      "step": 14719
+    },
+    {
+      "epoch": 0.5353311270320399,
+      "grad_norm": 0.0,
+      "learning_rate": 9.34205095894063e-06,
+      "loss": 0.9432,
+      "step": 14720
+    },
+    {
+      "epoch": 0.5353674946357785,
+      "grad_norm": 0.0,
+      "learning_rate": 9.34087565406334e-06,
+      "loss": 1.005,
+      "step": 14721
+    },
+    {
+      "epoch": 0.5354038622395171,
+      "grad_norm": 0.0,
+      "learning_rate": 9.339700358330464e-06,
+      "loss": 0.9694,
+      "step": 14722
+    },
+    {
+      "epoch": 0.5354402298432557,
+      "grad_norm": 0.0,
+      "learning_rate": 9.33852507175831e-06,
+      "loss": 0.9232,
+      "step": 14723
+    },
+    {
+      "epoch": 0.5354765974469942,
+      "grad_norm": 0.0,
+      "learning_rate": 9.33734979436318e-06,
+      "loss": 0.9413,
+      "step": 14724
+    },
+    {
+      "epoch": 0.5355129650507328,
+      "grad_norm": 0.0,
+      "learning_rate": 9.336174526161386e-06,
+      "loss": 0.9156,
+      "step": 14725
+    },
+    {
+      "epoch": 0.5355493326544714,
+      "grad_norm": 0.0,
+      "learning_rate": 9.334999267169223e-06,
+      "loss": 0.9443,
+      "step": 14726
+    },
+    {
+      "epoch": 0.53558570025821,
+      "grad_norm": 0.0,
+      "learning_rate": 9.333824017403004e-06,
+      "loss": 0.9046,
+      "step": 14727
+    },
+    {
+      "epoch": 0.5356220678619485,
+      "grad_norm": 0.0,
+      "learning_rate": 9.332648776879031e-06,
+      "loss": 0.9902,
+      "step": 14728
+    },
+    {
+      "epoch": 0.5356584354656871,
+      "grad_norm": 0.0,
+      "learning_rate": 9.331473545613611e-06,
+      "loss": 1.0027,
+      "step": 14729
+    },
+    {
+      "epoch": 0.5356948030694257,
+      "grad_norm": 0.0,
+      "learning_rate": 9.330298323623045e-06,
+      "loss": 1.0627,
+      "step": 14730
+    },
+    {
+      "epoch": 0.5357311706731643,
+      "grad_norm": 0.0,
+      "learning_rate": 9.329123110923643e-06,
+      "loss": 1.0303,
+      "step": 14731
+    },
+    {
+      "epoch": 0.5357675382769029,
+      "grad_norm": 0.0,
+      "learning_rate": 9.3279479075317e-06,
+      "loss": 1.0487,
+      "step": 14732
+    },
+    {
+      "epoch": 0.5358039058806415,
+      "grad_norm": 0.0,
+      "learning_rate": 9.326772713463534e-06,
+      "loss": 0.9825,
+      "step": 14733
+    },
+    {
+      "epoch": 0.5358402734843801,
+      "grad_norm": 0.0,
+      "learning_rate": 9.325597528735436e-06,
+      "loss": 0.8693,
+      "step": 14734
+    },
+    {
+      "epoch": 0.5358766410881187,
+      "grad_norm": 0.0,
+      "learning_rate": 9.324422353363718e-06,
+      "loss": 1.0066,
+      "step": 14735
+    },
+    {
+      "epoch": 0.5359130086918573,
+      "grad_norm": 0.0,
+      "learning_rate": 9.323247187364682e-06,
+      "loss": 0.8566,
+      "step": 14736
+    },
+    {
+      "epoch": 0.5359493762955959,
+      "grad_norm": 0.0,
+      "learning_rate": 9.322072030754626e-06,
+      "loss": 0.9873,
+      "step": 14737
+    },
+    {
+      "epoch": 0.5359857438993345,
+      "grad_norm": 0.0,
+      "learning_rate": 9.320896883549864e-06,
+      "loss": 0.9876,
+      "step": 14738
+    },
+    {
+      "epoch": 0.5360221115030731,
+      "grad_norm": 0.0,
+      "learning_rate": 9.319721745766693e-06,
+      "loss": 0.9164,
+      "step": 14739
+    },
+    {
+      "epoch": 0.5360584791068117,
+      "grad_norm": 0.0,
+      "learning_rate": 9.318546617421417e-06,
+      "loss": 0.9617,
+      "step": 14740
+    },
+    {
+      "epoch": 0.5360948467105502,
+      "grad_norm": 0.0,
+      "learning_rate": 9.317371498530337e-06,
+      "loss": 0.952,
+      "step": 14741
+    },
+    {
+      "epoch": 0.5361312143142888,
+      "grad_norm": 0.0,
+      "learning_rate": 9.316196389109763e-06,
+      "loss": 0.9764,
+      "step": 14742
+    },
+    {
+      "epoch": 0.5361675819180274,
+      "grad_norm": 0.0,
+      "learning_rate": 9.31502128917599e-06,
+      "loss": 0.969,
+      "step": 14743
+    },
+    {
+      "epoch": 0.536203949521766,
+      "grad_norm": 0.0,
+      "learning_rate": 9.31384619874533e-06,
+      "loss": 0.9633,
+      "step": 14744
+    },
+    {
+      "epoch": 0.5362403171255046,
+      "grad_norm": 0.0,
+      "learning_rate": 9.312671117834076e-06,
+      "loss": 0.9234,
+      "step": 14745
+    },
+    {
+      "epoch": 0.5362766847292432,
+      "grad_norm": 0.0,
+      "learning_rate": 9.311496046458539e-06,
+      "loss": 0.9857,
+      "step": 14746
+    },
+    {
+      "epoch": 0.5363130523329818,
+      "grad_norm": 0.0,
+      "learning_rate": 9.310320984635014e-06,
+      "loss": 1.0659,
+      "step": 14747
+    },
+    {
+      "epoch": 0.5363494199367204,
+      "grad_norm": 0.0,
+      "learning_rate": 9.309145932379808e-06,
+      "loss": 0.9805,
+      "step": 14748
+    },
+    {
+      "epoch": 0.536385787540459,
+      "grad_norm": 0.0,
+      "learning_rate": 9.307970889709221e-06,
+      "loss": 1.0174,
+      "step": 14749
+    },
+    {
+      "epoch": 0.5364221551441976,
+      "grad_norm": 0.0,
+      "learning_rate": 9.306795856639558e-06,
+      "loss": 0.8191,
+      "step": 14750
+    },
+    {
+      "epoch": 0.5364585227479362,
+      "grad_norm": 0.0,
+      "learning_rate": 9.305620833187117e-06,
+      "loss": 0.9295,
+      "step": 14751
+    },
+    {
+      "epoch": 0.5364948903516747,
+      "grad_norm": 0.0,
+      "learning_rate": 9.304445819368206e-06,
+      "loss": 0.884,
+      "step": 14752
+    },
+    {
+      "epoch": 0.5365312579554133,
+      "grad_norm": 0.0,
+      "learning_rate": 9.303270815199117e-06,
+      "loss": 1.0509,
+      "step": 14753
+    },
+    {
+      "epoch": 0.5365676255591519,
+      "grad_norm": 0.0,
+      "learning_rate": 9.30209582069616e-06,
+      "loss": 1.119,
+      "step": 14754
+    },
+    {
+      "epoch": 0.5366039931628905,
+      "grad_norm": 0.0,
+      "learning_rate": 9.300920835875633e-06,
+      "loss": 0.9389,
+      "step": 14755
+    },
+    {
+      "epoch": 0.5366403607666291,
+      "grad_norm": 0.0,
+      "learning_rate": 9.299745860753836e-06,
+      "loss": 0.9597,
+      "step": 14756
+    },
+    {
+      "epoch": 0.5366767283703677,
+      "grad_norm": 0.0,
+      "learning_rate": 9.298570895347075e-06,
+      "loss": 0.9317,
+      "step": 14757
+    },
+    {
+      "epoch": 0.5367130959741062,
+      "grad_norm": 0.0,
+      "learning_rate": 9.297395939671645e-06,
+      "loss": 1.0408,
+      "step": 14758
+    },
+    {
+      "epoch": 0.5367494635778448,
+      "grad_norm": 0.0,
+      "learning_rate": 9.29622099374385e-06,
+      "loss": 0.9713,
+      "step": 14759
+    },
+    {
+      "epoch": 0.5367858311815834,
+      "grad_norm": 0.0,
+      "learning_rate": 9.295046057579988e-06,
+      "loss": 1.0561,
+      "step": 14760
+    },
+    {
+      "epoch": 0.536822198785322,
+      "grad_norm": 0.0,
+      "learning_rate": 9.293871131196364e-06,
+      "loss": 0.9683,
+      "step": 14761
+    },
+    {
+      "epoch": 0.5368585663890606,
+      "grad_norm": 0.0,
+      "learning_rate": 9.292696214609274e-06,
+      "loss": 1.0309,
+      "step": 14762
+    },
+    {
+      "epoch": 0.5368949339927992,
+      "grad_norm": 0.0,
+      "learning_rate": 9.291521307835026e-06,
+      "loss": 0.9596,
+      "step": 14763
+    },
+    {
+      "epoch": 0.5369313015965378,
+      "grad_norm": 0.0,
+      "learning_rate": 9.290346410889907e-06,
+      "loss": 0.9776,
+      "step": 14764
+    },
+    {
+      "epoch": 0.5369676692002764,
+      "grad_norm": 0.0,
+      "learning_rate": 9.289171523790231e-06,
+      "loss": 0.8767,
+      "step": 14765
+    },
+    {
+      "epoch": 0.537004036804015,
+      "grad_norm": 0.0,
+      "learning_rate": 9.287996646552289e-06,
+      "loss": 0.9852,
+      "step": 14766
+    },
+    {
+      "epoch": 0.5370404044077536,
+      "grad_norm": 0.0,
+      "learning_rate": 9.286821779192382e-06,
+      "loss": 0.9629,
+      "step": 14767
+    },
+    {
+      "epoch": 0.5370767720114922,
+      "grad_norm": 0.0,
+      "learning_rate": 9.285646921726813e-06,
+      "loss": 0.9919,
+      "step": 14768
+    },
+    {
+      "epoch": 0.5371131396152308,
+      "grad_norm": 0.0,
+      "learning_rate": 9.284472074171879e-06,
+      "loss": 1.175,
+      "step": 14769
+    },
+    {
+      "epoch": 0.5371495072189694,
+      "grad_norm": 0.0,
+      "learning_rate": 9.28329723654388e-06,
+      "loss": 0.9442,
+      "step": 14770
+    },
+    {
+      "epoch": 0.537185874822708,
+      "grad_norm": 0.0,
+      "learning_rate": 9.282122408859116e-06,
+      "loss": 1.0203,
+      "step": 14771
+    },
+    {
+      "epoch": 0.5372222424264466,
+      "grad_norm": 0.0,
+      "learning_rate": 9.28094759113388e-06,
+      "loss": 0.9578,
+      "step": 14772
+    },
+    {
+      "epoch": 0.5372586100301852,
+      "grad_norm": 0.0,
+      "learning_rate": 9.279772783384483e-06,
+      "loss": 1.1027,
+      "step": 14773
+    },
+    {
+      "epoch": 0.5372949776339238,
+      "grad_norm": 0.0,
+      "learning_rate": 9.278597985627212e-06,
+      "loss": 1.0006,
+      "step": 14774
+    },
+    {
+      "epoch": 0.5373313452376622,
+      "grad_norm": 0.0,
+      "learning_rate": 9.277423197878372e-06,
+      "loss": 0.9568,
+      "step": 14775
+    },
+    {
+      "epoch": 0.5373677128414008,
+      "grad_norm": 0.0,
+      "learning_rate": 9.276248420154263e-06,
+      "loss": 0.8935,
+      "step": 14776
+    },
+    {
+      "epoch": 0.5374040804451394,
+      "grad_norm": 0.0,
+      "learning_rate": 9.275073652471173e-06,
+      "loss": 0.9081,
+      "step": 14777
+    },
+    {
+      "epoch": 0.537440448048878,
+      "grad_norm": 0.0,
+      "learning_rate": 9.273898894845415e-06,
+      "loss": 1.0961,
+      "step": 14778
+    },
+    {
+      "epoch": 0.5374768156526166,
+      "grad_norm": 0.0,
+      "learning_rate": 9.272724147293276e-06,
+      "loss": 0.8745,
+      "step": 14779
+    },
+    {
+      "epoch": 0.5375131832563552,
+      "grad_norm": 0.0,
+      "learning_rate": 9.271549409831057e-06,
+      "loss": 1.0959,
+      "step": 14780
+    },
+    {
+      "epoch": 0.5375495508600938,
+      "grad_norm": 0.0,
+      "learning_rate": 9.270374682475056e-06,
+      "loss": 1.0223,
+      "step": 14781
+    },
+    {
+      "epoch": 0.5375859184638324,
+      "grad_norm": 0.0,
+      "learning_rate": 9.269199965241574e-06,
+      "loss": 1.0282,
+      "step": 14782
+    },
+    {
+      "epoch": 0.537622286067571,
+      "grad_norm": 0.0,
+      "learning_rate": 9.268025258146903e-06,
+      "loss": 1.0797,
+      "step": 14783
+    },
+    {
+      "epoch": 0.5376586536713096,
+      "grad_norm": 0.0,
+      "learning_rate": 9.266850561207345e-06,
+      "loss": 1.1064,
+      "step": 14784
+    },
+    {
+      "epoch": 0.5376950212750482,
+      "grad_norm": 0.0,
+      "learning_rate": 9.265675874439195e-06,
+      "loss": 0.9714,
+      "step": 14785
+    },
+    {
+      "epoch": 0.5377313888787868,
+      "grad_norm": 0.0,
+      "learning_rate": 9.264501197858748e-06,
+      "loss": 0.9467,
+      "step": 14786
+    },
+    {
+      "epoch": 0.5377677564825254,
+      "grad_norm": 0.0,
+      "learning_rate": 9.263326531482304e-06,
+      "loss": 1.0247,
+      "step": 14787
+    },
+    {
+      "epoch": 0.537804124086264,
+      "grad_norm": 0.0,
+      "learning_rate": 9.26215187532616e-06,
+      "loss": 0.9312,
+      "step": 14788
+    },
+    {
+      "epoch": 0.5378404916900026,
+      "grad_norm": 0.0,
+      "learning_rate": 9.26097722940661e-06,
+      "loss": 0.9366,
+      "step": 14789
+    },
+    {
+      "epoch": 0.5378768592937412,
+      "grad_norm": 0.0,
+      "learning_rate": 9.25980259373996e-06,
+      "loss": 0.9975,
+      "step": 14790
+    },
+    {
+      "epoch": 0.5379132268974798,
+      "grad_norm": 0.0,
+      "learning_rate": 9.258627968342489e-06,
+      "loss": 1.1028,
+      "step": 14791
+    },
+    {
+      "epoch": 0.5379495945012183,
+      "grad_norm": 0.0,
+      "learning_rate": 9.25745335323051e-06,
+      "loss": 0.9836,
+      "step": 14792
+    },
+    {
+      "epoch": 0.5379859621049569,
+      "grad_norm": 0.0,
+      "learning_rate": 9.256278748420307e-06,
+      "loss": 1.0161,
+      "step": 14793
+    },
+    {
+      "epoch": 0.5380223297086955,
+      "grad_norm": 0.0,
+      "learning_rate": 9.255104153928185e-06,
+      "loss": 0.9951,
+      "step": 14794
+    },
+    {
+      "epoch": 0.538058697312434,
+      "grad_norm": 0.0,
+      "learning_rate": 9.253929569770436e-06,
+      "loss": 0.9442,
+      "step": 14795
+    },
+    {
+      "epoch": 0.5380950649161726,
+      "grad_norm": 0.0,
+      "learning_rate": 9.252754995963352e-06,
+      "loss": 0.9856,
+      "step": 14796
+    },
+    {
+      "epoch": 0.5381314325199112,
+      "grad_norm": 0.0,
+      "learning_rate": 9.251580432523237e-06,
+      "loss": 0.9105,
+      "step": 14797
+    },
+    {
+      "epoch": 0.5381678001236498,
+      "grad_norm": 0.0,
+      "learning_rate": 9.250405879466378e-06,
+      "loss": 0.801,
+      "step": 14798
+    },
+    {
+      "epoch": 0.5382041677273884,
+      "grad_norm": 0.0,
+      "learning_rate": 9.249231336809076e-06,
+      "loss": 0.9813,
+      "step": 14799
+    },
+    {
+      "epoch": 0.538240535331127,
+      "grad_norm": 0.0,
+      "learning_rate": 9.248056804567621e-06,
+      "loss": 0.9867,
+      "step": 14800
+    },
+    {
+      "epoch": 0.5382769029348656,
+      "grad_norm": 0.0,
+      "learning_rate": 9.246882282758314e-06,
+      "loss": 1.0732,
+      "step": 14801
+    },
+    {
+      "epoch": 0.5383132705386042,
+      "grad_norm": 0.0,
+      "learning_rate": 9.245707771397445e-06,
+      "loss": 0.9694,
+      "step": 14802
+    },
+    {
+      "epoch": 0.5383496381423428,
+      "grad_norm": 0.0,
+      "learning_rate": 9.244533270501313e-06,
+      "loss": 0.9554,
+      "step": 14803
+    },
+    {
+      "epoch": 0.5383860057460814,
+      "grad_norm": 0.0,
+      "learning_rate": 9.243358780086205e-06,
+      "loss": 0.9007,
+      "step": 14804
+    },
+    {
+      "epoch": 0.53842237334982,
+      "grad_norm": 0.0,
+      "learning_rate": 9.242184300168426e-06,
+      "loss": 1.0317,
+      "step": 14805
+    },
+    {
+      "epoch": 0.5384587409535586,
+      "grad_norm": 0.0,
+      "learning_rate": 9.241009830764261e-06,
+      "loss": 0.9229,
+      "step": 14806
+    },
+    {
+      "epoch": 0.5384951085572972,
+      "grad_norm": 0.0,
+      "learning_rate": 9.23983537189001e-06,
+      "loss": 1.0343,
+      "step": 14807
+    },
+    {
+      "epoch": 0.5385314761610358,
+      "grad_norm": 0.0,
+      "learning_rate": 9.238660923561963e-06,
+      "loss": 0.9933,
+      "step": 14808
+    },
+    {
+      "epoch": 0.5385678437647743,
+      "grad_norm": 0.0,
+      "learning_rate": 9.237486485796417e-06,
+      "loss": 0.9474,
+      "step": 14809
+    },
+    {
+      "epoch": 0.5386042113685129,
+      "grad_norm": 0.0,
+      "learning_rate": 9.236312058609662e-06,
+      "loss": 0.998,
+      "step": 14810
+    },
+    {
+      "epoch": 0.5386405789722515,
+      "grad_norm": 0.0,
+      "learning_rate": 9.235137642017998e-06,
+      "loss": 0.9569,
+      "step": 14811
+    },
+    {
+      "epoch": 0.5386769465759901,
+      "grad_norm": 0.0,
+      "learning_rate": 9.233963236037708e-06,
+      "loss": 1.0096,
+      "step": 14812
+    },
+    {
+      "epoch": 0.5387133141797287,
+      "grad_norm": 0.0,
+      "learning_rate": 9.232788840685096e-06,
+      "loss": 1.1584,
+      "step": 14813
+    },
+    {
+      "epoch": 0.5387496817834673,
+      "grad_norm": 0.0,
+      "learning_rate": 9.231614455976448e-06,
+      "loss": 1.0963,
+      "step": 14814
+    },
+    {
+      "epoch": 0.5387860493872059,
+      "grad_norm": 0.0,
+      "learning_rate": 9.230440081928059e-06,
+      "loss": 0.8898,
+      "step": 14815
+    },
+    {
+      "epoch": 0.5388224169909445,
+      "grad_norm": 0.0,
+      "learning_rate": 9.229265718556226e-06,
+      "loss": 1.0052,
+      "step": 14816
+    },
+    {
+      "epoch": 0.5388587845946831,
+      "grad_norm": 0.0,
+      "learning_rate": 9.228091365877231e-06,
+      "loss": 0.9608,
+      "step": 14817
+    },
+    {
+      "epoch": 0.5388951521984217,
+      "grad_norm": 0.0,
+      "learning_rate": 9.226917023907379e-06,
+      "loss": 0.9625,
+      "step": 14818
+    },
+    {
+      "epoch": 0.5389315198021603,
+      "grad_norm": 0.0,
+      "learning_rate": 9.225742692662953e-06,
+      "loss": 1.0697,
+      "step": 14819
+    },
+    {
+      "epoch": 0.5389678874058988,
+      "grad_norm": 0.0,
+      "learning_rate": 9.22456837216025e-06,
+      "loss": 0.959,
+      "step": 14820
+    },
+    {
+      "epoch": 0.5390042550096374,
+      "grad_norm": 0.0,
+      "learning_rate": 9.223394062415558e-06,
+      "loss": 0.9874,
+      "step": 14821
+    },
+    {
+      "epoch": 0.539040622613376,
+      "grad_norm": 0.0,
+      "learning_rate": 9.222219763445174e-06,
+      "loss": 0.9692,
+      "step": 14822
+    },
+    {
+      "epoch": 0.5390769902171146,
+      "grad_norm": 0.0,
+      "learning_rate": 9.221045475265386e-06,
+      "loss": 0.9561,
+      "step": 14823
+    },
+    {
+      "epoch": 0.5391133578208532,
+      "grad_norm": 0.0,
+      "learning_rate": 9.21987119789249e-06,
+      "loss": 0.9801,
+      "step": 14824
+    },
+    {
+      "epoch": 0.5391497254245917,
+      "grad_norm": 0.0,
+      "learning_rate": 9.218696931342773e-06,
+      "loss": 1.0744,
+      "step": 14825
+    },
+    {
+      "epoch": 0.5391860930283303,
+      "grad_norm": 0.0,
+      "learning_rate": 9.217522675632527e-06,
+      "loss": 0.935,
+      "step": 14826
+    },
+    {
+      "epoch": 0.5392224606320689,
+      "grad_norm": 0.0,
+      "learning_rate": 9.216348430778041e-06,
+      "loss": 1.024,
+      "step": 14827
+    },
+    {
+      "epoch": 0.5392588282358075,
+      "grad_norm": 0.0,
+      "learning_rate": 9.215174196795612e-06,
+      "loss": 0.9519,
+      "step": 14828
+    },
+    {
+      "epoch": 0.5392951958395461,
+      "grad_norm": 0.0,
+      "learning_rate": 9.213999973701525e-06,
+      "loss": 0.9174,
+      "step": 14829
+    },
+    {
+      "epoch": 0.5393315634432847,
+      "grad_norm": 0.0,
+      "learning_rate": 9.212825761512077e-06,
+      "loss": 0.9807,
+      "step": 14830
+    },
+    {
+      "epoch": 0.5393679310470233,
+      "grad_norm": 0.0,
+      "learning_rate": 9.211651560243549e-06,
+      "loss": 0.9983,
+      "step": 14831
+    },
+    {
+      "epoch": 0.5394042986507619,
+      "grad_norm": 0.0,
+      "learning_rate": 9.210477369912244e-06,
+      "loss": 0.976,
+      "step": 14832
+    },
+    {
+      "epoch": 0.5394406662545005,
+      "grad_norm": 0.0,
+      "learning_rate": 9.209303190534442e-06,
+      "loss": 0.8891,
+      "step": 14833
+    },
+    {
+      "epoch": 0.5394770338582391,
+      "grad_norm": 0.0,
+      "learning_rate": 9.208129022126435e-06,
+      "loss": 1.0749,
+      "step": 14834
+    },
+    {
+      "epoch": 0.5395134014619777,
+      "grad_norm": 0.0,
+      "learning_rate": 9.206954864704519e-06,
+      "loss": 0.9485,
+      "step": 14835
+    },
+    {
+      "epoch": 0.5395497690657163,
+      "grad_norm": 0.0,
+      "learning_rate": 9.20578071828497e-06,
+      "loss": 1.0349,
+      "step": 14836
+    },
+    {
+      "epoch": 0.5395861366694549,
+      "grad_norm": 0.0,
+      "learning_rate": 9.204606582884096e-06,
+      "loss": 1.0197,
+      "step": 14837
+    },
+    {
+      "epoch": 0.5396225042731935,
+      "grad_norm": 0.0,
+      "learning_rate": 9.203432458518172e-06,
+      "loss": 0.9556,
+      "step": 14838
+    },
+    {
+      "epoch": 0.5396588718769321,
+      "grad_norm": 0.0,
+      "learning_rate": 9.202258345203494e-06,
+      "loss": 1.0398,
+      "step": 14839
+    },
+    {
+      "epoch": 0.5396952394806707,
+      "grad_norm": 0.0,
+      "learning_rate": 9.201084242956348e-06,
+      "loss": 1.012,
+      "step": 14840
+    },
+    {
+      "epoch": 0.5397316070844093,
+      "grad_norm": 0.0,
+      "learning_rate": 9.199910151793026e-06,
+      "loss": 0.9987,
+      "step": 14841
+    },
+    {
+      "epoch": 0.5397679746881477,
+      "grad_norm": 0.0,
+      "learning_rate": 9.198736071729815e-06,
+      "loss": 0.9858,
+      "step": 14842
+    },
+    {
+      "epoch": 0.5398043422918863,
+      "grad_norm": 0.0,
+      "learning_rate": 9.197562002783009e-06,
+      "loss": 1.0787,
+      "step": 14843
+    },
+    {
+      "epoch": 0.5398407098956249,
+      "grad_norm": 0.0,
+      "learning_rate": 9.196387944968882e-06,
+      "loss": 0.9513,
+      "step": 14844
+    },
+    {
+      "epoch": 0.5398770774993635,
+      "grad_norm": 0.0,
+      "learning_rate": 9.195213898303742e-06,
+      "loss": 1.0307,
+      "step": 14845
+    },
+    {
+      "epoch": 0.5399134451031021,
+      "grad_norm": 0.0,
+      "learning_rate": 9.19403986280386e-06,
+      "loss": 0.9815,
+      "step": 14846
+    },
+    {
+      "epoch": 0.5399498127068407,
+      "grad_norm": 0.0,
+      "learning_rate": 9.192865838485535e-06,
+      "loss": 0.9925,
+      "step": 14847
+    },
+    {
+      "epoch": 0.5399861803105793,
+      "grad_norm": 0.0,
+      "learning_rate": 9.191691825365049e-06,
+      "loss": 0.9921,
+      "step": 14848
+    },
+    {
+      "epoch": 0.5400225479143179,
+      "grad_norm": 0.0,
+      "learning_rate": 9.190517823458695e-06,
+      "loss": 0.9374,
+      "step": 14849
+    },
+    {
+      "epoch": 0.5400589155180565,
+      "grad_norm": 0.0,
+      "learning_rate": 9.189343832782752e-06,
+      "loss": 0.8722,
+      "step": 14850
+    },
+    {
+      "epoch": 0.5400952831217951,
+      "grad_norm": 0.0,
+      "learning_rate": 9.188169853353522e-06,
+      "loss": 0.973,
+      "step": 14851
+    },
+    {
+      "epoch": 0.5401316507255337,
+      "grad_norm": 0.0,
+      "learning_rate": 9.18699588518728e-06,
+      "loss": 0.8943,
+      "step": 14852
+    },
+    {
+      "epoch": 0.5401680183292723,
+      "grad_norm": 0.0,
+      "learning_rate": 9.185821928300314e-06,
+      "loss": 1.0782,
+      "step": 14853
+    },
+    {
+      "epoch": 0.5402043859330109,
+      "grad_norm": 0.0,
+      "learning_rate": 9.184647982708915e-06,
+      "loss": 1.0193,
+      "step": 14854
+    },
+    {
+      "epoch": 0.5402407535367495,
+      "grad_norm": 0.0,
+      "learning_rate": 9.183474048429368e-06,
+      "loss": 1.1285,
+      "step": 14855
+    },
+    {
+      "epoch": 0.5402771211404881,
+      "grad_norm": 0.0,
+      "learning_rate": 9.182300125477963e-06,
+      "loss": 1.0701,
+      "step": 14856
+    },
+    {
+      "epoch": 0.5403134887442267,
+      "grad_norm": 0.0,
+      "learning_rate": 9.18112621387098e-06,
+      "loss": 0.9623,
+      "step": 14857
+    },
+    {
+      "epoch": 0.5403498563479653,
+      "grad_norm": 0.0,
+      "learning_rate": 9.17995231362471e-06,
+      "loss": 0.9888,
+      "step": 14858
+    },
+    {
+      "epoch": 0.5403862239517038,
+      "grad_norm": 0.0,
+      "learning_rate": 9.178778424755436e-06,
+      "loss": 0.9385,
+      "step": 14859
+    },
+    {
+      "epoch": 0.5404225915554424,
+      "grad_norm": 0.0,
+      "learning_rate": 9.17760454727945e-06,
+      "loss": 0.9451,
+      "step": 14860
+    },
+    {
+      "epoch": 0.540458959159181,
+      "grad_norm": 0.0,
+      "learning_rate": 9.17643068121303e-06,
+      "loss": 1.019,
+      "step": 14861
+    },
+    {
+      "epoch": 0.5404953267629196,
+      "grad_norm": 0.0,
+      "learning_rate": 9.175256826572471e-06,
+      "loss": 0.8912,
+      "step": 14862
+    },
+    {
+      "epoch": 0.5405316943666582,
+      "grad_norm": 0.0,
+      "learning_rate": 9.174082983374049e-06,
+      "loss": 0.9307,
+      "step": 14863
+    },
+    {
+      "epoch": 0.5405680619703968,
+      "grad_norm": 0.0,
+      "learning_rate": 9.172909151634057e-06,
+      "loss": 0.9786,
+      "step": 14864
+    },
+    {
+      "epoch": 0.5406044295741353,
+      "grad_norm": 0.0,
+      "learning_rate": 9.171735331368774e-06,
+      "loss": 0.9683,
+      "step": 14865
+    },
+    {
+      "epoch": 0.5406407971778739,
+      "grad_norm": 0.0,
+      "learning_rate": 9.170561522594492e-06,
+      "loss": 0.9763,
+      "step": 14866
+    },
+    {
+      "epoch": 0.5406771647816125,
+      "grad_norm": 0.0,
+      "learning_rate": 9.169387725327488e-06,
+      "loss": 1.0236,
+      "step": 14867
+    },
+    {
+      "epoch": 0.5407135323853511,
+      "grad_norm": 0.0,
+      "learning_rate": 9.168213939584052e-06,
+      "loss": 1.0179,
+      "step": 14868
+    },
+    {
+      "epoch": 0.5407498999890897,
+      "grad_norm": 0.0,
+      "learning_rate": 9.167040165380466e-06,
+      "loss": 0.9705,
+      "step": 14869
+    },
+    {
+      "epoch": 0.5407862675928283,
+      "grad_norm": 0.0,
+      "learning_rate": 9.165866402733021e-06,
+      "loss": 1.0542,
+      "step": 14870
+    },
+    {
+      "epoch": 0.5408226351965669,
+      "grad_norm": 0.0,
+      "learning_rate": 9.164692651657993e-06,
+      "loss": 1.0481,
+      "step": 14871
+    },
+    {
+      "epoch": 0.5408590028003055,
+      "grad_norm": 0.0,
+      "learning_rate": 9.163518912171669e-06,
+      "loss": 1.032,
+      "step": 14872
+    },
+    {
+      "epoch": 0.5408953704040441,
+      "grad_norm": 0.0,
+      "learning_rate": 9.162345184290334e-06,
+      "loss": 0.9528,
+      "step": 14873
+    },
+    {
+      "epoch": 0.5409317380077827,
+      "grad_norm": 0.0,
+      "learning_rate": 9.16117146803027e-06,
+      "loss": 0.9246,
+      "step": 14874
+    },
+    {
+      "epoch": 0.5409681056115213,
+      "grad_norm": 0.0,
+      "learning_rate": 9.159997763407765e-06,
+      "loss": 0.9893,
+      "step": 14875
+    },
+    {
+      "epoch": 0.5410044732152598,
+      "grad_norm": 0.0,
+      "learning_rate": 9.158824070439092e-06,
+      "loss": 1.1039,
+      "step": 14876
+    },
+    {
+      "epoch": 0.5410408408189984,
+      "grad_norm": 0.0,
+      "learning_rate": 9.15765038914055e-06,
+      "loss": 0.8806,
+      "step": 14877
+    },
+    {
+      "epoch": 0.541077208422737,
+      "grad_norm": 0.0,
+      "learning_rate": 9.156476719528409e-06,
+      "loss": 1.0292,
+      "step": 14878
+    },
+    {
+      "epoch": 0.5411135760264756,
+      "grad_norm": 0.0,
+      "learning_rate": 9.155303061618957e-06,
+      "loss": 1.0006,
+      "step": 14879
+    },
+    {
+      "epoch": 0.5411499436302142,
+      "grad_norm": 0.0,
+      "learning_rate": 9.154129415428477e-06,
+      "loss": 1.0056,
+      "step": 14880
+    },
+    {
+      "epoch": 0.5411863112339528,
+      "grad_norm": 0.0,
+      "learning_rate": 9.152955780973251e-06,
+      "loss": 1.1246,
+      "step": 14881
+    },
+    {
+      "epoch": 0.5412226788376914,
+      "grad_norm": 0.0,
+      "learning_rate": 9.151782158269562e-06,
+      "loss": 0.9878,
+      "step": 14882
+    },
+    {
+      "epoch": 0.54125904644143,
+      "grad_norm": 0.0,
+      "learning_rate": 9.150608547333695e-06,
+      "loss": 1.0799,
+      "step": 14883
+    },
+    {
+      "epoch": 0.5412954140451686,
+      "grad_norm": 0.0,
+      "learning_rate": 9.149434948181924e-06,
+      "loss": 1.0641,
+      "step": 14884
+    },
+    {
+      "epoch": 0.5413317816489072,
+      "grad_norm": 0.0,
+      "learning_rate": 9.14826136083054e-06,
+      "loss": 1.0704,
+      "step": 14885
+    },
+    {
+      "epoch": 0.5413681492526458,
+      "grad_norm": 0.0,
+      "learning_rate": 9.147087785295818e-06,
+      "loss": 1.0128,
+      "step": 14886
+    },
+    {
+      "epoch": 0.5414045168563844,
+      "grad_norm": 0.0,
+      "learning_rate": 9.145914221594046e-06,
+      "loss": 1.0389,
+      "step": 14887
+    },
+    {
+      "epoch": 0.541440884460123,
+      "grad_norm": 0.0,
+      "learning_rate": 9.1447406697415e-06,
+      "loss": 1.0332,
+      "step": 14888
+    },
+    {
+      "epoch": 0.5414772520638615,
+      "grad_norm": 0.0,
+      "learning_rate": 9.143567129754465e-06,
+      "loss": 0.9843,
+      "step": 14889
+    },
+    {
+      "epoch": 0.5415136196676001,
+      "grad_norm": 0.0,
+      "learning_rate": 9.142393601649224e-06,
+      "loss": 1.0599,
+      "step": 14890
+    },
+    {
+      "epoch": 0.5415499872713387,
+      "grad_norm": 0.0,
+      "learning_rate": 9.141220085442051e-06,
+      "loss": 0.8804,
+      "step": 14891
+    },
+    {
+      "epoch": 0.5415863548750773,
+      "grad_norm": 0.0,
+      "learning_rate": 9.140046581149232e-06,
+      "loss": 1.0099,
+      "step": 14892
+    },
+    {
+      "epoch": 0.5416227224788158,
+      "grad_norm": 0.0,
+      "learning_rate": 9.138873088787045e-06,
+      "loss": 1.0485,
+      "step": 14893
+    },
+    {
+      "epoch": 0.5416590900825544,
+      "grad_norm": 0.0,
+      "learning_rate": 9.137699608371775e-06,
+      "loss": 0.9737,
+      "step": 14894
+    },
+    {
+      "epoch": 0.541695457686293,
+      "grad_norm": 0.0,
+      "learning_rate": 9.136526139919696e-06,
+      "loss": 0.9398,
+      "step": 14895
+    },
+    {
+      "epoch": 0.5417318252900316,
+      "grad_norm": 0.0,
+      "learning_rate": 9.135352683447096e-06,
+      "loss": 0.983,
+      "step": 14896
+    },
+    {
+      "epoch": 0.5417681928937702,
+      "grad_norm": 0.0,
+      "learning_rate": 9.134179238970247e-06,
+      "loss": 1.0354,
+      "step": 14897
+    },
+    {
+      "epoch": 0.5418045604975088,
+      "grad_norm": 0.0,
+      "learning_rate": 9.133005806505435e-06,
+      "loss": 1.019,
+      "step": 14898
+    },
+    {
+      "epoch": 0.5418409281012474,
+      "grad_norm": 0.0,
+      "learning_rate": 9.131832386068934e-06,
+      "loss": 1.0434,
+      "step": 14899
+    },
+    {
+      "epoch": 0.541877295704986,
+      "grad_norm": 0.0,
+      "learning_rate": 9.130658977677031e-06,
+      "loss": 0.94,
+      "step": 14900
+    },
+    {
+      "epoch": 0.5419136633087246,
+      "grad_norm": 0.0,
+      "learning_rate": 9.129485581345997e-06,
+      "loss": 1.1027,
+      "step": 14901
+    },
+    {
+      "epoch": 0.5419500309124632,
+      "grad_norm": 0.0,
+      "learning_rate": 9.128312197092122e-06,
+      "loss": 0.9664,
+      "step": 14902
+    },
+    {
+      "epoch": 0.5419863985162018,
+      "grad_norm": 0.0,
+      "learning_rate": 9.127138824931669e-06,
+      "loss": 0.9016,
+      "step": 14903
+    },
+    {
+      "epoch": 0.5420227661199404,
+      "grad_norm": 0.0,
+      "learning_rate": 9.125965464880936e-06,
+      "loss": 0.9421,
+      "step": 14904
+    },
+    {
+      "epoch": 0.542059133723679,
+      "grad_norm": 0.0,
+      "learning_rate": 9.124792116956186e-06,
+      "loss": 1.1503,
+      "step": 14905
+    },
+    {
+      "epoch": 0.5420955013274176,
+      "grad_norm": 0.0,
+      "learning_rate": 9.123618781173706e-06,
+      "loss": 1.0809,
+      "step": 14906
+    },
+    {
+      "epoch": 0.5421318689311562,
+      "grad_norm": 0.0,
+      "learning_rate": 9.12244545754977e-06,
+      "loss": 0.9156,
+      "step": 14907
+    },
+    {
+      "epoch": 0.5421682365348948,
+      "grad_norm": 0.0,
+      "learning_rate": 9.121272146100661e-06,
+      "loss": 0.8621,
+      "step": 14908
+    },
+    {
+      "epoch": 0.5422046041386333,
+      "grad_norm": 0.0,
+      "learning_rate": 9.120098846842655e-06,
+      "loss": 1.0245,
+      "step": 14909
+    },
+    {
+      "epoch": 0.5422409717423718,
+      "grad_norm": 0.0,
+      "learning_rate": 9.118925559792025e-06,
+      "loss": 1.025,
+      "step": 14910
+    },
+    {
+      "epoch": 0.5422773393461104,
+      "grad_norm": 0.0,
+      "learning_rate": 9.117752284965056e-06,
+      "loss": 0.941,
+      "step": 14911
+    },
+    {
+      "epoch": 0.542313706949849,
+      "grad_norm": 0.0,
+      "learning_rate": 9.11657902237802e-06,
+      "loss": 1.0482,
+      "step": 14912
+    },
+    {
+      "epoch": 0.5423500745535876,
+      "grad_norm": 0.0,
+      "learning_rate": 9.115405772047199e-06,
+      "loss": 0.9861,
+      "step": 14913
+    },
+    {
+      "epoch": 0.5423864421573262,
+      "grad_norm": 0.0,
+      "learning_rate": 9.114232533988864e-06,
+      "loss": 0.9987,
+      "step": 14914
+    },
+    {
+      "epoch": 0.5424228097610648,
+      "grad_norm": 0.0,
+      "learning_rate": 9.113059308219301e-06,
+      "loss": 0.9119,
+      "step": 14915
+    },
+    {
+      "epoch": 0.5424591773648034,
+      "grad_norm": 0.0,
+      "learning_rate": 9.111886094754776e-06,
+      "loss": 0.9452,
+      "step": 14916
+    },
+    {
+      "epoch": 0.542495544968542,
+      "grad_norm": 0.0,
+      "learning_rate": 9.110712893611577e-06,
+      "loss": 0.9321,
+      "step": 14917
+    },
+    {
+      "epoch": 0.5425319125722806,
+      "grad_norm": 0.0,
+      "learning_rate": 9.109539704805971e-06,
+      "loss": 1.014,
+      "step": 14918
+    },
+    {
+      "epoch": 0.5425682801760192,
+      "grad_norm": 0.0,
+      "learning_rate": 9.10836652835424e-06,
+      "loss": 0.9389,
+      "step": 14919
+    },
+    {
+      "epoch": 0.5426046477797578,
+      "grad_norm": 0.0,
+      "learning_rate": 9.107193364272657e-06,
+      "loss": 1.0162,
+      "step": 14920
+    },
+    {
+      "epoch": 0.5426410153834964,
+      "grad_norm": 0.0,
+      "learning_rate": 9.106020212577501e-06,
+      "loss": 1.0705,
+      "step": 14921
+    },
+    {
+      "epoch": 0.542677382987235,
+      "grad_norm": 0.0,
+      "learning_rate": 9.104847073285043e-06,
+      "loss": 0.9853,
+      "step": 14922
+    },
+    {
+      "epoch": 0.5427137505909736,
+      "grad_norm": 0.0,
+      "learning_rate": 9.103673946411569e-06,
+      "loss": 0.9291,
+      "step": 14923
+    },
+    {
+      "epoch": 0.5427501181947122,
+      "grad_norm": 0.0,
+      "learning_rate": 9.10250083197334e-06,
+      "loss": 1.0039,
+      "step": 14924
+    },
+    {
+      "epoch": 0.5427864857984508,
+      "grad_norm": 0.0,
+      "learning_rate": 9.101327729986642e-06,
+      "loss": 1.0494,
+      "step": 14925
+    },
+    {
+      "epoch": 0.5428228534021893,
+      "grad_norm": 0.0,
+      "learning_rate": 9.100154640467745e-06,
+      "loss": 1.0256,
+      "step": 14926
+    },
+    {
+      "epoch": 0.5428592210059279,
+      "grad_norm": 0.0,
+      "learning_rate": 9.098981563432927e-06,
+      "loss": 1.0019,
+      "step": 14927
+    },
+    {
+      "epoch": 0.5428955886096665,
+      "grad_norm": 0.0,
+      "learning_rate": 9.097808498898462e-06,
+      "loss": 1.0053,
+      "step": 14928
+    },
+    {
+      "epoch": 0.5429319562134051,
+      "grad_norm": 0.0,
+      "learning_rate": 9.096635446880623e-06,
+      "loss": 1.0315,
+      "step": 14929
+    },
+    {
+      "epoch": 0.5429683238171437,
+      "grad_norm": 0.0,
+      "learning_rate": 9.095462407395686e-06,
+      "loss": 0.9651,
+      "step": 14930
+    },
+    {
+      "epoch": 0.5430046914208823,
+      "grad_norm": 0.0,
+      "learning_rate": 9.094289380459922e-06,
+      "loss": 0.9954,
+      "step": 14931
+    },
+    {
+      "epoch": 0.5430410590246209,
+      "grad_norm": 0.0,
+      "learning_rate": 9.09311636608961e-06,
+      "loss": 0.9503,
+      "step": 14932
+    },
+    {
+      "epoch": 0.5430774266283594,
+      "grad_norm": 0.0,
+      "learning_rate": 9.09194336430102e-06,
+      "loss": 0.8849,
+      "step": 14933
+    },
+    {
+      "epoch": 0.543113794232098,
+      "grad_norm": 0.0,
+      "learning_rate": 9.090770375110432e-06,
+      "loss": 1.0574,
+      "step": 14934
+    },
+    {
+      "epoch": 0.5431501618358366,
+      "grad_norm": 0.0,
+      "learning_rate": 9.08959739853411e-06,
+      "loss": 1.0219,
+      "step": 14935
+    },
+    {
+      "epoch": 0.5431865294395752,
+      "grad_norm": 0.0,
+      "learning_rate": 9.088424434588336e-06,
+      "loss": 1.0783,
+      "step": 14936
+    },
+    {
+      "epoch": 0.5432228970433138,
+      "grad_norm": 0.0,
+      "learning_rate": 9.087251483289376e-06,
+      "loss": 1.0355,
+      "step": 14937
+    },
+    {
+      "epoch": 0.5432592646470524,
+      "grad_norm": 0.0,
+      "learning_rate": 9.08607854465351e-06,
+      "loss": 1.0737,
+      "step": 14938
+    },
+    {
+      "epoch": 0.543295632250791,
+      "grad_norm": 0.0,
+      "learning_rate": 9.084905618697006e-06,
+      "loss": 0.9411,
+      "step": 14939
+    },
+    {
+      "epoch": 0.5433319998545296,
+      "grad_norm": 0.0,
+      "learning_rate": 9.083732705436138e-06,
+      "loss": 1.0682,
+      "step": 14940
+    },
+    {
+      "epoch": 0.5433683674582682,
+      "grad_norm": 0.0,
+      "learning_rate": 9.082559804887178e-06,
+      "loss": 0.9514,
+      "step": 14941
+    },
+    {
+      "epoch": 0.5434047350620068,
+      "grad_norm": 0.0,
+      "learning_rate": 9.081386917066403e-06,
+      "loss": 0.997,
+      "step": 14942
+    },
+    {
+      "epoch": 0.5434411026657453,
+      "grad_norm": 0.0,
+      "learning_rate": 9.080214041990074e-06,
+      "loss": 0.9471,
+      "step": 14943
+    },
+    {
+      "epoch": 0.5434774702694839,
+      "grad_norm": 0.0,
+      "learning_rate": 9.079041179674478e-06,
+      "loss": 0.9303,
+      "step": 14944
+    },
+    {
+      "epoch": 0.5435138378732225,
+      "grad_norm": 0.0,
+      "learning_rate": 9.077868330135874e-06,
+      "loss": 1.0392,
+      "step": 14945
+    },
+    {
+      "epoch": 0.5435502054769611,
+      "grad_norm": 0.0,
+      "learning_rate": 9.076695493390541e-06,
+      "loss": 1.0064,
+      "step": 14946
+    },
+    {
+      "epoch": 0.5435865730806997,
+      "grad_norm": 0.0,
+      "learning_rate": 9.07552266945475e-06,
+      "loss": 1.0004,
+      "step": 14947
+    },
+    {
+      "epoch": 0.5436229406844383,
+      "grad_norm": 0.0,
+      "learning_rate": 9.074349858344763e-06,
+      "loss": 1.0651,
+      "step": 14948
+    },
+    {
+      "epoch": 0.5436593082881769,
+      "grad_norm": 0.0,
+      "learning_rate": 9.073177060076866e-06,
+      "loss": 1.0211,
+      "step": 14949
+    },
+    {
+      "epoch": 0.5436956758919155,
+      "grad_norm": 0.0,
+      "learning_rate": 9.072004274667318e-06,
+      "loss": 0.9866,
+      "step": 14950
+    },
+    {
+      "epoch": 0.5437320434956541,
+      "grad_norm": 0.0,
+      "learning_rate": 9.070831502132396e-06,
+      "loss": 0.9959,
+      "step": 14951
+    },
+    {
+      "epoch": 0.5437684110993927,
+      "grad_norm": 0.0,
+      "learning_rate": 9.069658742488366e-06,
+      "loss": 0.9902,
+      "step": 14952
+    },
+    {
+      "epoch": 0.5438047787031313,
+      "grad_norm": 0.0,
+      "learning_rate": 9.068485995751504e-06,
+      "loss": 1.0902,
+      "step": 14953
+    },
+    {
+      "epoch": 0.5438411463068699,
+      "grad_norm": 0.0,
+      "learning_rate": 9.067313261938074e-06,
+      "loss": 0.9216,
+      "step": 14954
+    },
+    {
+      "epoch": 0.5438775139106085,
+      "grad_norm": 0.0,
+      "learning_rate": 9.066140541064354e-06,
+      "loss": 0.8671,
+      "step": 14955
+    },
+    {
+      "epoch": 0.543913881514347,
+      "grad_norm": 0.0,
+      "learning_rate": 9.064967833146605e-06,
+      "loss": 1.0266,
+      "step": 14956
+    },
+    {
+      "epoch": 0.5439502491180856,
+      "grad_norm": 0.0,
+      "learning_rate": 9.063795138201102e-06,
+      "loss": 0.892,
+      "step": 14957
+    },
+    {
+      "epoch": 0.5439866167218242,
+      "grad_norm": 0.0,
+      "learning_rate": 9.06262245624411e-06,
+      "loss": 0.9476,
+      "step": 14958
+    },
+    {
+      "epoch": 0.5440229843255628,
+      "grad_norm": 0.0,
+      "learning_rate": 9.061449787291907e-06,
+      "loss": 1.0608,
+      "step": 14959
+    },
+    {
+      "epoch": 0.5440593519293013,
+      "grad_norm": 0.0,
+      "learning_rate": 9.06027713136075e-06,
+      "loss": 0.9621,
+      "step": 14960
+    },
+    {
+      "epoch": 0.5440957195330399,
+      "grad_norm": 0.0,
+      "learning_rate": 9.05910448846692e-06,
+      "loss": 0.9017,
+      "step": 14961
+    },
+    {
+      "epoch": 0.5441320871367785,
+      "grad_norm": 0.0,
+      "learning_rate": 9.057931858626677e-06,
+      "loss": 0.9233,
+      "step": 14962
+    },
+    {
+      "epoch": 0.5441684547405171,
+      "grad_norm": 0.0,
+      "learning_rate": 9.056759241856298e-06,
+      "loss": 1.0052,
+      "step": 14963
+    },
+    {
+      "epoch": 0.5442048223442557,
+      "grad_norm": 0.0,
+      "learning_rate": 9.055586638172041e-06,
+      "loss": 0.9613,
+      "step": 14964
+    },
+    {
+      "epoch": 0.5442411899479943,
+      "grad_norm": 0.0,
+      "learning_rate": 9.054414047590183e-06,
+      "loss": 0.9272,
+      "step": 14965
+    },
+    {
+      "epoch": 0.5442775575517329,
+      "grad_norm": 0.0,
+      "learning_rate": 9.053241470126986e-06,
+      "loss": 0.893,
+      "step": 14966
+    },
+    {
+      "epoch": 0.5443139251554715,
+      "grad_norm": 0.0,
+      "learning_rate": 9.05206890579872e-06,
+      "loss": 1.1211,
+      "step": 14967
+    },
+    {
+      "epoch": 0.5443502927592101,
+      "grad_norm": 0.0,
+      "learning_rate": 9.050896354621657e-06,
+      "loss": 0.9473,
+      "step": 14968
+    },
+    {
+      "epoch": 0.5443866603629487,
+      "grad_norm": 0.0,
+      "learning_rate": 9.049723816612056e-06,
+      "loss": 1.1282,
+      "step": 14969
+    },
+    {
+      "epoch": 0.5444230279666873,
+      "grad_norm": 0.0,
+      "learning_rate": 9.04855129178619e-06,
+      "loss": 0.838,
+      "step": 14970
+    },
+    {
+      "epoch": 0.5444593955704259,
+      "grad_norm": 0.0,
+      "learning_rate": 9.047378780160324e-06,
+      "loss": 1.0051,
+      "step": 14971
+    },
+    {
+      "epoch": 0.5444957631741645,
+      "grad_norm": 0.0,
+      "learning_rate": 9.046206281750727e-06,
+      "loss": 0.9275,
+      "step": 14972
+    },
+    {
+      "epoch": 0.5445321307779031,
+      "grad_norm": 0.0,
+      "learning_rate": 9.045033796573664e-06,
+      "loss": 0.9774,
+      "step": 14973
+    },
+    {
+      "epoch": 0.5445684983816417,
+      "grad_norm": 0.0,
+      "learning_rate": 9.043861324645406e-06,
+      "loss": 0.9942,
+      "step": 14974
+    },
+    {
+      "epoch": 0.5446048659853803,
+      "grad_norm": 0.0,
+      "learning_rate": 9.042688865982207e-06,
+      "loss": 0.9399,
+      "step": 14975
+    },
+    {
+      "epoch": 0.5446412335891189,
+      "grad_norm": 0.0,
+      "learning_rate": 9.041516420600351e-06,
+      "loss": 1.1301,
+      "step": 14976
+    },
+    {
+      "epoch": 0.5446776011928574,
+      "grad_norm": 0.0,
+      "learning_rate": 9.040343988516089e-06,
+      "loss": 0.9762,
+      "step": 14977
+    },
+    {
+      "epoch": 0.544713968796596,
+      "grad_norm": 0.0,
+      "learning_rate": 9.039171569745694e-06,
+      "loss": 1.0296,
+      "step": 14978
+    },
+    {
+      "epoch": 0.5447503364003345,
+      "grad_norm": 0.0,
+      "learning_rate": 9.037999164305431e-06,
+      "loss": 0.7973,
+      "step": 14979
+    },
+    {
+      "epoch": 0.5447867040040731,
+      "grad_norm": 0.0,
+      "learning_rate": 9.036826772211565e-06,
+      "loss": 0.9326,
+      "step": 14980
+    },
+    {
+      "epoch": 0.5448230716078117,
+      "grad_norm": 0.0,
+      "learning_rate": 9.035654393480358e-06,
+      "loss": 1.0601,
+      "step": 14981
+    },
+    {
+      "epoch": 0.5448594392115503,
+      "grad_norm": 0.0,
+      "learning_rate": 9.034482028128082e-06,
+      "loss": 1.0576,
+      "step": 14982
+    },
+    {
+      "epoch": 0.5448958068152889,
+      "grad_norm": 0.0,
+      "learning_rate": 9.033309676170994e-06,
+      "loss": 0.9894,
+      "step": 14983
+    },
+    {
+      "epoch": 0.5449321744190275,
+      "grad_norm": 0.0,
+      "learning_rate": 9.032137337625367e-06,
+      "loss": 0.9868,
+      "step": 14984
+    },
+    {
+      "epoch": 0.5449685420227661,
+      "grad_norm": 0.0,
+      "learning_rate": 9.030965012507461e-06,
+      "loss": 1.035,
+      "step": 14985
+    },
+    {
+      "epoch": 0.5450049096265047,
+      "grad_norm": 0.0,
+      "learning_rate": 9.029792700833537e-06,
+      "loss": 1.057,
+      "step": 14986
+    },
+    {
+      "epoch": 0.5450412772302433,
+      "grad_norm": 0.0,
+      "learning_rate": 9.028620402619868e-06,
+      "loss": 0.9839,
+      "step": 14987
+    },
+    {
+      "epoch": 0.5450776448339819,
+      "grad_norm": 0.0,
+      "learning_rate": 9.027448117882704e-06,
+      "loss": 0.9877,
+      "step": 14988
+    },
+    {
+      "epoch": 0.5451140124377205,
+      "grad_norm": 0.0,
+      "learning_rate": 9.026275846638327e-06,
+      "loss": 0.9505,
+      "step": 14989
+    },
+    {
+      "epoch": 0.5451503800414591,
+      "grad_norm": 0.0,
+      "learning_rate": 9.025103588902985e-06,
+      "loss": 0.9595,
+      "step": 14990
+    },
+    {
+      "epoch": 0.5451867476451977,
+      "grad_norm": 0.0,
+      "learning_rate": 9.02393134469295e-06,
+      "loss": 1.0412,
+      "step": 14991
+    },
+    {
+      "epoch": 0.5452231152489363,
+      "grad_norm": 0.0,
+      "learning_rate": 9.02275911402448e-06,
+      "loss": 0.9008,
+      "step": 14992
+    },
+    {
+      "epoch": 0.5452594828526749,
+      "grad_norm": 0.0,
+      "learning_rate": 9.021586896913844e-06,
+      "loss": 0.9515,
+      "step": 14993
+    },
+    {
+      "epoch": 0.5452958504564134,
+      "grad_norm": 0.0,
+      "learning_rate": 9.020414693377299e-06,
+      "loss": 0.9723,
+      "step": 14994
+    },
+    {
+      "epoch": 0.545332218060152,
+      "grad_norm": 0.0,
+      "learning_rate": 9.019242503431113e-06,
+      "loss": 0.8644,
+      "step": 14995
+    },
+    {
+      "epoch": 0.5453685856638906,
+      "grad_norm": 0.0,
+      "learning_rate": 9.018070327091545e-06,
+      "loss": 1.0053,
+      "step": 14996
+    },
+    {
+      "epoch": 0.5454049532676292,
+      "grad_norm": 0.0,
+      "learning_rate": 9.016898164374858e-06,
+      "loss": 1.0412,
+      "step": 14997
+    },
+    {
+      "epoch": 0.5454413208713678,
+      "grad_norm": 0.0,
+      "learning_rate": 9.015726015297312e-06,
+      "loss": 0.9942,
+      "step": 14998
+    },
+    {
+      "epoch": 0.5454776884751064,
+      "grad_norm": 0.0,
+      "learning_rate": 9.014553879875172e-06,
+      "loss": 1.0106,
+      "step": 14999
+    },
+    {
+      "epoch": 0.545514056078845,
+      "grad_norm": 0.0,
+      "learning_rate": 9.013381758124699e-06,
+      "loss": 0.9983,
+      "step": 15000
+    },
+    {
+      "epoch": 0.5455504236825836,
+      "grad_norm": 0.0,
+      "learning_rate": 9.012209650062157e-06,
+      "loss": 1.023,
+      "step": 15001
+    },
+    {
+      "epoch": 0.5455867912863221,
+      "grad_norm": 0.0,
+      "learning_rate": 9.011037555703799e-06,
+      "loss": 0.9775,
+      "step": 15002
+    },
+    {
+      "epoch": 0.5456231588900607,
+      "grad_norm": 0.0,
+      "learning_rate": 9.009865475065897e-06,
+      "loss": 1.0358,
+      "step": 15003
+    },
+    {
+      "epoch": 0.5456595264937993,
+      "grad_norm": 0.0,
+      "learning_rate": 9.008693408164704e-06,
+      "loss": 1.0008,
+      "step": 15004
+    },
+    {
+      "epoch": 0.5456958940975379,
+      "grad_norm": 0.0,
+      "learning_rate": 9.007521355016486e-06,
+      "loss": 0.9699,
+      "step": 15005
+    },
+    {
+      "epoch": 0.5457322617012765,
+      "grad_norm": 0.0,
+      "learning_rate": 9.006349315637501e-06,
+      "loss": 0.9581,
+      "step": 15006
+    },
+    {
+      "epoch": 0.5457686293050151,
+      "grad_norm": 0.0,
+      "learning_rate": 9.005177290044004e-06,
+      "loss": 1.0397,
+      "step": 15007
+    },
+    {
+      "epoch": 0.5458049969087537,
+      "grad_norm": 0.0,
+      "learning_rate": 9.004005278252267e-06,
+      "loss": 1.0,
+      "step": 15008
+    },
+    {
+      "epoch": 0.5458413645124923,
+      "grad_norm": 0.0,
+      "learning_rate": 9.00283328027854e-06,
+      "loss": 1.0115,
+      "step": 15009
+    },
+    {
+      "epoch": 0.5458777321162308,
+      "grad_norm": 0.0,
+      "learning_rate": 9.001661296139089e-06,
+      "loss": 0.9444,
+      "step": 15010
+    },
+    {
+      "epoch": 0.5459140997199694,
+      "grad_norm": 0.0,
+      "learning_rate": 9.000489325850168e-06,
+      "loss": 0.9633,
+      "step": 15011
+    },
+    {
+      "epoch": 0.545950467323708,
+      "grad_norm": 0.0,
+      "learning_rate": 8.99931736942804e-06,
+      "loss": 0.9729,
+      "step": 15012
+    },
+    {
+      "epoch": 0.5459868349274466,
+      "grad_norm": 0.0,
+      "learning_rate": 8.998145426888963e-06,
+      "loss": 1.0296,
+      "step": 15013
+    },
+    {
+      "epoch": 0.5460232025311852,
+      "grad_norm": 0.0,
+      "learning_rate": 8.996973498249201e-06,
+      "loss": 1.0235,
+      "step": 15014
+    },
+    {
+      "epoch": 0.5460595701349238,
+      "grad_norm": 0.0,
+      "learning_rate": 8.995801583525003e-06,
+      "loss": 0.9297,
+      "step": 15015
+    },
+    {
+      "epoch": 0.5460959377386624,
+      "grad_norm": 0.0,
+      "learning_rate": 8.99462968273264e-06,
+      "loss": 1.0081,
+      "step": 15016
+    },
+    {
+      "epoch": 0.546132305342401,
+      "grad_norm": 0.0,
+      "learning_rate": 8.993457795888358e-06,
+      "loss": 0.942,
+      "step": 15017
+    },
+    {
+      "epoch": 0.5461686729461396,
+      "grad_norm": 0.0,
+      "learning_rate": 8.992285923008424e-06,
+      "loss": 1.1597,
+      "step": 15018
+    },
+    {
+      "epoch": 0.5462050405498782,
+      "grad_norm": 0.0,
+      "learning_rate": 8.99111406410909e-06,
+      "loss": 0.9146,
+      "step": 15019
+    },
+    {
+      "epoch": 0.5462414081536168,
+      "grad_norm": 0.0,
+      "learning_rate": 8.98994221920662e-06,
+      "loss": 0.9716,
+      "step": 15020
+    },
+    {
+      "epoch": 0.5462777757573554,
+      "grad_norm": 0.0,
+      "learning_rate": 8.988770388317266e-06,
+      "loss": 0.9738,
+      "step": 15021
+    },
+    {
+      "epoch": 0.546314143361094,
+      "grad_norm": 0.0,
+      "learning_rate": 8.987598571457293e-06,
+      "loss": 1.0186,
+      "step": 15022
+    },
+    {
+      "epoch": 0.5463505109648326,
+      "grad_norm": 0.0,
+      "learning_rate": 8.986426768642949e-06,
+      "loss": 1.065,
+      "step": 15023
+    },
+    {
+      "epoch": 0.5463868785685712,
+      "grad_norm": 0.0,
+      "learning_rate": 8.985254979890498e-06,
+      "loss": 0.922,
+      "step": 15024
+    },
+    {
+      "epoch": 0.5464232461723098,
+      "grad_norm": 0.0,
+      "learning_rate": 8.984083205216193e-06,
+      "loss": 0.9903,
+      "step": 15025
+    },
+    {
+      "epoch": 0.5464596137760483,
+      "grad_norm": 0.0,
+      "learning_rate": 8.982911444636292e-06,
+      "loss": 0.9495,
+      "step": 15026
+    },
+    {
+      "epoch": 0.5464959813797868,
+      "grad_norm": 0.0,
+      "learning_rate": 8.981739698167053e-06,
+      "loss": 1.0645,
+      "step": 15027
+    },
+    {
+      "epoch": 0.5465323489835254,
+      "grad_norm": 0.0,
+      "learning_rate": 8.980567965824729e-06,
+      "loss": 0.8833,
+      "step": 15028
+    },
+    {
+      "epoch": 0.546568716587264,
+      "grad_norm": 0.0,
+      "learning_rate": 8.979396247625579e-06,
+      "loss": 0.9645,
+      "step": 15029
+    },
+    {
+      "epoch": 0.5466050841910026,
+      "grad_norm": 0.0,
+      "learning_rate": 8.978224543585858e-06,
+      "loss": 1.0461,
+      "step": 15030
+    },
+    {
+      "epoch": 0.5466414517947412,
+      "grad_norm": 0.0,
+      "learning_rate": 8.977052853721822e-06,
+      "loss": 0.8536,
+      "step": 15031
+    },
+    {
+      "epoch": 0.5466778193984798,
+      "grad_norm": 0.0,
+      "learning_rate": 8.975881178049723e-06,
+      "loss": 0.8933,
+      "step": 15032
+    },
+    {
+      "epoch": 0.5467141870022184,
+      "grad_norm": 0.0,
+      "learning_rate": 8.974709516585825e-06,
+      "loss": 1.0449,
+      "step": 15033
+    },
+    {
+      "epoch": 0.546750554605957,
+      "grad_norm": 0.0,
+      "learning_rate": 8.973537869346373e-06,
+      "loss": 1.0712,
+      "step": 15034
+    },
+    {
+      "epoch": 0.5467869222096956,
+      "grad_norm": 0.0,
+      "learning_rate": 8.972366236347631e-06,
+      "loss": 0.9702,
+      "step": 15035
+    },
+    {
+      "epoch": 0.5468232898134342,
+      "grad_norm": 0.0,
+      "learning_rate": 8.971194617605847e-06,
+      "loss": 0.986,
+      "step": 15036
+    },
+    {
+      "epoch": 0.5468596574171728,
+      "grad_norm": 0.0,
+      "learning_rate": 8.97002301313728e-06,
+      "loss": 0.9843,
+      "step": 15037
+    },
+    {
+      "epoch": 0.5468960250209114,
+      "grad_norm": 0.0,
+      "learning_rate": 8.968851422958178e-06,
+      "loss": 0.9415,
+      "step": 15038
+    },
+    {
+      "epoch": 0.54693239262465,
+      "grad_norm": 0.0,
+      "learning_rate": 8.967679847084804e-06,
+      "loss": 0.9506,
+      "step": 15039
+    },
+    {
+      "epoch": 0.5469687602283886,
+      "grad_norm": 0.0,
+      "learning_rate": 8.966508285533404e-06,
+      "loss": 1.0904,
+      "step": 15040
+    },
+    {
+      "epoch": 0.5470051278321272,
+      "grad_norm": 0.0,
+      "learning_rate": 8.96533673832024e-06,
+      "loss": 0.9733,
+      "step": 15041
+    },
+    {
+      "epoch": 0.5470414954358658,
+      "grad_norm": 0.0,
+      "learning_rate": 8.964165205461554e-06,
+      "loss": 0.9065,
+      "step": 15042
+    },
+    {
+      "epoch": 0.5470778630396044,
+      "grad_norm": 0.0,
+      "learning_rate": 8.962993686973612e-06,
+      "loss": 1.0583,
+      "step": 15043
+    },
+    {
+      "epoch": 0.5471142306433429,
+      "grad_norm": 0.0,
+      "learning_rate": 8.96182218287266e-06,
+      "loss": 1.0878,
+      "step": 15044
+    },
+    {
+      "epoch": 0.5471505982470815,
+      "grad_norm": 0.0,
+      "learning_rate": 8.96065069317495e-06,
+      "loss": 1.0076,
+      "step": 15045
+    },
+    {
+      "epoch": 0.54718696585082,
+      "grad_norm": 0.0,
+      "learning_rate": 8.959479217896741e-06,
+      "loss": 0.9131,
+      "step": 15046
+    },
+    {
+      "epoch": 0.5472233334545586,
+      "grad_norm": 0.0,
+      "learning_rate": 8.958307757054275e-06,
+      "loss": 0.8869,
+      "step": 15047
+    },
+    {
+      "epoch": 0.5472597010582972,
+      "grad_norm": 0.0,
+      "learning_rate": 8.957136310663819e-06,
+      "loss": 0.9345,
+      "step": 15048
+    },
+    {
+      "epoch": 0.5472960686620358,
+      "grad_norm": 0.0,
+      "learning_rate": 8.95596487874161e-06,
+      "loss": 0.9519,
+      "step": 15049
+    },
+    {
+      "epoch": 0.5473324362657744,
+      "grad_norm": 0.0,
+      "learning_rate": 8.954793461303911e-06,
+      "loss": 0.8441,
+      "step": 15050
+    },
+    {
+      "epoch": 0.547368803869513,
+      "grad_norm": 0.0,
+      "learning_rate": 8.953622058366967e-06,
+      "loss": 0.9999,
+      "step": 15051
+    },
+    {
+      "epoch": 0.5474051714732516,
+      "grad_norm": 0.0,
+      "learning_rate": 8.952450669947034e-06,
+      "loss": 0.9657,
+      "step": 15052
+    },
+    {
+      "epoch": 0.5474415390769902,
+      "grad_norm": 0.0,
+      "learning_rate": 8.951279296060361e-06,
+      "loss": 0.9465,
+      "step": 15053
+    },
+    {
+      "epoch": 0.5474779066807288,
+      "grad_norm": 0.0,
+      "learning_rate": 8.950107936723203e-06,
+      "loss": 0.9733,
+      "step": 15054
+    },
+    {
+      "epoch": 0.5475142742844674,
+      "grad_norm": 0.0,
+      "learning_rate": 8.948936591951803e-06,
+      "loss": 1.032,
+      "step": 15055
+    },
+    {
+      "epoch": 0.547550641888206,
+      "grad_norm": 0.0,
+      "learning_rate": 8.94776526176242e-06,
+      "loss": 0.9676,
+      "step": 15056
+    },
+    {
+      "epoch": 0.5475870094919446,
+      "grad_norm": 0.0,
+      "learning_rate": 8.946593946171299e-06,
+      "loss": 0.8853,
+      "step": 15057
+    },
+    {
+      "epoch": 0.5476233770956832,
+      "grad_norm": 0.0,
+      "learning_rate": 8.945422645194695e-06,
+      "loss": 1.0782,
+      "step": 15058
+    },
+    {
+      "epoch": 0.5476597446994218,
+      "grad_norm": 0.0,
+      "learning_rate": 8.944251358848852e-06,
+      "loss": 0.8642,
+      "step": 15059
+    },
+    {
+      "epoch": 0.5476961123031604,
+      "grad_norm": 0.0,
+      "learning_rate": 8.943080087150025e-06,
+      "loss": 0.9572,
+      "step": 15060
+    },
+    {
+      "epoch": 0.5477324799068989,
+      "grad_norm": 0.0,
+      "learning_rate": 8.941908830114462e-06,
+      "loss": 0.9524,
+      "step": 15061
+    },
+    {
+      "epoch": 0.5477688475106375,
+      "grad_norm": 0.0,
+      "learning_rate": 8.940737587758415e-06,
+      "loss": 0.9988,
+      "step": 15062
+    },
+    {
+      "epoch": 0.5478052151143761,
+      "grad_norm": 0.0,
+      "learning_rate": 8.939566360098131e-06,
+      "loss": 0.878,
+      "step": 15063
+    },
+    {
+      "epoch": 0.5478415827181147,
+      "grad_norm": 0.0,
+      "learning_rate": 8.938395147149857e-06,
+      "loss": 0.9373,
+      "step": 15064
+    },
+    {
+      "epoch": 0.5478779503218533,
+      "grad_norm": 0.0,
+      "learning_rate": 8.937223948929845e-06,
+      "loss": 0.9813,
+      "step": 15065
+    },
+    {
+      "epoch": 0.5479143179255919,
+      "grad_norm": 0.0,
+      "learning_rate": 8.936052765454341e-06,
+      "loss": 0.9369,
+      "step": 15066
+    },
+    {
+      "epoch": 0.5479506855293305,
+      "grad_norm": 0.0,
+      "learning_rate": 8.9348815967396e-06,
+      "loss": 1.0743,
+      "step": 15067
+    },
+    {
+      "epoch": 0.5479870531330691,
+      "grad_norm": 0.0,
+      "learning_rate": 8.93371044280186e-06,
+      "loss": 0.967,
+      "step": 15068
+    },
+    {
+      "epoch": 0.5480234207368077,
+      "grad_norm": 0.0,
+      "learning_rate": 8.932539303657378e-06,
+      "loss": 1.0065,
+      "step": 15069
+    },
+    {
+      "epoch": 0.5480597883405463,
+      "grad_norm": 0.0,
+      "learning_rate": 8.931368179322396e-06,
+      "loss": 0.9382,
+      "step": 15070
+    },
+    {
+      "epoch": 0.5480961559442848,
+      "grad_norm": 0.0,
+      "learning_rate": 8.930197069813167e-06,
+      "loss": 1.1415,
+      "step": 15071
+    },
+    {
+      "epoch": 0.5481325235480234,
+      "grad_norm": 0.0,
+      "learning_rate": 8.929025975145934e-06,
+      "loss": 0.9045,
+      "step": 15072
+    },
+    {
+      "epoch": 0.548168891151762,
+      "grad_norm": 0.0,
+      "learning_rate": 8.92785489533695e-06,
+      "loss": 1.041,
+      "step": 15073
+    },
+    {
+      "epoch": 0.5482052587555006,
+      "grad_norm": 0.0,
+      "learning_rate": 8.92668383040245e-06,
+      "loss": 0.9712,
+      "step": 15074
+    },
+    {
+      "epoch": 0.5482416263592392,
+      "grad_norm": 0.0,
+      "learning_rate": 8.925512780358699e-06,
+      "loss": 1.0645,
+      "step": 15075
+    },
+    {
+      "epoch": 0.5482779939629778,
+      "grad_norm": 0.0,
+      "learning_rate": 8.924341745221926e-06,
+      "loss": 1.0032,
+      "step": 15076
+    },
+    {
+      "epoch": 0.5483143615667164,
+      "grad_norm": 0.0,
+      "learning_rate": 8.923170725008389e-06,
+      "loss": 1.1079,
+      "step": 15077
+    },
+    {
+      "epoch": 0.5483507291704549,
+      "grad_norm": 0.0,
+      "learning_rate": 8.921999719734328e-06,
+      "loss": 1.0676,
+      "step": 15078
+    },
+    {
+      "epoch": 0.5483870967741935,
+      "grad_norm": 0.0,
+      "learning_rate": 8.920828729415993e-06,
+      "loss": 0.9849,
+      "step": 15079
+    },
+    {
+      "epoch": 0.5484234643779321,
+      "grad_norm": 0.0,
+      "learning_rate": 8.919657754069627e-06,
+      "loss": 1.0388,
+      "step": 15080
+    },
+    {
+      "epoch": 0.5484598319816707,
+      "grad_norm": 0.0,
+      "learning_rate": 8.91848679371148e-06,
+      "loss": 0.9547,
+      "step": 15081
+    },
+    {
+      "epoch": 0.5484961995854093,
+      "grad_norm": 0.0,
+      "learning_rate": 8.917315848357794e-06,
+      "loss": 0.8923,
+      "step": 15082
+    },
+    {
+      "epoch": 0.5485325671891479,
+      "grad_norm": 0.0,
+      "learning_rate": 8.916144918024811e-06,
+      "loss": 0.9539,
+      "step": 15083
+    },
+    {
+      "epoch": 0.5485689347928865,
+      "grad_norm": 0.0,
+      "learning_rate": 8.914974002728782e-06,
+      "loss": 0.9451,
+      "step": 15084
+    },
+    {
+      "epoch": 0.5486053023966251,
+      "grad_norm": 0.0,
+      "learning_rate": 8.913803102485946e-06,
+      "loss": 1.0334,
+      "step": 15085
+    },
+    {
+      "epoch": 0.5486416700003637,
+      "grad_norm": 0.0,
+      "learning_rate": 8.912632217312556e-06,
+      "loss": 1.0003,
+      "step": 15086
+    },
+    {
+      "epoch": 0.5486780376041023,
+      "grad_norm": 0.0,
+      "learning_rate": 8.911461347224846e-06,
+      "loss": 0.946,
+      "step": 15087
+    },
+    {
+      "epoch": 0.5487144052078409,
+      "grad_norm": 0.0,
+      "learning_rate": 8.91029049223907e-06,
+      "loss": 0.8606,
+      "step": 15088
+    },
+    {
+      "epoch": 0.5487507728115795,
+      "grad_norm": 0.0,
+      "learning_rate": 8.909119652371463e-06,
+      "loss": 0.9078,
+      "step": 15089
+    },
+    {
+      "epoch": 0.5487871404153181,
+      "grad_norm": 0.0,
+      "learning_rate": 8.907948827638276e-06,
+      "loss": 1.0004,
+      "step": 15090
+    },
+    {
+      "epoch": 0.5488235080190567,
+      "grad_norm": 0.0,
+      "learning_rate": 8.906778018055748e-06,
+      "loss": 0.9372,
+      "step": 15091
+    },
+    {
+      "epoch": 0.5488598756227953,
+      "grad_norm": 0.0,
+      "learning_rate": 8.905607223640124e-06,
+      "loss": 1.0415,
+      "step": 15092
+    },
+    {
+      "epoch": 0.5488962432265339,
+      "grad_norm": 0.0,
+      "learning_rate": 8.904436444407648e-06,
+      "loss": 0.9555,
+      "step": 15093
+    },
+    {
+      "epoch": 0.5489326108302723,
+      "grad_norm": 0.0,
+      "learning_rate": 8.903265680374564e-06,
+      "loss": 0.8962,
+      "step": 15094
+    },
+    {
+      "epoch": 0.5489689784340109,
+      "grad_norm": 0.0,
+      "learning_rate": 8.902094931557107e-06,
+      "loss": 1.023,
+      "step": 15095
+    },
+    {
+      "epoch": 0.5490053460377495,
+      "grad_norm": 0.0,
+      "learning_rate": 8.900924197971531e-06,
+      "loss": 0.955,
+      "step": 15096
+    },
+    {
+      "epoch": 0.5490417136414881,
+      "grad_norm": 0.0,
+      "learning_rate": 8.899753479634067e-06,
+      "loss": 0.9741,
+      "step": 15097
+    },
+    {
+      "epoch": 0.5490780812452267,
+      "grad_norm": 0.0,
+      "learning_rate": 8.898582776560967e-06,
+      "loss": 1.0951,
+      "step": 15098
+    },
+    {
+      "epoch": 0.5491144488489653,
+      "grad_norm": 0.0,
+      "learning_rate": 8.897412088768465e-06,
+      "loss": 0.8485,
+      "step": 15099
+    },
+    {
+      "epoch": 0.5491508164527039,
+      "grad_norm": 0.0,
+      "learning_rate": 8.89624141627281e-06,
+      "loss": 1.0667,
+      "step": 15100
+    },
+    {
+      "epoch": 0.5491871840564425,
+      "grad_norm": 0.0,
+      "learning_rate": 8.895070759090238e-06,
+      "loss": 0.9978,
+      "step": 15101
+    },
+    {
+      "epoch": 0.5492235516601811,
+      "grad_norm": 0.0,
+      "learning_rate": 8.893900117236988e-06,
+      "loss": 0.9909,
+      "step": 15102
+    },
+    {
+      "epoch": 0.5492599192639197,
+      "grad_norm": 0.0,
+      "learning_rate": 8.892729490729307e-06,
+      "loss": 0.8937,
+      "step": 15103
+    },
+    {
+      "epoch": 0.5492962868676583,
+      "grad_norm": 0.0,
+      "learning_rate": 8.891558879583432e-06,
+      "loss": 1.0053,
+      "step": 15104
+    },
+    {
+      "epoch": 0.5493326544713969,
+      "grad_norm": 0.0,
+      "learning_rate": 8.890388283815607e-06,
+      "loss": 0.9471,
+      "step": 15105
+    },
+    {
+      "epoch": 0.5493690220751355,
+      "grad_norm": 0.0,
+      "learning_rate": 8.889217703442066e-06,
+      "loss": 0.9796,
+      "step": 15106
+    },
+    {
+      "epoch": 0.5494053896788741,
+      "grad_norm": 0.0,
+      "learning_rate": 8.88804713847906e-06,
+      "loss": 0.9383,
+      "step": 15107
+    },
+    {
+      "epoch": 0.5494417572826127,
+      "grad_norm": 0.0,
+      "learning_rate": 8.886876588942817e-06,
+      "loss": 1.0178,
+      "step": 15108
+    },
+    {
+      "epoch": 0.5494781248863513,
+      "grad_norm": 0.0,
+      "learning_rate": 8.885706054849585e-06,
+      "loss": 1.1,
+      "step": 15109
+    },
+    {
+      "epoch": 0.5495144924900899,
+      "grad_norm": 0.0,
+      "learning_rate": 8.884535536215598e-06,
+      "loss": 0.9565,
+      "step": 15110
+    },
+    {
+      "epoch": 0.5495508600938284,
+      "grad_norm": 0.0,
+      "learning_rate": 8.8833650330571e-06,
+      "loss": 0.8716,
+      "step": 15111
+    },
+    {
+      "epoch": 0.549587227697567,
+      "grad_norm": 0.0,
+      "learning_rate": 8.882194545390327e-06,
+      "loss": 0.9977,
+      "step": 15112
+    },
+    {
+      "epoch": 0.5496235953013056,
+      "grad_norm": 0.0,
+      "learning_rate": 8.88102407323152e-06,
+      "loss": 0.8918,
+      "step": 15113
+    },
+    {
+      "epoch": 0.5496599629050442,
+      "grad_norm": 0.0,
+      "learning_rate": 8.879853616596912e-06,
+      "loss": 1.0072,
+      "step": 15114
+    },
+    {
+      "epoch": 0.5496963305087827,
+      "grad_norm": 0.0,
+      "learning_rate": 8.878683175502752e-06,
+      "loss": 1.0068,
+      "step": 15115
+    },
+    {
+      "epoch": 0.5497326981125213,
+      "grad_norm": 0.0,
+      "learning_rate": 8.877512749965267e-06,
+      "loss": 1.0821,
+      "step": 15116
+    },
+    {
+      "epoch": 0.5497690657162599,
+      "grad_norm": 0.0,
+      "learning_rate": 8.876342340000702e-06,
+      "loss": 1.0519,
+      "step": 15117
+    },
+    {
+      "epoch": 0.5498054333199985,
+      "grad_norm": 0.0,
+      "learning_rate": 8.875171945625292e-06,
+      "loss": 1.0647,
+      "step": 15118
+    },
+    {
+      "epoch": 0.5498418009237371,
+      "grad_norm": 0.0,
+      "learning_rate": 8.874001566855276e-06,
+      "loss": 0.9161,
+      "step": 15119
+    },
+    {
+      "epoch": 0.5498781685274757,
+      "grad_norm": 0.0,
+      "learning_rate": 8.872831203706893e-06,
+      "loss": 1.0193,
+      "step": 15120
+    },
+    {
+      "epoch": 0.5499145361312143,
+      "grad_norm": 0.0,
+      "learning_rate": 8.871660856196374e-06,
+      "loss": 0.921,
+      "step": 15121
+    },
+    {
+      "epoch": 0.5499509037349529,
+      "grad_norm": 0.0,
+      "learning_rate": 8.870490524339963e-06,
+      "loss": 1.0136,
+      "step": 15122
+    },
+    {
+      "epoch": 0.5499872713386915,
+      "grad_norm": 0.0,
+      "learning_rate": 8.86932020815389e-06,
+      "loss": 1.0155,
+      "step": 15123
+    },
+    {
+      "epoch": 0.5500236389424301,
+      "grad_norm": 0.0,
+      "learning_rate": 8.868149907654397e-06,
+      "loss": 0.9659,
+      "step": 15124
+    },
+    {
+      "epoch": 0.5500600065461687,
+      "grad_norm": 0.0,
+      "learning_rate": 8.866979622857715e-06,
+      "loss": 1.1408,
+      "step": 15125
+    },
+    {
+      "epoch": 0.5500963741499073,
+      "grad_norm": 0.0,
+      "learning_rate": 8.865809353780089e-06,
+      "loss": 0.9492,
+      "step": 15126
+    },
+    {
+      "epoch": 0.5501327417536459,
+      "grad_norm": 0.0,
+      "learning_rate": 8.864639100437741e-06,
+      "loss": 0.974,
+      "step": 15127
+    },
+    {
+      "epoch": 0.5501691093573844,
+      "grad_norm": 0.0,
+      "learning_rate": 8.863468862846924e-06,
+      "loss": 1.0585,
+      "step": 15128
+    },
+    {
+      "epoch": 0.550205476961123,
+      "grad_norm": 0.0,
+      "learning_rate": 8.862298641023858e-06,
+      "loss": 0.9235,
+      "step": 15129
+    },
+    {
+      "epoch": 0.5502418445648616,
+      "grad_norm": 0.0,
+      "learning_rate": 8.861128434984785e-06,
+      "loss": 0.8078,
+      "step": 15130
+    },
+    {
+      "epoch": 0.5502782121686002,
+      "grad_norm": 0.0,
+      "learning_rate": 8.859958244745938e-06,
+      "loss": 0.9125,
+      "step": 15131
+    },
+    {
+      "epoch": 0.5503145797723388,
+      "grad_norm": 0.0,
+      "learning_rate": 8.858788070323556e-06,
+      "loss": 0.9972,
+      "step": 15132
+    },
+    {
+      "epoch": 0.5503509473760774,
+      "grad_norm": 0.0,
+      "learning_rate": 8.857617911733868e-06,
+      "loss": 0.9296,
+      "step": 15133
+    },
+    {
+      "epoch": 0.550387314979816,
+      "grad_norm": 0.0,
+      "learning_rate": 8.856447768993114e-06,
+      "loss": 0.9473,
+      "step": 15134
+    },
+    {
+      "epoch": 0.5504236825835546,
+      "grad_norm": 0.0,
+      "learning_rate": 8.85527764211752e-06,
+      "loss": 1.0272,
+      "step": 15135
+    },
+    {
+      "epoch": 0.5504600501872932,
+      "grad_norm": 0.0,
+      "learning_rate": 8.854107531123329e-06,
+      "loss": 0.8209,
+      "step": 15136
+    },
+    {
+      "epoch": 0.5504964177910318,
+      "grad_norm": 0.0,
+      "learning_rate": 8.852937436026766e-06,
+      "loss": 1.066,
+      "step": 15137
+    },
+    {
+      "epoch": 0.5505327853947704,
+      "grad_norm": 0.0,
+      "learning_rate": 8.851767356844071e-06,
+      "loss": 0.9205,
+      "step": 15138
+    },
+    {
+      "epoch": 0.550569152998509,
+      "grad_norm": 0.0,
+      "learning_rate": 8.850597293591477e-06,
+      "loss": 0.7795,
+      "step": 15139
+    },
+    {
+      "epoch": 0.5506055206022475,
+      "grad_norm": 0.0,
+      "learning_rate": 8.849427246285212e-06,
+      "loss": 0.9036,
+      "step": 15140
+    },
+    {
+      "epoch": 0.5506418882059861,
+      "grad_norm": 0.0,
+      "learning_rate": 8.84825721494151e-06,
+      "loss": 1.0324,
+      "step": 15141
+    },
+    {
+      "epoch": 0.5506782558097247,
+      "grad_norm": 0.0,
+      "learning_rate": 8.847087199576607e-06,
+      "loss": 1.0327,
+      "step": 15142
+    },
+    {
+      "epoch": 0.5507146234134633,
+      "grad_norm": 0.0,
+      "learning_rate": 8.845917200206734e-06,
+      "loss": 0.9317,
+      "step": 15143
+    },
+    {
+      "epoch": 0.5507509910172019,
+      "grad_norm": 0.0,
+      "learning_rate": 8.84474721684812e-06,
+      "loss": 1.0188,
+      "step": 15144
+    },
+    {
+      "epoch": 0.5507873586209404,
+      "grad_norm": 0.0,
+      "learning_rate": 8.843577249517004e-06,
+      "loss": 1.0285,
+      "step": 15145
+    },
+    {
+      "epoch": 0.550823726224679,
+      "grad_norm": 0.0,
+      "learning_rate": 8.842407298229605e-06,
+      "loss": 0.9894,
+      "step": 15146
+    },
+    {
+      "epoch": 0.5508600938284176,
+      "grad_norm": 0.0,
+      "learning_rate": 8.84123736300217e-06,
+      "loss": 1.0535,
+      "step": 15147
+    },
+    {
+      "epoch": 0.5508964614321562,
+      "grad_norm": 0.0,
+      "learning_rate": 8.840067443850917e-06,
+      "loss": 0.9948,
+      "step": 15148
+    },
+    {
+      "epoch": 0.5509328290358948,
+      "grad_norm": 0.0,
+      "learning_rate": 8.838897540792086e-06,
+      "loss": 0.9618,
+      "step": 15149
+    },
+    {
+      "epoch": 0.5509691966396334,
+      "grad_norm": 0.0,
+      "learning_rate": 8.837727653841904e-06,
+      "loss": 0.9554,
+      "step": 15150
+    },
+    {
+      "epoch": 0.551005564243372,
+      "grad_norm": 0.0,
+      "learning_rate": 8.836557783016601e-06,
+      "loss": 1.0061,
+      "step": 15151
+    },
+    {
+      "epoch": 0.5510419318471106,
+      "grad_norm": 0.0,
+      "learning_rate": 8.835387928332406e-06,
+      "loss": 0.898,
+      "step": 15152
+    },
+    {
+      "epoch": 0.5510782994508492,
+      "grad_norm": 0.0,
+      "learning_rate": 8.834218089805557e-06,
+      "loss": 0.8957,
+      "step": 15153
+    },
+    {
+      "epoch": 0.5511146670545878,
+      "grad_norm": 0.0,
+      "learning_rate": 8.833048267452269e-06,
+      "loss": 0.9527,
+      "step": 15154
+    },
+    {
+      "epoch": 0.5511510346583264,
+      "grad_norm": 0.0,
+      "learning_rate": 8.831878461288789e-06,
+      "loss": 0.9641,
+      "step": 15155
+    },
+    {
+      "epoch": 0.551187402262065,
+      "grad_norm": 0.0,
+      "learning_rate": 8.830708671331335e-06,
+      "loss": 0.9568,
+      "step": 15156
+    },
+    {
+      "epoch": 0.5512237698658036,
+      "grad_norm": 0.0,
+      "learning_rate": 8.829538897596138e-06,
+      "loss": 0.9093,
+      "step": 15157
+    },
+    {
+      "epoch": 0.5512601374695422,
+      "grad_norm": 0.0,
+      "learning_rate": 8.828369140099433e-06,
+      "loss": 0.9752,
+      "step": 15158
+    },
+    {
+      "epoch": 0.5512965050732808,
+      "grad_norm": 0.0,
+      "learning_rate": 8.827199398857435e-06,
+      "loss": 1.0065,
+      "step": 15159
+    },
+    {
+      "epoch": 0.5513328726770194,
+      "grad_norm": 0.0,
+      "learning_rate": 8.82602967388639e-06,
+      "loss": 1.2627,
+      "step": 15160
+    },
+    {
+      "epoch": 0.551369240280758,
+      "grad_norm": 0.0,
+      "learning_rate": 8.82485996520251e-06,
+      "loss": 0.965,
+      "step": 15161
+    },
+    {
+      "epoch": 0.5514056078844964,
+      "grad_norm": 0.0,
+      "learning_rate": 8.823690272822037e-06,
+      "loss": 1.0454,
+      "step": 15162
+    },
+    {
+      "epoch": 0.551441975488235,
+      "grad_norm": 0.0,
+      "learning_rate": 8.822520596761188e-06,
+      "loss": 1.0134,
+      "step": 15163
+    },
+    {
+      "epoch": 0.5514783430919736,
+      "grad_norm": 0.0,
+      "learning_rate": 8.821350937036198e-06,
+      "loss": 0.992,
+      "step": 15164
+    },
+    {
+      "epoch": 0.5515147106957122,
+      "grad_norm": 0.0,
+      "learning_rate": 8.820181293663289e-06,
+      "loss": 0.9764,
+      "step": 15165
+    },
+    {
+      "epoch": 0.5515510782994508,
+      "grad_norm": 0.0,
+      "learning_rate": 8.819011666658696e-06,
+      "loss": 1.0051,
+      "step": 15166
+    },
+    {
+      "epoch": 0.5515874459031894,
+      "grad_norm": 0.0,
+      "learning_rate": 8.817842056038635e-06,
+      "loss": 1.0836,
+      "step": 15167
+    },
+    {
+      "epoch": 0.551623813506928,
+      "grad_norm": 0.0,
+      "learning_rate": 8.81667246181934e-06,
+      "loss": 0.7923,
+      "step": 15168
+    },
+    {
+      "epoch": 0.5516601811106666,
+      "grad_norm": 0.0,
+      "learning_rate": 8.815502884017036e-06,
+      "loss": 0.9286,
+      "step": 15169
+    },
+    {
+      "epoch": 0.5516965487144052,
+      "grad_norm": 0.0,
+      "learning_rate": 8.814333322647948e-06,
+      "loss": 0.9905,
+      "step": 15170
+    },
+    {
+      "epoch": 0.5517329163181438,
+      "grad_norm": 0.0,
+      "learning_rate": 8.813163777728303e-06,
+      "loss": 1.0271,
+      "step": 15171
+    },
+    {
+      "epoch": 0.5517692839218824,
+      "grad_norm": 0.0,
+      "learning_rate": 8.811994249274329e-06,
+      "loss": 0.8771,
+      "step": 15172
+    },
+    {
+      "epoch": 0.551805651525621,
+      "grad_norm": 0.0,
+      "learning_rate": 8.810824737302244e-06,
+      "loss": 0.989,
+      "step": 15173
+    },
+    {
+      "epoch": 0.5518420191293596,
+      "grad_norm": 0.0,
+      "learning_rate": 8.809655241828284e-06,
+      "loss": 0.9744,
+      "step": 15174
+    },
+    {
+      "epoch": 0.5518783867330982,
+      "grad_norm": 0.0,
+      "learning_rate": 8.808485762868666e-06,
+      "loss": 0.8695,
+      "step": 15175
+    },
+    {
+      "epoch": 0.5519147543368368,
+      "grad_norm": 0.0,
+      "learning_rate": 8.807316300439618e-06,
+      "loss": 1.0092,
+      "step": 15176
+    },
+    {
+      "epoch": 0.5519511219405754,
+      "grad_norm": 0.0,
+      "learning_rate": 8.806146854557364e-06,
+      "loss": 1.0351,
+      "step": 15177
+    },
+    {
+      "epoch": 0.551987489544314,
+      "grad_norm": 0.0,
+      "learning_rate": 8.804977425238127e-06,
+      "loss": 1.0223,
+      "step": 15178
+    },
+    {
+      "epoch": 0.5520238571480525,
+      "grad_norm": 0.0,
+      "learning_rate": 8.803808012498138e-06,
+      "loss": 1.0386,
+      "step": 15179
+    },
+    {
+      "epoch": 0.5520602247517911,
+      "grad_norm": 0.0,
+      "learning_rate": 8.802638616353612e-06,
+      "loss": 0.8872,
+      "step": 15180
+    },
+    {
+      "epoch": 0.5520965923555297,
+      "grad_norm": 0.0,
+      "learning_rate": 8.801469236820777e-06,
+      "loss": 1.0178,
+      "step": 15181
+    },
+    {
+      "epoch": 0.5521329599592683,
+      "grad_norm": 0.0,
+      "learning_rate": 8.800299873915854e-06,
+      "loss": 1.0609,
+      "step": 15182
+    },
+    {
+      "epoch": 0.5521693275630069,
+      "grad_norm": 0.0,
+      "learning_rate": 8.79913052765507e-06,
+      "loss": 0.9926,
+      "step": 15183
+    },
+    {
+      "epoch": 0.5522056951667454,
+      "grad_norm": 0.0,
+      "learning_rate": 8.797961198054642e-06,
+      "loss": 0.9276,
+      "step": 15184
+    },
+    {
+      "epoch": 0.552242062770484,
+      "grad_norm": 0.0,
+      "learning_rate": 8.796791885130804e-06,
+      "loss": 0.9858,
+      "step": 15185
+    },
+    {
+      "epoch": 0.5522784303742226,
+      "grad_norm": 0.0,
+      "learning_rate": 8.795622588899764e-06,
+      "loss": 1.0757,
+      "step": 15186
+    },
+    {
+      "epoch": 0.5523147979779612,
+      "grad_norm": 0.0,
+      "learning_rate": 8.794453309377759e-06,
+      "loss": 0.8847,
+      "step": 15187
+    },
+    {
+      "epoch": 0.5523511655816998,
+      "grad_norm": 0.0,
+      "learning_rate": 8.793284046581e-06,
+      "loss": 0.9335,
+      "step": 15188
+    },
+    {
+      "epoch": 0.5523875331854384,
+      "grad_norm": 0.0,
+      "learning_rate": 8.792114800525714e-06,
+      "loss": 0.9665,
+      "step": 15189
+    },
+    {
+      "epoch": 0.552423900789177,
+      "grad_norm": 0.0,
+      "learning_rate": 8.790945571228119e-06,
+      "loss": 0.9992,
+      "step": 15190
+    },
+    {
+      "epoch": 0.5524602683929156,
+      "grad_norm": 0.0,
+      "learning_rate": 8.789776358704441e-06,
+      "loss": 0.984,
+      "step": 15191
+    },
+    {
+      "epoch": 0.5524966359966542,
+      "grad_norm": 0.0,
+      "learning_rate": 8.788607162970897e-06,
+      "loss": 1.0321,
+      "step": 15192
+    },
+    {
+      "epoch": 0.5525330036003928,
+      "grad_norm": 0.0,
+      "learning_rate": 8.787437984043714e-06,
+      "loss": 1.0004,
+      "step": 15193
+    },
+    {
+      "epoch": 0.5525693712041314,
+      "grad_norm": 0.0,
+      "learning_rate": 8.786268821939105e-06,
+      "loss": 1.0494,
+      "step": 15194
+    },
+    {
+      "epoch": 0.5526057388078699,
+      "grad_norm": 0.0,
+      "learning_rate": 8.785099676673296e-06,
+      "loss": 1.0286,
+      "step": 15195
+    },
+    {
+      "epoch": 0.5526421064116085,
+      "grad_norm": 0.0,
+      "learning_rate": 8.783930548262503e-06,
+      "loss": 0.8671,
+      "step": 15196
+    },
+    {
+      "epoch": 0.5526784740153471,
+      "grad_norm": 0.0,
+      "learning_rate": 8.782761436722949e-06,
+      "loss": 0.9578,
+      "step": 15197
+    },
+    {
+      "epoch": 0.5527148416190857,
+      "grad_norm": 0.0,
+      "learning_rate": 8.781592342070856e-06,
+      "loss": 1.0859,
+      "step": 15198
+    },
+    {
+      "epoch": 0.5527512092228243,
+      "grad_norm": 0.0,
+      "learning_rate": 8.780423264322434e-06,
+      "loss": 0.9658,
+      "step": 15199
+    },
+    {
+      "epoch": 0.5527875768265629,
+      "grad_norm": 0.0,
+      "learning_rate": 8.779254203493916e-06,
+      "loss": 1.0015,
+      "step": 15200
+    },
+    {
+      "epoch": 0.5528239444303015,
+      "grad_norm": 0.0,
+      "learning_rate": 8.77808515960151e-06,
+      "loss": 0.8173,
+      "step": 15201
+    },
+    {
+      "epoch": 0.5528603120340401,
+      "grad_norm": 0.0,
+      "learning_rate": 8.776916132661439e-06,
+      "loss": 1.0216,
+      "step": 15202
+    },
+    {
+      "epoch": 0.5528966796377787,
+      "grad_norm": 0.0,
+      "learning_rate": 8.775747122689922e-06,
+      "loss": 0.9195,
+      "step": 15203
+    },
+    {
+      "epoch": 0.5529330472415173,
+      "grad_norm": 0.0,
+      "learning_rate": 8.774578129703176e-06,
+      "loss": 0.9429,
+      "step": 15204
+    },
+    {
+      "epoch": 0.5529694148452559,
+      "grad_norm": 0.0,
+      "learning_rate": 8.77340915371742e-06,
+      "loss": 0.9718,
+      "step": 15205
+    },
+    {
+      "epoch": 0.5530057824489945,
+      "grad_norm": 0.0,
+      "learning_rate": 8.772240194748875e-06,
+      "loss": 1.009,
+      "step": 15206
+    },
+    {
+      "epoch": 0.553042150052733,
+      "grad_norm": 0.0,
+      "learning_rate": 8.77107125281375e-06,
+      "loss": 1.0064,
+      "step": 15207
+    },
+    {
+      "epoch": 0.5530785176564716,
+      "grad_norm": 0.0,
+      "learning_rate": 8.76990232792827e-06,
+      "loss": 0.9399,
+      "step": 15208
+    },
+    {
+      "epoch": 0.5531148852602102,
+      "grad_norm": 0.0,
+      "learning_rate": 8.76873342010865e-06,
+      "loss": 1.0,
+      "step": 15209
+    },
+    {
+      "epoch": 0.5531512528639488,
+      "grad_norm": 0.0,
+      "learning_rate": 8.767564529371106e-06,
+      "loss": 1.0469,
+      "step": 15210
+    },
+    {
+      "epoch": 0.5531876204676874,
+      "grad_norm": 0.0,
+      "learning_rate": 8.766395655731856e-06,
+      "loss": 0.8578,
+      "step": 15211
+    },
+    {
+      "epoch": 0.5532239880714259,
+      "grad_norm": 0.0,
+      "learning_rate": 8.765226799207119e-06,
+      "loss": 1.0685,
+      "step": 15212
+    },
+    {
+      "epoch": 0.5532603556751645,
+      "grad_norm": 0.0,
+      "learning_rate": 8.764057959813103e-06,
+      "loss": 0.7305,
+      "step": 15213
+    },
+    {
+      "epoch": 0.5532967232789031,
+      "grad_norm": 0.0,
+      "learning_rate": 8.762889137566033e-06,
+      "loss": 0.9185,
+      "step": 15214
+    },
+    {
+      "epoch": 0.5533330908826417,
+      "grad_norm": 0.0,
+      "learning_rate": 8.761720332482122e-06,
+      "loss": 0.9313,
+      "step": 15215
+    },
+    {
+      "epoch": 0.5533694584863803,
+      "grad_norm": 0.0,
+      "learning_rate": 8.76055154457758e-06,
+      "loss": 1.0146,
+      "step": 15216
+    },
+    {
+      "epoch": 0.5534058260901189,
+      "grad_norm": 0.0,
+      "learning_rate": 8.759382773868634e-06,
+      "loss": 0.9877,
+      "step": 15217
+    },
+    {
+      "epoch": 0.5534421936938575,
+      "grad_norm": 0.0,
+      "learning_rate": 8.758214020371481e-06,
+      "loss": 0.994,
+      "step": 15218
+    },
+    {
+      "epoch": 0.5534785612975961,
+      "grad_norm": 0.0,
+      "learning_rate": 8.757045284102356e-06,
+      "loss": 0.9135,
+      "step": 15219
+    },
+    {
+      "epoch": 0.5535149289013347,
+      "grad_norm": 0.0,
+      "learning_rate": 8.75587656507746e-06,
+      "loss": 0.991,
+      "step": 15220
+    },
+    {
+      "epoch": 0.5535512965050733,
+      "grad_norm": 0.0,
+      "learning_rate": 8.754707863313012e-06,
+      "loss": 0.9954,
+      "step": 15221
+    },
+    {
+      "epoch": 0.5535876641088119,
+      "grad_norm": 0.0,
+      "learning_rate": 8.753539178825225e-06,
+      "loss": 1.0518,
+      "step": 15222
+    },
+    {
+      "epoch": 0.5536240317125505,
+      "grad_norm": 0.0,
+      "learning_rate": 8.752370511630315e-06,
+      "loss": 0.9915,
+      "step": 15223
+    },
+    {
+      "epoch": 0.5536603993162891,
+      "grad_norm": 0.0,
+      "learning_rate": 8.751201861744491e-06,
+      "loss": 0.9619,
+      "step": 15224
+    },
+    {
+      "epoch": 0.5536967669200277,
+      "grad_norm": 0.0,
+      "learning_rate": 8.750033229183976e-06,
+      "loss": 1.016,
+      "step": 15225
+    },
+    {
+      "epoch": 0.5537331345237663,
+      "grad_norm": 0.0,
+      "learning_rate": 8.748864613964968e-06,
+      "loss": 1.0179,
+      "step": 15226
+    },
+    {
+      "epoch": 0.5537695021275049,
+      "grad_norm": 0.0,
+      "learning_rate": 8.747696016103696e-06,
+      "loss": 1.007,
+      "step": 15227
+    },
+    {
+      "epoch": 0.5538058697312435,
+      "grad_norm": 0.0,
+      "learning_rate": 8.746527435616358e-06,
+      "loss": 0.8649,
+      "step": 15228
+    },
+    {
+      "epoch": 0.553842237334982,
+      "grad_norm": 0.0,
+      "learning_rate": 8.745358872519179e-06,
+      "loss": 0.976,
+      "step": 15229
+    },
+    {
+      "epoch": 0.5538786049387205,
+      "grad_norm": 0.0,
+      "learning_rate": 8.744190326828362e-06,
+      "loss": 0.996,
+      "step": 15230
+    },
+    {
+      "epoch": 0.5539149725424591,
+      "grad_norm": 0.0,
+      "learning_rate": 8.743021798560123e-06,
+      "loss": 0.9053,
+      "step": 15231
+    },
+    {
+      "epoch": 0.5539513401461977,
+      "grad_norm": 0.0,
+      "learning_rate": 8.741853287730673e-06,
+      "loss": 0.9572,
+      "step": 15232
+    },
+    {
+      "epoch": 0.5539877077499363,
+      "grad_norm": 0.0,
+      "learning_rate": 8.740684794356226e-06,
+      "loss": 1.0152,
+      "step": 15233
+    },
+    {
+      "epoch": 0.5540240753536749,
+      "grad_norm": 0.0,
+      "learning_rate": 8.739516318452989e-06,
+      "loss": 0.9943,
+      "step": 15234
+    },
+    {
+      "epoch": 0.5540604429574135,
+      "grad_norm": 0.0,
+      "learning_rate": 8.738347860037176e-06,
+      "loss": 0.9258,
+      "step": 15235
+    },
+    {
+      "epoch": 0.5540968105611521,
+      "grad_norm": 0.0,
+      "learning_rate": 8.737179419124995e-06,
+      "loss": 0.9634,
+      "step": 15236
+    },
+    {
+      "epoch": 0.5541331781648907,
+      "grad_norm": 0.0,
+      "learning_rate": 8.736010995732657e-06,
+      "loss": 0.9744,
+      "step": 15237
+    },
+    {
+      "epoch": 0.5541695457686293,
+      "grad_norm": 0.0,
+      "learning_rate": 8.734842589876377e-06,
+      "loss": 0.9484,
+      "step": 15238
+    },
+    {
+      "epoch": 0.5542059133723679,
+      "grad_norm": 0.0,
+      "learning_rate": 8.733674201572357e-06,
+      "loss": 0.8997,
+      "step": 15239
+    },
+    {
+      "epoch": 0.5542422809761065,
+      "grad_norm": 0.0,
+      "learning_rate": 8.732505830836814e-06,
+      "loss": 0.9901,
+      "step": 15240
+    },
+    {
+      "epoch": 0.5542786485798451,
+      "grad_norm": 0.0,
+      "learning_rate": 8.73133747768595e-06,
+      "loss": 0.9153,
+      "step": 15241
+    },
+    {
+      "epoch": 0.5543150161835837,
+      "grad_norm": 0.0,
+      "learning_rate": 8.730169142135981e-06,
+      "loss": 0.9759,
+      "step": 15242
+    },
+    {
+      "epoch": 0.5543513837873223,
+      "grad_norm": 0.0,
+      "learning_rate": 8.729000824203111e-06,
+      "loss": 1.0062,
+      "step": 15243
+    },
+    {
+      "epoch": 0.5543877513910609,
+      "grad_norm": 0.0,
+      "learning_rate": 8.727832523903556e-06,
+      "loss": 0.9637,
+      "step": 15244
+    },
+    {
+      "epoch": 0.5544241189947995,
+      "grad_norm": 0.0,
+      "learning_rate": 8.726664241253514e-06,
+      "loss": 0.9278,
+      "step": 15245
+    },
+    {
+      "epoch": 0.554460486598538,
+      "grad_norm": 0.0,
+      "learning_rate": 8.725495976269205e-06,
+      "loss": 1.0363,
+      "step": 15246
+    },
+    {
+      "epoch": 0.5544968542022766,
+      "grad_norm": 0.0,
+      "learning_rate": 8.724327728966827e-06,
+      "loss": 0.9595,
+      "step": 15247
+    },
+    {
+      "epoch": 0.5545332218060152,
+      "grad_norm": 0.0,
+      "learning_rate": 8.723159499362595e-06,
+      "loss": 0.9815,
+      "step": 15248
+    },
+    {
+      "epoch": 0.5545695894097538,
+      "grad_norm": 0.0,
+      "learning_rate": 8.721991287472709e-06,
+      "loss": 1.0482,
+      "step": 15249
+    },
+    {
+      "epoch": 0.5546059570134924,
+      "grad_norm": 0.0,
+      "learning_rate": 8.720823093313384e-06,
+      "loss": 1.0715,
+      "step": 15250
+    },
+    {
+      "epoch": 0.554642324617231,
+      "grad_norm": 0.0,
+      "learning_rate": 8.719654916900822e-06,
+      "loss": 0.9869,
+      "step": 15251
+    },
+    {
+      "epoch": 0.5546786922209695,
+      "grad_norm": 0.0,
+      "learning_rate": 8.718486758251234e-06,
+      "loss": 1.0001,
+      "step": 15252
+    },
+    {
+      "epoch": 0.5547150598247081,
+      "grad_norm": 0.0,
+      "learning_rate": 8.717318617380819e-06,
+      "loss": 1.0381,
+      "step": 15253
+    },
+    {
+      "epoch": 0.5547514274284467,
+      "grad_norm": 0.0,
+      "learning_rate": 8.716150494305795e-06,
+      "loss": 1.0481,
+      "step": 15254
+    },
+    {
+      "epoch": 0.5547877950321853,
+      "grad_norm": 0.0,
+      "learning_rate": 8.714982389042358e-06,
+      "loss": 0.893,
+      "step": 15255
+    },
+    {
+      "epoch": 0.5548241626359239,
+      "grad_norm": 0.0,
+      "learning_rate": 8.713814301606716e-06,
+      "loss": 0.9903,
+      "step": 15256
+    },
+    {
+      "epoch": 0.5548605302396625,
+      "grad_norm": 0.0,
+      "learning_rate": 8.71264623201508e-06,
+      "loss": 1.0076,
+      "step": 15257
+    },
+    {
+      "epoch": 0.5548968978434011,
+      "grad_norm": 0.0,
+      "learning_rate": 8.711478180283644e-06,
+      "loss": 1.0492,
+      "step": 15258
+    },
+    {
+      "epoch": 0.5549332654471397,
+      "grad_norm": 0.0,
+      "learning_rate": 8.710310146428626e-06,
+      "loss": 0.9787,
+      "step": 15259
+    },
+    {
+      "epoch": 0.5549696330508783,
+      "grad_norm": 0.0,
+      "learning_rate": 8.70914213046622e-06,
+      "loss": 0.9702,
+      "step": 15260
+    },
+    {
+      "epoch": 0.5550060006546169,
+      "grad_norm": 0.0,
+      "learning_rate": 8.70797413241264e-06,
+      "loss": 0.8564,
+      "step": 15261
+    },
+    {
+      "epoch": 0.5550423682583555,
+      "grad_norm": 0.0,
+      "learning_rate": 8.706806152284084e-06,
+      "loss": 0.9946,
+      "step": 15262
+    },
+    {
+      "epoch": 0.555078735862094,
+      "grad_norm": 0.0,
+      "learning_rate": 8.705638190096757e-06,
+      "loss": 0.92,
+      "step": 15263
+    },
+    {
+      "epoch": 0.5551151034658326,
+      "grad_norm": 0.0,
+      "learning_rate": 8.704470245866863e-06,
+      "loss": 0.9237,
+      "step": 15264
+    },
+    {
+      "epoch": 0.5551514710695712,
+      "grad_norm": 0.0,
+      "learning_rate": 8.70330231961061e-06,
+      "loss": 1.0412,
+      "step": 15265
+    },
+    {
+      "epoch": 0.5551878386733098,
+      "grad_norm": 0.0,
+      "learning_rate": 8.702134411344194e-06,
+      "loss": 0.9135,
+      "step": 15266
+    },
+    {
+      "epoch": 0.5552242062770484,
+      "grad_norm": 0.0,
+      "learning_rate": 8.700966521083824e-06,
+      "loss": 0.9702,
+      "step": 15267
+    },
+    {
+      "epoch": 0.555260573880787,
+      "grad_norm": 0.0,
+      "learning_rate": 8.699798648845698e-06,
+      "loss": 0.8868,
+      "step": 15268
+    },
+    {
+      "epoch": 0.5552969414845256,
+      "grad_norm": 0.0,
+      "learning_rate": 8.698630794646023e-06,
+      "loss": 0.9186,
+      "step": 15269
+    },
+    {
+      "epoch": 0.5553333090882642,
+      "grad_norm": 0.0,
+      "learning_rate": 8.697462958500998e-06,
+      "loss": 0.9251,
+      "step": 15270
+    },
+    {
+      "epoch": 0.5553696766920028,
+      "grad_norm": 0.0,
+      "learning_rate": 8.696295140426827e-06,
+      "loss": 1.096,
+      "step": 15271
+    },
+    {
+      "epoch": 0.5554060442957414,
+      "grad_norm": 0.0,
+      "learning_rate": 8.695127340439709e-06,
+      "loss": 0.939,
+      "step": 15272
+    },
+    {
+      "epoch": 0.55544241189948,
+      "grad_norm": 0.0,
+      "learning_rate": 8.693959558555853e-06,
+      "loss": 1.0237,
+      "step": 15273
+    },
+    {
+      "epoch": 0.5554787795032186,
+      "grad_norm": 0.0,
+      "learning_rate": 8.692791794791453e-06,
+      "loss": 0.9964,
+      "step": 15274
+    },
+    {
+      "epoch": 0.5555151471069572,
+      "grad_norm": 0.0,
+      "learning_rate": 8.69162404916271e-06,
+      "loss": 0.9728,
+      "step": 15275
+    },
+    {
+      "epoch": 0.5555515147106957,
+      "grad_norm": 0.0,
+      "learning_rate": 8.690456321685828e-06,
+      "loss": 0.9666,
+      "step": 15276
+    },
+    {
+      "epoch": 0.5555878823144343,
+      "grad_norm": 0.0,
+      "learning_rate": 8.689288612377006e-06,
+      "loss": 1.0241,
+      "step": 15277
+    },
+    {
+      "epoch": 0.5556242499181729,
+      "grad_norm": 0.0,
+      "learning_rate": 8.688120921252448e-06,
+      "loss": 0.8485,
+      "step": 15278
+    },
+    {
+      "epoch": 0.5556606175219114,
+      "grad_norm": 0.0,
+      "learning_rate": 8.686953248328347e-06,
+      "loss": 0.9651,
+      "step": 15279
+    },
+    {
+      "epoch": 0.55569698512565,
+      "grad_norm": 0.0,
+      "learning_rate": 8.68578559362091e-06,
+      "loss": 0.8716,
+      "step": 15280
+    },
+    {
+      "epoch": 0.5557333527293886,
+      "grad_norm": 0.0,
+      "learning_rate": 8.684617957146329e-06,
+      "loss": 0.9938,
+      "step": 15281
+    },
+    {
+      "epoch": 0.5557697203331272,
+      "grad_norm": 0.0,
+      "learning_rate": 8.68345033892081e-06,
+      "loss": 1.0047,
+      "step": 15282
+    },
+    {
+      "epoch": 0.5558060879368658,
+      "grad_norm": 0.0,
+      "learning_rate": 8.682282738960548e-06,
+      "loss": 0.9353,
+      "step": 15283
+    },
+    {
+      "epoch": 0.5558424555406044,
+      "grad_norm": 0.0,
+      "learning_rate": 8.681115157281748e-06,
+      "loss": 1.0458,
+      "step": 15284
+    },
+    {
+      "epoch": 0.555878823144343,
+      "grad_norm": 0.0,
+      "learning_rate": 8.679947593900597e-06,
+      "loss": 0.9372,
+      "step": 15285
+    },
+    {
+      "epoch": 0.5559151907480816,
+      "grad_norm": 0.0,
+      "learning_rate": 8.678780048833309e-06,
+      "loss": 0.9063,
+      "step": 15286
+    },
+    {
+      "epoch": 0.5559515583518202,
+      "grad_norm": 0.0,
+      "learning_rate": 8.677612522096066e-06,
+      "loss": 1.0672,
+      "step": 15287
+    },
+    {
+      "epoch": 0.5559879259555588,
+      "grad_norm": 0.0,
+      "learning_rate": 8.676445013705076e-06,
+      "loss": 0.9768,
+      "step": 15288
+    },
+    {
+      "epoch": 0.5560242935592974,
+      "grad_norm": 0.0,
+      "learning_rate": 8.675277523676532e-06,
+      "loss": 0.9208,
+      "step": 15289
+    },
+    {
+      "epoch": 0.556060661163036,
+      "grad_norm": 0.0,
+      "learning_rate": 8.674110052026633e-06,
+      "loss": 0.9743,
+      "step": 15290
+    },
+    {
+      "epoch": 0.5560970287667746,
+      "grad_norm": 0.0,
+      "learning_rate": 8.672942598771576e-06,
+      "loss": 0.8998,
+      "step": 15291
+    },
+    {
+      "epoch": 0.5561333963705132,
+      "grad_norm": 0.0,
+      "learning_rate": 8.671775163927562e-06,
+      "loss": 0.9216,
+      "step": 15292
+    },
+    {
+      "epoch": 0.5561697639742518,
+      "grad_norm": 0.0,
+      "learning_rate": 8.67060774751078e-06,
+      "loss": 0.9698,
+      "step": 15293
+    },
+    {
+      "epoch": 0.5562061315779904,
+      "grad_norm": 0.0,
+      "learning_rate": 8.669440349537427e-06,
+      "loss": 1.017,
+      "step": 15294
+    },
+    {
+      "epoch": 0.556242499181729,
+      "grad_norm": 0.0,
+      "learning_rate": 8.668272970023703e-06,
+      "loss": 0.9245,
+      "step": 15295
+    },
+    {
+      "epoch": 0.5562788667854675,
+      "grad_norm": 0.0,
+      "learning_rate": 8.6671056089858e-06,
+      "loss": 0.9881,
+      "step": 15296
+    },
+    {
+      "epoch": 0.556315234389206,
+      "grad_norm": 0.0,
+      "learning_rate": 8.665938266439921e-06,
+      "loss": 0.8407,
+      "step": 15297
+    },
+    {
+      "epoch": 0.5563516019929446,
+      "grad_norm": 0.0,
+      "learning_rate": 8.66477094240225e-06,
+      "loss": 1.0259,
+      "step": 15298
+    },
+    {
+      "epoch": 0.5563879695966832,
+      "grad_norm": 0.0,
+      "learning_rate": 8.663603636888994e-06,
+      "loss": 1.1349,
+      "step": 15299
+    },
+    {
+      "epoch": 0.5564243372004218,
+      "grad_norm": 0.0,
+      "learning_rate": 8.662436349916335e-06,
+      "loss": 0.9497,
+      "step": 15300
+    },
+    {
+      "epoch": 0.5564607048041604,
+      "grad_norm": 0.0,
+      "learning_rate": 8.661269081500477e-06,
+      "loss": 0.9869,
+      "step": 15301
+    },
+    {
+      "epoch": 0.556497072407899,
+      "grad_norm": 0.0,
+      "learning_rate": 8.660101831657608e-06,
+      "loss": 0.954,
+      "step": 15302
+    },
+    {
+      "epoch": 0.5565334400116376,
+      "grad_norm": 0.0,
+      "learning_rate": 8.65893460040393e-06,
+      "loss": 1.0846,
+      "step": 15303
+    },
+    {
+      "epoch": 0.5565698076153762,
+      "grad_norm": 0.0,
+      "learning_rate": 8.657767387755627e-06,
+      "loss": 1.0003,
+      "step": 15304
+    },
+    {
+      "epoch": 0.5566061752191148,
+      "grad_norm": 0.0,
+      "learning_rate": 8.656600193728903e-06,
+      "loss": 0.9803,
+      "step": 15305
+    },
+    {
+      "epoch": 0.5566425428228534,
+      "grad_norm": 0.0,
+      "learning_rate": 8.65543301833994e-06,
+      "loss": 0.9785,
+      "step": 15306
+    },
+    {
+      "epoch": 0.556678910426592,
+      "grad_norm": 0.0,
+      "learning_rate": 8.654265861604938e-06,
+      "loss": 1.0126,
+      "step": 15307
+    },
+    {
+      "epoch": 0.5567152780303306,
+      "grad_norm": 0.0,
+      "learning_rate": 8.653098723540087e-06,
+      "loss": 0.9367,
+      "step": 15308
+    },
+    {
+      "epoch": 0.5567516456340692,
+      "grad_norm": 0.0,
+      "learning_rate": 8.65193160416158e-06,
+      "loss": 1.0506,
+      "step": 15309
+    },
+    {
+      "epoch": 0.5567880132378078,
+      "grad_norm": 0.0,
+      "learning_rate": 8.65076450348561e-06,
+      "loss": 0.9887,
+      "step": 15310
+    },
+    {
+      "epoch": 0.5568243808415464,
+      "grad_norm": 0.0,
+      "learning_rate": 8.64959742152837e-06,
+      "loss": 0.8881,
+      "step": 15311
+    },
+    {
+      "epoch": 0.556860748445285,
+      "grad_norm": 0.0,
+      "learning_rate": 8.64843035830605e-06,
+      "loss": 0.9642,
+      "step": 15312
+    },
+    {
+      "epoch": 0.5568971160490235,
+      "grad_norm": 0.0,
+      "learning_rate": 8.647263313834838e-06,
+      "loss": 0.9188,
+      "step": 15313
+    },
+    {
+      "epoch": 0.5569334836527621,
+      "grad_norm": 0.0,
+      "learning_rate": 8.64609628813093e-06,
+      "loss": 1.0263,
+      "step": 15314
+    },
+    {
+      "epoch": 0.5569698512565007,
+      "grad_norm": 0.0,
+      "learning_rate": 8.644929281210514e-06,
+      "loss": 0.9853,
+      "step": 15315
+    },
+    {
+      "epoch": 0.5570062188602393,
+      "grad_norm": 0.0,
+      "learning_rate": 8.643762293089783e-06,
+      "loss": 1.0344,
+      "step": 15316
+    },
+    {
+      "epoch": 0.5570425864639779,
+      "grad_norm": 0.0,
+      "learning_rate": 8.642595323784924e-06,
+      "loss": 0.9489,
+      "step": 15317
+    },
+    {
+      "epoch": 0.5570789540677165,
+      "grad_norm": 0.0,
+      "learning_rate": 8.641428373312133e-06,
+      "loss": 0.857,
+      "step": 15318
+    },
+    {
+      "epoch": 0.557115321671455,
+      "grad_norm": 0.0,
+      "learning_rate": 8.640261441687591e-06,
+      "loss": 0.9941,
+      "step": 15319
+    },
+    {
+      "epoch": 0.5571516892751937,
+      "grad_norm": 0.0,
+      "learning_rate": 8.639094528927495e-06,
+      "loss": 1.0054,
+      "step": 15320
+    },
+    {
+      "epoch": 0.5571880568789322,
+      "grad_norm": 0.0,
+      "learning_rate": 8.637927635048029e-06,
+      "loss": 0.867,
+      "step": 15321
+    },
+    {
+      "epoch": 0.5572244244826708,
+      "grad_norm": 0.0,
+      "learning_rate": 8.636760760065387e-06,
+      "loss": 0.9459,
+      "step": 15322
+    },
+    {
+      "epoch": 0.5572607920864094,
+      "grad_norm": 0.0,
+      "learning_rate": 8.635593903995754e-06,
+      "loss": 0.839,
+      "step": 15323
+    },
+    {
+      "epoch": 0.557297159690148,
+      "grad_norm": 0.0,
+      "learning_rate": 8.634427066855322e-06,
+      "loss": 0.9422,
+      "step": 15324
+    },
+    {
+      "epoch": 0.5573335272938866,
+      "grad_norm": 0.0,
+      "learning_rate": 8.633260248660272e-06,
+      "loss": 1.0943,
+      "step": 15325
+    },
+    {
+      "epoch": 0.5573698948976252,
+      "grad_norm": 0.0,
+      "learning_rate": 8.632093449426804e-06,
+      "loss": 0.987,
+      "step": 15326
+    },
+    {
+      "epoch": 0.5574062625013638,
+      "grad_norm": 0.0,
+      "learning_rate": 8.630926669171092e-06,
+      "loss": 0.9885,
+      "step": 15327
+    },
+    {
+      "epoch": 0.5574426301051024,
+      "grad_norm": 0.0,
+      "learning_rate": 8.629759907909333e-06,
+      "loss": 0.9204,
+      "step": 15328
+    },
+    {
+      "epoch": 0.557478997708841,
+      "grad_norm": 0.0,
+      "learning_rate": 8.62859316565771e-06,
+      "loss": 0.8979,
+      "step": 15329
+    },
+    {
+      "epoch": 0.5575153653125795,
+      "grad_norm": 0.0,
+      "learning_rate": 8.627426442432413e-06,
+      "loss": 0.8032,
+      "step": 15330
+    },
+    {
+      "epoch": 0.5575517329163181,
+      "grad_norm": 0.0,
+      "learning_rate": 8.62625973824963e-06,
+      "loss": 0.9782,
+      "step": 15331
+    },
+    {
+      "epoch": 0.5575881005200567,
+      "grad_norm": 0.0,
+      "learning_rate": 8.625093053125538e-06,
+      "loss": 0.9465,
+      "step": 15332
+    },
+    {
+      "epoch": 0.5576244681237953,
+      "grad_norm": 0.0,
+      "learning_rate": 8.623926387076332e-06,
+      "loss": 1.1068,
+      "step": 15333
+    },
+    {
+      "epoch": 0.5576608357275339,
+      "grad_norm": 0.0,
+      "learning_rate": 8.62275974011819e-06,
+      "loss": 1.0033,
+      "step": 15334
+    },
+    {
+      "epoch": 0.5576972033312725,
+      "grad_norm": 0.0,
+      "learning_rate": 8.621593112267309e-06,
+      "loss": 1.0549,
+      "step": 15335
+    },
+    {
+      "epoch": 0.5577335709350111,
+      "grad_norm": 0.0,
+      "learning_rate": 8.620426503539863e-06,
+      "loss": 0.9269,
+      "step": 15336
+    },
+    {
+      "epoch": 0.5577699385387497,
+      "grad_norm": 0.0,
+      "learning_rate": 8.619259913952047e-06,
+      "loss": 1.0124,
+      "step": 15337
+    },
+    {
+      "epoch": 0.5578063061424883,
+      "grad_norm": 0.0,
+      "learning_rate": 8.618093343520037e-06,
+      "loss": 0.9536,
+      "step": 15338
+    },
+    {
+      "epoch": 0.5578426737462269,
+      "grad_norm": 0.0,
+      "learning_rate": 8.616926792260022e-06,
+      "loss": 0.9623,
+      "step": 15339
+    },
+    {
+      "epoch": 0.5578790413499655,
+      "grad_norm": 0.0,
+      "learning_rate": 8.615760260188184e-06,
+      "loss": 0.9806,
+      "step": 15340
+    },
+    {
+      "epoch": 0.5579154089537041,
+      "grad_norm": 0.0,
+      "learning_rate": 8.614593747320711e-06,
+      "loss": 1.0381,
+      "step": 15341
+    },
+    {
+      "epoch": 0.5579517765574427,
+      "grad_norm": 0.0,
+      "learning_rate": 8.613427253673783e-06,
+      "loss": 1.0345,
+      "step": 15342
+    },
+    {
+      "epoch": 0.5579881441611813,
+      "grad_norm": 0.0,
+      "learning_rate": 8.612260779263584e-06,
+      "loss": 0.9805,
+      "step": 15343
+    },
+    {
+      "epoch": 0.5580245117649199,
+      "grad_norm": 0.0,
+      "learning_rate": 8.611094324106296e-06,
+      "loss": 0.8932,
+      "step": 15344
+    },
+    {
+      "epoch": 0.5580608793686584,
+      "grad_norm": 0.0,
+      "learning_rate": 8.60992788821811e-06,
+      "loss": 0.9541,
+      "step": 15345
+    },
+    {
+      "epoch": 0.558097246972397,
+      "grad_norm": 0.0,
+      "learning_rate": 8.608761471615197e-06,
+      "loss": 0.9884,
+      "step": 15346
+    },
+    {
+      "epoch": 0.5581336145761355,
+      "grad_norm": 0.0,
+      "learning_rate": 8.607595074313746e-06,
+      "loss": 1.0919,
+      "step": 15347
+    },
+    {
+      "epoch": 0.5581699821798741,
+      "grad_norm": 0.0,
+      "learning_rate": 8.606428696329936e-06,
+      "loss": 0.9645,
+      "step": 15348
+    },
+    {
+      "epoch": 0.5582063497836127,
+      "grad_norm": 0.0,
+      "learning_rate": 8.605262337679954e-06,
+      "loss": 1.0082,
+      "step": 15349
+    },
+    {
+      "epoch": 0.5582427173873513,
+      "grad_norm": 0.0,
+      "learning_rate": 8.604095998379979e-06,
+      "loss": 0.9769,
+      "step": 15350
+    },
+    {
+      "epoch": 0.5582790849910899,
+      "grad_norm": 0.0,
+      "learning_rate": 8.602929678446189e-06,
+      "loss": 1.0373,
+      "step": 15351
+    },
+    {
+      "epoch": 0.5583154525948285,
+      "grad_norm": 0.0,
+      "learning_rate": 8.601763377894767e-06,
+      "loss": 0.9579,
+      "step": 15352
+    },
+    {
+      "epoch": 0.5583518201985671,
+      "grad_norm": 0.0,
+      "learning_rate": 8.600597096741893e-06,
+      "loss": 0.9709,
+      "step": 15353
+    },
+    {
+      "epoch": 0.5583881878023057,
+      "grad_norm": 0.0,
+      "learning_rate": 8.599430835003751e-06,
+      "loss": 0.9748,
+      "step": 15354
+    },
+    {
+      "epoch": 0.5584245554060443,
+      "grad_norm": 0.0,
+      "learning_rate": 8.598264592696518e-06,
+      "loss": 1.1232,
+      "step": 15355
+    },
+    {
+      "epoch": 0.5584609230097829,
+      "grad_norm": 0.0,
+      "learning_rate": 8.597098369836378e-06,
+      "loss": 1.0353,
+      "step": 15356
+    },
+    {
+      "epoch": 0.5584972906135215,
+      "grad_norm": 0.0,
+      "learning_rate": 8.595932166439502e-06,
+      "loss": 0.958,
+      "step": 15357
+    },
+    {
+      "epoch": 0.5585336582172601,
+      "grad_norm": 0.0,
+      "learning_rate": 8.594765982522081e-06,
+      "loss": 1.0503,
+      "step": 15358
+    },
+    {
+      "epoch": 0.5585700258209987,
+      "grad_norm": 0.0,
+      "learning_rate": 8.593599818100285e-06,
+      "loss": 1.0412,
+      "step": 15359
+    },
+    {
+      "epoch": 0.5586063934247373,
+      "grad_norm": 0.0,
+      "learning_rate": 8.592433673190297e-06,
+      "loss": 0.8976,
+      "step": 15360
+    },
+    {
+      "epoch": 0.5586427610284759,
+      "grad_norm": 0.0,
+      "learning_rate": 8.591267547808293e-06,
+      "loss": 0.8566,
+      "step": 15361
+    },
+    {
+      "epoch": 0.5586791286322145,
+      "grad_norm": 0.0,
+      "learning_rate": 8.590101441970456e-06,
+      "loss": 0.9482,
+      "step": 15362
+    },
+    {
+      "epoch": 0.5587154962359531,
+      "grad_norm": 0.0,
+      "learning_rate": 8.588935355692957e-06,
+      "loss": 1.0359,
+      "step": 15363
+    },
+    {
+      "epoch": 0.5587518638396916,
+      "grad_norm": 0.0,
+      "learning_rate": 8.587769288991983e-06,
+      "loss": 1.08,
+      "step": 15364
+    },
+    {
+      "epoch": 0.5587882314434302,
+      "grad_norm": 0.0,
+      "learning_rate": 8.5866032418837e-06,
+      "loss": 0.9769,
+      "step": 15365
+    },
+    {
+      "epoch": 0.5588245990471687,
+      "grad_norm": 0.0,
+      "learning_rate": 8.585437214384299e-06,
+      "loss": 0.8908,
+      "step": 15366
+    },
+    {
+      "epoch": 0.5588609666509073,
+      "grad_norm": 0.0,
+      "learning_rate": 8.584271206509944e-06,
+      "loss": 0.8944,
+      "step": 15367
+    },
+    {
+      "epoch": 0.5588973342546459,
+      "grad_norm": 0.0,
+      "learning_rate": 8.583105218276822e-06,
+      "loss": 1.0212,
+      "step": 15368
+    },
+    {
+      "epoch": 0.5589337018583845,
+      "grad_norm": 0.0,
+      "learning_rate": 8.581939249701105e-06,
+      "loss": 0.9709,
+      "step": 15369
+    },
+    {
+      "epoch": 0.5589700694621231,
+      "grad_norm": 0.0,
+      "learning_rate": 8.58077330079896e-06,
+      "loss": 0.9555,
+      "step": 15370
+    },
+    {
+      "epoch": 0.5590064370658617,
+      "grad_norm": 0.0,
+      "learning_rate": 8.579607371586582e-06,
+      "loss": 0.9374,
+      "step": 15371
+    },
+    {
+      "epoch": 0.5590428046696003,
+      "grad_norm": 0.0,
+      "learning_rate": 8.578441462080132e-06,
+      "loss": 1.066,
+      "step": 15372
+    },
+    {
+      "epoch": 0.5590791722733389,
+      "grad_norm": 0.0,
+      "learning_rate": 8.577275572295791e-06,
+      "loss": 0.9356,
+      "step": 15373
+    },
+    {
+      "epoch": 0.5591155398770775,
+      "grad_norm": 0.0,
+      "learning_rate": 8.576109702249731e-06,
+      "loss": 0.9336,
+      "step": 15374
+    },
+    {
+      "epoch": 0.5591519074808161,
+      "grad_norm": 0.0,
+      "learning_rate": 8.574943851958131e-06,
+      "loss": 0.8968,
+      "step": 15375
+    },
+    {
+      "epoch": 0.5591882750845547,
+      "grad_norm": 0.0,
+      "learning_rate": 8.57377802143716e-06,
+      "loss": 0.9344,
+      "step": 15376
+    },
+    {
+      "epoch": 0.5592246426882933,
+      "grad_norm": 0.0,
+      "learning_rate": 8.572612210703e-06,
+      "loss": 1.0607,
+      "step": 15377
+    },
+    {
+      "epoch": 0.5592610102920319,
+      "grad_norm": 0.0,
+      "learning_rate": 8.571446419771815e-06,
+      "loss": 1.0812,
+      "step": 15378
+    },
+    {
+      "epoch": 0.5592973778957705,
+      "grad_norm": 0.0,
+      "learning_rate": 8.57028064865979e-06,
+      "loss": 0.9169,
+      "step": 15379
+    },
+    {
+      "epoch": 0.559333745499509,
+      "grad_norm": 0.0,
+      "learning_rate": 8.569114897383087e-06,
+      "loss": 1.1224,
+      "step": 15380
+    },
+    {
+      "epoch": 0.5593701131032476,
+      "grad_norm": 0.0,
+      "learning_rate": 8.567949165957888e-06,
+      "loss": 0.9843,
+      "step": 15381
+    },
+    {
+      "epoch": 0.5594064807069862,
+      "grad_norm": 0.0,
+      "learning_rate": 8.56678345440036e-06,
+      "loss": 0.9925,
+      "step": 15382
+    },
+    {
+      "epoch": 0.5594428483107248,
+      "grad_norm": 0.0,
+      "learning_rate": 8.565617762726682e-06,
+      "loss": 0.8219,
+      "step": 15383
+    },
+    {
+      "epoch": 0.5594792159144634,
+      "grad_norm": 0.0,
+      "learning_rate": 8.564452090953017e-06,
+      "loss": 0.8781,
+      "step": 15384
+    },
+    {
+      "epoch": 0.559515583518202,
+      "grad_norm": 0.0,
+      "learning_rate": 8.563286439095549e-06,
+      "loss": 0.9879,
+      "step": 15385
+    },
+    {
+      "epoch": 0.5595519511219406,
+      "grad_norm": 0.0,
+      "learning_rate": 8.562120807170437e-06,
+      "loss": 0.9115,
+      "step": 15386
+    },
+    {
+      "epoch": 0.5595883187256792,
+      "grad_norm": 0.0,
+      "learning_rate": 8.560955195193862e-06,
+      "loss": 0.9346,
+      "step": 15387
+    },
+    {
+      "epoch": 0.5596246863294178,
+      "grad_norm": 0.0,
+      "learning_rate": 8.559789603181993e-06,
+      "loss": 1.0223,
+      "step": 15388
+    },
+    {
+      "epoch": 0.5596610539331563,
+      "grad_norm": 0.0,
+      "learning_rate": 8.558624031150997e-06,
+      "loss": 0.9188,
+      "step": 15389
+    },
+    {
+      "epoch": 0.559697421536895,
+      "grad_norm": 0.0,
+      "learning_rate": 8.557458479117052e-06,
+      "loss": 0.9378,
+      "step": 15390
+    },
+    {
+      "epoch": 0.5597337891406335,
+      "grad_norm": 0.0,
+      "learning_rate": 8.556292947096318e-06,
+      "loss": 0.929,
+      "step": 15391
+    },
+    {
+      "epoch": 0.5597701567443721,
+      "grad_norm": 0.0,
+      "learning_rate": 8.555127435104975e-06,
+      "loss": 0.9995,
+      "step": 15392
+    },
+    {
+      "epoch": 0.5598065243481107,
+      "grad_norm": 0.0,
+      "learning_rate": 8.553961943159186e-06,
+      "loss": 1.0971,
+      "step": 15393
+    },
+    {
+      "epoch": 0.5598428919518493,
+      "grad_norm": 0.0,
+      "learning_rate": 8.552796471275125e-06,
+      "loss": 0.8709,
+      "step": 15394
+    },
+    {
+      "epoch": 0.5598792595555879,
+      "grad_norm": 0.0,
+      "learning_rate": 8.551631019468958e-06,
+      "loss": 0.869,
+      "step": 15395
+    },
+    {
+      "epoch": 0.5599156271593265,
+      "grad_norm": 0.0,
+      "learning_rate": 8.550465587756862e-06,
+      "loss": 0.9815,
+      "step": 15396
+    },
+    {
+      "epoch": 0.559951994763065,
+      "grad_norm": 0.0,
+      "learning_rate": 8.54930017615499e-06,
+      "loss": 0.9784,
+      "step": 15397
+    },
+    {
+      "epoch": 0.5599883623668036,
+      "grad_norm": 0.0,
+      "learning_rate": 8.548134784679527e-06,
+      "loss": 1.0443,
+      "step": 15398
+    },
+    {
+      "epoch": 0.5600247299705422,
+      "grad_norm": 0.0,
+      "learning_rate": 8.54696941334663e-06,
+      "loss": 0.9487,
+      "step": 15399
+    },
+    {
+      "epoch": 0.5600610975742808,
+      "grad_norm": 0.0,
+      "learning_rate": 8.545804062172472e-06,
+      "loss": 0.9885,
+      "step": 15400
+    },
+    {
+      "epoch": 0.5600974651780194,
+      "grad_norm": 0.0,
+      "learning_rate": 8.544638731173218e-06,
+      "loss": 1.056,
+      "step": 15401
+    },
+    {
+      "epoch": 0.560133832781758,
+      "grad_norm": 0.0,
+      "learning_rate": 8.54347342036504e-06,
+      "loss": 1.0135,
+      "step": 15402
+    },
+    {
+      "epoch": 0.5601702003854966,
+      "grad_norm": 0.0,
+      "learning_rate": 8.542308129764097e-06,
+      "loss": 0.8995,
+      "step": 15403
+    },
+    {
+      "epoch": 0.5602065679892352,
+      "grad_norm": 0.0,
+      "learning_rate": 8.541142859386567e-06,
+      "loss": 0.9339,
+      "step": 15404
+    },
+    {
+      "epoch": 0.5602429355929738,
+      "grad_norm": 0.0,
+      "learning_rate": 8.539977609248605e-06,
+      "loss": 1.01,
+      "step": 15405
+    },
+    {
+      "epoch": 0.5602793031967124,
+      "grad_norm": 0.0,
+      "learning_rate": 8.538812379366385e-06,
+      "loss": 1.008,
+      "step": 15406
+    },
+    {
+      "epoch": 0.560315670800451,
+      "grad_norm": 0.0,
+      "learning_rate": 8.53764716975607e-06,
+      "loss": 1.0114,
+      "step": 15407
+    },
+    {
+      "epoch": 0.5603520384041896,
+      "grad_norm": 0.0,
+      "learning_rate": 8.536481980433821e-06,
+      "loss": 0.8632,
+      "step": 15408
+    },
+    {
+      "epoch": 0.5603884060079282,
+      "grad_norm": 0.0,
+      "learning_rate": 8.535316811415815e-06,
+      "loss": 1.0727,
+      "step": 15409
+    },
+    {
+      "epoch": 0.5604247736116668,
+      "grad_norm": 0.0,
+      "learning_rate": 8.534151662718205e-06,
+      "loss": 0.976,
+      "step": 15410
+    },
+    {
+      "epoch": 0.5604611412154054,
+      "grad_norm": 0.0,
+      "learning_rate": 8.532986534357163e-06,
+      "loss": 0.9865,
+      "step": 15411
+    },
+    {
+      "epoch": 0.560497508819144,
+      "grad_norm": 0.0,
+      "learning_rate": 8.53182142634885e-06,
+      "loss": 1.0648,
+      "step": 15412
+    },
+    {
+      "epoch": 0.5605338764228825,
+      "grad_norm": 0.0,
+      "learning_rate": 8.530656338709433e-06,
+      "loss": 0.9966,
+      "step": 15413
+    },
+    {
+      "epoch": 0.560570244026621,
+      "grad_norm": 0.0,
+      "learning_rate": 8.529491271455073e-06,
+      "loss": 1.0085,
+      "step": 15414
+    },
+    {
+      "epoch": 0.5606066116303596,
+      "grad_norm": 0.0,
+      "learning_rate": 8.528326224601937e-06,
+      "loss": 1.0126,
+      "step": 15415
+    },
+    {
+      "epoch": 0.5606429792340982,
+      "grad_norm": 0.0,
+      "learning_rate": 8.527161198166184e-06,
+      "loss": 0.992,
+      "step": 15416
+    },
+    {
+      "epoch": 0.5606793468378368,
+      "grad_norm": 0.0,
+      "learning_rate": 8.525996192163985e-06,
+      "loss": 0.9302,
+      "step": 15417
+    },
+    {
+      "epoch": 0.5607157144415754,
+      "grad_norm": 0.0,
+      "learning_rate": 8.524831206611492e-06,
+      "loss": 0.9613,
+      "step": 15418
+    },
+    {
+      "epoch": 0.560752082045314,
+      "grad_norm": 0.0,
+      "learning_rate": 8.523666241524877e-06,
+      "loss": 0.9557,
+      "step": 15419
+    },
+    {
+      "epoch": 0.5607884496490526,
+      "grad_norm": 0.0,
+      "learning_rate": 8.522501296920294e-06,
+      "loss": 0.9796,
+      "step": 15420
+    },
+    {
+      "epoch": 0.5608248172527912,
+      "grad_norm": 0.0,
+      "learning_rate": 8.521336372813913e-06,
+      "loss": 0.8923,
+      "step": 15421
+    },
+    {
+      "epoch": 0.5608611848565298,
+      "grad_norm": 0.0,
+      "learning_rate": 8.520171469221888e-06,
+      "loss": 1.036,
+      "step": 15422
+    },
+    {
+      "epoch": 0.5608975524602684,
+      "grad_norm": 0.0,
+      "learning_rate": 8.51900658616039e-06,
+      "loss": 0.9225,
+      "step": 15423
+    },
+    {
+      "epoch": 0.560933920064007,
+      "grad_norm": 0.0,
+      "learning_rate": 8.517841723645569e-06,
+      "loss": 1.0473,
+      "step": 15424
+    },
+    {
+      "epoch": 0.5609702876677456,
+      "grad_norm": 0.0,
+      "learning_rate": 8.516676881693595e-06,
+      "loss": 1.0016,
+      "step": 15425
+    },
+    {
+      "epoch": 0.5610066552714842,
+      "grad_norm": 0.0,
+      "learning_rate": 8.515512060320622e-06,
+      "loss": 1.0568,
+      "step": 15426
+    },
+    {
+      "epoch": 0.5610430228752228,
+      "grad_norm": 0.0,
+      "learning_rate": 8.514347259542814e-06,
+      "loss": 1.0438,
+      "step": 15427
+    },
+    {
+      "epoch": 0.5610793904789614,
+      "grad_norm": 0.0,
+      "learning_rate": 8.513182479376332e-06,
+      "loss": 1.0078,
+      "step": 15428
+    },
+    {
+      "epoch": 0.5611157580827,
+      "grad_norm": 0.0,
+      "learning_rate": 8.512017719837329e-06,
+      "loss": 0.9929,
+      "step": 15429
+    },
+    {
+      "epoch": 0.5611521256864386,
+      "grad_norm": 0.0,
+      "learning_rate": 8.510852980941974e-06,
+      "loss": 0.9792,
+      "step": 15430
+    },
+    {
+      "epoch": 0.5611884932901771,
+      "grad_norm": 0.0,
+      "learning_rate": 8.509688262706416e-06,
+      "loss": 0.897,
+      "step": 15431
+    },
+    {
+      "epoch": 0.5612248608939157,
+      "grad_norm": 0.0,
+      "learning_rate": 8.508523565146823e-06,
+      "loss": 0.9571,
+      "step": 15432
+    },
+    {
+      "epoch": 0.5612612284976543,
+      "grad_norm": 0.0,
+      "learning_rate": 8.507358888279346e-06,
+      "loss": 0.9615,
+      "step": 15433
+    },
+    {
+      "epoch": 0.5612975961013928,
+      "grad_norm": 0.0,
+      "learning_rate": 8.506194232120148e-06,
+      "loss": 1.0688,
+      "step": 15434
+    },
+    {
+      "epoch": 0.5613339637051314,
+      "grad_norm": 0.0,
+      "learning_rate": 8.505029596685385e-06,
+      "loss": 0.9287,
+      "step": 15435
+    },
+    {
+      "epoch": 0.56137033130887,
+      "grad_norm": 0.0,
+      "learning_rate": 8.50386498199122e-06,
+      "loss": 0.8617,
+      "step": 15436
+    },
+    {
+      "epoch": 0.5614066989126086,
+      "grad_norm": 0.0,
+      "learning_rate": 8.502700388053797e-06,
+      "loss": 0.9684,
+      "step": 15437
+    },
+    {
+      "epoch": 0.5614430665163472,
+      "grad_norm": 0.0,
+      "learning_rate": 8.501535814889291e-06,
+      "loss": 0.8825,
+      "step": 15438
+    },
+    {
+      "epoch": 0.5614794341200858,
+      "grad_norm": 0.0,
+      "learning_rate": 8.500371262513843e-06,
+      "loss": 1.0097,
+      "step": 15439
+    },
+    {
+      "epoch": 0.5615158017238244,
+      "grad_norm": 0.0,
+      "learning_rate": 8.499206730943619e-06,
+      "loss": 0.9921,
+      "step": 15440
+    },
+    {
+      "epoch": 0.561552169327563,
+      "grad_norm": 0.0,
+      "learning_rate": 8.49804222019477e-06,
+      "loss": 0.9708,
+      "step": 15441
+    },
+    {
+      "epoch": 0.5615885369313016,
+      "grad_norm": 0.0,
+      "learning_rate": 8.496877730283456e-06,
+      "loss": 0.9848,
+      "step": 15442
+    },
+    {
+      "epoch": 0.5616249045350402,
+      "grad_norm": 0.0,
+      "learning_rate": 8.495713261225828e-06,
+      "loss": 0.9265,
+      "step": 15443
+    },
+    {
+      "epoch": 0.5616612721387788,
+      "grad_norm": 0.0,
+      "learning_rate": 8.49454881303805e-06,
+      "loss": 0.9232,
+      "step": 15444
+    },
+    {
+      "epoch": 0.5616976397425174,
+      "grad_norm": 0.0,
+      "learning_rate": 8.49338438573627e-06,
+      "loss": 1.0079,
+      "step": 15445
+    },
+    {
+      "epoch": 0.561734007346256,
+      "grad_norm": 0.0,
+      "learning_rate": 8.49221997933664e-06,
+      "loss": 0.9745,
+      "step": 15446
+    },
+    {
+      "epoch": 0.5617703749499946,
+      "grad_norm": 0.0,
+      "learning_rate": 8.491055593855322e-06,
+      "loss": 0.9447,
+      "step": 15447
+    },
+    {
+      "epoch": 0.5618067425537331,
+      "grad_norm": 0.0,
+      "learning_rate": 8.489891229308462e-06,
+      "loss": 0.9589,
+      "step": 15448
+    },
+    {
+      "epoch": 0.5618431101574717,
+      "grad_norm": 0.0,
+      "learning_rate": 8.488726885712226e-06,
+      "loss": 0.8988,
+      "step": 15449
+    },
+    {
+      "epoch": 0.5618794777612103,
+      "grad_norm": 0.0,
+      "learning_rate": 8.487562563082754e-06,
+      "loss": 0.9734,
+      "step": 15450
+    },
+    {
+      "epoch": 0.5619158453649489,
+      "grad_norm": 0.0,
+      "learning_rate": 8.486398261436208e-06,
+      "loss": 0.9458,
+      "step": 15451
+    },
+    {
+      "epoch": 0.5619522129686875,
+      "grad_norm": 0.0,
+      "learning_rate": 8.485233980788738e-06,
+      "loss": 0.9958,
+      "step": 15452
+    },
+    {
+      "epoch": 0.5619885805724261,
+      "grad_norm": 0.0,
+      "learning_rate": 8.484069721156498e-06,
+      "loss": 0.9662,
+      "step": 15453
+    },
+    {
+      "epoch": 0.5620249481761647,
+      "grad_norm": 0.0,
+      "learning_rate": 8.482905482555637e-06,
+      "loss": 0.8975,
+      "step": 15454
+    },
+    {
+      "epoch": 0.5620613157799033,
+      "grad_norm": 0.0,
+      "learning_rate": 8.481741265002315e-06,
+      "loss": 1.0548,
+      "step": 15455
+    },
+    {
+      "epoch": 0.5620976833836419,
+      "grad_norm": 0.0,
+      "learning_rate": 8.480577068512673e-06,
+      "loss": 1.096,
+      "step": 15456
+    },
+    {
+      "epoch": 0.5621340509873805,
+      "grad_norm": 0.0,
+      "learning_rate": 8.479412893102874e-06,
+      "loss": 1.0211,
+      "step": 15457
+    },
+    {
+      "epoch": 0.562170418591119,
+      "grad_norm": 0.0,
+      "learning_rate": 8.478248738789061e-06,
+      "loss": 0.941,
+      "step": 15458
+    },
+    {
+      "epoch": 0.5622067861948576,
+      "grad_norm": 0.0,
+      "learning_rate": 8.477084605587388e-06,
+      "loss": 1.068,
+      "step": 15459
+    },
+    {
+      "epoch": 0.5622431537985962,
+      "grad_norm": 0.0,
+      "learning_rate": 8.475920493514003e-06,
+      "loss": 0.9387,
+      "step": 15460
+    },
+    {
+      "epoch": 0.5622795214023348,
+      "grad_norm": 0.0,
+      "learning_rate": 8.474756402585063e-06,
+      "loss": 0.8473,
+      "step": 15461
+    },
+    {
+      "epoch": 0.5623158890060734,
+      "grad_norm": 0.0,
+      "learning_rate": 8.47359233281671e-06,
+      "loss": 0.9183,
+      "step": 15462
+    },
+    {
+      "epoch": 0.562352256609812,
+      "grad_norm": 0.0,
+      "learning_rate": 8.472428284225102e-06,
+      "loss": 1.0242,
+      "step": 15463
+    },
+    {
+      "epoch": 0.5623886242135505,
+      "grad_norm": 0.0,
+      "learning_rate": 8.471264256826384e-06,
+      "loss": 1.107,
+      "step": 15464
+    },
+    {
+      "epoch": 0.5624249918172891,
+      "grad_norm": 0.0,
+      "learning_rate": 8.4701002506367e-06,
+      "loss": 1.1409,
+      "step": 15465
+    },
+    {
+      "epoch": 0.5624613594210277,
+      "grad_norm": 0.0,
+      "learning_rate": 8.468936265672209e-06,
+      "loss": 1.0482,
+      "step": 15466
+    },
+    {
+      "epoch": 0.5624977270247663,
+      "grad_norm": 0.0,
+      "learning_rate": 8.467772301949054e-06,
+      "loss": 1.0242,
+      "step": 15467
+    },
+    {
+      "epoch": 0.5625340946285049,
+      "grad_norm": 0.0,
+      "learning_rate": 8.466608359483386e-06,
+      "loss": 1.0591,
+      "step": 15468
+    },
+    {
+      "epoch": 0.5625704622322435,
+      "grad_norm": 0.0,
+      "learning_rate": 8.465444438291346e-06,
+      "loss": 0.9241,
+      "step": 15469
+    },
+    {
+      "epoch": 0.5626068298359821,
+      "grad_norm": 0.0,
+      "learning_rate": 8.464280538389093e-06,
+      "loss": 0.9082,
+      "step": 15470
+    },
+    {
+      "epoch": 0.5626431974397207,
+      "grad_norm": 0.0,
+      "learning_rate": 8.463116659792764e-06,
+      "loss": 0.9625,
+      "step": 15471
+    },
+    {
+      "epoch": 0.5626795650434593,
+      "grad_norm": 0.0,
+      "learning_rate": 8.461952802518515e-06,
+      "loss": 1.1188,
+      "step": 15472
+    },
+    {
+      "epoch": 0.5627159326471979,
+      "grad_norm": 0.0,
+      "learning_rate": 8.460788966582485e-06,
+      "loss": 0.9101,
+      "step": 15473
+    },
+    {
+      "epoch": 0.5627523002509365,
+      "grad_norm": 0.0,
+      "learning_rate": 8.459625152000826e-06,
+      "loss": 0.9696,
+      "step": 15474
+    },
+    {
+      "epoch": 0.5627886678546751,
+      "grad_norm": 0.0,
+      "learning_rate": 8.45846135878968e-06,
+      "loss": 0.9452,
+      "step": 15475
+    },
+    {
+      "epoch": 0.5628250354584137,
+      "grad_norm": 0.0,
+      "learning_rate": 8.457297586965202e-06,
+      "loss": 0.9735,
+      "step": 15476
+    },
+    {
+      "epoch": 0.5628614030621523,
+      "grad_norm": 0.0,
+      "learning_rate": 8.456133836543527e-06,
+      "loss": 1.0086,
+      "step": 15477
+    },
+    {
+      "epoch": 0.5628977706658909,
+      "grad_norm": 0.0,
+      "learning_rate": 8.454970107540805e-06,
+      "loss": 1.0277,
+      "step": 15478
+    },
+    {
+      "epoch": 0.5629341382696295,
+      "grad_norm": 0.0,
+      "learning_rate": 8.453806399973178e-06,
+      "loss": 0.9842,
+      "step": 15479
+    },
+    {
+      "epoch": 0.562970505873368,
+      "grad_norm": 0.0,
+      "learning_rate": 8.452642713856797e-06,
+      "loss": 0.9989,
+      "step": 15480
+    },
+    {
+      "epoch": 0.5630068734771065,
+      "grad_norm": 0.0,
+      "learning_rate": 8.451479049207801e-06,
+      "loss": 0.9257,
+      "step": 15481
+    },
+    {
+      "epoch": 0.5630432410808451,
+      "grad_norm": 0.0,
+      "learning_rate": 8.450315406042339e-06,
+      "loss": 0.9206,
+      "step": 15482
+    },
+    {
+      "epoch": 0.5630796086845837,
+      "grad_norm": 0.0,
+      "learning_rate": 8.449151784376546e-06,
+      "loss": 1.0031,
+      "step": 15483
+    },
+    {
+      "epoch": 0.5631159762883223,
+      "grad_norm": 0.0,
+      "learning_rate": 8.44798818422658e-06,
+      "loss": 0.943,
+      "step": 15484
+    },
+    {
+      "epoch": 0.5631523438920609,
+      "grad_norm": 0.0,
+      "learning_rate": 8.44682460560857e-06,
+      "loss": 1.0063,
+      "step": 15485
+    },
+    {
+      "epoch": 0.5631887114957995,
+      "grad_norm": 0.0,
+      "learning_rate": 8.445661048538665e-06,
+      "loss": 1.0412,
+      "step": 15486
+    },
+    {
+      "epoch": 0.5632250790995381,
+      "grad_norm": 0.0,
+      "learning_rate": 8.44449751303301e-06,
+      "loss": 0.9645,
+      "step": 15487
+    },
+    {
+      "epoch": 0.5632614467032767,
+      "grad_norm": 0.0,
+      "learning_rate": 8.443333999107743e-06,
+      "loss": 1.0772,
+      "step": 15488
+    },
+    {
+      "epoch": 0.5632978143070153,
+      "grad_norm": 0.0,
+      "learning_rate": 8.442170506779011e-06,
+      "loss": 0.984,
+      "step": 15489
+    },
+    {
+      "epoch": 0.5633341819107539,
+      "grad_norm": 0.0,
+      "learning_rate": 8.44100703606295e-06,
+      "loss": 1.0292,
+      "step": 15490
+    },
+    {
+      "epoch": 0.5633705495144925,
+      "grad_norm": 0.0,
+      "learning_rate": 8.439843586975707e-06,
+      "loss": 1.0381,
+      "step": 15491
+    },
+    {
+      "epoch": 0.5634069171182311,
+      "grad_norm": 0.0,
+      "learning_rate": 8.438680159533417e-06,
+      "loss": 1.0011,
+      "step": 15492
+    },
+    {
+      "epoch": 0.5634432847219697,
+      "grad_norm": 0.0,
+      "learning_rate": 8.437516753752228e-06,
+      "loss": 1.0011,
+      "step": 15493
+    },
+    {
+      "epoch": 0.5634796523257083,
+      "grad_norm": 0.0,
+      "learning_rate": 8.436353369648274e-06,
+      "loss": 1.0465,
+      "step": 15494
+    },
+    {
+      "epoch": 0.5635160199294469,
+      "grad_norm": 0.0,
+      "learning_rate": 8.435190007237703e-06,
+      "loss": 0.8604,
+      "step": 15495
+    },
+    {
+      "epoch": 0.5635523875331855,
+      "grad_norm": 0.0,
+      "learning_rate": 8.434026666536644e-06,
+      "loss": 0.9944,
+      "step": 15496
+    },
+    {
+      "epoch": 0.5635887551369241,
+      "grad_norm": 0.0,
+      "learning_rate": 8.432863347561251e-06,
+      "loss": 1.0265,
+      "step": 15497
+    },
+    {
+      "epoch": 0.5636251227406626,
+      "grad_norm": 0.0,
+      "learning_rate": 8.43170005032765e-06,
+      "loss": 1.0197,
+      "step": 15498
+    },
+    {
+      "epoch": 0.5636614903444012,
+      "grad_norm": 0.0,
+      "learning_rate": 8.430536774851988e-06,
+      "loss": 1.0409,
+      "step": 15499
+    },
+    {
+      "epoch": 0.5636978579481398,
+      "grad_norm": 0.0,
+      "learning_rate": 8.4293735211504e-06,
+      "loss": 1.0161,
+      "step": 15500
+    },
+    {
+      "epoch": 0.5637342255518784,
+      "grad_norm": 0.0,
+      "learning_rate": 8.428210289239026e-06,
+      "loss": 1.0981,
+      "step": 15501
+    },
+    {
+      "epoch": 0.563770593155617,
+      "grad_norm": 0.0,
+      "learning_rate": 8.427047079134003e-06,
+      "loss": 0.9652,
+      "step": 15502
+    },
+    {
+      "epoch": 0.5638069607593555,
+      "grad_norm": 0.0,
+      "learning_rate": 8.425883890851476e-06,
+      "loss": 1.1384,
+      "step": 15503
+    },
+    {
+      "epoch": 0.5638433283630941,
+      "grad_norm": 0.0,
+      "learning_rate": 8.424720724407573e-06,
+      "loss": 1.1316,
+      "step": 15504
+    },
+    {
+      "epoch": 0.5638796959668327,
+      "grad_norm": 0.0,
+      "learning_rate": 8.423557579818433e-06,
+      "loss": 0.9397,
+      "step": 15505
+    },
+    {
+      "epoch": 0.5639160635705713,
+      "grad_norm": 0.0,
+      "learning_rate": 8.422394457100198e-06,
+      "loss": 1.1693,
+      "step": 15506
+    },
+    {
+      "epoch": 0.5639524311743099,
+      "grad_norm": 0.0,
+      "learning_rate": 8.421231356268998e-06,
+      "loss": 1.0398,
+      "step": 15507
+    },
+    {
+      "epoch": 0.5639887987780485,
+      "grad_norm": 0.0,
+      "learning_rate": 8.420068277340977e-06,
+      "loss": 0.9854,
+      "step": 15508
+    },
+    {
+      "epoch": 0.5640251663817871,
+      "grad_norm": 0.0,
+      "learning_rate": 8.418905220332262e-06,
+      "loss": 0.8549,
+      "step": 15509
+    },
+    {
+      "epoch": 0.5640615339855257,
+      "grad_norm": 0.0,
+      "learning_rate": 8.417742185259e-06,
+      "loss": 0.9259,
+      "step": 15510
+    },
+    {
+      "epoch": 0.5640979015892643,
+      "grad_norm": 0.0,
+      "learning_rate": 8.416579172137314e-06,
+      "loss": 1.0312,
+      "step": 15511
+    },
+    {
+      "epoch": 0.5641342691930029,
+      "grad_norm": 0.0,
+      "learning_rate": 8.41541618098335e-06,
+      "loss": 0.9608,
+      "step": 15512
+    },
+    {
+      "epoch": 0.5641706367967415,
+      "grad_norm": 0.0,
+      "learning_rate": 8.414253211813234e-06,
+      "loss": 1.0456,
+      "step": 15513
+    },
+    {
+      "epoch": 0.5642070044004801,
+      "grad_norm": 0.0,
+      "learning_rate": 8.413090264643108e-06,
+      "loss": 0.9005,
+      "step": 15514
+    },
+    {
+      "epoch": 0.5642433720042186,
+      "grad_norm": 0.0,
+      "learning_rate": 8.411927339489101e-06,
+      "loss": 1.0392,
+      "step": 15515
+    },
+    {
+      "epoch": 0.5642797396079572,
+      "grad_norm": 0.0,
+      "learning_rate": 8.410764436367355e-06,
+      "loss": 0.9727,
+      "step": 15516
+    },
+    {
+      "epoch": 0.5643161072116958,
+      "grad_norm": 0.0,
+      "learning_rate": 8.409601555293991e-06,
+      "loss": 1.0526,
+      "step": 15517
+    },
+    {
+      "epoch": 0.5643524748154344,
+      "grad_norm": 0.0,
+      "learning_rate": 8.408438696285153e-06,
+      "loss": 1.0023,
+      "step": 15518
+    },
+    {
+      "epoch": 0.564388842419173,
+      "grad_norm": 0.0,
+      "learning_rate": 8.407275859356967e-06,
+      "loss": 0.9725,
+      "step": 15519
+    },
+    {
+      "epoch": 0.5644252100229116,
+      "grad_norm": 0.0,
+      "learning_rate": 8.406113044525569e-06,
+      "loss": 1.0653,
+      "step": 15520
+    },
+    {
+      "epoch": 0.5644615776266502,
+      "grad_norm": 0.0,
+      "learning_rate": 8.40495025180709e-06,
+      "loss": 1.1265,
+      "step": 15521
+    },
+    {
+      "epoch": 0.5644979452303888,
+      "grad_norm": 0.0,
+      "learning_rate": 8.403787481217668e-06,
+      "loss": 0.976,
+      "step": 15522
+    },
+    {
+      "epoch": 0.5645343128341274,
+      "grad_norm": 0.0,
+      "learning_rate": 8.402624732773428e-06,
+      "loss": 0.89,
+      "step": 15523
+    },
+    {
+      "epoch": 0.564570680437866,
+      "grad_norm": 0.0,
+      "learning_rate": 8.4014620064905e-06,
+      "loss": 1.0248,
+      "step": 15524
+    },
+    {
+      "epoch": 0.5646070480416046,
+      "grad_norm": 0.0,
+      "learning_rate": 8.400299302385021e-06,
+      "loss": 0.9987,
+      "step": 15525
+    },
+    {
+      "epoch": 0.5646434156453431,
+      "grad_norm": 0.0,
+      "learning_rate": 8.399136620473119e-06,
+      "loss": 0.9933,
+      "step": 15526
+    },
+    {
+      "epoch": 0.5646797832490817,
+      "grad_norm": 0.0,
+      "learning_rate": 8.397973960770928e-06,
+      "loss": 0.9975,
+      "step": 15527
+    },
+    {
+      "epoch": 0.5647161508528203,
+      "grad_norm": 0.0,
+      "learning_rate": 8.39681132329457e-06,
+      "loss": 0.9913,
+      "step": 15528
+    },
+    {
+      "epoch": 0.5647525184565589,
+      "grad_norm": 0.0,
+      "learning_rate": 8.395648708060187e-06,
+      "loss": 1.0432,
+      "step": 15529
+    },
+    {
+      "epoch": 0.5647888860602975,
+      "grad_norm": 0.0,
+      "learning_rate": 8.394486115083896e-06,
+      "loss": 0.9963,
+      "step": 15530
+    },
+    {
+      "epoch": 0.5648252536640361,
+      "grad_norm": 0.0,
+      "learning_rate": 8.393323544381835e-06,
+      "loss": 0.9836,
+      "step": 15531
+    },
+    {
+      "epoch": 0.5648616212677746,
+      "grad_norm": 0.0,
+      "learning_rate": 8.392160995970128e-06,
+      "loss": 0.9934,
+      "step": 15532
+    },
+    {
+      "epoch": 0.5648979888715132,
+      "grad_norm": 0.0,
+      "learning_rate": 8.390998469864908e-06,
+      "loss": 0.9875,
+      "step": 15533
+    },
+    {
+      "epoch": 0.5649343564752518,
+      "grad_norm": 0.0,
+      "learning_rate": 8.3898359660823e-06,
+      "loss": 0.7822,
+      "step": 15534
+    },
+    {
+      "epoch": 0.5649707240789904,
+      "grad_norm": 0.0,
+      "learning_rate": 8.388673484638436e-06,
+      "loss": 1.1124,
+      "step": 15535
+    },
+    {
+      "epoch": 0.565007091682729,
+      "grad_norm": 0.0,
+      "learning_rate": 8.387511025549437e-06,
+      "loss": 1.1361,
+      "step": 15536
+    },
+    {
+      "epoch": 0.5650434592864676,
+      "grad_norm": 0.0,
+      "learning_rate": 8.38634858883144e-06,
+      "loss": 0.9195,
+      "step": 15537
+    },
+    {
+      "epoch": 0.5650798268902062,
+      "grad_norm": 0.0,
+      "learning_rate": 8.385186174500562e-06,
+      "loss": 0.904,
+      "step": 15538
+    },
+    {
+      "epoch": 0.5651161944939448,
+      "grad_norm": 0.0,
+      "learning_rate": 8.384023782572938e-06,
+      "loss": 0.8678,
+      "step": 15539
+    },
+    {
+      "epoch": 0.5651525620976834,
+      "grad_norm": 0.0,
+      "learning_rate": 8.382861413064687e-06,
+      "loss": 1.0977,
+      "step": 15540
+    },
+    {
+      "epoch": 0.565188929701422,
+      "grad_norm": 0.0,
+      "learning_rate": 8.381699065991944e-06,
+      "loss": 0.8975,
+      "step": 15541
+    },
+    {
+      "epoch": 0.5652252973051606,
+      "grad_norm": 0.0,
+      "learning_rate": 8.380536741370831e-06,
+      "loss": 1.0254,
+      "step": 15542
+    },
+    {
+      "epoch": 0.5652616649088992,
+      "grad_norm": 0.0,
+      "learning_rate": 8.37937443921747e-06,
+      "loss": 1.0151,
+      "step": 15543
+    },
+    {
+      "epoch": 0.5652980325126378,
+      "grad_norm": 0.0,
+      "learning_rate": 8.378212159547992e-06,
+      "loss": 0.9887,
+      "step": 15544
+    },
+    {
+      "epoch": 0.5653344001163764,
+      "grad_norm": 0.0,
+      "learning_rate": 8.377049902378514e-06,
+      "loss": 1.041,
+      "step": 15545
+    },
+    {
+      "epoch": 0.565370767720115,
+      "grad_norm": 0.0,
+      "learning_rate": 8.375887667725171e-06,
+      "loss": 1.0992,
+      "step": 15546
+    },
+    {
+      "epoch": 0.5654071353238536,
+      "grad_norm": 0.0,
+      "learning_rate": 8.374725455604079e-06,
+      "loss": 1.0044,
+      "step": 15547
+    },
+    {
+      "epoch": 0.565443502927592,
+      "grad_norm": 0.0,
+      "learning_rate": 8.37356326603137e-06,
+      "loss": 1.0014,
+      "step": 15548
+    },
+    {
+      "epoch": 0.5654798705313306,
+      "grad_norm": 0.0,
+      "learning_rate": 8.372401099023158e-06,
+      "loss": 0.9635,
+      "step": 15549
+    },
+    {
+      "epoch": 0.5655162381350692,
+      "grad_norm": 0.0,
+      "learning_rate": 8.371238954595575e-06,
+      "loss": 1.0368,
+      "step": 15550
+    },
+    {
+      "epoch": 0.5655526057388078,
+      "grad_norm": 0.0,
+      "learning_rate": 8.370076832764737e-06,
+      "loss": 1.0539,
+      "step": 15551
+    },
+    {
+      "epoch": 0.5655889733425464,
+      "grad_norm": 0.0,
+      "learning_rate": 8.368914733546774e-06,
+      "loss": 1.0204,
+      "step": 15552
+    },
+    {
+      "epoch": 0.565625340946285,
+      "grad_norm": 0.0,
+      "learning_rate": 8.367752656957802e-06,
+      "loss": 0.9321,
+      "step": 15553
+    },
+    {
+      "epoch": 0.5656617085500236,
+      "grad_norm": 0.0,
+      "learning_rate": 8.36659060301395e-06,
+      "loss": 0.9892,
+      "step": 15554
+    },
+    {
+      "epoch": 0.5656980761537622,
+      "grad_norm": 0.0,
+      "learning_rate": 8.36542857173133e-06,
+      "loss": 1.0959,
+      "step": 15555
+    },
+    {
+      "epoch": 0.5657344437575008,
+      "grad_norm": 0.0,
+      "learning_rate": 8.364266563126076e-06,
+      "loss": 0.9954,
+      "step": 15556
+    },
+    {
+      "epoch": 0.5657708113612394,
+      "grad_norm": 0.0,
+      "learning_rate": 8.363104577214297e-06,
+      "loss": 1.0178,
+      "step": 15557
+    },
+    {
+      "epoch": 0.565807178964978,
+      "grad_norm": 0.0,
+      "learning_rate": 8.361942614012124e-06,
+      "loss": 1.0706,
+      "step": 15558
+    },
+    {
+      "epoch": 0.5658435465687166,
+      "grad_norm": 0.0,
+      "learning_rate": 8.36078067353567e-06,
+      "loss": 1.0529,
+      "step": 15559
+    },
+    {
+      "epoch": 0.5658799141724552,
+      "grad_norm": 0.0,
+      "learning_rate": 8.35961875580106e-06,
+      "loss": 0.9564,
+      "step": 15560
+    },
+    {
+      "epoch": 0.5659162817761938,
+      "grad_norm": 0.0,
+      "learning_rate": 8.358456860824413e-06,
+      "loss": 0.9814,
+      "step": 15561
+    },
+    {
+      "epoch": 0.5659526493799324,
+      "grad_norm": 0.0,
+      "learning_rate": 8.357294988621845e-06,
+      "loss": 1.0548,
+      "step": 15562
+    },
+    {
+      "epoch": 0.565989016983671,
+      "grad_norm": 0.0,
+      "learning_rate": 8.356133139209482e-06,
+      "loss": 0.9987,
+      "step": 15563
+    },
+    {
+      "epoch": 0.5660253845874096,
+      "grad_norm": 0.0,
+      "learning_rate": 8.354971312603436e-06,
+      "loss": 0.9482,
+      "step": 15564
+    },
+    {
+      "epoch": 0.5660617521911481,
+      "grad_norm": 0.0,
+      "learning_rate": 8.353809508819831e-06,
+      "loss": 0.8439,
+      "step": 15565
+    },
+    {
+      "epoch": 0.5660981197948867,
+      "grad_norm": 0.0,
+      "learning_rate": 8.35264772787478e-06,
+      "loss": 0.9066,
+      "step": 15566
+    },
+    {
+      "epoch": 0.5661344873986253,
+      "grad_norm": 0.0,
+      "learning_rate": 8.351485969784412e-06,
+      "loss": 0.8941,
+      "step": 15567
+    },
+    {
+      "epoch": 0.5661708550023639,
+      "grad_norm": 0.0,
+      "learning_rate": 8.350324234564829e-06,
+      "loss": 1.0077,
+      "step": 15568
+    },
+    {
+      "epoch": 0.5662072226061025,
+      "grad_norm": 0.0,
+      "learning_rate": 8.349162522232164e-06,
+      "loss": 1.0451,
+      "step": 15569
+    },
+    {
+      "epoch": 0.566243590209841,
+      "grad_norm": 0.0,
+      "learning_rate": 8.348000832802521e-06,
+      "loss": 1.0132,
+      "step": 15570
+    },
+    {
+      "epoch": 0.5662799578135796,
+      "grad_norm": 0.0,
+      "learning_rate": 8.346839166292025e-06,
+      "loss": 0.9122,
+      "step": 15571
+    },
+    {
+      "epoch": 0.5663163254173182,
+      "grad_norm": 0.0,
+      "learning_rate": 8.345677522716789e-06,
+      "loss": 1.0185,
+      "step": 15572
+    },
+    {
+      "epoch": 0.5663526930210568,
+      "grad_norm": 0.0,
+      "learning_rate": 8.344515902092932e-06,
+      "loss": 1.0297,
+      "step": 15573
+    },
+    {
+      "epoch": 0.5663890606247954,
+      "grad_norm": 0.0,
+      "learning_rate": 8.343354304436563e-06,
+      "loss": 0.9636,
+      "step": 15574
+    },
+    {
+      "epoch": 0.566425428228534,
+      "grad_norm": 0.0,
+      "learning_rate": 8.34219272976381e-06,
+      "loss": 0.8783,
+      "step": 15575
+    },
+    {
+      "epoch": 0.5664617958322726,
+      "grad_norm": 0.0,
+      "learning_rate": 8.341031178090778e-06,
+      "loss": 1.0223,
+      "step": 15576
+    },
+    {
+      "epoch": 0.5664981634360112,
+      "grad_norm": 0.0,
+      "learning_rate": 8.339869649433582e-06,
+      "loss": 1.0776,
+      "step": 15577
+    },
+    {
+      "epoch": 0.5665345310397498,
+      "grad_norm": 0.0,
+      "learning_rate": 8.338708143808341e-06,
+      "loss": 0.8362,
+      "step": 15578
+    },
+    {
+      "epoch": 0.5665708986434884,
+      "grad_norm": 0.0,
+      "learning_rate": 8.337546661231168e-06,
+      "loss": 1.0687,
+      "step": 15579
+    },
+    {
+      "epoch": 0.566607266247227,
+      "grad_norm": 0.0,
+      "learning_rate": 8.336385201718178e-06,
+      "loss": 1.045,
+      "step": 15580
+    },
+    {
+      "epoch": 0.5666436338509656,
+      "grad_norm": 0.0,
+      "learning_rate": 8.335223765285478e-06,
+      "loss": 0.9841,
+      "step": 15581
+    },
+    {
+      "epoch": 0.5666800014547041,
+      "grad_norm": 0.0,
+      "learning_rate": 8.334062351949193e-06,
+      "loss": 1.0641,
+      "step": 15582
+    },
+    {
+      "epoch": 0.5667163690584427,
+      "grad_norm": 0.0,
+      "learning_rate": 8.332900961725423e-06,
+      "loss": 0.9358,
+      "step": 15583
+    },
+    {
+      "epoch": 0.5667527366621813,
+      "grad_norm": 0.0,
+      "learning_rate": 8.33173959463029e-06,
+      "loss": 0.9553,
+      "step": 15584
+    },
+    {
+      "epoch": 0.5667891042659199,
+      "grad_norm": 0.0,
+      "learning_rate": 8.330578250679901e-06,
+      "loss": 0.949,
+      "step": 15585
+    },
+    {
+      "epoch": 0.5668254718696585,
+      "grad_norm": 0.0,
+      "learning_rate": 8.329416929890373e-06,
+      "loss": 0.9739,
+      "step": 15586
+    },
+    {
+      "epoch": 0.5668618394733971,
+      "grad_norm": 0.0,
+      "learning_rate": 8.328255632277811e-06,
+      "loss": 0.9559,
+      "step": 15587
+    },
+    {
+      "epoch": 0.5668982070771357,
+      "grad_norm": 0.0,
+      "learning_rate": 8.327094357858336e-06,
+      "loss": 0.9153,
+      "step": 15588
+    },
+    {
+      "epoch": 0.5669345746808743,
+      "grad_norm": 0.0,
+      "learning_rate": 8.325933106648048e-06,
+      "loss": 0.8966,
+      "step": 15589
+    },
+    {
+      "epoch": 0.5669709422846129,
+      "grad_norm": 0.0,
+      "learning_rate": 8.324771878663064e-06,
+      "loss": 1.07,
+      "step": 15590
+    },
+    {
+      "epoch": 0.5670073098883515,
+      "grad_norm": 0.0,
+      "learning_rate": 8.323610673919494e-06,
+      "loss": 0.9811,
+      "step": 15591
+    },
+    {
+      "epoch": 0.5670436774920901,
+      "grad_norm": 0.0,
+      "learning_rate": 8.322449492433448e-06,
+      "loss": 0.9837,
+      "step": 15592
+    },
+    {
+      "epoch": 0.5670800450958287,
+      "grad_norm": 0.0,
+      "learning_rate": 8.321288334221034e-06,
+      "loss": 1.1385,
+      "step": 15593
+    },
+    {
+      "epoch": 0.5671164126995673,
+      "grad_norm": 0.0,
+      "learning_rate": 8.320127199298366e-06,
+      "loss": 0.9391,
+      "step": 15594
+    },
+    {
+      "epoch": 0.5671527803033058,
+      "grad_norm": 0.0,
+      "learning_rate": 8.318966087681544e-06,
+      "loss": 1.0056,
+      "step": 15595
+    },
+    {
+      "epoch": 0.5671891479070444,
+      "grad_norm": 0.0,
+      "learning_rate": 8.317804999386688e-06,
+      "loss": 0.9276,
+      "step": 15596
+    },
+    {
+      "epoch": 0.567225515510783,
+      "grad_norm": 0.0,
+      "learning_rate": 8.316643934429897e-06,
+      "loss": 0.9117,
+      "step": 15597
+    },
+    {
+      "epoch": 0.5672618831145216,
+      "grad_norm": 0.0,
+      "learning_rate": 8.315482892827285e-06,
+      "loss": 1.0607,
+      "step": 15598
+    },
+    {
+      "epoch": 0.5672982507182601,
+      "grad_norm": 0.0,
+      "learning_rate": 8.314321874594959e-06,
+      "loss": 1.0065,
+      "step": 15599
+    },
+    {
+      "epoch": 0.5673346183219987,
+      "grad_norm": 0.0,
+      "learning_rate": 8.31316087974902e-06,
+      "loss": 0.9331,
+      "step": 15600
+    },
+    {
+      "epoch": 0.5673709859257373,
+      "grad_norm": 0.0,
+      "learning_rate": 8.311999908305588e-06,
+      "loss": 0.9838,
+      "step": 15601
+    },
+    {
+      "epoch": 0.5674073535294759,
+      "grad_norm": 0.0,
+      "learning_rate": 8.310838960280757e-06,
+      "loss": 0.9852,
+      "step": 15602
+    },
+    {
+      "epoch": 0.5674437211332145,
+      "grad_norm": 0.0,
+      "learning_rate": 8.30967803569064e-06,
+      "loss": 0.8937,
+      "step": 15603
+    },
+    {
+      "epoch": 0.5674800887369531,
+      "grad_norm": 0.0,
+      "learning_rate": 8.308517134551341e-06,
+      "loss": 1.0653,
+      "step": 15604
+    },
+    {
+      "epoch": 0.5675164563406917,
+      "grad_norm": 0.0,
+      "learning_rate": 8.30735625687897e-06,
+      "loss": 1.0119,
+      "step": 15605
+    },
+    {
+      "epoch": 0.5675528239444303,
+      "grad_norm": 0.0,
+      "learning_rate": 8.306195402689624e-06,
+      "loss": 1.0611,
+      "step": 15606
+    },
+    {
+      "epoch": 0.5675891915481689,
+      "grad_norm": 0.0,
+      "learning_rate": 8.30503457199942e-06,
+      "loss": 1.0451,
+      "step": 15607
+    },
+    {
+      "epoch": 0.5676255591519075,
+      "grad_norm": 0.0,
+      "learning_rate": 8.30387376482445e-06,
+      "loss": 1.0264,
+      "step": 15608
+    },
+    {
+      "epoch": 0.5676619267556461,
+      "grad_norm": 0.0,
+      "learning_rate": 8.302712981180832e-06,
+      "loss": 0.9907,
+      "step": 15609
+    },
+    {
+      "epoch": 0.5676982943593847,
+      "grad_norm": 0.0,
+      "learning_rate": 8.301552221084658e-06,
+      "loss": 0.9439,
+      "step": 15610
+    },
+    {
+      "epoch": 0.5677346619631233,
+      "grad_norm": 0.0,
+      "learning_rate": 8.30039148455204e-06,
+      "loss": 0.9937,
+      "step": 15611
+    },
+    {
+      "epoch": 0.5677710295668619,
+      "grad_norm": 0.0,
+      "learning_rate": 8.299230771599077e-06,
+      "loss": 0.9692,
+      "step": 15612
+    },
+    {
+      "epoch": 0.5678073971706005,
+      "grad_norm": 0.0,
+      "learning_rate": 8.298070082241874e-06,
+      "loss": 1.0779,
+      "step": 15613
+    },
+    {
+      "epoch": 0.5678437647743391,
+      "grad_norm": 0.0,
+      "learning_rate": 8.296909416496532e-06,
+      "loss": 0.9953,
+      "step": 15614
+    },
+    {
+      "epoch": 0.5678801323780777,
+      "grad_norm": 0.0,
+      "learning_rate": 8.295748774379163e-06,
+      "loss": 0.9358,
+      "step": 15615
+    },
+    {
+      "epoch": 0.5679164999818161,
+      "grad_norm": 0.0,
+      "learning_rate": 8.294588155905854e-06,
+      "loss": 0.8847,
+      "step": 15616
+    },
+    {
+      "epoch": 0.5679528675855547,
+      "grad_norm": 0.0,
+      "learning_rate": 8.293427561092718e-06,
+      "loss": 1.0435,
+      "step": 15617
+    },
+    {
+      "epoch": 0.5679892351892933,
+      "grad_norm": 0.0,
+      "learning_rate": 8.292266989955853e-06,
+      "loss": 1.0235,
+      "step": 15618
+    },
+    {
+      "epoch": 0.5680256027930319,
+      "grad_norm": 0.0,
+      "learning_rate": 8.29110644251136e-06,
+      "loss": 0.9617,
+      "step": 15619
+    },
+    {
+      "epoch": 0.5680619703967705,
+      "grad_norm": 0.0,
+      "learning_rate": 8.289945918775343e-06,
+      "loss": 0.9535,
+      "step": 15620
+    },
+    {
+      "epoch": 0.5680983380005091,
+      "grad_norm": 0.0,
+      "learning_rate": 8.288785418763895e-06,
+      "loss": 0.9531,
+      "step": 15621
+    },
+    {
+      "epoch": 0.5681347056042477,
+      "grad_norm": 0.0,
+      "learning_rate": 8.287624942493126e-06,
+      "loss": 0.9579,
+      "step": 15622
+    },
+    {
+      "epoch": 0.5681710732079863,
+      "grad_norm": 0.0,
+      "learning_rate": 8.286464489979128e-06,
+      "loss": 0.8925,
+      "step": 15623
+    },
+    {
+      "epoch": 0.5682074408117249,
+      "grad_norm": 0.0,
+      "learning_rate": 8.285304061238006e-06,
+      "loss": 1.0,
+      "step": 15624
+    },
+    {
+      "epoch": 0.5682438084154635,
+      "grad_norm": 0.0,
+      "learning_rate": 8.284143656285858e-06,
+      "loss": 0.9846,
+      "step": 15625
+    },
+    {
+      "epoch": 0.5682801760192021,
+      "grad_norm": 0.0,
+      "learning_rate": 8.28298327513878e-06,
+      "loss": 0.9467,
+      "step": 15626
+    },
+    {
+      "epoch": 0.5683165436229407,
+      "grad_norm": 0.0,
+      "learning_rate": 8.281822917812873e-06,
+      "loss": 0.8121,
+      "step": 15627
+    },
+    {
+      "epoch": 0.5683529112266793,
+      "grad_norm": 0.0,
+      "learning_rate": 8.28066258432424e-06,
+      "loss": 1.0062,
+      "step": 15628
+    },
+    {
+      "epoch": 0.5683892788304179,
+      "grad_norm": 0.0,
+      "learning_rate": 8.279502274688968e-06,
+      "loss": 0.964,
+      "step": 15629
+    },
+    {
+      "epoch": 0.5684256464341565,
+      "grad_norm": 0.0,
+      "learning_rate": 8.278341988923164e-06,
+      "loss": 1.0172,
+      "step": 15630
+    },
+    {
+      "epoch": 0.5684620140378951,
+      "grad_norm": 0.0,
+      "learning_rate": 8.277181727042921e-06,
+      "loss": 0.9304,
+      "step": 15631
+    },
+    {
+      "epoch": 0.5684983816416337,
+      "grad_norm": 0.0,
+      "learning_rate": 8.27602148906434e-06,
+      "loss": 0.9669,
+      "step": 15632
+    },
+    {
+      "epoch": 0.5685347492453722,
+      "grad_norm": 0.0,
+      "learning_rate": 8.27486127500351e-06,
+      "loss": 0.9933,
+      "step": 15633
+    },
+    {
+      "epoch": 0.5685711168491108,
+      "grad_norm": 0.0,
+      "learning_rate": 8.273701084876538e-06,
+      "loss": 0.9792,
+      "step": 15634
+    },
+    {
+      "epoch": 0.5686074844528494,
+      "grad_norm": 0.0,
+      "learning_rate": 8.272540918699507e-06,
+      "loss": 0.9551,
+      "step": 15635
+    },
+    {
+      "epoch": 0.568643852056588,
+      "grad_norm": 0.0,
+      "learning_rate": 8.271380776488527e-06,
+      "loss": 1.0385,
+      "step": 15636
+    },
+    {
+      "epoch": 0.5686802196603266,
+      "grad_norm": 0.0,
+      "learning_rate": 8.270220658259684e-06,
+      "loss": 0.9368,
+      "step": 15637
+    },
+    {
+      "epoch": 0.5687165872640652,
+      "grad_norm": 0.0,
+      "learning_rate": 8.269060564029073e-06,
+      "loss": 1.0354,
+      "step": 15638
+    },
+    {
+      "epoch": 0.5687529548678038,
+      "grad_norm": 0.0,
+      "learning_rate": 8.267900493812794e-06,
+      "loss": 1.0286,
+      "step": 15639
+    },
+    {
+      "epoch": 0.5687893224715423,
+      "grad_norm": 0.0,
+      "learning_rate": 8.266740447626932e-06,
+      "loss": 0.9478,
+      "step": 15640
+    },
+    {
+      "epoch": 0.5688256900752809,
+      "grad_norm": 0.0,
+      "learning_rate": 8.265580425487593e-06,
+      "loss": 1.0422,
+      "step": 15641
+    },
+    {
+      "epoch": 0.5688620576790195,
+      "grad_norm": 0.0,
+      "learning_rate": 8.264420427410861e-06,
+      "loss": 1.0788,
+      "step": 15642
+    },
+    {
+      "epoch": 0.5688984252827581,
+      "grad_norm": 0.0,
+      "learning_rate": 8.263260453412837e-06,
+      "loss": 1.008,
+      "step": 15643
+    },
+    {
+      "epoch": 0.5689347928864967,
+      "grad_norm": 0.0,
+      "learning_rate": 8.262100503509606e-06,
+      "loss": 0.8738,
+      "step": 15644
+    },
+    {
+      "epoch": 0.5689711604902353,
+      "grad_norm": 0.0,
+      "learning_rate": 8.260940577717268e-06,
+      "loss": 1.1758,
+      "step": 15645
+    },
+    {
+      "epoch": 0.5690075280939739,
+      "grad_norm": 0.0,
+      "learning_rate": 8.25978067605191e-06,
+      "loss": 0.9336,
+      "step": 15646
+    },
+    {
+      "epoch": 0.5690438956977125,
+      "grad_norm": 0.0,
+      "learning_rate": 8.258620798529629e-06,
+      "loss": 0.9673,
+      "step": 15647
+    },
+    {
+      "epoch": 0.5690802633014511,
+      "grad_norm": 0.0,
+      "learning_rate": 8.25746094516651e-06,
+      "loss": 1.0031,
+      "step": 15648
+    },
+    {
+      "epoch": 0.5691166309051896,
+      "grad_norm": 0.0,
+      "learning_rate": 8.256301115978652e-06,
+      "loss": 0.9683,
+      "step": 15649
+    },
+    {
+      "epoch": 0.5691529985089282,
+      "grad_norm": 0.0,
+      "learning_rate": 8.255141310982138e-06,
+      "loss": 1.0332,
+      "step": 15650
+    },
+    {
+      "epoch": 0.5691893661126668,
+      "grad_norm": 0.0,
+      "learning_rate": 8.253981530193064e-06,
+      "loss": 0.9506,
+      "step": 15651
+    },
+    {
+      "epoch": 0.5692257337164054,
+      "grad_norm": 0.0,
+      "learning_rate": 8.25282177362752e-06,
+      "loss": 0.9141,
+      "step": 15652
+    },
+    {
+      "epoch": 0.569262101320144,
+      "grad_norm": 0.0,
+      "learning_rate": 8.251662041301595e-06,
+      "loss": 1.0701,
+      "step": 15653
+    },
+    {
+      "epoch": 0.5692984689238826,
+      "grad_norm": 0.0,
+      "learning_rate": 8.250502333231376e-06,
+      "loss": 1.0113,
+      "step": 15654
+    },
+    {
+      "epoch": 0.5693348365276212,
+      "grad_norm": 0.0,
+      "learning_rate": 8.24934264943296e-06,
+      "loss": 1.0083,
+      "step": 15655
+    },
+    {
+      "epoch": 0.5693712041313598,
+      "grad_norm": 0.0,
+      "learning_rate": 8.248182989922429e-06,
+      "loss": 0.9501,
+      "step": 15656
+    },
+    {
+      "epoch": 0.5694075717350984,
+      "grad_norm": 0.0,
+      "learning_rate": 8.24702335471587e-06,
+      "loss": 0.9712,
+      "step": 15657
+    },
+    {
+      "epoch": 0.569443939338837,
+      "grad_norm": 0.0,
+      "learning_rate": 8.24586374382938e-06,
+      "loss": 1.0376,
+      "step": 15658
+    },
+    {
+      "epoch": 0.5694803069425756,
+      "grad_norm": 0.0,
+      "learning_rate": 8.244704157279036e-06,
+      "loss": 1.0701,
+      "step": 15659
+    },
+    {
+      "epoch": 0.5695166745463142,
+      "grad_norm": 0.0,
+      "learning_rate": 8.243544595080937e-06,
+      "loss": 0.962,
+      "step": 15660
+    },
+    {
+      "epoch": 0.5695530421500528,
+      "grad_norm": 0.0,
+      "learning_rate": 8.242385057251162e-06,
+      "loss": 1.1231,
+      "step": 15661
+    },
+    {
+      "epoch": 0.5695894097537914,
+      "grad_norm": 0.0,
+      "learning_rate": 8.2412255438058e-06,
+      "loss": 1.0794,
+      "step": 15662
+    },
+    {
+      "epoch": 0.56962577735753,
+      "grad_norm": 0.0,
+      "learning_rate": 8.240066054760937e-06,
+      "loss": 0.9622,
+      "step": 15663
+    },
+    {
+      "epoch": 0.5696621449612685,
+      "grad_norm": 0.0,
+      "learning_rate": 8.238906590132662e-06,
+      "loss": 1.0737,
+      "step": 15664
+    },
+    {
+      "epoch": 0.5696985125650071,
+      "grad_norm": 0.0,
+      "learning_rate": 8.237747149937059e-06,
+      "loss": 1.1173,
+      "step": 15665
+    },
+    {
+      "epoch": 0.5697348801687456,
+      "grad_norm": 0.0,
+      "learning_rate": 8.236587734190218e-06,
+      "loss": 0.9481,
+      "step": 15666
+    },
+    {
+      "epoch": 0.5697712477724842,
+      "grad_norm": 0.0,
+      "learning_rate": 8.235428342908213e-06,
+      "loss": 1.0132,
+      "step": 15667
+    },
+    {
+      "epoch": 0.5698076153762228,
+      "grad_norm": 0.0,
+      "learning_rate": 8.234268976107142e-06,
+      "loss": 0.9685,
+      "step": 15668
+    },
+    {
+      "epoch": 0.5698439829799614,
+      "grad_norm": 0.0,
+      "learning_rate": 8.233109633803081e-06,
+      "loss": 1.0117,
+      "step": 15669
+    },
+    {
+      "epoch": 0.5698803505837,
+      "grad_norm": 0.0,
+      "learning_rate": 8.231950316012118e-06,
+      "loss": 1.0358,
+      "step": 15670
+    },
+    {
+      "epoch": 0.5699167181874386,
+      "grad_norm": 0.0,
+      "learning_rate": 8.230791022750336e-06,
+      "loss": 0.9138,
+      "step": 15671
+    },
+    {
+      "epoch": 0.5699530857911772,
+      "grad_norm": 0.0,
+      "learning_rate": 8.229631754033818e-06,
+      "loss": 1.044,
+      "step": 15672
+    },
+    {
+      "epoch": 0.5699894533949158,
+      "grad_norm": 0.0,
+      "learning_rate": 8.228472509878646e-06,
+      "loss": 0.9738,
+      "step": 15673
+    },
+    {
+      "epoch": 0.5700258209986544,
+      "grad_norm": 0.0,
+      "learning_rate": 8.227313290300907e-06,
+      "loss": 0.9452,
+      "step": 15674
+    },
+    {
+      "epoch": 0.570062188602393,
+      "grad_norm": 0.0,
+      "learning_rate": 8.226154095316682e-06,
+      "loss": 1.0219,
+      "step": 15675
+    },
+    {
+      "epoch": 0.5700985562061316,
+      "grad_norm": 0.0,
+      "learning_rate": 8.224994924942049e-06,
+      "loss": 0.9652,
+      "step": 15676
+    },
+    {
+      "epoch": 0.5701349238098702,
+      "grad_norm": 0.0,
+      "learning_rate": 8.223835779193094e-06,
+      "loss": 0.9778,
+      "step": 15677
+    },
+    {
+      "epoch": 0.5701712914136088,
+      "grad_norm": 0.0,
+      "learning_rate": 8.222676658085897e-06,
+      "loss": 1.0637,
+      "step": 15678
+    },
+    {
+      "epoch": 0.5702076590173474,
+      "grad_norm": 0.0,
+      "learning_rate": 8.221517561636543e-06,
+      "loss": 0.8923,
+      "step": 15679
+    },
+    {
+      "epoch": 0.570244026621086,
+      "grad_norm": 0.0,
+      "learning_rate": 8.220358489861102e-06,
+      "loss": 0.9818,
+      "step": 15680
+    },
+    {
+      "epoch": 0.5702803942248246,
+      "grad_norm": 0.0,
+      "learning_rate": 8.21919944277567e-06,
+      "loss": 1.0104,
+      "step": 15681
+    },
+    {
+      "epoch": 0.5703167618285632,
+      "grad_norm": 0.0,
+      "learning_rate": 8.218040420396315e-06,
+      "loss": 0.9698,
+      "step": 15682
+    },
+    {
+      "epoch": 0.5703531294323017,
+      "grad_norm": 0.0,
+      "learning_rate": 8.216881422739123e-06,
+      "loss": 0.9431,
+      "step": 15683
+    },
+    {
+      "epoch": 0.5703894970360402,
+      "grad_norm": 0.0,
+      "learning_rate": 8.215722449820167e-06,
+      "loss": 0.8937,
+      "step": 15684
+    },
+    {
+      "epoch": 0.5704258646397788,
+      "grad_norm": 0.0,
+      "learning_rate": 8.214563501655535e-06,
+      "loss": 1.1656,
+      "step": 15685
+    },
+    {
+      "epoch": 0.5704622322435174,
+      "grad_norm": 0.0,
+      "learning_rate": 8.213404578261298e-06,
+      "loss": 1.0778,
+      "step": 15686
+    },
+    {
+      "epoch": 0.570498599847256,
+      "grad_norm": 0.0,
+      "learning_rate": 8.212245679653544e-06,
+      "loss": 0.8486,
+      "step": 15687
+    },
+    {
+      "epoch": 0.5705349674509946,
+      "grad_norm": 0.0,
+      "learning_rate": 8.211086805848337e-06,
+      "loss": 0.9961,
+      "step": 15688
+    },
+    {
+      "epoch": 0.5705713350547332,
+      "grad_norm": 0.0,
+      "learning_rate": 8.209927956861766e-06,
+      "loss": 0.9875,
+      "step": 15689
+    },
+    {
+      "epoch": 0.5706077026584718,
+      "grad_norm": 0.0,
+      "learning_rate": 8.208769132709904e-06,
+      "loss": 1.008,
+      "step": 15690
+    },
+    {
+      "epoch": 0.5706440702622104,
+      "grad_norm": 0.0,
+      "learning_rate": 8.207610333408829e-06,
+      "loss": 1.0369,
+      "step": 15691
+    },
+    {
+      "epoch": 0.570680437865949,
+      "grad_norm": 0.0,
+      "learning_rate": 8.206451558974616e-06,
+      "loss": 1.0449,
+      "step": 15692
+    },
+    {
+      "epoch": 0.5707168054696876,
+      "grad_norm": 0.0,
+      "learning_rate": 8.205292809423348e-06,
+      "loss": 1.0505,
+      "step": 15693
+    },
+    {
+      "epoch": 0.5707531730734262,
+      "grad_norm": 0.0,
+      "learning_rate": 8.204134084771094e-06,
+      "loss": 1.0257,
+      "step": 15694
+    },
+    {
+      "epoch": 0.5707895406771648,
+      "grad_norm": 0.0,
+      "learning_rate": 8.202975385033928e-06,
+      "loss": 0.8704,
+      "step": 15695
+    },
+    {
+      "epoch": 0.5708259082809034,
+      "grad_norm": 0.0,
+      "learning_rate": 8.201816710227933e-06,
+      "loss": 1.0067,
+      "step": 15696
+    },
+    {
+      "epoch": 0.570862275884642,
+      "grad_norm": 0.0,
+      "learning_rate": 8.200658060369176e-06,
+      "loss": 0.9354,
+      "step": 15697
+    },
+    {
+      "epoch": 0.5708986434883806,
+      "grad_norm": 0.0,
+      "learning_rate": 8.199499435473739e-06,
+      "loss": 0.9897,
+      "step": 15698
+    },
+    {
+      "epoch": 0.5709350110921192,
+      "grad_norm": 0.0,
+      "learning_rate": 8.198340835557687e-06,
+      "loss": 1.0266,
+      "step": 15699
+    },
+    {
+      "epoch": 0.5709713786958577,
+      "grad_norm": 0.0,
+      "learning_rate": 8.197182260637107e-06,
+      "loss": 1.0994,
+      "step": 15700
+    },
+    {
+      "epoch": 0.5710077462995963,
+      "grad_norm": 0.0,
+      "learning_rate": 8.196023710728062e-06,
+      "loss": 0.9704,
+      "step": 15701
+    },
+    {
+      "epoch": 0.5710441139033349,
+      "grad_norm": 0.0,
+      "learning_rate": 8.19486518584663e-06,
+      "loss": 1.0632,
+      "step": 15702
+    },
+    {
+      "epoch": 0.5710804815070735,
+      "grad_norm": 0.0,
+      "learning_rate": 8.193706686008878e-06,
+      "loss": 1.0356,
+      "step": 15703
+    },
+    {
+      "epoch": 0.5711168491108121,
+      "grad_norm": 0.0,
+      "learning_rate": 8.192548211230886e-06,
+      "loss": 0.9788,
+      "step": 15704
+    },
+    {
+      "epoch": 0.5711532167145507,
+      "grad_norm": 0.0,
+      "learning_rate": 8.191389761528723e-06,
+      "loss": 1.0847,
+      "step": 15705
+    },
+    {
+      "epoch": 0.5711895843182893,
+      "grad_norm": 0.0,
+      "learning_rate": 8.190231336918464e-06,
+      "loss": 0.9574,
+      "step": 15706
+    },
+    {
+      "epoch": 0.5712259519220279,
+      "grad_norm": 0.0,
+      "learning_rate": 8.18907293741617e-06,
+      "loss": 0.9596,
+      "step": 15707
+    },
+    {
+      "epoch": 0.5712623195257664,
+      "grad_norm": 0.0,
+      "learning_rate": 8.187914563037928e-06,
+      "loss": 1.0307,
+      "step": 15708
+    },
+    {
+      "epoch": 0.571298687129505,
+      "grad_norm": 0.0,
+      "learning_rate": 8.186756213799795e-06,
+      "loss": 0.9677,
+      "step": 15709
+    },
+    {
+      "epoch": 0.5713350547332436,
+      "grad_norm": 0.0,
+      "learning_rate": 8.18559788971785e-06,
+      "loss": 0.9773,
+      "step": 15710
+    },
+    {
+      "epoch": 0.5713714223369822,
+      "grad_norm": 0.0,
+      "learning_rate": 8.184439590808155e-06,
+      "loss": 0.8836,
+      "step": 15711
+    },
+    {
+      "epoch": 0.5714077899407208,
+      "grad_norm": 0.0,
+      "learning_rate": 8.18328131708679e-06,
+      "loss": 0.9487,
+      "step": 15712
+    },
+    {
+      "epoch": 0.5714441575444594,
+      "grad_norm": 0.0,
+      "learning_rate": 8.182123068569815e-06,
+      "loss": 0.9497,
+      "step": 15713
+    },
+    {
+      "epoch": 0.571480525148198,
+      "grad_norm": 0.0,
+      "learning_rate": 8.180964845273309e-06,
+      "loss": 1.0667,
+      "step": 15714
+    },
+    {
+      "epoch": 0.5715168927519366,
+      "grad_norm": 0.0,
+      "learning_rate": 8.179806647213332e-06,
+      "loss": 0.9814,
+      "step": 15715
+    },
+    {
+      "epoch": 0.5715532603556752,
+      "grad_norm": 0.0,
+      "learning_rate": 8.178648474405952e-06,
+      "loss": 0.767,
+      "step": 15716
+    },
+    {
+      "epoch": 0.5715896279594137,
+      "grad_norm": 0.0,
+      "learning_rate": 8.177490326867245e-06,
+      "loss": 1.0421,
+      "step": 15717
+    },
+    {
+      "epoch": 0.5716259955631523,
+      "grad_norm": 0.0,
+      "learning_rate": 8.176332204613269e-06,
+      "loss": 1.1487,
+      "step": 15718
+    },
+    {
+      "epoch": 0.5716623631668909,
+      "grad_norm": 0.0,
+      "learning_rate": 8.175174107660103e-06,
+      "loss": 0.9387,
+      "step": 15719
+    },
+    {
+      "epoch": 0.5716987307706295,
+      "grad_norm": 0.0,
+      "learning_rate": 8.174016036023802e-06,
+      "loss": 0.8577,
+      "step": 15720
+    },
+    {
+      "epoch": 0.5717350983743681,
+      "grad_norm": 0.0,
+      "learning_rate": 8.17285798972044e-06,
+      "loss": 0.9949,
+      "step": 15721
+    },
+    {
+      "epoch": 0.5717714659781067,
+      "grad_norm": 0.0,
+      "learning_rate": 8.171699968766079e-06,
+      "loss": 0.9916,
+      "step": 15722
+    },
+    {
+      "epoch": 0.5718078335818453,
+      "grad_norm": 0.0,
+      "learning_rate": 8.170541973176788e-06,
+      "loss": 1.0747,
+      "step": 15723
+    },
+    {
+      "epoch": 0.5718442011855839,
+      "grad_norm": 0.0,
+      "learning_rate": 8.169384002968628e-06,
+      "loss": 0.9443,
+      "step": 15724
+    },
+    {
+      "epoch": 0.5718805687893225,
+      "grad_norm": 0.0,
+      "learning_rate": 8.168226058157674e-06,
+      "loss": 0.9291,
+      "step": 15725
+    },
+    {
+      "epoch": 0.5719169363930611,
+      "grad_norm": 0.0,
+      "learning_rate": 8.16706813875998e-06,
+      "loss": 0.9992,
+      "step": 15726
+    },
+    {
+      "epoch": 0.5719533039967997,
+      "grad_norm": 0.0,
+      "learning_rate": 8.165910244791619e-06,
+      "loss": 0.9502,
+      "step": 15727
+    },
+    {
+      "epoch": 0.5719896716005383,
+      "grad_norm": 0.0,
+      "learning_rate": 8.164752376268646e-06,
+      "loss": 0.9178,
+      "step": 15728
+    },
+    {
+      "epoch": 0.5720260392042769,
+      "grad_norm": 0.0,
+      "learning_rate": 8.163594533207133e-06,
+      "loss": 1.1651,
+      "step": 15729
+    },
+    {
+      "epoch": 0.5720624068080155,
+      "grad_norm": 0.0,
+      "learning_rate": 8.16243671562314e-06,
+      "loss": 0.9004,
+      "step": 15730
+    },
+    {
+      "epoch": 0.572098774411754,
+      "grad_norm": 0.0,
+      "learning_rate": 8.161278923532729e-06,
+      "loss": 0.9047,
+      "step": 15731
+    },
+    {
+      "epoch": 0.5721351420154926,
+      "grad_norm": 0.0,
+      "learning_rate": 8.160121156951965e-06,
+      "loss": 0.9995,
+      "step": 15732
+    },
+    {
+      "epoch": 0.5721715096192311,
+      "grad_norm": 0.0,
+      "learning_rate": 8.158963415896912e-06,
+      "loss": 0.9595,
+      "step": 15733
+    },
+    {
+      "epoch": 0.5722078772229697,
+      "grad_norm": 0.0,
+      "learning_rate": 8.157805700383628e-06,
+      "loss": 1.0441,
+      "step": 15734
+    },
+    {
+      "epoch": 0.5722442448267083,
+      "grad_norm": 0.0,
+      "learning_rate": 8.156648010428175e-06,
+      "loss": 1.012,
+      "step": 15735
+    },
+    {
+      "epoch": 0.5722806124304469,
+      "grad_norm": 0.0,
+      "learning_rate": 8.155490346046618e-06,
+      "loss": 0.9958,
+      "step": 15736
+    },
+    {
+      "epoch": 0.5723169800341855,
+      "grad_norm": 0.0,
+      "learning_rate": 8.154332707255011e-06,
+      "loss": 0.9717,
+      "step": 15737
+    },
+    {
+      "epoch": 0.5723533476379241,
+      "grad_norm": 0.0,
+      "learning_rate": 8.153175094069424e-06,
+      "loss": 0.9903,
+      "step": 15738
+    },
+    {
+      "epoch": 0.5723897152416627,
+      "grad_norm": 0.0,
+      "learning_rate": 8.152017506505906e-06,
+      "loss": 1.0523,
+      "step": 15739
+    },
+    {
+      "epoch": 0.5724260828454013,
+      "grad_norm": 0.0,
+      "learning_rate": 8.15085994458053e-06,
+      "loss": 0.8997,
+      "step": 15740
+    },
+    {
+      "epoch": 0.5724624504491399,
+      "grad_norm": 0.0,
+      "learning_rate": 8.149702408309345e-06,
+      "loss": 1.0128,
+      "step": 15741
+    },
+    {
+      "epoch": 0.5724988180528785,
+      "grad_norm": 0.0,
+      "learning_rate": 8.148544897708415e-06,
+      "loss": 1.1072,
+      "step": 15742
+    },
+    {
+      "epoch": 0.5725351856566171,
+      "grad_norm": 0.0,
+      "learning_rate": 8.147387412793794e-06,
+      "loss": 1.0868,
+      "step": 15743
+    },
+    {
+      "epoch": 0.5725715532603557,
+      "grad_norm": 0.0,
+      "learning_rate": 8.14622995358155e-06,
+      "loss": 1.0204,
+      "step": 15744
+    },
+    {
+      "epoch": 0.5726079208640943,
+      "grad_norm": 0.0,
+      "learning_rate": 8.14507252008773e-06,
+      "loss": 0.9551,
+      "step": 15745
+    },
+    {
+      "epoch": 0.5726442884678329,
+      "grad_norm": 0.0,
+      "learning_rate": 8.143915112328403e-06,
+      "loss": 0.9654,
+      "step": 15746
+    },
+    {
+      "epoch": 0.5726806560715715,
+      "grad_norm": 0.0,
+      "learning_rate": 8.142757730319612e-06,
+      "loss": 0.9396,
+      "step": 15747
+    },
+    {
+      "epoch": 0.5727170236753101,
+      "grad_norm": 0.0,
+      "learning_rate": 8.14160037407743e-06,
+      "loss": 1.0497,
+      "step": 15748
+    },
+    {
+      "epoch": 0.5727533912790487,
+      "grad_norm": 0.0,
+      "learning_rate": 8.140443043617902e-06,
+      "loss": 0.983,
+      "step": 15749
+    },
+    {
+      "epoch": 0.5727897588827872,
+      "grad_norm": 0.0,
+      "learning_rate": 8.13928573895709e-06,
+      "loss": 0.9515,
+      "step": 15750
+    },
+    {
+      "epoch": 0.5728261264865258,
+      "grad_norm": 0.0,
+      "learning_rate": 8.138128460111047e-06,
+      "loss": 1.0554,
+      "step": 15751
+    },
+    {
+      "epoch": 0.5728624940902644,
+      "grad_norm": 0.0,
+      "learning_rate": 8.136971207095833e-06,
+      "loss": 0.9005,
+      "step": 15752
+    },
+    {
+      "epoch": 0.572898861694003,
+      "grad_norm": 0.0,
+      "learning_rate": 8.135813979927502e-06,
+      "loss": 0.925,
+      "step": 15753
+    },
+    {
+      "epoch": 0.5729352292977415,
+      "grad_norm": 0.0,
+      "learning_rate": 8.1346567786221e-06,
+      "loss": 1.0075,
+      "step": 15754
+    },
+    {
+      "epoch": 0.5729715969014801,
+      "grad_norm": 0.0,
+      "learning_rate": 8.133499603195694e-06,
+      "loss": 1.004,
+      "step": 15755
+    },
+    {
+      "epoch": 0.5730079645052187,
+      "grad_norm": 0.0,
+      "learning_rate": 8.13234245366433e-06,
+      "loss": 0.982,
+      "step": 15756
+    },
+    {
+      "epoch": 0.5730443321089573,
+      "grad_norm": 0.0,
+      "learning_rate": 8.131185330044065e-06,
+      "loss": 1.1029,
+      "step": 15757
+    },
+    {
+      "epoch": 0.5730806997126959,
+      "grad_norm": 0.0,
+      "learning_rate": 8.130028232350951e-06,
+      "loss": 0.9213,
+      "step": 15758
+    },
+    {
+      "epoch": 0.5731170673164345,
+      "grad_norm": 0.0,
+      "learning_rate": 8.128871160601047e-06,
+      "loss": 1.0454,
+      "step": 15759
+    },
+    {
+      "epoch": 0.5731534349201731,
+      "grad_norm": 0.0,
+      "learning_rate": 8.127714114810395e-06,
+      "loss": 0.9563,
+      "step": 15760
+    },
+    {
+      "epoch": 0.5731898025239117,
+      "grad_norm": 0.0,
+      "learning_rate": 8.126557094995059e-06,
+      "loss": 1.0129,
+      "step": 15761
+    },
+    {
+      "epoch": 0.5732261701276503,
+      "grad_norm": 0.0,
+      "learning_rate": 8.125400101171079e-06,
+      "loss": 0.9492,
+      "step": 15762
+    },
+    {
+      "epoch": 0.5732625377313889,
+      "grad_norm": 0.0,
+      "learning_rate": 8.124243133354516e-06,
+      "loss": 0.9285,
+      "step": 15763
+    },
+    {
+      "epoch": 0.5732989053351275,
+      "grad_norm": 0.0,
+      "learning_rate": 8.123086191561417e-06,
+      "loss": 0.9768,
+      "step": 15764
+    },
+    {
+      "epoch": 0.5733352729388661,
+      "grad_norm": 0.0,
+      "learning_rate": 8.121929275807838e-06,
+      "loss": 0.9089,
+      "step": 15765
+    },
+    {
+      "epoch": 0.5733716405426047,
+      "grad_norm": 0.0,
+      "learning_rate": 8.120772386109818e-06,
+      "loss": 0.98,
+      "step": 15766
+    },
+    {
+      "epoch": 0.5734080081463432,
+      "grad_norm": 0.0,
+      "learning_rate": 8.119615522483424e-06,
+      "loss": 0.9215,
+      "step": 15767
+    },
+    {
+      "epoch": 0.5734443757500818,
+      "grad_norm": 0.0,
+      "learning_rate": 8.11845868494469e-06,
+      "loss": 0.9727,
+      "step": 15768
+    },
+    {
+      "epoch": 0.5734807433538204,
+      "grad_norm": 0.0,
+      "learning_rate": 8.117301873509675e-06,
+      "loss": 1.0402,
+      "step": 15769
+    },
+    {
+      "epoch": 0.573517110957559,
+      "grad_norm": 0.0,
+      "learning_rate": 8.116145088194424e-06,
+      "loss": 1.0872,
+      "step": 15770
+    },
+    {
+      "epoch": 0.5735534785612976,
+      "grad_norm": 0.0,
+      "learning_rate": 8.114988329014988e-06,
+      "loss": 0.9722,
+      "step": 15771
+    },
+    {
+      "epoch": 0.5735898461650362,
+      "grad_norm": 0.0,
+      "learning_rate": 8.113831595987418e-06,
+      "loss": 1.0547,
+      "step": 15772
+    },
+    {
+      "epoch": 0.5736262137687748,
+      "grad_norm": 0.0,
+      "learning_rate": 8.112674889127752e-06,
+      "loss": 1.0381,
+      "step": 15773
+    },
+    {
+      "epoch": 0.5736625813725134,
+      "grad_norm": 0.0,
+      "learning_rate": 8.111518208452048e-06,
+      "loss": 1.0505,
+      "step": 15774
+    },
+    {
+      "epoch": 0.573698948976252,
+      "grad_norm": 0.0,
+      "learning_rate": 8.110361553976347e-06,
+      "loss": 1.0413,
+      "step": 15775
+    },
+    {
+      "epoch": 0.5737353165799906,
+      "grad_norm": 0.0,
+      "learning_rate": 8.1092049257167e-06,
+      "loss": 0.936,
+      "step": 15776
+    },
+    {
+      "epoch": 0.5737716841837291,
+      "grad_norm": 0.0,
+      "learning_rate": 8.108048323689152e-06,
+      "loss": 1.0289,
+      "step": 15777
+    },
+    {
+      "epoch": 0.5738080517874677,
+      "grad_norm": 0.0,
+      "learning_rate": 8.10689174790975e-06,
+      "loss": 0.9989,
+      "step": 15778
+    },
+    {
+      "epoch": 0.5738444193912063,
+      "grad_norm": 0.0,
+      "learning_rate": 8.105735198394535e-06,
+      "loss": 0.9953,
+      "step": 15779
+    },
+    {
+      "epoch": 0.5738807869949449,
+      "grad_norm": 0.0,
+      "learning_rate": 8.104578675159564e-06,
+      "loss": 1.0231,
+      "step": 15780
+    },
+    {
+      "epoch": 0.5739171545986835,
+      "grad_norm": 0.0,
+      "learning_rate": 8.103422178220868e-06,
+      "loss": 1.0067,
+      "step": 15781
+    },
+    {
+      "epoch": 0.5739535222024221,
+      "grad_norm": 0.0,
+      "learning_rate": 8.102265707594502e-06,
+      "loss": 0.8631,
+      "step": 15782
+    },
+    {
+      "epoch": 0.5739898898061607,
+      "grad_norm": 0.0,
+      "learning_rate": 8.101109263296503e-06,
+      "loss": 1.034,
+      "step": 15783
+    },
+    {
+      "epoch": 0.5740262574098992,
+      "grad_norm": 0.0,
+      "learning_rate": 8.099952845342922e-06,
+      "loss": 1.0772,
+      "step": 15784
+    },
+    {
+      "epoch": 0.5740626250136378,
+      "grad_norm": 0.0,
+      "learning_rate": 8.098796453749797e-06,
+      "loss": 1.0293,
+      "step": 15785
+    },
+    {
+      "epoch": 0.5740989926173764,
+      "grad_norm": 0.0,
+      "learning_rate": 8.09764008853318e-06,
+      "loss": 1.0344,
+      "step": 15786
+    },
+    {
+      "epoch": 0.574135360221115,
+      "grad_norm": 0.0,
+      "learning_rate": 8.0964837497091e-06,
+      "loss": 0.9939,
+      "step": 15787
+    },
+    {
+      "epoch": 0.5741717278248536,
+      "grad_norm": 0.0,
+      "learning_rate": 8.09532743729361e-06,
+      "loss": 1.0309,
+      "step": 15788
+    },
+    {
+      "epoch": 0.5742080954285922,
+      "grad_norm": 0.0,
+      "learning_rate": 8.09417115130275e-06,
+      "loss": 0.9958,
+      "step": 15789
+    },
+    {
+      "epoch": 0.5742444630323308,
+      "grad_norm": 0.0,
+      "learning_rate": 8.09301489175256e-06,
+      "loss": 0.9772,
+      "step": 15790
+    },
+    {
+      "epoch": 0.5742808306360694,
+      "grad_norm": 0.0,
+      "learning_rate": 8.091858658659086e-06,
+      "loss": 0.85,
+      "step": 15791
+    },
+    {
+      "epoch": 0.574317198239808,
+      "grad_norm": 0.0,
+      "learning_rate": 8.090702452038362e-06,
+      "loss": 0.9535,
+      "step": 15792
+    },
+    {
+      "epoch": 0.5743535658435466,
+      "grad_norm": 0.0,
+      "learning_rate": 8.089546271906436e-06,
+      "loss": 1.0792,
+      "step": 15793
+    },
+    {
+      "epoch": 0.5743899334472852,
+      "grad_norm": 0.0,
+      "learning_rate": 8.088390118279342e-06,
+      "loss": 1.1933,
+      "step": 15794
+    },
+    {
+      "epoch": 0.5744263010510238,
+      "grad_norm": 0.0,
+      "learning_rate": 8.087233991173123e-06,
+      "loss": 1.0269,
+      "step": 15795
+    },
+    {
+      "epoch": 0.5744626686547624,
+      "grad_norm": 0.0,
+      "learning_rate": 8.086077890603818e-06,
+      "loss": 1.011,
+      "step": 15796
+    },
+    {
+      "epoch": 0.574499036258501,
+      "grad_norm": 0.0,
+      "learning_rate": 8.084921816587468e-06,
+      "loss": 0.91,
+      "step": 15797
+    },
+    {
+      "epoch": 0.5745354038622396,
+      "grad_norm": 0.0,
+      "learning_rate": 8.08376576914011e-06,
+      "loss": 0.9713,
+      "step": 15798
+    },
+    {
+      "epoch": 0.5745717714659782,
+      "grad_norm": 0.0,
+      "learning_rate": 8.082609748277787e-06,
+      "loss": 0.9554,
+      "step": 15799
+    },
+    {
+      "epoch": 0.5746081390697168,
+      "grad_norm": 0.0,
+      "learning_rate": 8.081453754016528e-06,
+      "loss": 0.9572,
+      "step": 15800
+    },
+    {
+      "epoch": 0.5746445066734552,
+      "grad_norm": 0.0,
+      "learning_rate": 8.08029778637238e-06,
+      "loss": 0.9975,
+      "step": 15801
+    },
+    {
+      "epoch": 0.5746808742771938,
+      "grad_norm": 0.0,
+      "learning_rate": 8.079141845361373e-06,
+      "loss": 1.1298,
+      "step": 15802
+    },
+    {
+      "epoch": 0.5747172418809324,
+      "grad_norm": 0.0,
+      "learning_rate": 8.07798593099955e-06,
+      "loss": 1.0376,
+      "step": 15803
+    },
+    {
+      "epoch": 0.574753609484671,
+      "grad_norm": 0.0,
+      "learning_rate": 8.076830043302944e-06,
+      "loss": 1.0995,
+      "step": 15804
+    },
+    {
+      "epoch": 0.5747899770884096,
+      "grad_norm": 0.0,
+      "learning_rate": 8.075674182287596e-06,
+      "loss": 1.0598,
+      "step": 15805
+    },
+    {
+      "epoch": 0.5748263446921482,
+      "grad_norm": 0.0,
+      "learning_rate": 8.074518347969532e-06,
+      "loss": 0.9663,
+      "step": 15806
+    },
+    {
+      "epoch": 0.5748627122958868,
+      "grad_norm": 0.0,
+      "learning_rate": 8.0733625403648e-06,
+      "loss": 1.0151,
+      "step": 15807
+    },
+    {
+      "epoch": 0.5748990798996254,
+      "grad_norm": 0.0,
+      "learning_rate": 8.072206759489424e-06,
+      "loss": 0.9208,
+      "step": 15808
+    },
+    {
+      "epoch": 0.574935447503364,
+      "grad_norm": 0.0,
+      "learning_rate": 8.071051005359447e-06,
+      "loss": 1.0163,
+      "step": 15809
+    },
+    {
+      "epoch": 0.5749718151071026,
+      "grad_norm": 0.0,
+      "learning_rate": 8.069895277990904e-06,
+      "loss": 1.143,
+      "step": 15810
+    },
+    {
+      "epoch": 0.5750081827108412,
+      "grad_norm": 0.0,
+      "learning_rate": 8.068739577399818e-06,
+      "loss": 1.0024,
+      "step": 15811
+    },
+    {
+      "epoch": 0.5750445503145798,
+      "grad_norm": 0.0,
+      "learning_rate": 8.067583903602238e-06,
+      "loss": 0.9821,
+      "step": 15812
+    },
+    {
+      "epoch": 0.5750809179183184,
+      "grad_norm": 0.0,
+      "learning_rate": 8.066428256614184e-06,
+      "loss": 1.0746,
+      "step": 15813
+    },
+    {
+      "epoch": 0.575117285522057,
+      "grad_norm": 0.0,
+      "learning_rate": 8.065272636451697e-06,
+      "loss": 0.9785,
+      "step": 15814
+    },
+    {
+      "epoch": 0.5751536531257956,
+      "grad_norm": 0.0,
+      "learning_rate": 8.064117043130805e-06,
+      "loss": 0.9577,
+      "step": 15815
+    },
+    {
+      "epoch": 0.5751900207295342,
+      "grad_norm": 0.0,
+      "learning_rate": 8.062961476667544e-06,
+      "loss": 0.9016,
+      "step": 15816
+    },
+    {
+      "epoch": 0.5752263883332728,
+      "grad_norm": 0.0,
+      "learning_rate": 8.061805937077943e-06,
+      "loss": 0.9415,
+      "step": 15817
+    },
+    {
+      "epoch": 0.5752627559370113,
+      "grad_norm": 0.0,
+      "learning_rate": 8.060650424378039e-06,
+      "loss": 1.0901,
+      "step": 15818
+    },
+    {
+      "epoch": 0.5752991235407499,
+      "grad_norm": 0.0,
+      "learning_rate": 8.059494938583851e-06,
+      "loss": 1.069,
+      "step": 15819
+    },
+    {
+      "epoch": 0.5753354911444885,
+      "grad_norm": 0.0,
+      "learning_rate": 8.058339479711425e-06,
+      "loss": 0.9163,
+      "step": 15820
+    },
+    {
+      "epoch": 0.575371858748227,
+      "grad_norm": 0.0,
+      "learning_rate": 8.057184047776779e-06,
+      "loss": 0.9927,
+      "step": 15821
+    },
+    {
+      "epoch": 0.5754082263519656,
+      "grad_norm": 0.0,
+      "learning_rate": 8.056028642795952e-06,
+      "loss": 0.975,
+      "step": 15822
+    },
+    {
+      "epoch": 0.5754445939557042,
+      "grad_norm": 0.0,
+      "learning_rate": 8.054873264784965e-06,
+      "loss": 0.9389,
+      "step": 15823
+    },
+    {
+      "epoch": 0.5754809615594428,
+      "grad_norm": 0.0,
+      "learning_rate": 8.053717913759853e-06,
+      "loss": 1.031,
+      "step": 15824
+    },
+    {
+      "epoch": 0.5755173291631814,
+      "grad_norm": 0.0,
+      "learning_rate": 8.052562589736645e-06,
+      "loss": 0.9136,
+      "step": 15825
+    },
+    {
+      "epoch": 0.57555369676692,
+      "grad_norm": 0.0,
+      "learning_rate": 8.051407292731371e-06,
+      "loss": 1.0385,
+      "step": 15826
+    },
+    {
+      "epoch": 0.5755900643706586,
+      "grad_norm": 0.0,
+      "learning_rate": 8.05025202276005e-06,
+      "loss": 1.1358,
+      "step": 15827
+    },
+    {
+      "epoch": 0.5756264319743972,
+      "grad_norm": 0.0,
+      "learning_rate": 8.04909677983872e-06,
+      "loss": 0.837,
+      "step": 15828
+    },
+    {
+      "epoch": 0.5756627995781358,
+      "grad_norm": 0.0,
+      "learning_rate": 8.047941563983402e-06,
+      "loss": 0.9961,
+      "step": 15829
+    },
+    {
+      "epoch": 0.5756991671818744,
+      "grad_norm": 0.0,
+      "learning_rate": 8.046786375210124e-06,
+      "loss": 0.8978,
+      "step": 15830
+    },
+    {
+      "epoch": 0.575735534785613,
+      "grad_norm": 0.0,
+      "learning_rate": 8.045631213534918e-06,
+      "loss": 0.9824,
+      "step": 15831
+    },
+    {
+      "epoch": 0.5757719023893516,
+      "grad_norm": 0.0,
+      "learning_rate": 8.044476078973803e-06,
+      "loss": 1.0069,
+      "step": 15832
+    },
+    {
+      "epoch": 0.5758082699930902,
+      "grad_norm": 0.0,
+      "learning_rate": 8.043320971542809e-06,
+      "loss": 0.9986,
+      "step": 15833
+    },
+    {
+      "epoch": 0.5758446375968287,
+      "grad_norm": 0.0,
+      "learning_rate": 8.042165891257959e-06,
+      "loss": 0.9991,
+      "step": 15834
+    },
+    {
+      "epoch": 0.5758810052005673,
+      "grad_norm": 0.0,
+      "learning_rate": 8.04101083813528e-06,
+      "loss": 1.0782,
+      "step": 15835
+    },
+    {
+      "epoch": 0.5759173728043059,
+      "grad_norm": 0.0,
+      "learning_rate": 8.039855812190795e-06,
+      "loss": 1.0072,
+      "step": 15836
+    },
+    {
+      "epoch": 0.5759537404080445,
+      "grad_norm": 0.0,
+      "learning_rate": 8.038700813440533e-06,
+      "loss": 0.8985,
+      "step": 15837
+    },
+    {
+      "epoch": 0.5759901080117831,
+      "grad_norm": 0.0,
+      "learning_rate": 8.03754584190051e-06,
+      "loss": 0.9478,
+      "step": 15838
+    },
+    {
+      "epoch": 0.5760264756155217,
+      "grad_norm": 0.0,
+      "learning_rate": 8.03639089758676e-06,
+      "loss": 1.0645,
+      "step": 15839
+    },
+    {
+      "epoch": 0.5760628432192603,
+      "grad_norm": 0.0,
+      "learning_rate": 8.035235980515295e-06,
+      "loss": 1.0368,
+      "step": 15840
+    },
+    {
+      "epoch": 0.5760992108229989,
+      "grad_norm": 0.0,
+      "learning_rate": 8.034081090702146e-06,
+      "loss": 0.8878,
+      "step": 15841
+    },
+    {
+      "epoch": 0.5761355784267375,
+      "grad_norm": 0.0,
+      "learning_rate": 8.032926228163329e-06,
+      "loss": 1.0594,
+      "step": 15842
+    },
+    {
+      "epoch": 0.5761719460304761,
+      "grad_norm": 0.0,
+      "learning_rate": 8.031771392914873e-06,
+      "loss": 1.0438,
+      "step": 15843
+    },
+    {
+      "epoch": 0.5762083136342147,
+      "grad_norm": 0.0,
+      "learning_rate": 8.030616584972794e-06,
+      "loss": 1.1395,
+      "step": 15844
+    },
+    {
+      "epoch": 0.5762446812379532,
+      "grad_norm": 0.0,
+      "learning_rate": 8.02946180435312e-06,
+      "loss": 1.0519,
+      "step": 15845
+    },
+    {
+      "epoch": 0.5762810488416918,
+      "grad_norm": 0.0,
+      "learning_rate": 8.028307051071862e-06,
+      "loss": 0.8921,
+      "step": 15846
+    },
+    {
+      "epoch": 0.5763174164454304,
+      "grad_norm": 0.0,
+      "learning_rate": 8.027152325145052e-06,
+      "loss": 0.9929,
+      "step": 15847
+    },
+    {
+      "epoch": 0.576353784049169,
+      "grad_norm": 0.0,
+      "learning_rate": 8.025997626588702e-06,
+      "loss": 1.001,
+      "step": 15848
+    },
+    {
+      "epoch": 0.5763901516529076,
+      "grad_norm": 0.0,
+      "learning_rate": 8.024842955418832e-06,
+      "loss": 0.8521,
+      "step": 15849
+    },
+    {
+      "epoch": 0.5764265192566462,
+      "grad_norm": 0.0,
+      "learning_rate": 8.02368831165147e-06,
+      "loss": 0.9501,
+      "step": 15850
+    },
+    {
+      "epoch": 0.5764628868603847,
+      "grad_norm": 0.0,
+      "learning_rate": 8.02253369530262e-06,
+      "loss": 1.0167,
+      "step": 15851
+    },
+    {
+      "epoch": 0.5764992544641233,
+      "grad_norm": 0.0,
+      "learning_rate": 8.021379106388316e-06,
+      "loss": 0.9714,
+      "step": 15852
+    },
+    {
+      "epoch": 0.5765356220678619,
+      "grad_norm": 0.0,
+      "learning_rate": 8.020224544924566e-06,
+      "loss": 0.9935,
+      "step": 15853
+    },
+    {
+      "epoch": 0.5765719896716005,
+      "grad_norm": 0.0,
+      "learning_rate": 8.019070010927393e-06,
+      "loss": 0.9321,
+      "step": 15854
+    },
+    {
+      "epoch": 0.5766083572753391,
+      "grad_norm": 0.0,
+      "learning_rate": 8.01791550441281e-06,
+      "loss": 0.9082,
+      "step": 15855
+    },
+    {
+      "epoch": 0.5766447248790777,
+      "grad_norm": 0.0,
+      "learning_rate": 8.01676102539684e-06,
+      "loss": 0.9868,
+      "step": 15856
+    },
+    {
+      "epoch": 0.5766810924828163,
+      "grad_norm": 0.0,
+      "learning_rate": 8.015606573895496e-06,
+      "loss": 0.9835,
+      "step": 15857
+    },
+    {
+      "epoch": 0.5767174600865549,
+      "grad_norm": 0.0,
+      "learning_rate": 8.014452149924799e-06,
+      "loss": 0.9968,
+      "step": 15858
+    },
+    {
+      "epoch": 0.5767538276902935,
+      "grad_norm": 0.0,
+      "learning_rate": 8.013297753500756e-06,
+      "loss": 0.8645,
+      "step": 15859
+    },
+    {
+      "epoch": 0.5767901952940321,
+      "grad_norm": 0.0,
+      "learning_rate": 8.012143384639391e-06,
+      "loss": 1.1216,
+      "step": 15860
+    },
+    {
+      "epoch": 0.5768265628977707,
+      "grad_norm": 0.0,
+      "learning_rate": 8.010989043356715e-06,
+      "loss": 1.0089,
+      "step": 15861
+    },
+    {
+      "epoch": 0.5768629305015093,
+      "grad_norm": 0.0,
+      "learning_rate": 8.009834729668744e-06,
+      "loss": 1.0789,
+      "step": 15862
+    },
+    {
+      "epoch": 0.5768992981052479,
+      "grad_norm": 0.0,
+      "learning_rate": 8.00868044359149e-06,
+      "loss": 0.9982,
+      "step": 15863
+    },
+    {
+      "epoch": 0.5769356657089865,
+      "grad_norm": 0.0,
+      "learning_rate": 8.007526185140974e-06,
+      "loss": 0.9227,
+      "step": 15864
+    },
+    {
+      "epoch": 0.5769720333127251,
+      "grad_norm": 0.0,
+      "learning_rate": 8.006371954333201e-06,
+      "loss": 0.9532,
+      "step": 15865
+    },
+    {
+      "epoch": 0.5770084009164637,
+      "grad_norm": 0.0,
+      "learning_rate": 8.005217751184194e-06,
+      "loss": 0.9927,
+      "step": 15866
+    },
+    {
+      "epoch": 0.5770447685202023,
+      "grad_norm": 0.0,
+      "learning_rate": 8.004063575709958e-06,
+      "loss": 1.0939,
+      "step": 15867
+    },
+    {
+      "epoch": 0.5770811361239407,
+      "grad_norm": 0.0,
+      "learning_rate": 8.002909427926505e-06,
+      "loss": 0.8754,
+      "step": 15868
+    },
+    {
+      "epoch": 0.5771175037276793,
+      "grad_norm": 0.0,
+      "learning_rate": 8.001755307849854e-06,
+      "loss": 0.9928,
+      "step": 15869
+    },
+    {
+      "epoch": 0.5771538713314179,
+      "grad_norm": 0.0,
+      "learning_rate": 8.00060121549601e-06,
+      "loss": 1.0243,
+      "step": 15870
+    },
+    {
+      "epoch": 0.5771902389351565,
+      "grad_norm": 0.0,
+      "learning_rate": 7.999447150880992e-06,
+      "loss": 1.115,
+      "step": 15871
+    },
+    {
+      "epoch": 0.5772266065388951,
+      "grad_norm": 0.0,
+      "learning_rate": 7.998293114020803e-06,
+      "loss": 0.9724,
+      "step": 15872
+    },
+    {
+      "epoch": 0.5772629741426337,
+      "grad_norm": 0.0,
+      "learning_rate": 7.997139104931458e-06,
+      "loss": 0.9604,
+      "step": 15873
+    },
+    {
+      "epoch": 0.5772993417463723,
+      "grad_norm": 0.0,
+      "learning_rate": 7.995985123628963e-06,
+      "loss": 1.0366,
+      "step": 15874
+    },
+    {
+      "epoch": 0.5773357093501109,
+      "grad_norm": 0.0,
+      "learning_rate": 7.994831170129335e-06,
+      "loss": 1.0365,
+      "step": 15875
+    },
+    {
+      "epoch": 0.5773720769538495,
+      "grad_norm": 0.0,
+      "learning_rate": 7.993677244448577e-06,
+      "loss": 0.9251,
+      "step": 15876
+    },
+    {
+      "epoch": 0.5774084445575881,
+      "grad_norm": 0.0,
+      "learning_rate": 7.992523346602704e-06,
+      "loss": 0.9692,
+      "step": 15877
+    },
+    {
+      "epoch": 0.5774448121613267,
+      "grad_norm": 0.0,
+      "learning_rate": 7.991369476607716e-06,
+      "loss": 0.9875,
+      "step": 15878
+    },
+    {
+      "epoch": 0.5774811797650653,
+      "grad_norm": 0.0,
+      "learning_rate": 7.990215634479632e-06,
+      "loss": 1.0108,
+      "step": 15879
+    },
+    {
+      "epoch": 0.5775175473688039,
+      "grad_norm": 0.0,
+      "learning_rate": 7.98906182023445e-06,
+      "loss": 0.9845,
+      "step": 15880
+    },
+    {
+      "epoch": 0.5775539149725425,
+      "grad_norm": 0.0,
+      "learning_rate": 7.987908033888186e-06,
+      "loss": 0.9971,
+      "step": 15881
+    },
+    {
+      "epoch": 0.5775902825762811,
+      "grad_norm": 0.0,
+      "learning_rate": 7.986754275456841e-06,
+      "loss": 0.9367,
+      "step": 15882
+    },
+    {
+      "epoch": 0.5776266501800197,
+      "grad_norm": 0.0,
+      "learning_rate": 7.985600544956424e-06,
+      "loss": 1.0615,
+      "step": 15883
+    },
+    {
+      "epoch": 0.5776630177837583,
+      "grad_norm": 0.0,
+      "learning_rate": 7.984446842402942e-06,
+      "loss": 0.8931,
+      "step": 15884
+    },
+    {
+      "epoch": 0.5776993853874968,
+      "grad_norm": 0.0,
+      "learning_rate": 7.983293167812402e-06,
+      "loss": 0.876,
+      "step": 15885
+    },
+    {
+      "epoch": 0.5777357529912354,
+      "grad_norm": 0.0,
+      "learning_rate": 7.982139521200809e-06,
+      "loss": 0.9911,
+      "step": 15886
+    },
+    {
+      "epoch": 0.577772120594974,
+      "grad_norm": 0.0,
+      "learning_rate": 7.980985902584162e-06,
+      "loss": 0.933,
+      "step": 15887
+    },
+    {
+      "epoch": 0.5778084881987126,
+      "grad_norm": 0.0,
+      "learning_rate": 7.979832311978474e-06,
+      "loss": 0.974,
+      "step": 15888
+    },
+    {
+      "epoch": 0.5778448558024512,
+      "grad_norm": 0.0,
+      "learning_rate": 7.978678749399744e-06,
+      "loss": 1.0148,
+      "step": 15889
+    },
+    {
+      "epoch": 0.5778812234061897,
+      "grad_norm": 0.0,
+      "learning_rate": 7.977525214863984e-06,
+      "loss": 0.96,
+      "step": 15890
+    },
+    {
+      "epoch": 0.5779175910099283,
+      "grad_norm": 0.0,
+      "learning_rate": 7.976371708387185e-06,
+      "loss": 1.0392,
+      "step": 15891
+    },
+    {
+      "epoch": 0.5779539586136669,
+      "grad_norm": 0.0,
+      "learning_rate": 7.975218229985365e-06,
+      "loss": 1.0854,
+      "step": 15892
+    },
+    {
+      "epoch": 0.5779903262174055,
+      "grad_norm": 0.0,
+      "learning_rate": 7.974064779674513e-06,
+      "loss": 1.09,
+      "step": 15893
+    },
+    {
+      "epoch": 0.5780266938211441,
+      "grad_norm": 0.0,
+      "learning_rate": 7.972911357470641e-06,
+      "loss": 1.0805,
+      "step": 15894
+    },
+    {
+      "epoch": 0.5780630614248827,
+      "grad_norm": 0.0,
+      "learning_rate": 7.971757963389744e-06,
+      "loss": 1.0029,
+      "step": 15895
+    },
+    {
+      "epoch": 0.5780994290286213,
+      "grad_norm": 0.0,
+      "learning_rate": 7.970604597447832e-06,
+      "loss": 0.9097,
+      "step": 15896
+    },
+    {
+      "epoch": 0.5781357966323599,
+      "grad_norm": 0.0,
+      "learning_rate": 7.9694512596609e-06,
+      "loss": 0.9894,
+      "step": 15897
+    },
+    {
+      "epoch": 0.5781721642360985,
+      "grad_norm": 0.0,
+      "learning_rate": 7.968297950044952e-06,
+      "loss": 1.105,
+      "step": 15898
+    },
+    {
+      "epoch": 0.5782085318398371,
+      "grad_norm": 0.0,
+      "learning_rate": 7.967144668615985e-06,
+      "loss": 0.8088,
+      "step": 15899
+    },
+    {
+      "epoch": 0.5782448994435757,
+      "grad_norm": 0.0,
+      "learning_rate": 7.965991415390004e-06,
+      "loss": 0.9821,
+      "step": 15900
+    },
+    {
+      "epoch": 0.5782812670473143,
+      "grad_norm": 0.0,
+      "learning_rate": 7.964838190383004e-06,
+      "loss": 0.9068,
+      "step": 15901
+    },
+    {
+      "epoch": 0.5783176346510528,
+      "grad_norm": 0.0,
+      "learning_rate": 7.963684993610988e-06,
+      "loss": 1.0553,
+      "step": 15902
+    },
+    {
+      "epoch": 0.5783540022547914,
+      "grad_norm": 0.0,
+      "learning_rate": 7.962531825089952e-06,
+      "loss": 0.8552,
+      "step": 15903
+    },
+    {
+      "epoch": 0.57839036985853,
+      "grad_norm": 0.0,
+      "learning_rate": 7.9613786848359e-06,
+      "loss": 0.9856,
+      "step": 15904
+    },
+    {
+      "epoch": 0.5784267374622686,
+      "grad_norm": 0.0,
+      "learning_rate": 7.960225572864825e-06,
+      "loss": 0.9317,
+      "step": 15905
+    },
+    {
+      "epoch": 0.5784631050660072,
+      "grad_norm": 0.0,
+      "learning_rate": 7.959072489192723e-06,
+      "loss": 0.9986,
+      "step": 15906
+    },
+    {
+      "epoch": 0.5784994726697458,
+      "grad_norm": 0.0,
+      "learning_rate": 7.957919433835597e-06,
+      "loss": 1.0917,
+      "step": 15907
+    },
+    {
+      "epoch": 0.5785358402734844,
+      "grad_norm": 0.0,
+      "learning_rate": 7.95676640680944e-06,
+      "loss": 0.9019,
+      "step": 15908
+    },
+    {
+      "epoch": 0.578572207877223,
+      "grad_norm": 0.0,
+      "learning_rate": 7.955613408130253e-06,
+      "loss": 0.943,
+      "step": 15909
+    },
+    {
+      "epoch": 0.5786085754809616,
+      "grad_norm": 0.0,
+      "learning_rate": 7.954460437814024e-06,
+      "loss": 1.022,
+      "step": 15910
+    },
+    {
+      "epoch": 0.5786449430847002,
+      "grad_norm": 0.0,
+      "learning_rate": 7.953307495876762e-06,
+      "loss": 1.0141,
+      "step": 15911
+    },
+    {
+      "epoch": 0.5786813106884388,
+      "grad_norm": 0.0,
+      "learning_rate": 7.952154582334447e-06,
+      "loss": 1.0121,
+      "step": 15912
+    },
+    {
+      "epoch": 0.5787176782921774,
+      "grad_norm": 0.0,
+      "learning_rate": 7.951001697203085e-06,
+      "loss": 0.9476,
+      "step": 15913
+    },
+    {
+      "epoch": 0.578754045895916,
+      "grad_norm": 0.0,
+      "learning_rate": 7.949848840498667e-06,
+      "loss": 1.135,
+      "step": 15914
+    },
+    {
+      "epoch": 0.5787904134996545,
+      "grad_norm": 0.0,
+      "learning_rate": 7.948696012237188e-06,
+      "loss": 0.9416,
+      "step": 15915
+    },
+    {
+      "epoch": 0.5788267811033931,
+      "grad_norm": 0.0,
+      "learning_rate": 7.947543212434639e-06,
+      "loss": 1.1068,
+      "step": 15916
+    },
+    {
+      "epoch": 0.5788631487071317,
+      "grad_norm": 0.0,
+      "learning_rate": 7.946390441107022e-06,
+      "loss": 0.931,
+      "step": 15917
+    },
+    {
+      "epoch": 0.5788995163108702,
+      "grad_norm": 0.0,
+      "learning_rate": 7.945237698270317e-06,
+      "loss": 0.9095,
+      "step": 15918
+    },
+    {
+      "epoch": 0.5789358839146088,
+      "grad_norm": 0.0,
+      "learning_rate": 7.944084983940528e-06,
+      "loss": 0.891,
+      "step": 15919
+    },
+    {
+      "epoch": 0.5789722515183474,
+      "grad_norm": 0.0,
+      "learning_rate": 7.94293229813364e-06,
+      "loss": 0.9695,
+      "step": 15920
+    },
+    {
+      "epoch": 0.579008619122086,
+      "grad_norm": 0.0,
+      "learning_rate": 7.94177964086565e-06,
+      "loss": 0.8884,
+      "step": 15921
+    },
+    {
+      "epoch": 0.5790449867258246,
+      "grad_norm": 0.0,
+      "learning_rate": 7.940627012152545e-06,
+      "loss": 0.9414,
+      "step": 15922
+    },
+    {
+      "epoch": 0.5790813543295632,
+      "grad_norm": 0.0,
+      "learning_rate": 7.93947441201032e-06,
+      "loss": 1.0313,
+      "step": 15923
+    },
+    {
+      "epoch": 0.5791177219333018,
+      "grad_norm": 0.0,
+      "learning_rate": 7.938321840454968e-06,
+      "loss": 0.9623,
+      "step": 15924
+    },
+    {
+      "epoch": 0.5791540895370404,
+      "grad_norm": 0.0,
+      "learning_rate": 7.93716929750247e-06,
+      "loss": 0.906,
+      "step": 15925
+    },
+    {
+      "epoch": 0.579190457140779,
+      "grad_norm": 0.0,
+      "learning_rate": 7.936016783168822e-06,
+      "loss": 1.0121,
+      "step": 15926
+    },
+    {
+      "epoch": 0.5792268247445176,
+      "grad_norm": 0.0,
+      "learning_rate": 7.934864297470011e-06,
+      "loss": 1.0228,
+      "step": 15927
+    },
+    {
+      "epoch": 0.5792631923482562,
+      "grad_norm": 0.0,
+      "learning_rate": 7.933711840422031e-06,
+      "loss": 1.0123,
+      "step": 15928
+    },
+    {
+      "epoch": 0.5792995599519948,
+      "grad_norm": 0.0,
+      "learning_rate": 7.932559412040865e-06,
+      "loss": 1.0134,
+      "step": 15929
+    },
+    {
+      "epoch": 0.5793359275557334,
+      "grad_norm": 0.0,
+      "learning_rate": 7.93140701234251e-06,
+      "loss": 1.0805,
+      "step": 15930
+    },
+    {
+      "epoch": 0.579372295159472,
+      "grad_norm": 0.0,
+      "learning_rate": 7.930254641342941e-06,
+      "loss": 1.1027,
+      "step": 15931
+    },
+    {
+      "epoch": 0.5794086627632106,
+      "grad_norm": 0.0,
+      "learning_rate": 7.929102299058154e-06,
+      "loss": 0.9059,
+      "step": 15932
+    },
+    {
+      "epoch": 0.5794450303669492,
+      "grad_norm": 0.0,
+      "learning_rate": 7.927949985504133e-06,
+      "loss": 0.869,
+      "step": 15933
+    },
+    {
+      "epoch": 0.5794813979706878,
+      "grad_norm": 0.0,
+      "learning_rate": 7.926797700696869e-06,
+      "loss": 0.9439,
+      "step": 15934
+    },
+    {
+      "epoch": 0.5795177655744262,
+      "grad_norm": 0.0,
+      "learning_rate": 7.925645444652342e-06,
+      "loss": 0.9979,
+      "step": 15935
+    },
+    {
+      "epoch": 0.5795541331781648,
+      "grad_norm": 0.0,
+      "learning_rate": 7.924493217386545e-06,
+      "loss": 0.977,
+      "step": 15936
+    },
+    {
+      "epoch": 0.5795905007819034,
+      "grad_norm": 0.0,
+      "learning_rate": 7.923341018915457e-06,
+      "loss": 0.99,
+      "step": 15937
+    },
+    {
+      "epoch": 0.579626868385642,
+      "grad_norm": 0.0,
+      "learning_rate": 7.922188849255071e-06,
+      "loss": 0.9863,
+      "step": 15938
+    },
+    {
+      "epoch": 0.5796632359893806,
+      "grad_norm": 0.0,
+      "learning_rate": 7.92103670842136e-06,
+      "loss": 0.9695,
+      "step": 15939
+    },
+    {
+      "epoch": 0.5796996035931192,
+      "grad_norm": 0.0,
+      "learning_rate": 7.91988459643032e-06,
+      "loss": 0.9772,
+      "step": 15940
+    },
+    {
+      "epoch": 0.5797359711968578,
+      "grad_norm": 0.0,
+      "learning_rate": 7.918732513297926e-06,
+      "loss": 0.9875,
+      "step": 15941
+    },
+    {
+      "epoch": 0.5797723388005964,
+      "grad_norm": 0.0,
+      "learning_rate": 7.917580459040169e-06,
+      "loss": 0.9483,
+      "step": 15942
+    },
+    {
+      "epoch": 0.579808706404335,
+      "grad_norm": 0.0,
+      "learning_rate": 7.91642843367303e-06,
+      "loss": 0.9195,
+      "step": 15943
+    },
+    {
+      "epoch": 0.5798450740080736,
+      "grad_norm": 0.0,
+      "learning_rate": 7.915276437212484e-06,
+      "loss": 1.0024,
+      "step": 15944
+    },
+    {
+      "epoch": 0.5798814416118122,
+      "grad_norm": 0.0,
+      "learning_rate": 7.914124469674524e-06,
+      "loss": 0.9132,
+      "step": 15945
+    },
+    {
+      "epoch": 0.5799178092155508,
+      "grad_norm": 0.0,
+      "learning_rate": 7.912972531075125e-06,
+      "loss": 1.06,
+      "step": 15946
+    },
+    {
+      "epoch": 0.5799541768192894,
+      "grad_norm": 0.0,
+      "learning_rate": 7.911820621430273e-06,
+      "loss": 0.9891,
+      "step": 15947
+    },
+    {
+      "epoch": 0.579990544423028,
+      "grad_norm": 0.0,
+      "learning_rate": 7.910668740755945e-06,
+      "loss": 0.8829,
+      "step": 15948
+    },
+    {
+      "epoch": 0.5800269120267666,
+      "grad_norm": 0.0,
+      "learning_rate": 7.909516889068128e-06,
+      "loss": 0.952,
+      "step": 15949
+    },
+    {
+      "epoch": 0.5800632796305052,
+      "grad_norm": 0.0,
+      "learning_rate": 7.908365066382793e-06,
+      "loss": 1.003,
+      "step": 15950
+    },
+    {
+      "epoch": 0.5800996472342438,
+      "grad_norm": 0.0,
+      "learning_rate": 7.90721327271593e-06,
+      "loss": 0.9969,
+      "step": 15951
+    },
+    {
+      "epoch": 0.5801360148379823,
+      "grad_norm": 0.0,
+      "learning_rate": 7.906061508083508e-06,
+      "loss": 1.0497,
+      "step": 15952
+    },
+    {
+      "epoch": 0.5801723824417209,
+      "grad_norm": 0.0,
+      "learning_rate": 7.904909772501516e-06,
+      "loss": 1.0047,
+      "step": 15953
+    },
+    {
+      "epoch": 0.5802087500454595,
+      "grad_norm": 0.0,
+      "learning_rate": 7.903758065985927e-06,
+      "loss": 0.8761,
+      "step": 15954
+    },
+    {
+      "epoch": 0.5802451176491981,
+      "grad_norm": 0.0,
+      "learning_rate": 7.902606388552721e-06,
+      "loss": 1.0397,
+      "step": 15955
+    },
+    {
+      "epoch": 0.5802814852529367,
+      "grad_norm": 0.0,
+      "learning_rate": 7.901454740217874e-06,
+      "loss": 1.0275,
+      "step": 15956
+    },
+    {
+      "epoch": 0.5803178528566753,
+      "grad_norm": 0.0,
+      "learning_rate": 7.90030312099737e-06,
+      "loss": 1.0164,
+      "step": 15957
+    },
+    {
+      "epoch": 0.5803542204604138,
+      "grad_norm": 0.0,
+      "learning_rate": 7.899151530907175e-06,
+      "loss": 0.9892,
+      "step": 15958
+    },
+    {
+      "epoch": 0.5803905880641524,
+      "grad_norm": 0.0,
+      "learning_rate": 7.897999969963276e-06,
+      "loss": 1.0195,
+      "step": 15959
+    },
+    {
+      "epoch": 0.580426955667891,
+      "grad_norm": 0.0,
+      "learning_rate": 7.896848438181641e-06,
+      "loss": 0.9883,
+      "step": 15960
+    },
+    {
+      "epoch": 0.5804633232716296,
+      "grad_norm": 0.0,
+      "learning_rate": 7.895696935578255e-06,
+      "loss": 0.9987,
+      "step": 15961
+    },
+    {
+      "epoch": 0.5804996908753682,
+      "grad_norm": 0.0,
+      "learning_rate": 7.894545462169083e-06,
+      "loss": 0.8891,
+      "step": 15962
+    },
+    {
+      "epoch": 0.5805360584791068,
+      "grad_norm": 0.0,
+      "learning_rate": 7.893394017970109e-06,
+      "loss": 1.0565,
+      "step": 15963
+    },
+    {
+      "epoch": 0.5805724260828454,
+      "grad_norm": 0.0,
+      "learning_rate": 7.892242602997306e-06,
+      "loss": 0.9236,
+      "step": 15964
+    },
+    {
+      "epoch": 0.580608793686584,
+      "grad_norm": 0.0,
+      "learning_rate": 7.891091217266641e-06,
+      "loss": 0.9556,
+      "step": 15965
+    },
+    {
+      "epoch": 0.5806451612903226,
+      "grad_norm": 0.0,
+      "learning_rate": 7.889939860794097e-06,
+      "loss": 1.0187,
+      "step": 15966
+    },
+    {
+      "epoch": 0.5806815288940612,
+      "grad_norm": 0.0,
+      "learning_rate": 7.88878853359564e-06,
+      "loss": 1.0939,
+      "step": 15967
+    },
+    {
+      "epoch": 0.5807178964977998,
+      "grad_norm": 0.0,
+      "learning_rate": 7.887637235687251e-06,
+      "loss": 0.8716,
+      "step": 15968
+    },
+    {
+      "epoch": 0.5807542641015383,
+      "grad_norm": 0.0,
+      "learning_rate": 7.886485967084896e-06,
+      "loss": 1.0711,
+      "step": 15969
+    },
+    {
+      "epoch": 0.5807906317052769,
+      "grad_norm": 0.0,
+      "learning_rate": 7.885334727804553e-06,
+      "loss": 0.9828,
+      "step": 15970
+    },
+    {
+      "epoch": 0.5808269993090155,
+      "grad_norm": 0.0,
+      "learning_rate": 7.884183517862186e-06,
+      "loss": 1.0442,
+      "step": 15971
+    },
+    {
+      "epoch": 0.5808633669127541,
+      "grad_norm": 0.0,
+      "learning_rate": 7.883032337273773e-06,
+      "loss": 1.0061,
+      "step": 15972
+    },
+    {
+      "epoch": 0.5808997345164927,
+      "grad_norm": 0.0,
+      "learning_rate": 7.88188118605528e-06,
+      "loss": 1.0677,
+      "step": 15973
+    },
+    {
+      "epoch": 0.5809361021202313,
+      "grad_norm": 0.0,
+      "learning_rate": 7.880730064222683e-06,
+      "loss": 1.0298,
+      "step": 15974
+    },
+    {
+      "epoch": 0.5809724697239699,
+      "grad_norm": 0.0,
+      "learning_rate": 7.879578971791949e-06,
+      "loss": 1.0961,
+      "step": 15975
+    },
+    {
+      "epoch": 0.5810088373277085,
+      "grad_norm": 0.0,
+      "learning_rate": 7.87842790877905e-06,
+      "loss": 0.931,
+      "step": 15976
+    },
+    {
+      "epoch": 0.5810452049314471,
+      "grad_norm": 0.0,
+      "learning_rate": 7.877276875199948e-06,
+      "loss": 1.069,
+      "step": 15977
+    },
+    {
+      "epoch": 0.5810815725351857,
+      "grad_norm": 0.0,
+      "learning_rate": 7.876125871070623e-06,
+      "loss": 0.9728,
+      "step": 15978
+    },
+    {
+      "epoch": 0.5811179401389243,
+      "grad_norm": 0.0,
+      "learning_rate": 7.874974896407035e-06,
+      "loss": 0.9627,
+      "step": 15979
+    },
+    {
+      "epoch": 0.5811543077426629,
+      "grad_norm": 0.0,
+      "learning_rate": 7.873823951225158e-06,
+      "loss": 0.9758,
+      "step": 15980
+    },
+    {
+      "epoch": 0.5811906753464015,
+      "grad_norm": 0.0,
+      "learning_rate": 7.872673035540955e-06,
+      "loss": 0.9642,
+      "step": 15981
+    },
+    {
+      "epoch": 0.58122704295014,
+      "grad_norm": 0.0,
+      "learning_rate": 7.871522149370397e-06,
+      "loss": 1.0811,
+      "step": 15982
+    },
+    {
+      "epoch": 0.5812634105538786,
+      "grad_norm": 0.0,
+      "learning_rate": 7.87037129272945e-06,
+      "loss": 0.9784,
+      "step": 15983
+    },
+    {
+      "epoch": 0.5812997781576172,
+      "grad_norm": 0.0,
+      "learning_rate": 7.869220465634076e-06,
+      "loss": 0.9737,
+      "step": 15984
+    },
+    {
+      "epoch": 0.5813361457613558,
+      "grad_norm": 0.0,
+      "learning_rate": 7.86806966810025e-06,
+      "loss": 1.034,
+      "step": 15985
+    },
+    {
+      "epoch": 0.5813725133650943,
+      "grad_norm": 0.0,
+      "learning_rate": 7.866918900143926e-06,
+      "loss": 0.9953,
+      "step": 15986
+    },
+    {
+      "epoch": 0.5814088809688329,
+      "grad_norm": 0.0,
+      "learning_rate": 7.86576816178108e-06,
+      "loss": 0.9185,
+      "step": 15987
+    },
+    {
+      "epoch": 0.5814452485725715,
+      "grad_norm": 0.0,
+      "learning_rate": 7.864617453027671e-06,
+      "loss": 1.0401,
+      "step": 15988
+    },
+    {
+      "epoch": 0.5814816161763101,
+      "grad_norm": 0.0,
+      "learning_rate": 7.863466773899669e-06,
+      "loss": 0.967,
+      "step": 15989
+    },
+    {
+      "epoch": 0.5815179837800487,
+      "grad_norm": 0.0,
+      "learning_rate": 7.862316124413027e-06,
+      "loss": 0.9521,
+      "step": 15990
+    },
+    {
+      "epoch": 0.5815543513837873,
+      "grad_norm": 0.0,
+      "learning_rate": 7.861165504583723e-06,
+      "loss": 0.8584,
+      "step": 15991
+    },
+    {
+      "epoch": 0.5815907189875259,
+      "grad_norm": 0.0,
+      "learning_rate": 7.860014914427709e-06,
+      "loss": 0.9923,
+      "step": 15992
+    },
+    {
+      "epoch": 0.5816270865912645,
+      "grad_norm": 0.0,
+      "learning_rate": 7.858864353960954e-06,
+      "loss": 0.8706,
+      "step": 15993
+    },
+    {
+      "epoch": 0.5816634541950031,
+      "grad_norm": 0.0,
+      "learning_rate": 7.857713823199415e-06,
+      "loss": 0.9396,
+      "step": 15994
+    },
+    {
+      "epoch": 0.5816998217987417,
+      "grad_norm": 0.0,
+      "learning_rate": 7.85656332215906e-06,
+      "loss": 0.9615,
+      "step": 15995
+    },
+    {
+      "epoch": 0.5817361894024803,
+      "grad_norm": 0.0,
+      "learning_rate": 7.855412850855846e-06,
+      "loss": 0.923,
+      "step": 15996
+    },
+    {
+      "epoch": 0.5817725570062189,
+      "grad_norm": 0.0,
+      "learning_rate": 7.854262409305739e-06,
+      "loss": 0.9421,
+      "step": 15997
+    },
+    {
+      "epoch": 0.5818089246099575,
+      "grad_norm": 0.0,
+      "learning_rate": 7.853111997524694e-06,
+      "loss": 0.8771,
+      "step": 15998
+    },
+    {
+      "epoch": 0.5818452922136961,
+      "grad_norm": 0.0,
+      "learning_rate": 7.851961615528675e-06,
+      "loss": 1.0718,
+      "step": 15999
+    },
+    {
+      "epoch": 0.5818816598174347,
+      "grad_norm": 0.0,
+      "learning_rate": 7.85081126333364e-06,
+      "loss": 0.9738,
+      "step": 16000
+    },
+    {
+      "epoch": 0.5819180274211733,
+      "grad_norm": 0.0,
+      "learning_rate": 7.84966094095555e-06,
+      "loss": 0.9345,
+      "step": 16001
+    },
+    {
+      "epoch": 0.5819543950249119,
+      "grad_norm": 0.0,
+      "learning_rate": 7.848510648410367e-06,
+      "loss": 0.9648,
+      "step": 16002
+    },
+    {
+      "epoch": 0.5819907626286503,
+      "grad_norm": 0.0,
+      "learning_rate": 7.847360385714041e-06,
+      "loss": 0.838,
+      "step": 16003
+    },
+    {
+      "epoch": 0.582027130232389,
+      "grad_norm": 0.0,
+      "learning_rate": 7.846210152882539e-06,
+      "loss": 0.9958,
+      "step": 16004
+    },
+    {
+      "epoch": 0.5820634978361275,
+      "grad_norm": 0.0,
+      "learning_rate": 7.84505994993181e-06,
+      "loss": 0.8913,
+      "step": 16005
+    },
+    {
+      "epoch": 0.5820998654398661,
+      "grad_norm": 0.0,
+      "learning_rate": 7.843909776877822e-06,
+      "loss": 0.9426,
+      "step": 16006
+    },
+    {
+      "epoch": 0.5821362330436047,
+      "grad_norm": 0.0,
+      "learning_rate": 7.842759633736522e-06,
+      "loss": 1.0263,
+      "step": 16007
+    },
+    {
+      "epoch": 0.5821726006473433,
+      "grad_norm": 0.0,
+      "learning_rate": 7.841609520523877e-06,
+      "loss": 1.0166,
+      "step": 16008
+    },
+    {
+      "epoch": 0.5822089682510819,
+      "grad_norm": 0.0,
+      "learning_rate": 7.840459437255833e-06,
+      "loss": 0.9985,
+      "step": 16009
+    },
+    {
+      "epoch": 0.5822453358548205,
+      "grad_norm": 0.0,
+      "learning_rate": 7.839309383948355e-06,
+      "loss": 0.9718,
+      "step": 16010
+    },
+    {
+      "epoch": 0.5822817034585591,
+      "grad_norm": 0.0,
+      "learning_rate": 7.83815936061739e-06,
+      "loss": 0.972,
+      "step": 16011
+    },
+    {
+      "epoch": 0.5823180710622977,
+      "grad_norm": 0.0,
+      "learning_rate": 7.837009367278898e-06,
+      "loss": 1.0385,
+      "step": 16012
+    },
+    {
+      "epoch": 0.5823544386660363,
+      "grad_norm": 0.0,
+      "learning_rate": 7.83585940394883e-06,
+      "loss": 0.9618,
+      "step": 16013
+    },
+    {
+      "epoch": 0.5823908062697749,
+      "grad_norm": 0.0,
+      "learning_rate": 7.834709470643145e-06,
+      "loss": 0.9863,
+      "step": 16014
+    },
+    {
+      "epoch": 0.5824271738735135,
+      "grad_norm": 0.0,
+      "learning_rate": 7.833559567377791e-06,
+      "loss": 1.0043,
+      "step": 16015
+    },
+    {
+      "epoch": 0.5824635414772521,
+      "grad_norm": 0.0,
+      "learning_rate": 7.83240969416873e-06,
+      "loss": 0.8809,
+      "step": 16016
+    },
+    {
+      "epoch": 0.5824999090809907,
+      "grad_norm": 0.0,
+      "learning_rate": 7.8312598510319e-06,
+      "loss": 0.9769,
+      "step": 16017
+    },
+    {
+      "epoch": 0.5825362766847293,
+      "grad_norm": 0.0,
+      "learning_rate": 7.83011003798327e-06,
+      "loss": 0.9142,
+      "step": 16018
+    },
+    {
+      "epoch": 0.5825726442884678,
+      "grad_norm": 0.0,
+      "learning_rate": 7.828960255038783e-06,
+      "loss": 1.0299,
+      "step": 16019
+    },
+    {
+      "epoch": 0.5826090118922064,
+      "grad_norm": 0.0,
+      "learning_rate": 7.82781050221439e-06,
+      "loss": 0.9559,
+      "step": 16020
+    },
+    {
+      "epoch": 0.582645379495945,
+      "grad_norm": 0.0,
+      "learning_rate": 7.826660779526049e-06,
+      "loss": 0.9259,
+      "step": 16021
+    },
+    {
+      "epoch": 0.5826817470996836,
+      "grad_norm": 0.0,
+      "learning_rate": 7.8255110869897e-06,
+      "loss": 0.9114,
+      "step": 16022
+    },
+    {
+      "epoch": 0.5827181147034222,
+      "grad_norm": 0.0,
+      "learning_rate": 7.824361424621305e-06,
+      "loss": 0.9712,
+      "step": 16023
+    },
+    {
+      "epoch": 0.5827544823071608,
+      "grad_norm": 0.0,
+      "learning_rate": 7.823211792436805e-06,
+      "loss": 0.9739,
+      "step": 16024
+    },
+    {
+      "epoch": 0.5827908499108994,
+      "grad_norm": 0.0,
+      "learning_rate": 7.822062190452153e-06,
+      "loss": 0.9878,
+      "step": 16025
+    },
+    {
+      "epoch": 0.582827217514638,
+      "grad_norm": 0.0,
+      "learning_rate": 7.820912618683298e-06,
+      "loss": 1.0527,
+      "step": 16026
+    },
+    {
+      "epoch": 0.5828635851183765,
+      "grad_norm": 0.0,
+      "learning_rate": 7.81976307714619e-06,
+      "loss": 0.9781,
+      "step": 16027
+    },
+    {
+      "epoch": 0.5828999527221151,
+      "grad_norm": 0.0,
+      "learning_rate": 7.818613565856775e-06,
+      "loss": 0.9502,
+      "step": 16028
+    },
+    {
+      "epoch": 0.5829363203258537,
+      "grad_norm": 0.0,
+      "learning_rate": 7.817464084831003e-06,
+      "loss": 0.8967,
+      "step": 16029
+    },
+    {
+      "epoch": 0.5829726879295923,
+      "grad_norm": 0.0,
+      "learning_rate": 7.816314634084819e-06,
+      "loss": 0.9797,
+      "step": 16030
+    },
+    {
+      "epoch": 0.5830090555333309,
+      "grad_norm": 0.0,
+      "learning_rate": 7.815165213634171e-06,
+      "loss": 0.9825,
+      "step": 16031
+    },
+    {
+      "epoch": 0.5830454231370695,
+      "grad_norm": 0.0,
+      "learning_rate": 7.814015823495006e-06,
+      "loss": 1.0472,
+      "step": 16032
+    },
+    {
+      "epoch": 0.5830817907408081,
+      "grad_norm": 0.0,
+      "learning_rate": 7.81286646368327e-06,
+      "loss": 0.9711,
+      "step": 16033
+    },
+    {
+      "epoch": 0.5831181583445467,
+      "grad_norm": 0.0,
+      "learning_rate": 7.811717134214908e-06,
+      "loss": 0.9274,
+      "step": 16034
+    },
+    {
+      "epoch": 0.5831545259482853,
+      "grad_norm": 0.0,
+      "learning_rate": 7.810567835105867e-06,
+      "loss": 1.0376,
+      "step": 16035
+    },
+    {
+      "epoch": 0.5831908935520238,
+      "grad_norm": 0.0,
+      "learning_rate": 7.80941856637209e-06,
+      "loss": 0.9884,
+      "step": 16036
+    },
+    {
+      "epoch": 0.5832272611557624,
+      "grad_norm": 0.0,
+      "learning_rate": 7.808269328029526e-06,
+      "loss": 0.8768,
+      "step": 16037
+    },
+    {
+      "epoch": 0.583263628759501,
+      "grad_norm": 0.0,
+      "learning_rate": 7.807120120094112e-06,
+      "loss": 0.9175,
+      "step": 16038
+    },
+    {
+      "epoch": 0.5832999963632396,
+      "grad_norm": 0.0,
+      "learning_rate": 7.805970942581795e-06,
+      "loss": 0.7984,
+      "step": 16039
+    },
+    {
+      "epoch": 0.5833363639669782,
+      "grad_norm": 0.0,
+      "learning_rate": 7.80482179550852e-06,
+      "loss": 1.0304,
+      "step": 16040
+    },
+    {
+      "epoch": 0.5833727315707168,
+      "grad_norm": 0.0,
+      "learning_rate": 7.803672678890225e-06,
+      "loss": 1.0326,
+      "step": 16041
+    },
+    {
+      "epoch": 0.5834090991744554,
+      "grad_norm": 0.0,
+      "learning_rate": 7.802523592742862e-06,
+      "loss": 0.9535,
+      "step": 16042
+    },
+    {
+      "epoch": 0.583445466778194,
+      "grad_norm": 0.0,
+      "learning_rate": 7.801374537082359e-06,
+      "loss": 0.8889,
+      "step": 16043
+    },
+    {
+      "epoch": 0.5834818343819326,
+      "grad_norm": 0.0,
+      "learning_rate": 7.800225511924669e-06,
+      "loss": 1.1599,
+      "step": 16044
+    },
+    {
+      "epoch": 0.5835182019856712,
+      "grad_norm": 0.0,
+      "learning_rate": 7.799076517285725e-06,
+      "loss": 0.9009,
+      "step": 16045
+    },
+    {
+      "epoch": 0.5835545695894098,
+      "grad_norm": 0.0,
+      "learning_rate": 7.797927553181474e-06,
+      "loss": 0.863,
+      "step": 16046
+    },
+    {
+      "epoch": 0.5835909371931484,
+      "grad_norm": 0.0,
+      "learning_rate": 7.796778619627853e-06,
+      "loss": 1.0684,
+      "step": 16047
+    },
+    {
+      "epoch": 0.583627304796887,
+      "grad_norm": 0.0,
+      "learning_rate": 7.795629716640807e-06,
+      "loss": 1.0577,
+      "step": 16048
+    },
+    {
+      "epoch": 0.5836636724006256,
+      "grad_norm": 0.0,
+      "learning_rate": 7.794480844236265e-06,
+      "loss": 1.0275,
+      "step": 16049
+    },
+    {
+      "epoch": 0.5837000400043642,
+      "grad_norm": 0.0,
+      "learning_rate": 7.793332002430178e-06,
+      "loss": 0.9395,
+      "step": 16050
+    },
+    {
+      "epoch": 0.5837364076081027,
+      "grad_norm": 0.0,
+      "learning_rate": 7.792183191238472e-06,
+      "loss": 0.9983,
+      "step": 16051
+    },
+    {
+      "epoch": 0.5837727752118413,
+      "grad_norm": 0.0,
+      "learning_rate": 7.791034410677096e-06,
+      "loss": 1.0528,
+      "step": 16052
+    },
+    {
+      "epoch": 0.5838091428155798,
+      "grad_norm": 0.0,
+      "learning_rate": 7.789885660761981e-06,
+      "loss": 1.0212,
+      "step": 16053
+    },
+    {
+      "epoch": 0.5838455104193184,
+      "grad_norm": 0.0,
+      "learning_rate": 7.78873694150907e-06,
+      "loss": 1.0445,
+      "step": 16054
+    },
+    {
+      "epoch": 0.583881878023057,
+      "grad_norm": 0.0,
+      "learning_rate": 7.787588252934293e-06,
+      "loss": 0.9464,
+      "step": 16055
+    },
+    {
+      "epoch": 0.5839182456267956,
+      "grad_norm": 0.0,
+      "learning_rate": 7.786439595053595e-06,
+      "loss": 0.8258,
+      "step": 16056
+    },
+    {
+      "epoch": 0.5839546132305342,
+      "grad_norm": 0.0,
+      "learning_rate": 7.7852909678829e-06,
+      "loss": 0.9858,
+      "step": 16057
+    },
+    {
+      "epoch": 0.5839909808342728,
+      "grad_norm": 0.0,
+      "learning_rate": 7.784142371438158e-06,
+      "loss": 0.9354,
+      "step": 16058
+    },
+    {
+      "epoch": 0.5840273484380114,
+      "grad_norm": 0.0,
+      "learning_rate": 7.782993805735294e-06,
+      "loss": 0.9467,
+      "step": 16059
+    },
+    {
+      "epoch": 0.58406371604175,
+      "grad_norm": 0.0,
+      "learning_rate": 7.781845270790242e-06,
+      "loss": 0.91,
+      "step": 16060
+    },
+    {
+      "epoch": 0.5841000836454886,
+      "grad_norm": 0.0,
+      "learning_rate": 7.780696766618947e-06,
+      "loss": 0.8144,
+      "step": 16061
+    },
+    {
+      "epoch": 0.5841364512492272,
+      "grad_norm": 0.0,
+      "learning_rate": 7.779548293237326e-06,
+      "loss": 1.0739,
+      "step": 16062
+    },
+    {
+      "epoch": 0.5841728188529658,
+      "grad_norm": 0.0,
+      "learning_rate": 7.77839985066133e-06,
+      "loss": 0.8304,
+      "step": 16063
+    },
+    {
+      "epoch": 0.5842091864567044,
+      "grad_norm": 0.0,
+      "learning_rate": 7.77725143890688e-06,
+      "loss": 1.0213,
+      "step": 16064
+    },
+    {
+      "epoch": 0.584245554060443,
+      "grad_norm": 0.0,
+      "learning_rate": 7.776103057989913e-06,
+      "loss": 0.9028,
+      "step": 16065
+    },
+    {
+      "epoch": 0.5842819216641816,
+      "grad_norm": 0.0,
+      "learning_rate": 7.774954707926361e-06,
+      "loss": 0.9138,
+      "step": 16066
+    },
+    {
+      "epoch": 0.5843182892679202,
+      "grad_norm": 0.0,
+      "learning_rate": 7.773806388732156e-06,
+      "loss": 0.9798,
+      "step": 16067
+    },
+    {
+      "epoch": 0.5843546568716588,
+      "grad_norm": 0.0,
+      "learning_rate": 7.772658100423228e-06,
+      "loss": 0.8724,
+      "step": 16068
+    },
+    {
+      "epoch": 0.5843910244753974,
+      "grad_norm": 0.0,
+      "learning_rate": 7.77150984301551e-06,
+      "loss": 1.0166,
+      "step": 16069
+    },
+    {
+      "epoch": 0.5844273920791359,
+      "grad_norm": 0.0,
+      "learning_rate": 7.77036161652493e-06,
+      "loss": 1.01,
+      "step": 16070
+    },
+    {
+      "epoch": 0.5844637596828745,
+      "grad_norm": 0.0,
+      "learning_rate": 7.76921342096742e-06,
+      "loss": 0.9321,
+      "step": 16071
+    },
+    {
+      "epoch": 0.584500127286613,
+      "grad_norm": 0.0,
+      "learning_rate": 7.768065256358908e-06,
+      "loss": 0.868,
+      "step": 16072
+    },
+    {
+      "epoch": 0.5845364948903516,
+      "grad_norm": 0.0,
+      "learning_rate": 7.766917122715325e-06,
+      "loss": 0.9566,
+      "step": 16073
+    },
+    {
+      "epoch": 0.5845728624940902,
+      "grad_norm": 0.0,
+      "learning_rate": 7.765769020052595e-06,
+      "loss": 1.0282,
+      "step": 16074
+    },
+    {
+      "epoch": 0.5846092300978288,
+      "grad_norm": 0.0,
+      "learning_rate": 7.764620948386655e-06,
+      "loss": 0.9625,
+      "step": 16075
+    },
+    {
+      "epoch": 0.5846455977015674,
+      "grad_norm": 0.0,
+      "learning_rate": 7.763472907733423e-06,
+      "loss": 0.9372,
+      "step": 16076
+    },
+    {
+      "epoch": 0.584681965305306,
+      "grad_norm": 0.0,
+      "learning_rate": 7.762324898108835e-06,
+      "loss": 0.9313,
+      "step": 16077
+    },
+    {
+      "epoch": 0.5847183329090446,
+      "grad_norm": 0.0,
+      "learning_rate": 7.761176919528814e-06,
+      "loss": 0.9067,
+      "step": 16078
+    },
+    {
+      "epoch": 0.5847547005127832,
+      "grad_norm": 0.0,
+      "learning_rate": 7.760028972009284e-06,
+      "loss": 0.902,
+      "step": 16079
+    },
+    {
+      "epoch": 0.5847910681165218,
+      "grad_norm": 0.0,
+      "learning_rate": 7.758881055566177e-06,
+      "loss": 0.8435,
+      "step": 16080
+    },
+    {
+      "epoch": 0.5848274357202604,
+      "grad_norm": 0.0,
+      "learning_rate": 7.757733170215412e-06,
+      "loss": 0.9034,
+      "step": 16081
+    },
+    {
+      "epoch": 0.584863803323999,
+      "grad_norm": 0.0,
+      "learning_rate": 7.756585315972922e-06,
+      "loss": 0.9518,
+      "step": 16082
+    },
+    {
+      "epoch": 0.5849001709277376,
+      "grad_norm": 0.0,
+      "learning_rate": 7.755437492854624e-06,
+      "loss": 1.0097,
+      "step": 16083
+    },
+    {
+      "epoch": 0.5849365385314762,
+      "grad_norm": 0.0,
+      "learning_rate": 7.75428970087645e-06,
+      "loss": 0.8618,
+      "step": 16084
+    },
+    {
+      "epoch": 0.5849729061352148,
+      "grad_norm": 0.0,
+      "learning_rate": 7.753141940054316e-06,
+      "loss": 1.0948,
+      "step": 16085
+    },
+    {
+      "epoch": 0.5850092737389534,
+      "grad_norm": 0.0,
+      "learning_rate": 7.751994210404153e-06,
+      "loss": 1.1211,
+      "step": 16086
+    },
+    {
+      "epoch": 0.5850456413426919,
+      "grad_norm": 0.0,
+      "learning_rate": 7.750846511941878e-06,
+      "loss": 1.067,
+      "step": 16087
+    },
+    {
+      "epoch": 0.5850820089464305,
+      "grad_norm": 0.0,
+      "learning_rate": 7.74969884468342e-06,
+      "loss": 1.0541,
+      "step": 16088
+    },
+    {
+      "epoch": 0.5851183765501691,
+      "grad_norm": 0.0,
+      "learning_rate": 7.748551208644693e-06,
+      "loss": 0.9027,
+      "step": 16089
+    },
+    {
+      "epoch": 0.5851547441539077,
+      "grad_norm": 0.0,
+      "learning_rate": 7.747403603841629e-06,
+      "loss": 1.017,
+      "step": 16090
+    },
+    {
+      "epoch": 0.5851911117576463,
+      "grad_norm": 0.0,
+      "learning_rate": 7.746256030290139e-06,
+      "loss": 1.1438,
+      "step": 16091
+    },
+    {
+      "epoch": 0.5852274793613849,
+      "grad_norm": 0.0,
+      "learning_rate": 7.745108488006152e-06,
+      "loss": 0.9458,
+      "step": 16092
+    },
+    {
+      "epoch": 0.5852638469651235,
+      "grad_norm": 0.0,
+      "learning_rate": 7.743960977005584e-06,
+      "loss": 1.0293,
+      "step": 16093
+    },
+    {
+      "epoch": 0.585300214568862,
+      "grad_norm": 0.0,
+      "learning_rate": 7.742813497304358e-06,
+      "loss": 0.8753,
+      "step": 16094
+    },
+    {
+      "epoch": 0.5853365821726007,
+      "grad_norm": 0.0,
+      "learning_rate": 7.74166604891839e-06,
+      "loss": 1.1365,
+      "step": 16095
+    },
+    {
+      "epoch": 0.5853729497763392,
+      "grad_norm": 0.0,
+      "learning_rate": 7.740518631863606e-06,
+      "loss": 1.0084,
+      "step": 16096
+    },
+    {
+      "epoch": 0.5854093173800778,
+      "grad_norm": 0.0,
+      "learning_rate": 7.739371246155918e-06,
+      "loss": 0.9813,
+      "step": 16097
+    },
+    {
+      "epoch": 0.5854456849838164,
+      "grad_norm": 0.0,
+      "learning_rate": 7.738223891811244e-06,
+      "loss": 1.1293,
+      "step": 16098
+    },
+    {
+      "epoch": 0.585482052587555,
+      "grad_norm": 0.0,
+      "learning_rate": 7.737076568845506e-06,
+      "loss": 0.888,
+      "step": 16099
+    },
+    {
+      "epoch": 0.5855184201912936,
+      "grad_norm": 0.0,
+      "learning_rate": 7.735929277274618e-06,
+      "loss": 1.0931,
+      "step": 16100
+    },
+    {
+      "epoch": 0.5855547877950322,
+      "grad_norm": 0.0,
+      "learning_rate": 7.734782017114506e-06,
+      "loss": 0.903,
+      "step": 16101
+    },
+    {
+      "epoch": 0.5855911553987708,
+      "grad_norm": 0.0,
+      "learning_rate": 7.73363478838107e-06,
+      "loss": 0.97,
+      "step": 16102
+    },
+    {
+      "epoch": 0.5856275230025093,
+      "grad_norm": 0.0,
+      "learning_rate": 7.732487591090243e-06,
+      "loss": 0.9312,
+      "step": 16103
+    },
+    {
+      "epoch": 0.5856638906062479,
+      "grad_norm": 0.0,
+      "learning_rate": 7.731340425257929e-06,
+      "loss": 0.945,
+      "step": 16104
+    },
+    {
+      "epoch": 0.5857002582099865,
+      "grad_norm": 0.0,
+      "learning_rate": 7.73019329090005e-06,
+      "loss": 0.8759,
+      "step": 16105
+    },
+    {
+      "epoch": 0.5857366258137251,
+      "grad_norm": 0.0,
+      "learning_rate": 7.729046188032518e-06,
+      "loss": 0.896,
+      "step": 16106
+    },
+    {
+      "epoch": 0.5857729934174637,
+      "grad_norm": 0.0,
+      "learning_rate": 7.72789911667125e-06,
+      "loss": 0.9821,
+      "step": 16107
+    },
+    {
+      "epoch": 0.5858093610212023,
+      "grad_norm": 0.0,
+      "learning_rate": 7.726752076832154e-06,
+      "loss": 1.0351,
+      "step": 16108
+    },
+    {
+      "epoch": 0.5858457286249409,
+      "grad_norm": 0.0,
+      "learning_rate": 7.725605068531154e-06,
+      "loss": 0.9956,
+      "step": 16109
+    },
+    {
+      "epoch": 0.5858820962286795,
+      "grad_norm": 0.0,
+      "learning_rate": 7.72445809178415e-06,
+      "loss": 0.974,
+      "step": 16110
+    },
+    {
+      "epoch": 0.5859184638324181,
+      "grad_norm": 0.0,
+      "learning_rate": 7.723311146607066e-06,
+      "loss": 1.083,
+      "step": 16111
+    },
+    {
+      "epoch": 0.5859548314361567,
+      "grad_norm": 0.0,
+      "learning_rate": 7.722164233015804e-06,
+      "loss": 1.0487,
+      "step": 16112
+    },
+    {
+      "epoch": 0.5859911990398953,
+      "grad_norm": 0.0,
+      "learning_rate": 7.721017351026287e-06,
+      "loss": 0.9391,
+      "step": 16113
+    },
+    {
+      "epoch": 0.5860275666436339,
+      "grad_norm": 0.0,
+      "learning_rate": 7.719870500654419e-06,
+      "loss": 1.084,
+      "step": 16114
+    },
+    {
+      "epoch": 0.5860639342473725,
+      "grad_norm": 0.0,
+      "learning_rate": 7.718723681916116e-06,
+      "loss": 0.9883,
+      "step": 16115
+    },
+    {
+      "epoch": 0.5861003018511111,
+      "grad_norm": 0.0,
+      "learning_rate": 7.717576894827283e-06,
+      "loss": 0.9666,
+      "step": 16116
+    },
+    {
+      "epoch": 0.5861366694548497,
+      "grad_norm": 0.0,
+      "learning_rate": 7.716430139403829e-06,
+      "loss": 1.0503,
+      "step": 16117
+    },
+    {
+      "epoch": 0.5861730370585883,
+      "grad_norm": 0.0,
+      "learning_rate": 7.71528341566167e-06,
+      "loss": 0.9443,
+      "step": 16118
+    },
+    {
+      "epoch": 0.5862094046623268,
+      "grad_norm": 0.0,
+      "learning_rate": 7.71413672361671e-06,
+      "loss": 1.0143,
+      "step": 16119
+    },
+    {
+      "epoch": 0.5862457722660653,
+      "grad_norm": 0.0,
+      "learning_rate": 7.712990063284864e-06,
+      "loss": 0.9996,
+      "step": 16120
+    },
+    {
+      "epoch": 0.5862821398698039,
+      "grad_norm": 0.0,
+      "learning_rate": 7.711843434682029e-06,
+      "loss": 0.859,
+      "step": 16121
+    },
+    {
+      "epoch": 0.5863185074735425,
+      "grad_norm": 0.0,
+      "learning_rate": 7.710696837824127e-06,
+      "loss": 1.0379,
+      "step": 16122
+    },
+    {
+      "epoch": 0.5863548750772811,
+      "grad_norm": 0.0,
+      "learning_rate": 7.709550272727053e-06,
+      "loss": 0.9888,
+      "step": 16123
+    },
+    {
+      "epoch": 0.5863912426810197,
+      "grad_norm": 0.0,
+      "learning_rate": 7.708403739406722e-06,
+      "loss": 0.9083,
+      "step": 16124
+    },
+    {
+      "epoch": 0.5864276102847583,
+      "grad_norm": 0.0,
+      "learning_rate": 7.707257237879036e-06,
+      "loss": 0.9301,
+      "step": 16125
+    },
+    {
+      "epoch": 0.5864639778884969,
+      "grad_norm": 0.0,
+      "learning_rate": 7.706110768159905e-06,
+      "loss": 0.9264,
+      "step": 16126
+    },
+    {
+      "epoch": 0.5865003454922355,
+      "grad_norm": 0.0,
+      "learning_rate": 7.70496433026523e-06,
+      "loss": 1.1289,
+      "step": 16127
+    },
+    {
+      "epoch": 0.5865367130959741,
+      "grad_norm": 0.0,
+      "learning_rate": 7.703817924210922e-06,
+      "loss": 0.967,
+      "step": 16128
+    },
+    {
+      "epoch": 0.5865730806997127,
+      "grad_norm": 0.0,
+      "learning_rate": 7.702671550012876e-06,
+      "loss": 0.9188,
+      "step": 16129
+    },
+    {
+      "epoch": 0.5866094483034513,
+      "grad_norm": 0.0,
+      "learning_rate": 7.701525207687011e-06,
+      "loss": 0.9201,
+      "step": 16130
+    },
+    {
+      "epoch": 0.5866458159071899,
+      "grad_norm": 0.0,
+      "learning_rate": 7.700378897249217e-06,
+      "loss": 1.0365,
+      "step": 16131
+    },
+    {
+      "epoch": 0.5866821835109285,
+      "grad_norm": 0.0,
+      "learning_rate": 7.699232618715406e-06,
+      "loss": 0.8559,
+      "step": 16132
+    },
+    {
+      "epoch": 0.5867185511146671,
+      "grad_norm": 0.0,
+      "learning_rate": 7.698086372101477e-06,
+      "loss": 0.9316,
+      "step": 16133
+    },
+    {
+      "epoch": 0.5867549187184057,
+      "grad_norm": 0.0,
+      "learning_rate": 7.696940157423333e-06,
+      "loss": 0.9659,
+      "step": 16134
+    },
+    {
+      "epoch": 0.5867912863221443,
+      "grad_norm": 0.0,
+      "learning_rate": 7.695793974696882e-06,
+      "loss": 0.9302,
+      "step": 16135
+    },
+    {
+      "epoch": 0.5868276539258829,
+      "grad_norm": 0.0,
+      "learning_rate": 7.694647823938012e-06,
+      "loss": 0.8618,
+      "step": 16136
+    },
+    {
+      "epoch": 0.5868640215296214,
+      "grad_norm": 0.0,
+      "learning_rate": 7.693501705162638e-06,
+      "loss": 0.9865,
+      "step": 16137
+    },
+    {
+      "epoch": 0.58690038913336,
+      "grad_norm": 0.0,
+      "learning_rate": 7.692355618386654e-06,
+      "loss": 1.0416,
+      "step": 16138
+    },
+    {
+      "epoch": 0.5869367567370986,
+      "grad_norm": 0.0,
+      "learning_rate": 7.69120956362596e-06,
+      "loss": 0.9087,
+      "step": 16139
+    },
+    {
+      "epoch": 0.5869731243408371,
+      "grad_norm": 0.0,
+      "learning_rate": 7.690063540896458e-06,
+      "loss": 0.9207,
+      "step": 16140
+    },
+    {
+      "epoch": 0.5870094919445757,
+      "grad_norm": 0.0,
+      "learning_rate": 7.688917550214051e-06,
+      "loss": 1.0607,
+      "step": 16141
+    },
+    {
+      "epoch": 0.5870458595483143,
+      "grad_norm": 0.0,
+      "learning_rate": 7.68777159159463e-06,
+      "loss": 0.9379,
+      "step": 16142
+    },
+    {
+      "epoch": 0.5870822271520529,
+      "grad_norm": 0.0,
+      "learning_rate": 7.686625665054099e-06,
+      "loss": 1.0446,
+      "step": 16143
+    },
+    {
+      "epoch": 0.5871185947557915,
+      "grad_norm": 0.0,
+      "learning_rate": 7.685479770608351e-06,
+      "loss": 0.9114,
+      "step": 16144
+    },
+    {
+      "epoch": 0.5871549623595301,
+      "grad_norm": 0.0,
+      "learning_rate": 7.684333908273291e-06,
+      "loss": 0.9467,
+      "step": 16145
+    },
+    {
+      "epoch": 0.5871913299632687,
+      "grad_norm": 0.0,
+      "learning_rate": 7.683188078064811e-06,
+      "loss": 1.0198,
+      "step": 16146
+    },
+    {
+      "epoch": 0.5872276975670073,
+      "grad_norm": 0.0,
+      "learning_rate": 7.682042279998815e-06,
+      "loss": 0.9664,
+      "step": 16147
+    },
+    {
+      "epoch": 0.5872640651707459,
+      "grad_norm": 0.0,
+      "learning_rate": 7.680896514091184e-06,
+      "loss": 1.0187,
+      "step": 16148
+    },
+    {
+      "epoch": 0.5873004327744845,
+      "grad_norm": 0.0,
+      "learning_rate": 7.679750780357831e-06,
+      "loss": 0.9702,
+      "step": 16149
+    },
+    {
+      "epoch": 0.5873368003782231,
+      "grad_norm": 0.0,
+      "learning_rate": 7.67860507881464e-06,
+      "loss": 0.9168,
+      "step": 16150
+    },
+    {
+      "epoch": 0.5873731679819617,
+      "grad_norm": 0.0,
+      "learning_rate": 7.677459409477513e-06,
+      "loss": 1.0259,
+      "step": 16151
+    },
+    {
+      "epoch": 0.5874095355857003,
+      "grad_norm": 0.0,
+      "learning_rate": 7.676313772362338e-06,
+      "loss": 0.9153,
+      "step": 16152
+    },
+    {
+      "epoch": 0.5874459031894389,
+      "grad_norm": 0.0,
+      "learning_rate": 7.675168167485014e-06,
+      "loss": 1.0134,
+      "step": 16153
+    },
+    {
+      "epoch": 0.5874822707931774,
+      "grad_norm": 0.0,
+      "learning_rate": 7.674022594861435e-06,
+      "loss": 1.0008,
+      "step": 16154
+    },
+    {
+      "epoch": 0.587518638396916,
+      "grad_norm": 0.0,
+      "learning_rate": 7.672877054507491e-06,
+      "loss": 0.8651,
+      "step": 16155
+    },
+    {
+      "epoch": 0.5875550060006546,
+      "grad_norm": 0.0,
+      "learning_rate": 7.671731546439075e-06,
+      "loss": 0.85,
+      "step": 16156
+    },
+    {
+      "epoch": 0.5875913736043932,
+      "grad_norm": 0.0,
+      "learning_rate": 7.67058607067208e-06,
+      "loss": 0.9756,
+      "step": 16157
+    },
+    {
+      "epoch": 0.5876277412081318,
+      "grad_norm": 0.0,
+      "learning_rate": 7.6694406272224e-06,
+      "loss": 0.9729,
+      "step": 16158
+    },
+    {
+      "epoch": 0.5876641088118704,
+      "grad_norm": 0.0,
+      "learning_rate": 7.668295216105924e-06,
+      "loss": 0.9051,
+      "step": 16159
+    },
+    {
+      "epoch": 0.587700476415609,
+      "grad_norm": 0.0,
+      "learning_rate": 7.667149837338546e-06,
+      "loss": 1.0003,
+      "step": 16160
+    },
+    {
+      "epoch": 0.5877368440193476,
+      "grad_norm": 0.0,
+      "learning_rate": 7.666004490936148e-06,
+      "loss": 0.8926,
+      "step": 16161
+    },
+    {
+      "epoch": 0.5877732116230862,
+      "grad_norm": 0.0,
+      "learning_rate": 7.664859176914633e-06,
+      "loss": 1.0052,
+      "step": 16162
+    },
+    {
+      "epoch": 0.5878095792268248,
+      "grad_norm": 0.0,
+      "learning_rate": 7.66371389528988e-06,
+      "loss": 1.0629,
+      "step": 16163
+    },
+    {
+      "epoch": 0.5878459468305633,
+      "grad_norm": 0.0,
+      "learning_rate": 7.662568646077783e-06,
+      "loss": 0.9514,
+      "step": 16164
+    },
+    {
+      "epoch": 0.587882314434302,
+      "grad_norm": 0.0,
+      "learning_rate": 7.661423429294228e-06,
+      "loss": 1.0629,
+      "step": 16165
+    },
+    {
+      "epoch": 0.5879186820380405,
+      "grad_norm": 0.0,
+      "learning_rate": 7.660278244955106e-06,
+      "loss": 1.0119,
+      "step": 16166
+    },
+    {
+      "epoch": 0.5879550496417791,
+      "grad_norm": 0.0,
+      "learning_rate": 7.659133093076303e-06,
+      "loss": 0.9358,
+      "step": 16167
+    },
+    {
+      "epoch": 0.5879914172455177,
+      "grad_norm": 0.0,
+      "learning_rate": 7.65798797367371e-06,
+      "loss": 0.8427,
+      "step": 16168
+    },
+    {
+      "epoch": 0.5880277848492563,
+      "grad_norm": 0.0,
+      "learning_rate": 7.65684288676321e-06,
+      "loss": 0.8993,
+      "step": 16169
+    },
+    {
+      "epoch": 0.5880641524529949,
+      "grad_norm": 0.0,
+      "learning_rate": 7.655697832360689e-06,
+      "loss": 1.0595,
+      "step": 16170
+    },
+    {
+      "epoch": 0.5881005200567334,
+      "grad_norm": 0.0,
+      "learning_rate": 7.654552810482034e-06,
+      "loss": 1.0888,
+      "step": 16171
+    },
+    {
+      "epoch": 0.588136887660472,
+      "grad_norm": 0.0,
+      "learning_rate": 7.653407821143132e-06,
+      "loss": 1.0947,
+      "step": 16172
+    },
+    {
+      "epoch": 0.5881732552642106,
+      "grad_norm": 0.0,
+      "learning_rate": 7.652262864359869e-06,
+      "loss": 1.0042,
+      "step": 16173
+    },
+    {
+      "epoch": 0.5882096228679492,
+      "grad_norm": 0.0,
+      "learning_rate": 7.651117940148123e-06,
+      "loss": 0.897,
+      "step": 16174
+    },
+    {
+      "epoch": 0.5882459904716878,
+      "grad_norm": 0.0,
+      "learning_rate": 7.64997304852379e-06,
+      "loss": 0.9272,
+      "step": 16175
+    },
+    {
+      "epoch": 0.5882823580754264,
+      "grad_norm": 0.0,
+      "learning_rate": 7.648828189502741e-06,
+      "loss": 1.1425,
+      "step": 16176
+    },
+    {
+      "epoch": 0.588318725679165,
+      "grad_norm": 0.0,
+      "learning_rate": 7.647683363100866e-06,
+      "loss": 0.994,
+      "step": 16177
+    },
+    {
+      "epoch": 0.5883550932829036,
+      "grad_norm": 0.0,
+      "learning_rate": 7.646538569334047e-06,
+      "loss": 0.9664,
+      "step": 16178
+    },
+    {
+      "epoch": 0.5883914608866422,
+      "grad_norm": 0.0,
+      "learning_rate": 7.645393808218166e-06,
+      "loss": 0.9816,
+      "step": 16179
+    },
+    {
+      "epoch": 0.5884278284903808,
+      "grad_norm": 0.0,
+      "learning_rate": 7.644249079769105e-06,
+      "loss": 0.9492,
+      "step": 16180
+    },
+    {
+      "epoch": 0.5884641960941194,
+      "grad_norm": 0.0,
+      "learning_rate": 7.643104384002748e-06,
+      "loss": 0.9098,
+      "step": 16181
+    },
+    {
+      "epoch": 0.588500563697858,
+      "grad_norm": 0.0,
+      "learning_rate": 7.641959720934971e-06,
+      "loss": 0.8961,
+      "step": 16182
+    },
+    {
+      "epoch": 0.5885369313015966,
+      "grad_norm": 0.0,
+      "learning_rate": 7.640815090581658e-06,
+      "loss": 0.8854,
+      "step": 16183
+    },
+    {
+      "epoch": 0.5885732989053352,
+      "grad_norm": 0.0,
+      "learning_rate": 7.639670492958688e-06,
+      "loss": 1.0482,
+      "step": 16184
+    },
+    {
+      "epoch": 0.5886096665090738,
+      "grad_norm": 0.0,
+      "learning_rate": 7.638525928081943e-06,
+      "loss": 0.9329,
+      "step": 16185
+    },
+    {
+      "epoch": 0.5886460341128124,
+      "grad_norm": 0.0,
+      "learning_rate": 7.637381395967298e-06,
+      "loss": 0.9733,
+      "step": 16186
+    },
+    {
+      "epoch": 0.588682401716551,
+      "grad_norm": 0.0,
+      "learning_rate": 7.636236896630638e-06,
+      "loss": 0.9818,
+      "step": 16187
+    },
+    {
+      "epoch": 0.5887187693202894,
+      "grad_norm": 0.0,
+      "learning_rate": 7.63509243008783e-06,
+      "loss": 0.976,
+      "step": 16188
+    },
+    {
+      "epoch": 0.588755136924028,
+      "grad_norm": 0.0,
+      "learning_rate": 7.633947996354766e-06,
+      "loss": 1.0474,
+      "step": 16189
+    },
+    {
+      "epoch": 0.5887915045277666,
+      "grad_norm": 0.0,
+      "learning_rate": 7.632803595447313e-06,
+      "loss": 1.026,
+      "step": 16190
+    },
+    {
+      "epoch": 0.5888278721315052,
+      "grad_norm": 0.0,
+      "learning_rate": 7.631659227381354e-06,
+      "loss": 0.9837,
+      "step": 16191
+    },
+    {
+      "epoch": 0.5888642397352438,
+      "grad_norm": 0.0,
+      "learning_rate": 7.63051489217276e-06,
+      "loss": 0.9434,
+      "step": 16192
+    },
+    {
+      "epoch": 0.5889006073389824,
+      "grad_norm": 0.0,
+      "learning_rate": 7.629370589837412e-06,
+      "loss": 0.975,
+      "step": 16193
+    },
+    {
+      "epoch": 0.588936974942721,
+      "grad_norm": 0.0,
+      "learning_rate": 7.628226320391185e-06,
+      "loss": 0.8907,
+      "step": 16194
+    },
+    {
+      "epoch": 0.5889733425464596,
+      "grad_norm": 0.0,
+      "learning_rate": 7.627082083849951e-06,
+      "loss": 1.0486,
+      "step": 16195
+    },
+    {
+      "epoch": 0.5890097101501982,
+      "grad_norm": 0.0,
+      "learning_rate": 7.625937880229586e-06,
+      "loss": 0.9498,
+      "step": 16196
+    },
+    {
+      "epoch": 0.5890460777539368,
+      "grad_norm": 0.0,
+      "learning_rate": 7.624793709545964e-06,
+      "loss": 1.0676,
+      "step": 16197
+    },
+    {
+      "epoch": 0.5890824453576754,
+      "grad_norm": 0.0,
+      "learning_rate": 7.62364957181496e-06,
+      "loss": 0.9984,
+      "step": 16198
+    },
+    {
+      "epoch": 0.589118812961414,
+      "grad_norm": 0.0,
+      "learning_rate": 7.622505467052446e-06,
+      "loss": 0.932,
+      "step": 16199
+    },
+    {
+      "epoch": 0.5891551805651526,
+      "grad_norm": 0.0,
+      "learning_rate": 7.621361395274298e-06,
+      "loss": 0.8682,
+      "step": 16200
+    },
+    {
+      "epoch": 0.5891915481688912,
+      "grad_norm": 0.0,
+      "learning_rate": 7.62021735649638e-06,
+      "loss": 0.9055,
+      "step": 16201
+    },
+    {
+      "epoch": 0.5892279157726298,
+      "grad_norm": 0.0,
+      "learning_rate": 7.619073350734576e-06,
+      "loss": 0.9952,
+      "step": 16202
+    },
+    {
+      "epoch": 0.5892642833763684,
+      "grad_norm": 0.0,
+      "learning_rate": 7.6179293780047485e-06,
+      "loss": 1.0203,
+      "step": 16203
+    },
+    {
+      "epoch": 0.5893006509801069,
+      "grad_norm": 0.0,
+      "learning_rate": 7.616785438322772e-06,
+      "loss": 0.9539,
+      "step": 16204
+    },
+    {
+      "epoch": 0.5893370185838455,
+      "grad_norm": 0.0,
+      "learning_rate": 7.615641531704514e-06,
+      "loss": 0.9622,
+      "step": 16205
+    },
+    {
+      "epoch": 0.5893733861875841,
+      "grad_norm": 0.0,
+      "learning_rate": 7.614497658165848e-06,
+      "loss": 0.9316,
+      "step": 16206
+    },
+    {
+      "epoch": 0.5894097537913227,
+      "grad_norm": 0.0,
+      "learning_rate": 7.613353817722641e-06,
+      "loss": 0.9256,
+      "step": 16207
+    },
+    {
+      "epoch": 0.5894461213950613,
+      "grad_norm": 0.0,
+      "learning_rate": 7.61221001039077e-06,
+      "loss": 1.0707,
+      "step": 16208
+    },
+    {
+      "epoch": 0.5894824889987998,
+      "grad_norm": 0.0,
+      "learning_rate": 7.6110662361860895e-06,
+      "loss": 0.8954,
+      "step": 16209
+    },
+    {
+      "epoch": 0.5895188566025384,
+      "grad_norm": 0.0,
+      "learning_rate": 7.60992249512448e-06,
+      "loss": 0.9954,
+      "step": 16210
+    },
+    {
+      "epoch": 0.589555224206277,
+      "grad_norm": 0.0,
+      "learning_rate": 7.6087787872218025e-06,
+      "loss": 1.0265,
+      "step": 16211
+    },
+    {
+      "epoch": 0.5895915918100156,
+      "grad_norm": 0.0,
+      "learning_rate": 7.607635112493927e-06,
+      "loss": 0.9122,
+      "step": 16212
+    },
+    {
+      "epoch": 0.5896279594137542,
+      "grad_norm": 0.0,
+      "learning_rate": 7.606491470956724e-06,
+      "loss": 1.0753,
+      "step": 16213
+    },
+    {
+      "epoch": 0.5896643270174928,
+      "grad_norm": 0.0,
+      "learning_rate": 7.605347862626051e-06,
+      "loss": 0.9118,
+      "step": 16214
+    },
+    {
+      "epoch": 0.5897006946212314,
+      "grad_norm": 0.0,
+      "learning_rate": 7.604204287517781e-06,
+      "loss": 1.1079,
+      "step": 16215
+    },
+    {
+      "epoch": 0.58973706222497,
+      "grad_norm": 0.0,
+      "learning_rate": 7.6030607456477755e-06,
+      "loss": 0.9343,
+      "step": 16216
+    },
+    {
+      "epoch": 0.5897734298287086,
+      "grad_norm": 0.0,
+      "learning_rate": 7.601917237031902e-06,
+      "loss": 1.0836,
+      "step": 16217
+    },
+    {
+      "epoch": 0.5898097974324472,
+      "grad_norm": 0.0,
+      "learning_rate": 7.600773761686025e-06,
+      "loss": 1.0443,
+      "step": 16218
+    },
+    {
+      "epoch": 0.5898461650361858,
+      "grad_norm": 0.0,
+      "learning_rate": 7.59963031962601e-06,
+      "loss": 0.9642,
+      "step": 16219
+    },
+    {
+      "epoch": 0.5898825326399244,
+      "grad_norm": 0.0,
+      "learning_rate": 7.598486910867713e-06,
+      "loss": 0.9289,
+      "step": 16220
+    },
+    {
+      "epoch": 0.5899189002436629,
+      "grad_norm": 0.0,
+      "learning_rate": 7.597343535427009e-06,
+      "loss": 1.0042,
+      "step": 16221
+    },
+    {
+      "epoch": 0.5899552678474015,
+      "grad_norm": 0.0,
+      "learning_rate": 7.59620019331975e-06,
+      "loss": 0.9011,
+      "step": 16222
+    },
+    {
+      "epoch": 0.5899916354511401,
+      "grad_norm": 0.0,
+      "learning_rate": 7.595056884561806e-06,
+      "loss": 0.9057,
+      "step": 16223
+    },
+    {
+      "epoch": 0.5900280030548787,
+      "grad_norm": 0.0,
+      "learning_rate": 7.593913609169032e-06,
+      "loss": 1.0483,
+      "step": 16224
+    },
+    {
+      "epoch": 0.5900643706586173,
+      "grad_norm": 0.0,
+      "learning_rate": 7.592770367157296e-06,
+      "loss": 1.0549,
+      "step": 16225
+    },
+    {
+      "epoch": 0.5901007382623559,
+      "grad_norm": 0.0,
+      "learning_rate": 7.591627158542454e-06,
+      "loss": 0.992,
+      "step": 16226
+    },
+    {
+      "epoch": 0.5901371058660945,
+      "grad_norm": 0.0,
+      "learning_rate": 7.590483983340371e-06,
+      "loss": 1.0275,
+      "step": 16227
+    },
+    {
+      "epoch": 0.5901734734698331,
+      "grad_norm": 0.0,
+      "learning_rate": 7.589340841566899e-06,
+      "loss": 1.034,
+      "step": 16228
+    },
+    {
+      "epoch": 0.5902098410735717,
+      "grad_norm": 0.0,
+      "learning_rate": 7.588197733237908e-06,
+      "loss": 1.0784,
+      "step": 16229
+    },
+    {
+      "epoch": 0.5902462086773103,
+      "grad_norm": 0.0,
+      "learning_rate": 7.587054658369247e-06,
+      "loss": 1.032,
+      "step": 16230
+    },
+    {
+      "epoch": 0.5902825762810489,
+      "grad_norm": 0.0,
+      "learning_rate": 7.585911616976782e-06,
+      "loss": 0.8522,
+      "step": 16231
+    },
+    {
+      "epoch": 0.5903189438847875,
+      "grad_norm": 0.0,
+      "learning_rate": 7.584768609076371e-06,
+      "loss": 0.9043,
+      "step": 16232
+    },
+    {
+      "epoch": 0.590355311488526,
+      "grad_norm": 0.0,
+      "learning_rate": 7.583625634683861e-06,
+      "loss": 0.9553,
+      "step": 16233
+    },
+    {
+      "epoch": 0.5903916790922646,
+      "grad_norm": 0.0,
+      "learning_rate": 7.582482693815124e-06,
+      "loss": 0.9353,
+      "step": 16234
+    },
+    {
+      "epoch": 0.5904280466960032,
+      "grad_norm": 0.0,
+      "learning_rate": 7.581339786486005e-06,
+      "loss": 0.8871,
+      "step": 16235
+    },
+    {
+      "epoch": 0.5904644142997418,
+      "grad_norm": 0.0,
+      "learning_rate": 7.580196912712368e-06,
+      "loss": 0.9807,
+      "step": 16236
+    },
+    {
+      "epoch": 0.5905007819034804,
+      "grad_norm": 0.0,
+      "learning_rate": 7.579054072510062e-06,
+      "loss": 0.9661,
+      "step": 16237
+    },
+    {
+      "epoch": 0.5905371495072189,
+      "grad_norm": 0.0,
+      "learning_rate": 7.577911265894948e-06,
+      "loss": 1.023,
+      "step": 16238
+    },
+    {
+      "epoch": 0.5905735171109575,
+      "grad_norm": 0.0,
+      "learning_rate": 7.576768492882877e-06,
+      "loss": 0.9344,
+      "step": 16239
+    },
+    {
+      "epoch": 0.5906098847146961,
+      "grad_norm": 0.0,
+      "learning_rate": 7.57562575348971e-06,
+      "loss": 1.0023,
+      "step": 16240
+    },
+    {
+      "epoch": 0.5906462523184347,
+      "grad_norm": 0.0,
+      "learning_rate": 7.574483047731291e-06,
+      "loss": 1.0202,
+      "step": 16241
+    },
+    {
+      "epoch": 0.5906826199221733,
+      "grad_norm": 0.0,
+      "learning_rate": 7.573340375623481e-06,
+      "loss": 1.0327,
+      "step": 16242
+    },
+    {
+      "epoch": 0.5907189875259119,
+      "grad_norm": 0.0,
+      "learning_rate": 7.572197737182129e-06,
+      "loss": 1.0581,
+      "step": 16243
+    },
+    {
+      "epoch": 0.5907553551296505,
+      "grad_norm": 0.0,
+      "learning_rate": 7.571055132423088e-06,
+      "loss": 1.0461,
+      "step": 16244
+    },
+    {
+      "epoch": 0.5907917227333891,
+      "grad_norm": 0.0,
+      "learning_rate": 7.569912561362212e-06,
+      "loss": 0.9801,
+      "step": 16245
+    },
+    {
+      "epoch": 0.5908280903371277,
+      "grad_norm": 0.0,
+      "learning_rate": 7.568770024015352e-06,
+      "loss": 0.9457,
+      "step": 16246
+    },
+    {
+      "epoch": 0.5908644579408663,
+      "grad_norm": 0.0,
+      "learning_rate": 7.567627520398356e-06,
+      "loss": 0.908,
+      "step": 16247
+    },
+    {
+      "epoch": 0.5909008255446049,
+      "grad_norm": 0.0,
+      "learning_rate": 7.566485050527082e-06,
+      "loss": 1.114,
+      "step": 16248
+    },
+    {
+      "epoch": 0.5909371931483435,
+      "grad_norm": 0.0,
+      "learning_rate": 7.56534261441737e-06,
+      "loss": 0.93,
+      "step": 16249
+    },
+    {
+      "epoch": 0.5909735607520821,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5642002120850775e-06,
+      "loss": 1.0913,
+      "step": 16250
+    },
+    {
+      "epoch": 0.5910099283558207,
+      "grad_norm": 0.0,
+      "learning_rate": 7.563057843546051e-06,
+      "loss": 1.0488,
+      "step": 16251
+    },
+    {
+      "epoch": 0.5910462959595593,
+      "grad_norm": 0.0,
+      "learning_rate": 7.561915508816138e-06,
+      "loss": 0.8899,
+      "step": 16252
+    },
+    {
+      "epoch": 0.5910826635632979,
+      "grad_norm": 0.0,
+      "learning_rate": 7.560773207911192e-06,
+      "loss": 1.0121,
+      "step": 16253
+    },
+    {
+      "epoch": 0.5911190311670365,
+      "grad_norm": 0.0,
+      "learning_rate": 7.559630940847053e-06,
+      "loss": 1.0128,
+      "step": 16254
+    },
+    {
+      "epoch": 0.5911553987707749,
+      "grad_norm": 0.0,
+      "learning_rate": 7.558488707639573e-06,
+      "loss": 0.9736,
+      "step": 16255
+    },
+    {
+      "epoch": 0.5911917663745135,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5573465083045975e-06,
+      "loss": 0.9014,
+      "step": 16256
+    },
+    {
+      "epoch": 0.5912281339782521,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5562043428579745e-06,
+      "loss": 1.0541,
+      "step": 16257
+    },
+    {
+      "epoch": 0.5912645015819907,
+      "grad_norm": 0.0,
+      "learning_rate": 7.555062211315547e-06,
+      "loss": 0.9806,
+      "step": 16258
+    },
+    {
+      "epoch": 0.5913008691857293,
+      "grad_norm": 0.0,
+      "learning_rate": 7.553920113693167e-06,
+      "loss": 0.942,
+      "step": 16259
+    },
+    {
+      "epoch": 0.5913372367894679,
+      "grad_norm": 0.0,
+      "learning_rate": 7.552778050006669e-06,
+      "loss": 1.0656,
+      "step": 16260
+    },
+    {
+      "epoch": 0.5913736043932065,
+      "grad_norm": 0.0,
+      "learning_rate": 7.55163602027191e-06,
+      "loss": 0.9948,
+      "step": 16261
+    },
+    {
+      "epoch": 0.5914099719969451,
+      "grad_norm": 0.0,
+      "learning_rate": 7.550494024504722e-06,
+      "loss": 1.0108,
+      "step": 16262
+    },
+    {
+      "epoch": 0.5914463396006837,
+      "grad_norm": 0.0,
+      "learning_rate": 7.549352062720956e-06,
+      "loss": 0.9592,
+      "step": 16263
+    },
+    {
+      "epoch": 0.5914827072044223,
+      "grad_norm": 0.0,
+      "learning_rate": 7.548210134936452e-06,
+      "loss": 0.916,
+      "step": 16264
+    },
+    {
+      "epoch": 0.5915190748081609,
+      "grad_norm": 0.0,
+      "learning_rate": 7.547068241167056e-06,
+      "loss": 0.9454,
+      "step": 16265
+    },
+    {
+      "epoch": 0.5915554424118995,
+      "grad_norm": 0.0,
+      "learning_rate": 7.545926381428605e-06,
+      "loss": 0.9889,
+      "step": 16266
+    },
+    {
+      "epoch": 0.5915918100156381,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5447845557369485e-06,
+      "loss": 0.9551,
+      "step": 16267
+    },
+    {
+      "epoch": 0.5916281776193767,
+      "grad_norm": 0.0,
+      "learning_rate": 7.543642764107918e-06,
+      "loss": 0.8137,
+      "step": 16268
+    },
+    {
+      "epoch": 0.5916645452231153,
+      "grad_norm": 0.0,
+      "learning_rate": 7.542501006557361e-06,
+      "loss": 0.9565,
+      "step": 16269
+    },
+    {
+      "epoch": 0.5917009128268539,
+      "grad_norm": 0.0,
+      "learning_rate": 7.541359283101116e-06,
+      "loss": 0.9323,
+      "step": 16270
+    },
+    {
+      "epoch": 0.5917372804305925,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5402175937550215e-06,
+      "loss": 0.8463,
+      "step": 16271
+    },
+    {
+      "epoch": 0.591773648034331,
+      "grad_norm": 0.0,
+      "learning_rate": 7.539075938534921e-06,
+      "loss": 0.9521,
+      "step": 16272
+    },
+    {
+      "epoch": 0.5918100156380696,
+      "grad_norm": 0.0,
+      "learning_rate": 7.537934317456643e-06,
+      "loss": 0.9559,
+      "step": 16273
+    },
+    {
+      "epoch": 0.5918463832418082,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5367927305360415e-06,
+      "loss": 1.0323,
+      "step": 16274
+    },
+    {
+      "epoch": 0.5918827508455468,
+      "grad_norm": 0.0,
+      "learning_rate": 7.535651177788941e-06,
+      "loss": 1.0131,
+      "step": 16275
+    },
+    {
+      "epoch": 0.5919191184492854,
+      "grad_norm": 0.0,
+      "learning_rate": 7.534509659231186e-06,
+      "loss": 0.946,
+      "step": 16276
+    },
+    {
+      "epoch": 0.591955486053024,
+      "grad_norm": 0.0,
+      "learning_rate": 7.533368174878609e-06,
+      "loss": 1.0223,
+      "step": 16277
+    },
+    {
+      "epoch": 0.5919918536567625,
+      "grad_norm": 0.0,
+      "learning_rate": 7.532226724747051e-06,
+      "loss": 0.9314,
+      "step": 16278
+    },
+    {
+      "epoch": 0.5920282212605011,
+      "grad_norm": 0.0,
+      "learning_rate": 7.531085308852344e-06,
+      "loss": 1.0153,
+      "step": 16279
+    },
+    {
+      "epoch": 0.5920645888642397,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5299439272103305e-06,
+      "loss": 1.0775,
+      "step": 16280
+    },
+    {
+      "epoch": 0.5921009564679783,
+      "grad_norm": 0.0,
+      "learning_rate": 7.528802579836835e-06,
+      "loss": 0.8447,
+      "step": 16281
+    },
+    {
+      "epoch": 0.5921373240717169,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5276612667477e-06,
+      "loss": 0.9437,
+      "step": 16282
+    },
+    {
+      "epoch": 0.5921736916754555,
+      "grad_norm": 0.0,
+      "learning_rate": 7.526519987958756e-06,
+      "loss": 1.0151,
+      "step": 16283
+    },
+    {
+      "epoch": 0.5922100592791941,
+      "grad_norm": 0.0,
+      "learning_rate": 7.525378743485838e-06,
+      "loss": 0.9998,
+      "step": 16284
+    },
+    {
+      "epoch": 0.5922464268829327,
+      "grad_norm": 0.0,
+      "learning_rate": 7.524237533344778e-06,
+      "loss": 0.8892,
+      "step": 16285
+    },
+    {
+      "epoch": 0.5922827944866713,
+      "grad_norm": 0.0,
+      "learning_rate": 7.523096357551414e-06,
+      "loss": 0.9579,
+      "step": 16286
+    },
+    {
+      "epoch": 0.5923191620904099,
+      "grad_norm": 0.0,
+      "learning_rate": 7.521955216121566e-06,
+      "loss": 1.0355,
+      "step": 16287
+    },
+    {
+      "epoch": 0.5923555296941484,
+      "grad_norm": 0.0,
+      "learning_rate": 7.520814109071082e-06,
+      "loss": 0.9901,
+      "step": 16288
+    },
+    {
+      "epoch": 0.592391897297887,
+      "grad_norm": 0.0,
+      "learning_rate": 7.519673036415781e-06,
+      "loss": 0.8514,
+      "step": 16289
+    },
+    {
+      "epoch": 0.5924282649016256,
+      "grad_norm": 0.0,
+      "learning_rate": 7.518531998171496e-06,
+      "loss": 1.0046,
+      "step": 16290
+    },
+    {
+      "epoch": 0.5924646325053642,
+      "grad_norm": 0.0,
+      "learning_rate": 7.517390994354063e-06,
+      "loss": 0.968,
+      "step": 16291
+    },
+    {
+      "epoch": 0.5925010001091028,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5162500249793e-06,
+      "loss": 1.0115,
+      "step": 16292
+    },
+    {
+      "epoch": 0.5925373677128414,
+      "grad_norm": 0.0,
+      "learning_rate": 7.515109090063052e-06,
+      "loss": 0.9149,
+      "step": 16293
+    },
+    {
+      "epoch": 0.59257373531658,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5139681896211344e-06,
+      "loss": 0.9001,
+      "step": 16294
+    },
+    {
+      "epoch": 0.5926101029203186,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5128273236693835e-06,
+      "loss": 0.9831,
+      "step": 16295
+    },
+    {
+      "epoch": 0.5926464705240572,
+      "grad_norm": 0.0,
+      "learning_rate": 7.511686492223624e-06,
+      "loss": 0.9831,
+      "step": 16296
+    },
+    {
+      "epoch": 0.5926828381277958,
+      "grad_norm": 0.0,
+      "learning_rate": 7.510545695299683e-06,
+      "loss": 0.9803,
+      "step": 16297
+    },
+    {
+      "epoch": 0.5927192057315344,
+      "grad_norm": 0.0,
+      "learning_rate": 7.509404932913389e-06,
+      "loss": 0.9616,
+      "step": 16298
+    },
+    {
+      "epoch": 0.592755573335273,
+      "grad_norm": 0.0,
+      "learning_rate": 7.508264205080571e-06,
+      "loss": 1.0001,
+      "step": 16299
+    },
+    {
+      "epoch": 0.5927919409390116,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5071235118170445e-06,
+      "loss": 1.0188,
+      "step": 16300
+    },
+    {
+      "epoch": 0.5928283085427501,
+      "grad_norm": 0.0,
+      "learning_rate": 7.50598285313865e-06,
+      "loss": 0.906,
+      "step": 16301
+    },
+    {
+      "epoch": 0.5928646761464887,
+      "grad_norm": 0.0,
+      "learning_rate": 7.504842229061201e-06,
+      "loss": 0.998,
+      "step": 16302
+    },
+    {
+      "epoch": 0.5929010437502273,
+      "grad_norm": 0.0,
+      "learning_rate": 7.503701639600527e-06,
+      "loss": 1.0369,
+      "step": 16303
+    },
+    {
+      "epoch": 0.5929374113539659,
+      "grad_norm": 0.0,
+      "learning_rate": 7.50256108477245e-06,
+      "loss": 0.9725,
+      "step": 16304
+    },
+    {
+      "epoch": 0.5929737789577044,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5014205645927955e-06,
+      "loss": 0.9542,
+      "step": 16305
+    },
+    {
+      "epoch": 0.593010146561443,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5002800790773835e-06,
+      "loss": 0.8373,
+      "step": 16306
+    },
+    {
+      "epoch": 0.5930465141651816,
+      "grad_norm": 0.0,
+      "learning_rate": 7.499139628242044e-06,
+      "loss": 0.9869,
+      "step": 16307
+    },
+    {
+      "epoch": 0.5930828817689202,
+      "grad_norm": 0.0,
+      "learning_rate": 7.497999212102591e-06,
+      "loss": 0.9682,
+      "step": 16308
+    },
+    {
+      "epoch": 0.5931192493726588,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4968588306748465e-06,
+      "loss": 0.9777,
+      "step": 16309
+    },
+    {
+      "epoch": 0.5931556169763974,
+      "grad_norm": 0.0,
+      "learning_rate": 7.495718483974638e-06,
+      "loss": 1.1809,
+      "step": 16310
+    },
+    {
+      "epoch": 0.593191984580136,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4945781720177794e-06,
+      "loss": 0.9252,
+      "step": 16311
+    },
+    {
+      "epoch": 0.5932283521838746,
+      "grad_norm": 0.0,
+      "learning_rate": 7.493437894820097e-06,
+      "loss": 1.037,
+      "step": 16312
+    },
+    {
+      "epoch": 0.5932647197876132,
+      "grad_norm": 0.0,
+      "learning_rate": 7.492297652397405e-06,
+      "loss": 0.8546,
+      "step": 16313
+    },
+    {
+      "epoch": 0.5933010873913518,
+      "grad_norm": 0.0,
+      "learning_rate": 7.491157444765527e-06,
+      "loss": 0.9583,
+      "step": 16314
+    },
+    {
+      "epoch": 0.5933374549950904,
+      "grad_norm": 0.0,
+      "learning_rate": 7.490017271940278e-06,
+      "loss": 0.998,
+      "step": 16315
+    },
+    {
+      "epoch": 0.593373822598829,
+      "grad_norm": 0.0,
+      "learning_rate": 7.488877133937478e-06,
+      "loss": 1.0549,
+      "step": 16316
+    },
+    {
+      "epoch": 0.5934101902025676,
+      "grad_norm": 0.0,
+      "learning_rate": 7.487737030772945e-06,
+      "loss": 1.0744,
+      "step": 16317
+    },
+    {
+      "epoch": 0.5934465578063062,
+      "grad_norm": 0.0,
+      "learning_rate": 7.486596962462498e-06,
+      "loss": 0.8434,
+      "step": 16318
+    },
+    {
+      "epoch": 0.5934829254100448,
+      "grad_norm": 0.0,
+      "learning_rate": 7.485456929021949e-06,
+      "loss": 1.011,
+      "step": 16319
+    },
+    {
+      "epoch": 0.5935192930137834,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4843169304671224e-06,
+      "loss": 0.883,
+      "step": 16320
+    },
+    {
+      "epoch": 0.593555660617522,
+      "grad_norm": 0.0,
+      "learning_rate": 7.483176966813825e-06,
+      "loss": 1.0323,
+      "step": 16321
+    },
+    {
+      "epoch": 0.5935920282212604,
+      "grad_norm": 0.0,
+      "learning_rate": 7.482037038077878e-06,
+      "loss": 0.9952,
+      "step": 16322
+    },
+    {
+      "epoch": 0.593628395824999,
+      "grad_norm": 0.0,
+      "learning_rate": 7.480897144275093e-06,
+      "loss": 1.0199,
+      "step": 16323
+    },
+    {
+      "epoch": 0.5936647634287376,
+      "grad_norm": 0.0,
+      "learning_rate": 7.479757285421288e-06,
+      "loss": 1.0549,
+      "step": 16324
+    },
+    {
+      "epoch": 0.5937011310324762,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4786174615322725e-06,
+      "loss": 0.7103,
+      "step": 16325
+    },
+    {
+      "epoch": 0.5937374986362148,
+      "grad_norm": 0.0,
+      "learning_rate": 7.477477672623867e-06,
+      "loss": 0.9446,
+      "step": 16326
+    },
+    {
+      "epoch": 0.5937738662399534,
+      "grad_norm": 0.0,
+      "learning_rate": 7.476337918711878e-06,
+      "loss": 0.9523,
+      "step": 16327
+    },
+    {
+      "epoch": 0.593810233843692,
+      "grad_norm": 0.0,
+      "learning_rate": 7.475198199812117e-06,
+      "loss": 0.9683,
+      "step": 16328
+    },
+    {
+      "epoch": 0.5938466014474306,
+      "grad_norm": 0.0,
+      "learning_rate": 7.474058515940401e-06,
+      "loss": 0.8514,
+      "step": 16329
+    },
+    {
+      "epoch": 0.5938829690511692,
+      "grad_norm": 0.0,
+      "learning_rate": 7.472918867112536e-06,
+      "loss": 0.8677,
+      "step": 16330
+    },
+    {
+      "epoch": 0.5939193366549078,
+      "grad_norm": 0.0,
+      "learning_rate": 7.471779253344342e-06,
+      "loss": 0.8976,
+      "step": 16331
+    },
+    {
+      "epoch": 0.5939557042586464,
+      "grad_norm": 0.0,
+      "learning_rate": 7.470639674651616e-06,
+      "loss": 0.9877,
+      "step": 16332
+    },
+    {
+      "epoch": 0.593992071862385,
+      "grad_norm": 0.0,
+      "learning_rate": 7.469500131050185e-06,
+      "loss": 0.9099,
+      "step": 16333
+    },
+    {
+      "epoch": 0.5940284394661236,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4683606225558425e-06,
+      "loss": 0.9039,
+      "step": 16334
+    },
+    {
+      "epoch": 0.5940648070698622,
+      "grad_norm": 0.0,
+      "learning_rate": 7.467221149184407e-06,
+      "loss": 0.9248,
+      "step": 16335
+    },
+    {
+      "epoch": 0.5941011746736008,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4660817109516825e-06,
+      "loss": 1.0731,
+      "step": 16336
+    },
+    {
+      "epoch": 0.5941375422773394,
+      "grad_norm": 0.0,
+      "learning_rate": 7.464942307873482e-06,
+      "loss": 1.0693,
+      "step": 16337
+    },
+    {
+      "epoch": 0.594173909881078,
+      "grad_norm": 0.0,
+      "learning_rate": 7.463802939965608e-06,
+      "loss": 0.887,
+      "step": 16338
+    },
+    {
+      "epoch": 0.5942102774848165,
+      "grad_norm": 0.0,
+      "learning_rate": 7.462663607243873e-06,
+      "loss": 1.1113,
+      "step": 16339
+    },
+    {
+      "epoch": 0.5942466450885551,
+      "grad_norm": 0.0,
+      "learning_rate": 7.461524309724077e-06,
+      "loss": 0.9487,
+      "step": 16340
+    },
+    {
+      "epoch": 0.5942830126922937,
+      "grad_norm": 0.0,
+      "learning_rate": 7.460385047422032e-06,
+      "loss": 0.9468,
+      "step": 16341
+    },
+    {
+      "epoch": 0.5943193802960323,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4592458203535374e-06,
+      "loss": 1.0603,
+      "step": 16342
+    },
+    {
+      "epoch": 0.5943557478997709,
+      "grad_norm": 0.0,
+      "learning_rate": 7.458106628534406e-06,
+      "loss": 1.0261,
+      "step": 16343
+    },
+    {
+      "epoch": 0.5943921155035095,
+      "grad_norm": 0.0,
+      "learning_rate": 7.456967471980437e-06,
+      "loss": 0.849,
+      "step": 16344
+    },
+    {
+      "epoch": 0.594428483107248,
+      "grad_norm": 0.0,
+      "learning_rate": 7.455828350707437e-06,
+      "loss": 0.9915,
+      "step": 16345
+    },
+    {
+      "epoch": 0.5944648507109866,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4546892647312105e-06,
+      "loss": 0.8612,
+      "step": 16346
+    },
+    {
+      "epoch": 0.5945012183147252,
+      "grad_norm": 0.0,
+      "learning_rate": 7.453550214067556e-06,
+      "loss": 1.0249,
+      "step": 16347
+    },
+    {
+      "epoch": 0.5945375859184638,
+      "grad_norm": 0.0,
+      "learning_rate": 7.452411198732281e-06,
+      "loss": 1.037,
+      "step": 16348
+    },
+    {
+      "epoch": 0.5945739535222024,
+      "grad_norm": 0.0,
+      "learning_rate": 7.451272218741184e-06,
+      "loss": 0.9545,
+      "step": 16349
+    },
+    {
+      "epoch": 0.594610321125941,
+      "grad_norm": 0.0,
+      "learning_rate": 7.450133274110072e-06,
+      "loss": 0.904,
+      "step": 16350
+    },
+    {
+      "epoch": 0.5946466887296796,
+      "grad_norm": 0.0,
+      "learning_rate": 7.448994364854741e-06,
+      "loss": 1.0005,
+      "step": 16351
+    },
+    {
+      "epoch": 0.5946830563334182,
+      "grad_norm": 0.0,
+      "learning_rate": 7.447855490990996e-06,
+      "loss": 1.1334,
+      "step": 16352
+    },
+    {
+      "epoch": 0.5947194239371568,
+      "grad_norm": 0.0,
+      "learning_rate": 7.446716652534631e-06,
+      "loss": 0.9055,
+      "step": 16353
+    },
+    {
+      "epoch": 0.5947557915408954,
+      "grad_norm": 0.0,
+      "learning_rate": 7.445577849501453e-06,
+      "loss": 0.9824,
+      "step": 16354
+    },
+    {
+      "epoch": 0.594792159144634,
+      "grad_norm": 0.0,
+      "learning_rate": 7.444439081907256e-06,
+      "loss": 0.9982,
+      "step": 16355
+    },
+    {
+      "epoch": 0.5948285267483725,
+      "grad_norm": 0.0,
+      "learning_rate": 7.443300349767844e-06,
+      "loss": 0.9641,
+      "step": 16356
+    },
+    {
+      "epoch": 0.5948648943521111,
+      "grad_norm": 0.0,
+      "learning_rate": 7.442161653099008e-06,
+      "loss": 1.1145,
+      "step": 16357
+    },
+    {
+      "epoch": 0.5949012619558497,
+      "grad_norm": 0.0,
+      "learning_rate": 7.441022991916555e-06,
+      "loss": 0.93,
+      "step": 16358
+    },
+    {
+      "epoch": 0.5949376295595883,
+      "grad_norm": 0.0,
+      "learning_rate": 7.439884366236272e-06,
+      "loss": 0.946,
+      "step": 16359
+    },
+    {
+      "epoch": 0.5949739971633269,
+      "grad_norm": 0.0,
+      "learning_rate": 7.438745776073966e-06,
+      "loss": 0.9767,
+      "step": 16360
+    },
+    {
+      "epoch": 0.5950103647670655,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4376072214454245e-06,
+      "loss": 0.9904,
+      "step": 16361
+    },
+    {
+      "epoch": 0.5950467323708041,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4364687023664485e-06,
+      "loss": 1.0609,
+      "step": 16362
+    },
+    {
+      "epoch": 0.5950830999745427,
+      "grad_norm": 0.0,
+      "learning_rate": 7.435330218852832e-06,
+      "loss": 1.0071,
+      "step": 16363
+    },
+    {
+      "epoch": 0.5951194675782813,
+      "grad_norm": 0.0,
+      "learning_rate": 7.43419177092037e-06,
+      "loss": 1.0532,
+      "step": 16364
+    },
+    {
+      "epoch": 0.5951558351820199,
+      "grad_norm": 0.0,
+      "learning_rate": 7.433053358584861e-06,
+      "loss": 0.9662,
+      "step": 16365
+    },
+    {
+      "epoch": 0.5951922027857585,
+      "grad_norm": 0.0,
+      "learning_rate": 7.431914981862088e-06,
+      "loss": 0.9075,
+      "step": 16366
+    },
+    {
+      "epoch": 0.5952285703894971,
+      "grad_norm": 0.0,
+      "learning_rate": 7.430776640767854e-06,
+      "loss": 1.0229,
+      "step": 16367
+    },
+    {
+      "epoch": 0.5952649379932357,
+      "grad_norm": 0.0,
+      "learning_rate": 7.429638335317948e-06,
+      "loss": 0.9607,
+      "step": 16368
+    },
+    {
+      "epoch": 0.5953013055969743,
+      "grad_norm": 0.0,
+      "learning_rate": 7.428500065528164e-06,
+      "loss": 1.0344,
+      "step": 16369
+    },
+    {
+      "epoch": 0.5953376732007128,
+      "grad_norm": 0.0,
+      "learning_rate": 7.427361831414292e-06,
+      "loss": 0.9526,
+      "step": 16370
+    },
+    {
+      "epoch": 0.5953740408044514,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4262236329921275e-06,
+      "loss": 0.9998,
+      "step": 16371
+    },
+    {
+      "epoch": 0.59541040840819,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4250854702774536e-06,
+      "loss": 1.0869,
+      "step": 16372
+    },
+    {
+      "epoch": 0.5954467760119285,
+      "grad_norm": 0.0,
+      "learning_rate": 7.423947343286071e-06,
+      "loss": 1.0056,
+      "step": 16373
+    },
+    {
+      "epoch": 0.5954831436156671,
+      "grad_norm": 0.0,
+      "learning_rate": 7.422809252033758e-06,
+      "loss": 0.9043,
+      "step": 16374
+    },
+    {
+      "epoch": 0.5955195112194057,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4216711965363156e-06,
+      "loss": 0.7948,
+      "step": 16375
+    },
+    {
+      "epoch": 0.5955558788231443,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4205331768095225e-06,
+      "loss": 1.0181,
+      "step": 16376
+    },
+    {
+      "epoch": 0.5955922464268829,
+      "grad_norm": 0.0,
+      "learning_rate": 7.419395192869174e-06,
+      "loss": 1.0403,
+      "step": 16377
+    },
+    {
+      "epoch": 0.5956286140306215,
+      "grad_norm": 0.0,
+      "learning_rate": 7.418257244731056e-06,
+      "loss": 0.9504,
+      "step": 16378
+    },
+    {
+      "epoch": 0.5956649816343601,
+      "grad_norm": 0.0,
+      "learning_rate": 7.417119332410958e-06,
+      "loss": 0.979,
+      "step": 16379
+    },
+    {
+      "epoch": 0.5957013492380987,
+      "grad_norm": 0.0,
+      "learning_rate": 7.415981455924661e-06,
+      "loss": 0.9226,
+      "step": 16380
+    },
+    {
+      "epoch": 0.5957377168418373,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4148436152879565e-06,
+      "loss": 0.9619,
+      "step": 16381
+    },
+    {
+      "epoch": 0.5957740844455759,
+      "grad_norm": 0.0,
+      "learning_rate": 7.413705810516627e-06,
+      "loss": 0.9731,
+      "step": 16382
+    },
+    {
+      "epoch": 0.5958104520493145,
+      "grad_norm": 0.0,
+      "learning_rate": 7.412568041626463e-06,
+      "loss": 0.9123,
+      "step": 16383
+    },
+    {
+      "epoch": 0.5958468196530531,
+      "grad_norm": 0.0,
+      "learning_rate": 7.411430308633247e-06,
+      "loss": 1.0137,
+      "step": 16384
+    },
+    {
+      "epoch": 0.5958831872567917,
+      "grad_norm": 0.0,
+      "learning_rate": 7.41029261155276e-06,
+      "loss": 0.9825,
+      "step": 16385
+    },
+    {
+      "epoch": 0.5959195548605303,
+      "grad_norm": 0.0,
+      "learning_rate": 7.40915495040079e-06,
+      "loss": 0.9456,
+      "step": 16386
+    },
+    {
+      "epoch": 0.5959559224642689,
+      "grad_norm": 0.0,
+      "learning_rate": 7.408017325193118e-06,
+      "loss": 0.8611,
+      "step": 16387
+    },
+    {
+      "epoch": 0.5959922900680075,
+      "grad_norm": 0.0,
+      "learning_rate": 7.406879735945529e-06,
+      "loss": 1.042,
+      "step": 16388
+    },
+    {
+      "epoch": 0.596028657671746,
+      "grad_norm": 0.0,
+      "learning_rate": 7.405742182673803e-06,
+      "loss": 0.9328,
+      "step": 16389
+    },
+    {
+      "epoch": 0.5960650252754846,
+      "grad_norm": 0.0,
+      "learning_rate": 7.404604665393724e-06,
+      "loss": 1.1376,
+      "step": 16390
+    },
+    {
+      "epoch": 0.5961013928792231,
+      "grad_norm": 0.0,
+      "learning_rate": 7.403467184121072e-06,
+      "loss": 0.9646,
+      "step": 16391
+    },
+    {
+      "epoch": 0.5961377604829617,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4023297388716345e-06,
+      "loss": 0.9534,
+      "step": 16392
+    },
+    {
+      "epoch": 0.5961741280867003,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4011923296611795e-06,
+      "loss": 0.9467,
+      "step": 16393
+    },
+    {
+      "epoch": 0.5962104956904389,
+      "grad_norm": 0.0,
+      "learning_rate": 7.400054956505498e-06,
+      "loss": 0.9156,
+      "step": 16394
+    },
+    {
+      "epoch": 0.5962468632941775,
+      "grad_norm": 0.0,
+      "learning_rate": 7.398917619420361e-06,
+      "loss": 0.9377,
+      "step": 16395
+    },
+    {
+      "epoch": 0.5962832308979161,
+      "grad_norm": 0.0,
+      "learning_rate": 7.397780318421554e-06,
+      "loss": 1.1371,
+      "step": 16396
+    },
+    {
+      "epoch": 0.5963195985016547,
+      "grad_norm": 0.0,
+      "learning_rate": 7.396643053524852e-06,
+      "loss": 0.9244,
+      "step": 16397
+    },
+    {
+      "epoch": 0.5963559661053933,
+      "grad_norm": 0.0,
+      "learning_rate": 7.395505824746038e-06,
+      "loss": 1.0303,
+      "step": 16398
+    },
+    {
+      "epoch": 0.5963923337091319,
+      "grad_norm": 0.0,
+      "learning_rate": 7.394368632100879e-06,
+      "loss": 0.9732,
+      "step": 16399
+    },
+    {
+      "epoch": 0.5964287013128705,
+      "grad_norm": 0.0,
+      "learning_rate": 7.393231475605165e-06,
+      "loss": 1.0245,
+      "step": 16400
+    },
+    {
+      "epoch": 0.5964650689166091,
+      "grad_norm": 0.0,
+      "learning_rate": 7.392094355274661e-06,
+      "loss": 0.9194,
+      "step": 16401
+    },
+    {
+      "epoch": 0.5965014365203477,
+      "grad_norm": 0.0,
+      "learning_rate": 7.39095727112515e-06,
+      "loss": 1.0009,
+      "step": 16402
+    },
+    {
+      "epoch": 0.5965378041240863,
+      "grad_norm": 0.0,
+      "learning_rate": 7.389820223172407e-06,
+      "loss": 0.9206,
+      "step": 16403
+    },
+    {
+      "epoch": 0.5965741717278249,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3886832114322e-06,
+      "loss": 1.0105,
+      "step": 16404
+    },
+    {
+      "epoch": 0.5966105393315635,
+      "grad_norm": 0.0,
+      "learning_rate": 7.387546235920312e-06,
+      "loss": 1.0033,
+      "step": 16405
+    },
+    {
+      "epoch": 0.596646906935302,
+      "grad_norm": 0.0,
+      "learning_rate": 7.38640929665251e-06,
+      "loss": 0.9904,
+      "step": 16406
+    },
+    {
+      "epoch": 0.5966832745390406,
+      "grad_norm": 0.0,
+      "learning_rate": 7.385272393644575e-06,
+      "loss": 0.8113,
+      "step": 16407
+    },
+    {
+      "epoch": 0.5967196421427792,
+      "grad_norm": 0.0,
+      "learning_rate": 7.384135526912272e-06,
+      "loss": 0.9259,
+      "step": 16408
+    },
+    {
+      "epoch": 0.5967560097465178,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3829986964713796e-06,
+      "loss": 0.9087,
+      "step": 16409
+    },
+    {
+      "epoch": 0.5967923773502564,
+      "grad_norm": 0.0,
+      "learning_rate": 7.381861902337665e-06,
+      "loss": 0.9427,
+      "step": 16410
+    },
+    {
+      "epoch": 0.596828744953995,
+      "grad_norm": 0.0,
+      "learning_rate": 7.380725144526905e-06,
+      "loss": 1.0483,
+      "step": 16411
+    },
+    {
+      "epoch": 0.5968651125577336,
+      "grad_norm": 0.0,
+      "learning_rate": 7.379588423054863e-06,
+      "loss": 0.9462,
+      "step": 16412
+    },
+    {
+      "epoch": 0.5969014801614722,
+      "grad_norm": 0.0,
+      "learning_rate": 7.37845173793732e-06,
+      "loss": 0.9863,
+      "step": 16413
+    },
+    {
+      "epoch": 0.5969378477652107,
+      "grad_norm": 0.0,
+      "learning_rate": 7.377315089190033e-06,
+      "loss": 0.8873,
+      "step": 16414
+    },
+    {
+      "epoch": 0.5969742153689493,
+      "grad_norm": 0.0,
+      "learning_rate": 7.376178476828784e-06,
+      "loss": 1.0583,
+      "step": 16415
+    },
+    {
+      "epoch": 0.5970105829726879,
+      "grad_norm": 0.0,
+      "learning_rate": 7.375041900869331e-06,
+      "loss": 1.0359,
+      "step": 16416
+    },
+    {
+      "epoch": 0.5970469505764265,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3739053613274495e-06,
+      "loss": 0.8809,
+      "step": 16417
+    },
+    {
+      "epoch": 0.5970833181801651,
+      "grad_norm": 0.0,
+      "learning_rate": 7.372768858218904e-06,
+      "loss": 1.0258,
+      "step": 16418
+    },
+    {
+      "epoch": 0.5971196857839037,
+      "grad_norm": 0.0,
+      "learning_rate": 7.371632391559467e-06,
+      "loss": 1.0492,
+      "step": 16419
+    },
+    {
+      "epoch": 0.5971560533876423,
+      "grad_norm": 0.0,
+      "learning_rate": 7.370495961364896e-06,
+      "loss": 0.9501,
+      "step": 16420
+    },
+    {
+      "epoch": 0.5971924209913809,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3693595676509665e-06,
+      "loss": 0.8517,
+      "step": 16421
+    },
+    {
+      "epoch": 0.5972287885951195,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3682232104334405e-06,
+      "loss": 0.9425,
+      "step": 16422
+    },
+    {
+      "epoch": 0.597265156198858,
+      "grad_norm": 0.0,
+      "learning_rate": 7.36708688972808e-06,
+      "loss": 0.9956,
+      "step": 16423
+    },
+    {
+      "epoch": 0.5973015238025966,
+      "grad_norm": 0.0,
+      "learning_rate": 7.365950605550658e-06,
+      "loss": 0.9102,
+      "step": 16424
+    },
+    {
+      "epoch": 0.5973378914063352,
+      "grad_norm": 0.0,
+      "learning_rate": 7.364814357916931e-06,
+      "loss": 0.9948,
+      "step": 16425
+    },
+    {
+      "epoch": 0.5973742590100738,
+      "grad_norm": 0.0,
+      "learning_rate": 7.363678146842668e-06,
+      "loss": 0.9115,
+      "step": 16426
+    },
+    {
+      "epoch": 0.5974106266138124,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3625419723436265e-06,
+      "loss": 0.9016,
+      "step": 16427
+    },
+    {
+      "epoch": 0.597446994217551,
+      "grad_norm": 0.0,
+      "learning_rate": 7.361405834435576e-06,
+      "loss": 0.8808,
+      "step": 16428
+    },
+    {
+      "epoch": 0.5974833618212896,
+      "grad_norm": 0.0,
+      "learning_rate": 7.360269733134273e-06,
+      "loss": 0.9105,
+      "step": 16429
+    },
+    {
+      "epoch": 0.5975197294250282,
+      "grad_norm": 0.0,
+      "learning_rate": 7.359133668455488e-06,
+      "loss": 1.0157,
+      "step": 16430
+    },
+    {
+      "epoch": 0.5975560970287668,
+      "grad_norm": 0.0,
+      "learning_rate": 7.357997640414969e-06,
+      "loss": 0.8918,
+      "step": 16431
+    },
+    {
+      "epoch": 0.5975924646325054,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3568616490284916e-06,
+      "loss": 0.9806,
+      "step": 16432
+    },
+    {
+      "epoch": 0.597628832236244,
+      "grad_norm": 0.0,
+      "learning_rate": 7.355725694311804e-06,
+      "loss": 0.801,
+      "step": 16433
+    },
+    {
+      "epoch": 0.5976651998399826,
+      "grad_norm": 0.0,
+      "learning_rate": 7.354589776280672e-06,
+      "loss": 0.937,
+      "step": 16434
+    },
+    {
+      "epoch": 0.5977015674437212,
+      "grad_norm": 0.0,
+      "learning_rate": 7.353453894950853e-06,
+      "loss": 1.0015,
+      "step": 16435
+    },
+    {
+      "epoch": 0.5977379350474598,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3523180503381075e-06,
+      "loss": 0.9662,
+      "step": 16436
+    },
+    {
+      "epoch": 0.5977743026511984,
+      "grad_norm": 0.0,
+      "learning_rate": 7.35118224245819e-06,
+      "loss": 0.9763,
+      "step": 16437
+    },
+    {
+      "epoch": 0.597810670254937,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3500464713268646e-06,
+      "loss": 1.0106,
+      "step": 16438
+    },
+    {
+      "epoch": 0.5978470378586755,
+      "grad_norm": 0.0,
+      "learning_rate": 7.348910736959879e-06,
+      "loss": 1.008,
+      "step": 16439
+    },
+    {
+      "epoch": 0.597883405462414,
+      "grad_norm": 0.0,
+      "learning_rate": 7.347775039373002e-06,
+      "loss": 0.9676,
+      "step": 16440
+    },
+    {
+      "epoch": 0.5979197730661526,
+      "grad_norm": 0.0,
+      "learning_rate": 7.346639378581979e-06,
+      "loss": 1.0154,
+      "step": 16441
+    },
+    {
+      "epoch": 0.5979561406698912,
+      "grad_norm": 0.0,
+      "learning_rate": 7.345503754602572e-06,
+      "loss": 1.0664,
+      "step": 16442
+    },
+    {
+      "epoch": 0.5979925082736298,
+      "grad_norm": 0.0,
+      "learning_rate": 7.344368167450535e-06,
+      "loss": 0.9735,
+      "step": 16443
+    },
+    {
+      "epoch": 0.5980288758773684,
+      "grad_norm": 0.0,
+      "learning_rate": 7.343232617141618e-06,
+      "loss": 1.024,
+      "step": 16444
+    },
+    {
+      "epoch": 0.598065243481107,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3420971036915835e-06,
+      "loss": 0.94,
+      "step": 16445
+    },
+    {
+      "epoch": 0.5981016110848456,
+      "grad_norm": 0.0,
+      "learning_rate": 7.340961627116177e-06,
+      "loss": 1.0571,
+      "step": 16446
+    },
+    {
+      "epoch": 0.5981379786885842,
+      "grad_norm": 0.0,
+      "learning_rate": 7.339826187431156e-06,
+      "loss": 0.9488,
+      "step": 16447
+    },
+    {
+      "epoch": 0.5981743462923228,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3386907846522714e-06,
+      "loss": 0.8225,
+      "step": 16448
+    },
+    {
+      "epoch": 0.5982107138960614,
+      "grad_norm": 0.0,
+      "learning_rate": 7.337555418795279e-06,
+      "loss": 0.9055,
+      "step": 16449
+    },
+    {
+      "epoch": 0.5982470814998,
+      "grad_norm": 0.0,
+      "learning_rate": 7.336420089875925e-06,
+      "loss": 0.8247,
+      "step": 16450
+    },
+    {
+      "epoch": 0.5982834491035386,
+      "grad_norm": 0.0,
+      "learning_rate": 7.335284797909967e-06,
+      "loss": 1.0859,
+      "step": 16451
+    },
+    {
+      "epoch": 0.5983198167072772,
+      "grad_norm": 0.0,
+      "learning_rate": 7.334149542913148e-06,
+      "loss": 1.0001,
+      "step": 16452
+    },
+    {
+      "epoch": 0.5983561843110158,
+      "grad_norm": 0.0,
+      "learning_rate": 7.333014324901224e-06,
+      "loss": 0.9852,
+      "step": 16453
+    },
+    {
+      "epoch": 0.5983925519147544,
+      "grad_norm": 0.0,
+      "learning_rate": 7.33187914388994e-06,
+      "loss": 1.0674,
+      "step": 16454
+    },
+    {
+      "epoch": 0.598428919518493,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3307439998950496e-06,
+      "loss": 0.982,
+      "step": 16455
+    },
+    {
+      "epoch": 0.5984652871222316,
+      "grad_norm": 0.0,
+      "learning_rate": 7.329608892932296e-06,
+      "loss": 0.9594,
+      "step": 16456
+    },
+    {
+      "epoch": 0.5985016547259701,
+      "grad_norm": 0.0,
+      "learning_rate": 7.328473823017434e-06,
+      "loss": 1.0451,
+      "step": 16457
+    },
+    {
+      "epoch": 0.5985380223297087,
+      "grad_norm": 0.0,
+      "learning_rate": 7.327338790166203e-06,
+      "loss": 0.986,
+      "step": 16458
+    },
+    {
+      "epoch": 0.5985743899334472,
+      "grad_norm": 0.0,
+      "learning_rate": 7.32620379439436e-06,
+      "loss": 0.8688,
+      "step": 16459
+    },
+    {
+      "epoch": 0.5986107575371858,
+      "grad_norm": 0.0,
+      "learning_rate": 7.325068835717641e-06,
+      "loss": 0.9327,
+      "step": 16460
+    },
+    {
+      "epoch": 0.5986471251409244,
+      "grad_norm": 0.0,
+      "learning_rate": 7.323933914151799e-06,
+      "loss": 0.8878,
+      "step": 16461
+    },
+    {
+      "epoch": 0.598683492744663,
+      "grad_norm": 0.0,
+      "learning_rate": 7.322799029712578e-06,
+      "loss": 1.0112,
+      "step": 16462
+    },
+    {
+      "epoch": 0.5987198603484016,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3216641824157174e-06,
+      "loss": 1.0233,
+      "step": 16463
+    },
+    {
+      "epoch": 0.5987562279521402,
+      "grad_norm": 0.0,
+      "learning_rate": 7.320529372276972e-06,
+      "loss": 0.9792,
+      "step": 16464
+    },
+    {
+      "epoch": 0.5987925955558788,
+      "grad_norm": 0.0,
+      "learning_rate": 7.319394599312075e-06,
+      "loss": 0.8668,
+      "step": 16465
+    },
+    {
+      "epoch": 0.5988289631596174,
+      "grad_norm": 0.0,
+      "learning_rate": 7.318259863536779e-06,
+      "loss": 0.9413,
+      "step": 16466
+    },
+    {
+      "epoch": 0.598865330763356,
+      "grad_norm": 0.0,
+      "learning_rate": 7.317125164966819e-06,
+      "loss": 0.9603,
+      "step": 16467
+    },
+    {
+      "epoch": 0.5989016983670946,
+      "grad_norm": 0.0,
+      "learning_rate": 7.315990503617941e-06,
+      "loss": 1.0975,
+      "step": 16468
+    },
+    {
+      "epoch": 0.5989380659708332,
+      "grad_norm": 0.0,
+      "learning_rate": 7.314855879505887e-06,
+      "loss": 0.8456,
+      "step": 16469
+    },
+    {
+      "epoch": 0.5989744335745718,
+      "grad_norm": 0.0,
+      "learning_rate": 7.313721292646401e-06,
+      "loss": 1.1313,
+      "step": 16470
+    },
+    {
+      "epoch": 0.5990108011783104,
+      "grad_norm": 0.0,
+      "learning_rate": 7.312586743055215e-06,
+      "loss": 0.9337,
+      "step": 16471
+    },
+    {
+      "epoch": 0.599047168782049,
+      "grad_norm": 0.0,
+      "learning_rate": 7.311452230748082e-06,
+      "loss": 0.9279,
+      "step": 16472
+    },
+    {
+      "epoch": 0.5990835363857875,
+      "grad_norm": 0.0,
+      "learning_rate": 7.310317755740728e-06,
+      "loss": 0.8192,
+      "step": 16473
+    },
+    {
+      "epoch": 0.5991199039895261,
+      "grad_norm": 0.0,
+      "learning_rate": 7.309183318048903e-06,
+      "loss": 0.969,
+      "step": 16474
+    },
+    {
+      "epoch": 0.5991562715932647,
+      "grad_norm": 0.0,
+      "learning_rate": 7.308048917688339e-06,
+      "loss": 0.9361,
+      "step": 16475
+    },
+    {
+      "epoch": 0.5991926391970033,
+      "grad_norm": 0.0,
+      "learning_rate": 7.30691455467478e-06,
+      "loss": 0.8763,
+      "step": 16476
+    },
+    {
+      "epoch": 0.5992290068007419,
+      "grad_norm": 0.0,
+      "learning_rate": 7.305780229023958e-06,
+      "loss": 0.9554,
+      "step": 16477
+    },
+    {
+      "epoch": 0.5992653744044805,
+      "grad_norm": 0.0,
+      "learning_rate": 7.304645940751616e-06,
+      "loss": 0.8914,
+      "step": 16478
+    },
+    {
+      "epoch": 0.5993017420082191,
+      "grad_norm": 0.0,
+      "learning_rate": 7.303511689873484e-06,
+      "loss": 0.8941,
+      "step": 16479
+    },
+    {
+      "epoch": 0.5993381096119577,
+      "grad_norm": 0.0,
+      "learning_rate": 7.302377476405303e-06,
+      "loss": 0.9435,
+      "step": 16480
+    },
+    {
+      "epoch": 0.5993744772156963,
+      "grad_norm": 0.0,
+      "learning_rate": 7.301243300362808e-06,
+      "loss": 0.9859,
+      "step": 16481
+    },
+    {
+      "epoch": 0.5994108448194349,
+      "grad_norm": 0.0,
+      "learning_rate": 7.30010916176173e-06,
+      "loss": 1.0546,
+      "step": 16482
+    },
+    {
+      "epoch": 0.5994472124231734,
+      "grad_norm": 0.0,
+      "learning_rate": 7.298975060617812e-06,
+      "loss": 0.9632,
+      "step": 16483
+    },
+    {
+      "epoch": 0.599483580026912,
+      "grad_norm": 0.0,
+      "learning_rate": 7.297840996946777e-06,
+      "loss": 1.0512,
+      "step": 16484
+    },
+    {
+      "epoch": 0.5995199476306506,
+      "grad_norm": 0.0,
+      "learning_rate": 7.296706970764369e-06,
+      "loss": 0.9389,
+      "step": 16485
+    },
+    {
+      "epoch": 0.5995563152343892,
+      "grad_norm": 0.0,
+      "learning_rate": 7.295572982086313e-06,
+      "loss": 1.0408,
+      "step": 16486
+    },
+    {
+      "epoch": 0.5995926828381278,
+      "grad_norm": 0.0,
+      "learning_rate": 7.294439030928347e-06,
+      "loss": 0.921,
+      "step": 16487
+    },
+    {
+      "epoch": 0.5996290504418664,
+      "grad_norm": 0.0,
+      "learning_rate": 7.293305117306197e-06,
+      "loss": 0.9967,
+      "step": 16488
+    },
+    {
+      "epoch": 0.599665418045605,
+      "grad_norm": 0.0,
+      "learning_rate": 7.292171241235603e-06,
+      "loss": 0.9841,
+      "step": 16489
+    },
+    {
+      "epoch": 0.5997017856493435,
+      "grad_norm": 0.0,
+      "learning_rate": 7.291037402732287e-06,
+      "loss": 0.9902,
+      "step": 16490
+    },
+    {
+      "epoch": 0.5997381532530821,
+      "grad_norm": 0.0,
+      "learning_rate": 7.289903601811989e-06,
+      "loss": 1.0488,
+      "step": 16491
+    },
+    {
+      "epoch": 0.5997745208568207,
+      "grad_norm": 0.0,
+      "learning_rate": 7.288769838490428e-06,
+      "loss": 1.0349,
+      "step": 16492
+    },
+    {
+      "epoch": 0.5998108884605593,
+      "grad_norm": 0.0,
+      "learning_rate": 7.28763611278334e-06,
+      "loss": 0.9378,
+      "step": 16493
+    },
+    {
+      "epoch": 0.5998472560642979,
+      "grad_norm": 0.0,
+      "learning_rate": 7.286502424706453e-06,
+      "loss": 1.0233,
+      "step": 16494
+    },
+    {
+      "epoch": 0.5998836236680365,
+      "grad_norm": 0.0,
+      "learning_rate": 7.285368774275497e-06,
+      "loss": 1.008,
+      "step": 16495
+    },
+    {
+      "epoch": 0.5999199912717751,
+      "grad_norm": 0.0,
+      "learning_rate": 7.284235161506193e-06,
+      "loss": 0.7517,
+      "step": 16496
+    },
+    {
+      "epoch": 0.5999563588755137,
+      "grad_norm": 0.0,
+      "learning_rate": 7.28310158641428e-06,
+      "loss": 0.8589,
+      "step": 16497
+    },
+    {
+      "epoch": 0.5999927264792523,
+      "grad_norm": 0.0,
+      "learning_rate": 7.281968049015469e-06,
+      "loss": 0.9493,
+      "step": 16498
+    },
+    {
+      "epoch": 0.6000290940829909,
+      "grad_norm": 0.0,
+      "learning_rate": 7.280834549325503e-06,
+      "loss": 0.9795,
+      "step": 16499
+    },
+    {
+      "epoch": 0.6000654616867295,
+      "grad_norm": 0.0,
+      "learning_rate": 7.279701087360096e-06,
+      "loss": 0.9443,
+      "step": 16500
+    },
+    {
+      "epoch": 0.6001018292904681,
+      "grad_norm": 0.0,
+      "learning_rate": 7.278567663134977e-06,
+      "loss": 0.911,
+      "step": 16501
+    },
+    {
+      "epoch": 0.6001381968942067,
+      "grad_norm": 0.0,
+      "learning_rate": 7.277434276665872e-06,
+      "loss": 1.0467,
+      "step": 16502
+    },
+    {
+      "epoch": 0.6001745644979453,
+      "grad_norm": 0.0,
+      "learning_rate": 7.276300927968498e-06,
+      "loss": 1.0951,
+      "step": 16503
+    },
+    {
+      "epoch": 0.6002109321016839,
+      "grad_norm": 0.0,
+      "learning_rate": 7.275167617058592e-06,
+      "loss": 1.2076,
+      "step": 16504
+    },
+    {
+      "epoch": 0.6002472997054225,
+      "grad_norm": 0.0,
+      "learning_rate": 7.274034343951863e-06,
+      "loss": 1.0175,
+      "step": 16505
+    },
+    {
+      "epoch": 0.600283667309161,
+      "grad_norm": 0.0,
+      "learning_rate": 7.272901108664045e-06,
+      "loss": 0.9607,
+      "step": 16506
+    },
+    {
+      "epoch": 0.6003200349128995,
+      "grad_norm": 0.0,
+      "learning_rate": 7.27176791121085e-06,
+      "loss": 0.9711,
+      "step": 16507
+    },
+    {
+      "epoch": 0.6003564025166381,
+      "grad_norm": 0.0,
+      "learning_rate": 7.270634751608008e-06,
+      "loss": 0.9171,
+      "step": 16508
+    },
+    {
+      "epoch": 0.6003927701203767,
+      "grad_norm": 0.0,
+      "learning_rate": 7.269501629871233e-06,
+      "loss": 0.9392,
+      "step": 16509
+    },
+    {
+      "epoch": 0.6004291377241153,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2683685460162535e-06,
+      "loss": 1.0152,
+      "step": 16510
+    },
+    {
+      "epoch": 0.6004655053278539,
+      "grad_norm": 0.0,
+      "learning_rate": 7.267235500058778e-06,
+      "loss": 1.0173,
+      "step": 16511
+    },
+    {
+      "epoch": 0.6005018729315925,
+      "grad_norm": 0.0,
+      "learning_rate": 7.266102492014539e-06,
+      "loss": 1.007,
+      "step": 16512
+    },
+    {
+      "epoch": 0.6005382405353311,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2649695218992455e-06,
+      "loss": 0.9175,
+      "step": 16513
+    },
+    {
+      "epoch": 0.6005746081390697,
+      "grad_norm": 0.0,
+      "learning_rate": 7.263836589728622e-06,
+      "loss": 1.0273,
+      "step": 16514
+    },
+    {
+      "epoch": 0.6006109757428083,
+      "grad_norm": 0.0,
+      "learning_rate": 7.26270369551838e-06,
+      "loss": 0.9637,
+      "step": 16515
+    },
+    {
+      "epoch": 0.6006473433465469,
+      "grad_norm": 0.0,
+      "learning_rate": 7.261570839284245e-06,
+      "loss": 0.9392,
+      "step": 16516
+    },
+    {
+      "epoch": 0.6006837109502855,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2604380210419245e-06,
+      "loss": 0.8255,
+      "step": 16517
+    },
+    {
+      "epoch": 0.6007200785540241,
+      "grad_norm": 0.0,
+      "learning_rate": 7.259305240807147e-06,
+      "loss": 0.9055,
+      "step": 16518
+    },
+    {
+      "epoch": 0.6007564461577627,
+      "grad_norm": 0.0,
+      "learning_rate": 7.258172498595615e-06,
+      "loss": 0.7814,
+      "step": 16519
+    },
+    {
+      "epoch": 0.6007928137615013,
+      "grad_norm": 0.0,
+      "learning_rate": 7.257039794423048e-06,
+      "loss": 0.9412,
+      "step": 16520
+    },
+    {
+      "epoch": 0.6008291813652399,
+      "grad_norm": 0.0,
+      "learning_rate": 7.255907128305166e-06,
+      "loss": 1.056,
+      "step": 16521
+    },
+    {
+      "epoch": 0.6008655489689785,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2547745002576765e-06,
+      "loss": 0.9591,
+      "step": 16522
+    },
+    {
+      "epoch": 0.6009019165727171,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2536419102963e-06,
+      "loss": 1.0364,
+      "step": 16523
+    },
+    {
+      "epoch": 0.6009382841764556,
+      "grad_norm": 0.0,
+      "learning_rate": 7.252509358436741e-06,
+      "loss": 1.02,
+      "step": 16524
+    },
+    {
+      "epoch": 0.6009746517801942,
+      "grad_norm": 0.0,
+      "learning_rate": 7.251376844694718e-06,
+      "loss": 0.9526,
+      "step": 16525
+    },
+    {
+      "epoch": 0.6010110193839328,
+      "grad_norm": 0.0,
+      "learning_rate": 7.25024436908594e-06,
+      "loss": 1.0199,
+      "step": 16526
+    },
+    {
+      "epoch": 0.6010473869876714,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2491119316261225e-06,
+      "loss": 0.9969,
+      "step": 16527
+    },
+    {
+      "epoch": 0.60108375459141,
+      "grad_norm": 0.0,
+      "learning_rate": 7.247979532330971e-06,
+      "loss": 0.871,
+      "step": 16528
+    },
+    {
+      "epoch": 0.6011201221951485,
+      "grad_norm": 0.0,
+      "learning_rate": 7.246847171216203e-06,
+      "loss": 1.0291,
+      "step": 16529
+    },
+    {
+      "epoch": 0.6011564897988871,
+      "grad_norm": 0.0,
+      "learning_rate": 7.245714848297519e-06,
+      "loss": 0.8871,
+      "step": 16530
+    },
+    {
+      "epoch": 0.6011928574026257,
+      "grad_norm": 0.0,
+      "learning_rate": 7.24458256359064e-06,
+      "loss": 0.8916,
+      "step": 16531
+    },
+    {
+      "epoch": 0.6012292250063643,
+      "grad_norm": 0.0,
+      "learning_rate": 7.243450317111262e-06,
+      "loss": 1.0087,
+      "step": 16532
+    },
+    {
+      "epoch": 0.6012655926101029,
+      "grad_norm": 0.0,
+      "learning_rate": 7.242318108875105e-06,
+      "loss": 1.0549,
+      "step": 16533
+    },
+    {
+      "epoch": 0.6013019602138415,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2411859388978675e-06,
+      "loss": 0.8406,
+      "step": 16534
+    },
+    {
+      "epoch": 0.6013383278175801,
+      "grad_norm": 0.0,
+      "learning_rate": 7.240053807195264e-06,
+      "loss": 1.0412,
+      "step": 16535
+    },
+    {
+      "epoch": 0.6013746954213187,
+      "grad_norm": 0.0,
+      "learning_rate": 7.238921713782996e-06,
+      "loss": 0.9884,
+      "step": 16536
+    },
+    {
+      "epoch": 0.6014110630250573,
+      "grad_norm": 0.0,
+      "learning_rate": 7.237789658676775e-06,
+      "loss": 1.0189,
+      "step": 16537
+    },
+    {
+      "epoch": 0.6014474306287959,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2366576418923014e-06,
+      "loss": 0.9264,
+      "step": 16538
+    },
+    {
+      "epoch": 0.6014837982325345,
+      "grad_norm": 0.0,
+      "learning_rate": 7.235525663445282e-06,
+      "loss": 0.9997,
+      "step": 16539
+    },
+    {
+      "epoch": 0.6015201658362731,
+      "grad_norm": 0.0,
+      "learning_rate": 7.234393723351423e-06,
+      "loss": 0.876,
+      "step": 16540
+    },
+    {
+      "epoch": 0.6015565334400116,
+      "grad_norm": 0.0,
+      "learning_rate": 7.233261821626424e-06,
+      "loss": 0.8985,
+      "step": 16541
+    },
+    {
+      "epoch": 0.6015929010437502,
+      "grad_norm": 0.0,
+      "learning_rate": 7.232129958285996e-06,
+      "loss": 0.9716,
+      "step": 16542
+    },
+    {
+      "epoch": 0.6016292686474888,
+      "grad_norm": 0.0,
+      "learning_rate": 7.230998133345832e-06,
+      "loss": 0.8457,
+      "step": 16543
+    },
+    {
+      "epoch": 0.6016656362512274,
+      "grad_norm": 0.0,
+      "learning_rate": 7.229866346821647e-06,
+      "loss": 1.0791,
+      "step": 16544
+    },
+    {
+      "epoch": 0.601702003854966,
+      "grad_norm": 0.0,
+      "learning_rate": 7.228734598729131e-06,
+      "loss": 1.084,
+      "step": 16545
+    },
+    {
+      "epoch": 0.6017383714587046,
+      "grad_norm": 0.0,
+      "learning_rate": 7.227602889083993e-06,
+      "loss": 1.09,
+      "step": 16546
+    },
+    {
+      "epoch": 0.6017747390624432,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2264712179019294e-06,
+      "loss": 1.0151,
+      "step": 16547
+    },
+    {
+      "epoch": 0.6018111066661818,
+      "grad_norm": 0.0,
+      "learning_rate": 7.225339585198645e-06,
+      "loss": 0.9238,
+      "step": 16548
+    },
+    {
+      "epoch": 0.6018474742699204,
+      "grad_norm": 0.0,
+      "learning_rate": 7.224207990989836e-06,
+      "loss": 0.9176,
+      "step": 16549
+    },
+    {
+      "epoch": 0.601883841873659,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2230764352912045e-06,
+      "loss": 0.9725,
+      "step": 16550
+    },
+    {
+      "epoch": 0.6019202094773975,
+      "grad_norm": 0.0,
+      "learning_rate": 7.221944918118444e-06,
+      "loss": 0.906,
+      "step": 16551
+    },
+    {
+      "epoch": 0.6019565770811361,
+      "grad_norm": 0.0,
+      "learning_rate": 7.22081343948726e-06,
+      "loss": 1.037,
+      "step": 16552
+    },
+    {
+      "epoch": 0.6019929446848747,
+      "grad_norm": 0.0,
+      "learning_rate": 7.219681999413341e-06,
+      "loss": 1.097,
+      "step": 16553
+    },
+    {
+      "epoch": 0.6020293122886133,
+      "grad_norm": 0.0,
+      "learning_rate": 7.218550597912395e-06,
+      "loss": 0.9065,
+      "step": 16554
+    },
+    {
+      "epoch": 0.6020656798923519,
+      "grad_norm": 0.0,
+      "learning_rate": 7.217419235000109e-06,
+      "loss": 0.9868,
+      "step": 16555
+    },
+    {
+      "epoch": 0.6021020474960905,
+      "grad_norm": 0.0,
+      "learning_rate": 7.216287910692185e-06,
+      "loss": 1.0113,
+      "step": 16556
+    },
+    {
+      "epoch": 0.6021384150998291,
+      "grad_norm": 0.0,
+      "learning_rate": 7.21515662500432e-06,
+      "loss": 0.8606,
+      "step": 16557
+    },
+    {
+      "epoch": 0.6021747827035676,
+      "grad_norm": 0.0,
+      "learning_rate": 7.214025377952203e-06,
+      "loss": 1.034,
+      "step": 16558
+    },
+    {
+      "epoch": 0.6022111503073062,
+      "grad_norm": 0.0,
+      "learning_rate": 7.21289416955153e-06,
+      "loss": 0.8999,
+      "step": 16559
+    },
+    {
+      "epoch": 0.6022475179110448,
+      "grad_norm": 0.0,
+      "learning_rate": 7.211762999817995e-06,
+      "loss": 0.9119,
+      "step": 16560
+    },
+    {
+      "epoch": 0.6022838855147834,
+      "grad_norm": 0.0,
+      "learning_rate": 7.210631868767295e-06,
+      "loss": 1.0206,
+      "step": 16561
+    },
+    {
+      "epoch": 0.602320253118522,
+      "grad_norm": 0.0,
+      "learning_rate": 7.209500776415118e-06,
+      "loss": 0.9762,
+      "step": 16562
+    },
+    {
+      "epoch": 0.6023566207222606,
+      "grad_norm": 0.0,
+      "learning_rate": 7.208369722777163e-06,
+      "loss": 0.9807,
+      "step": 16563
+    },
+    {
+      "epoch": 0.6023929883259992,
+      "grad_norm": 0.0,
+      "learning_rate": 7.207238707869113e-06,
+      "loss": 1.0636,
+      "step": 16564
+    },
+    {
+      "epoch": 0.6024293559297378,
+      "grad_norm": 0.0,
+      "learning_rate": 7.206107731706664e-06,
+      "loss": 1.0349,
+      "step": 16565
+    },
+    {
+      "epoch": 0.6024657235334764,
+      "grad_norm": 0.0,
+      "learning_rate": 7.204976794305506e-06,
+      "loss": 1.1228,
+      "step": 16566
+    },
+    {
+      "epoch": 0.602502091137215,
+      "grad_norm": 0.0,
+      "learning_rate": 7.203845895681332e-06,
+      "loss": 0.8658,
+      "step": 16567
+    },
+    {
+      "epoch": 0.6025384587409536,
+      "grad_norm": 0.0,
+      "learning_rate": 7.202715035849826e-06,
+      "loss": 1.0057,
+      "step": 16568
+    },
+    {
+      "epoch": 0.6025748263446922,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2015842148266825e-06,
+      "loss": 1.0839,
+      "step": 16569
+    },
+    {
+      "epoch": 0.6026111939484308,
+      "grad_norm": 0.0,
+      "learning_rate": 7.200453432627583e-06,
+      "loss": 0.9475,
+      "step": 16570
+    },
+    {
+      "epoch": 0.6026475615521694,
+      "grad_norm": 0.0,
+      "learning_rate": 7.199322689268227e-06,
+      "loss": 0.9713,
+      "step": 16571
+    },
+    {
+      "epoch": 0.602683929155908,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1981919847642885e-06,
+      "loss": 0.8948,
+      "step": 16572
+    },
+    {
+      "epoch": 0.6027202967596466,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1970613191314655e-06,
+      "loss": 1.0072,
+      "step": 16573
+    },
+    {
+      "epoch": 0.602756664363385,
+      "grad_norm": 0.0,
+      "learning_rate": 7.195930692385436e-06,
+      "loss": 0.9229,
+      "step": 16574
+    },
+    {
+      "epoch": 0.6027930319671236,
+      "grad_norm": 0.0,
+      "learning_rate": 7.194800104541892e-06,
+      "loss": 0.9053,
+      "step": 16575
+    },
+    {
+      "epoch": 0.6028293995708622,
+      "grad_norm": 0.0,
+      "learning_rate": 7.193669555616521e-06,
+      "loss": 0.9561,
+      "step": 16576
+    },
+    {
+      "epoch": 0.6028657671746008,
+      "grad_norm": 0.0,
+      "learning_rate": 7.192539045624996e-06,
+      "loss": 1.0524,
+      "step": 16577
+    },
+    {
+      "epoch": 0.6029021347783394,
+      "grad_norm": 0.0,
+      "learning_rate": 7.191408574583012e-06,
+      "loss": 0.9796,
+      "step": 16578
+    },
+    {
+      "epoch": 0.602938502382078,
+      "grad_norm": 0.0,
+      "learning_rate": 7.190278142506247e-06,
+      "loss": 0.9854,
+      "step": 16579
+    },
+    {
+      "epoch": 0.6029748699858166,
+      "grad_norm": 0.0,
+      "learning_rate": 7.189147749410388e-06,
+      "loss": 0.9071,
+      "step": 16580
+    },
+    {
+      "epoch": 0.6030112375895552,
+      "grad_norm": 0.0,
+      "learning_rate": 7.188017395311115e-06,
+      "loss": 0.9237,
+      "step": 16581
+    },
+    {
+      "epoch": 0.6030476051932938,
+      "grad_norm": 0.0,
+      "learning_rate": 7.186887080224114e-06,
+      "loss": 0.9577,
+      "step": 16582
+    },
+    {
+      "epoch": 0.6030839727970324,
+      "grad_norm": 0.0,
+      "learning_rate": 7.185756804165057e-06,
+      "loss": 1.0352,
+      "step": 16583
+    },
+    {
+      "epoch": 0.603120340400771,
+      "grad_norm": 0.0,
+      "learning_rate": 7.184626567149639e-06,
+      "loss": 0.9959,
+      "step": 16584
+    },
+    {
+      "epoch": 0.6031567080045096,
+      "grad_norm": 0.0,
+      "learning_rate": 7.183496369193528e-06,
+      "loss": 1.0485,
+      "step": 16585
+    },
+    {
+      "epoch": 0.6031930756082482,
+      "grad_norm": 0.0,
+      "learning_rate": 7.18236621031241e-06,
+      "loss": 0.941,
+      "step": 16586
+    },
+    {
+      "epoch": 0.6032294432119868,
+      "grad_norm": 0.0,
+      "learning_rate": 7.181236090521961e-06,
+      "loss": 0.9931,
+      "step": 16587
+    },
+    {
+      "epoch": 0.6032658108157254,
+      "grad_norm": 0.0,
+      "learning_rate": 7.180106009837866e-06,
+      "loss": 0.9346,
+      "step": 16588
+    },
+    {
+      "epoch": 0.603302178419464,
+      "grad_norm": 0.0,
+      "learning_rate": 7.178975968275795e-06,
+      "loss": 1.0065,
+      "step": 16589
+    },
+    {
+      "epoch": 0.6033385460232026,
+      "grad_norm": 0.0,
+      "learning_rate": 7.177845965851436e-06,
+      "loss": 0.9859,
+      "step": 16590
+    },
+    {
+      "epoch": 0.6033749136269411,
+      "grad_norm": 0.0,
+      "learning_rate": 7.176716002580455e-06,
+      "loss": 0.8549,
+      "step": 16591
+    },
+    {
+      "epoch": 0.6034112812306797,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1755860784785356e-06,
+      "loss": 0.9977,
+      "step": 16592
+    },
+    {
+      "epoch": 0.6034476488344183,
+      "grad_norm": 0.0,
+      "learning_rate": 7.17445619356135e-06,
+      "loss": 0.954,
+      "step": 16593
+    },
+    {
+      "epoch": 0.6034840164381569,
+      "grad_norm": 0.0,
+      "learning_rate": 7.173326347844578e-06,
+      "loss": 0.874,
+      "step": 16594
+    },
+    {
+      "epoch": 0.6035203840418955,
+      "grad_norm": 0.0,
+      "learning_rate": 7.172196541343894e-06,
+      "loss": 1.0375,
+      "step": 16595
+    },
+    {
+      "epoch": 0.603556751645634,
+      "grad_norm": 0.0,
+      "learning_rate": 7.171066774074967e-06,
+      "loss": 0.9748,
+      "step": 16596
+    },
+    {
+      "epoch": 0.6035931192493726,
+      "grad_norm": 0.0,
+      "learning_rate": 7.169937046053478e-06,
+      "loss": 0.8589,
+      "step": 16597
+    },
+    {
+      "epoch": 0.6036294868531112,
+      "grad_norm": 0.0,
+      "learning_rate": 7.168807357295094e-06,
+      "loss": 0.9745,
+      "step": 16598
+    },
+    {
+      "epoch": 0.6036658544568498,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1676777078154925e-06,
+      "loss": 0.8445,
+      "step": 16599
+    },
+    {
+      "epoch": 0.6037022220605884,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1665480976303435e-06,
+      "loss": 1.0534,
+      "step": 16600
+    },
+    {
+      "epoch": 0.603738589664327,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1654185267553235e-06,
+      "loss": 1.0176,
+      "step": 16601
+    },
+    {
+      "epoch": 0.6037749572680656,
+      "grad_norm": 0.0,
+      "learning_rate": 7.164288995206093e-06,
+      "loss": 1.0433,
+      "step": 16602
+    },
+    {
+      "epoch": 0.6038113248718042,
+      "grad_norm": 0.0,
+      "learning_rate": 7.163159502998334e-06,
+      "loss": 1.0319,
+      "step": 16603
+    },
+    {
+      "epoch": 0.6038476924755428,
+      "grad_norm": 0.0,
+      "learning_rate": 7.16203005014771e-06,
+      "loss": 0.8896,
+      "step": 16604
+    },
+    {
+      "epoch": 0.6038840600792814,
+      "grad_norm": 0.0,
+      "learning_rate": 7.160900636669894e-06,
+      "loss": 0.9783,
+      "step": 16605
+    },
+    {
+      "epoch": 0.60392042768302,
+      "grad_norm": 0.0,
+      "learning_rate": 7.159771262580552e-06,
+      "loss": 0.9562,
+      "step": 16606
+    },
+    {
+      "epoch": 0.6039567952867586,
+      "grad_norm": 0.0,
+      "learning_rate": 7.158641927895356e-06,
+      "loss": 0.9964,
+      "step": 16607
+    },
+    {
+      "epoch": 0.6039931628904971,
+      "grad_norm": 0.0,
+      "learning_rate": 7.157512632629968e-06,
+      "loss": 0.8944,
+      "step": 16608
+    },
+    {
+      "epoch": 0.6040295304942357,
+      "grad_norm": 0.0,
+      "learning_rate": 7.156383376800065e-06,
+      "loss": 0.9896,
+      "step": 16609
+    },
+    {
+      "epoch": 0.6040658980979743,
+      "grad_norm": 0.0,
+      "learning_rate": 7.155254160421301e-06,
+      "loss": 0.9737,
+      "step": 16610
+    },
+    {
+      "epoch": 0.6041022657017129,
+      "grad_norm": 0.0,
+      "learning_rate": 7.154124983509356e-06,
+      "loss": 1.1167,
+      "step": 16611
+    },
+    {
+      "epoch": 0.6041386333054515,
+      "grad_norm": 0.0,
+      "learning_rate": 7.152995846079886e-06,
+      "loss": 0.9841,
+      "step": 16612
+    },
+    {
+      "epoch": 0.6041750009091901,
+      "grad_norm": 0.0,
+      "learning_rate": 7.15186674814856e-06,
+      "loss": 1.0182,
+      "step": 16613
+    },
+    {
+      "epoch": 0.6042113685129287,
+      "grad_norm": 0.0,
+      "learning_rate": 7.150737689731044e-06,
+      "loss": 1.0296,
+      "step": 16614
+    },
+    {
+      "epoch": 0.6042477361166673,
+      "grad_norm": 0.0,
+      "learning_rate": 7.149608670842995e-06,
+      "loss": 0.8087,
+      "step": 16615
+    },
+    {
+      "epoch": 0.6042841037204059,
+      "grad_norm": 0.0,
+      "learning_rate": 7.148479691500086e-06,
+      "loss": 0.9269,
+      "step": 16616
+    },
+    {
+      "epoch": 0.6043204713241445,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1473507517179715e-06,
+      "loss": 1.015,
+      "step": 16617
+    },
+    {
+      "epoch": 0.604356838927883,
+      "grad_norm": 0.0,
+      "learning_rate": 7.146221851512322e-06,
+      "loss": 1.0189,
+      "step": 16618
+    },
+    {
+      "epoch": 0.6043932065316217,
+      "grad_norm": 0.0,
+      "learning_rate": 7.145092990898792e-06,
+      "loss": 0.9582,
+      "step": 16619
+    },
+    {
+      "epoch": 0.6044295741353602,
+      "grad_norm": 0.0,
+      "learning_rate": 7.143964169893048e-06,
+      "loss": 0.9402,
+      "step": 16620
+    },
+    {
+      "epoch": 0.6044659417390988,
+      "grad_norm": 0.0,
+      "learning_rate": 7.142835388510747e-06,
+      "loss": 0.917,
+      "step": 16621
+    },
+    {
+      "epoch": 0.6045023093428374,
+      "grad_norm": 0.0,
+      "learning_rate": 7.141706646767555e-06,
+      "loss": 0.9682,
+      "step": 16622
+    },
+    {
+      "epoch": 0.604538676946576,
+      "grad_norm": 0.0,
+      "learning_rate": 7.140577944679124e-06,
+      "loss": 1.1154,
+      "step": 16623
+    },
+    {
+      "epoch": 0.6045750445503146,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1394492822611195e-06,
+      "loss": 0.9847,
+      "step": 16624
+    },
+    {
+      "epoch": 0.6046114121540531,
+      "grad_norm": 0.0,
+      "learning_rate": 7.138320659529195e-06,
+      "loss": 0.8818,
+      "step": 16625
+    },
+    {
+      "epoch": 0.6046477797577917,
+      "grad_norm": 0.0,
+      "learning_rate": 7.137192076499015e-06,
+      "loss": 0.8854,
+      "step": 16626
+    },
+    {
+      "epoch": 0.6046841473615303,
+      "grad_norm": 0.0,
+      "learning_rate": 7.136063533186229e-06,
+      "loss": 0.9001,
+      "step": 16627
+    },
+    {
+      "epoch": 0.6047205149652689,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1349350296065e-06,
+      "loss": 1.0005,
+      "step": 16628
+    },
+    {
+      "epoch": 0.6047568825690075,
+      "grad_norm": 0.0,
+      "learning_rate": 7.133806565775482e-06,
+      "loss": 0.9725,
+      "step": 16629
+    },
+    {
+      "epoch": 0.6047932501727461,
+      "grad_norm": 0.0,
+      "learning_rate": 7.132678141708834e-06,
+      "loss": 0.9645,
+      "step": 16630
+    },
+    {
+      "epoch": 0.6048296177764847,
+      "grad_norm": 0.0,
+      "learning_rate": 7.131549757422205e-06,
+      "loss": 0.9736,
+      "step": 16631
+    },
+    {
+      "epoch": 0.6048659853802233,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1304214129312565e-06,
+      "loss": 1.0295,
+      "step": 16632
+    },
+    {
+      "epoch": 0.6049023529839619,
+      "grad_norm": 0.0,
+      "learning_rate": 7.129293108251638e-06,
+      "loss": 1.1151,
+      "step": 16633
+    },
+    {
+      "epoch": 0.6049387205877005,
+      "grad_norm": 0.0,
+      "learning_rate": 7.128164843399004e-06,
+      "loss": 0.9572,
+      "step": 16634
+    },
+    {
+      "epoch": 0.6049750881914391,
+      "grad_norm": 0.0,
+      "learning_rate": 7.12703661838901e-06,
+      "loss": 1.0704,
+      "step": 16635
+    },
+    {
+      "epoch": 0.6050114557951777,
+      "grad_norm": 0.0,
+      "learning_rate": 7.125908433237304e-06,
+      "loss": 0.9311,
+      "step": 16636
+    },
+    {
+      "epoch": 0.6050478233989163,
+      "grad_norm": 0.0,
+      "learning_rate": 7.124780287959543e-06,
+      "loss": 1.0121,
+      "step": 16637
+    },
+    {
+      "epoch": 0.6050841910026549,
+      "grad_norm": 0.0,
+      "learning_rate": 7.123652182571373e-06,
+      "loss": 0.8913,
+      "step": 16638
+    },
+    {
+      "epoch": 0.6051205586063935,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1225241170884515e-06,
+      "loss": 0.9074,
+      "step": 16639
+    },
+    {
+      "epoch": 0.6051569262101321,
+      "grad_norm": 0.0,
+      "learning_rate": 7.121396091526422e-06,
+      "loss": 0.9082,
+      "step": 16640
+    },
+    {
+      "epoch": 0.6051932938138707,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1202681059009424e-06,
+      "loss": 0.946,
+      "step": 16641
+    },
+    {
+      "epoch": 0.6052296614176091,
+      "grad_norm": 0.0,
+      "learning_rate": 7.119140160227651e-06,
+      "loss": 1.0079,
+      "step": 16642
+    },
+    {
+      "epoch": 0.6052660290213477,
+      "grad_norm": 0.0,
+      "learning_rate": 7.118012254522208e-06,
+      "loss": 0.9563,
+      "step": 16643
+    },
+    {
+      "epoch": 0.6053023966250863,
+      "grad_norm": 0.0,
+      "learning_rate": 7.1168843888002514e-06,
+      "loss": 0.8357,
+      "step": 16644
+    },
+    {
+      "epoch": 0.6053387642288249,
+      "grad_norm": 0.0,
+      "learning_rate": 7.115756563077436e-06,
+      "loss": 0.7978,
+      "step": 16645
+    },
+    {
+      "epoch": 0.6053751318325635,
+      "grad_norm": 0.0,
+      "learning_rate": 7.114628777369405e-06,
+      "loss": 0.9454,
+      "step": 16646
+    },
+    {
+      "epoch": 0.6054114994363021,
+      "grad_norm": 0.0,
+      "learning_rate": 7.113501031691806e-06,
+      "loss": 1.0403,
+      "step": 16647
+    },
+    {
+      "epoch": 0.6054478670400407,
+      "grad_norm": 0.0,
+      "learning_rate": 7.112373326060285e-06,
+      "loss": 0.9094,
+      "step": 16648
+    },
+    {
+      "epoch": 0.6054842346437793,
+      "grad_norm": 0.0,
+      "learning_rate": 7.111245660490489e-06,
+      "loss": 0.8675,
+      "step": 16649
+    },
+    {
+      "epoch": 0.6055206022475179,
+      "grad_norm": 0.0,
+      "learning_rate": 7.110118034998055e-06,
+      "loss": 0.9762,
+      "step": 16650
+    },
+    {
+      "epoch": 0.6055569698512565,
+      "grad_norm": 0.0,
+      "learning_rate": 7.108990449598641e-06,
+      "loss": 1.0738,
+      "step": 16651
+    },
+    {
+      "epoch": 0.6055933374549951,
+      "grad_norm": 0.0,
+      "learning_rate": 7.107862904307878e-06,
+      "loss": 0.8178,
+      "step": 16652
+    },
+    {
+      "epoch": 0.6056297050587337,
+      "grad_norm": 0.0,
+      "learning_rate": 7.106735399141413e-06,
+      "loss": 1.047,
+      "step": 16653
+    },
+    {
+      "epoch": 0.6056660726624723,
+      "grad_norm": 0.0,
+      "learning_rate": 7.105607934114895e-06,
+      "loss": 0.8155,
+      "step": 16654
+    },
+    {
+      "epoch": 0.6057024402662109,
+      "grad_norm": 0.0,
+      "learning_rate": 7.104480509243952e-06,
+      "loss": 0.9962,
+      "step": 16655
+    },
+    {
+      "epoch": 0.6057388078699495,
+      "grad_norm": 0.0,
+      "learning_rate": 7.103353124544241e-06,
+      "loss": 0.9537,
+      "step": 16656
+    },
+    {
+      "epoch": 0.6057751754736881,
+      "grad_norm": 0.0,
+      "learning_rate": 7.102225780031391e-06,
+      "loss": 1.0061,
+      "step": 16657
+    },
+    {
+      "epoch": 0.6058115430774266,
+      "grad_norm": 0.0,
+      "learning_rate": 7.101098475721049e-06,
+      "loss": 1.0503,
+      "step": 16658
+    },
+    {
+      "epoch": 0.6058479106811652,
+      "grad_norm": 0.0,
+      "learning_rate": 7.099971211628851e-06,
+      "loss": 0.9448,
+      "step": 16659
+    },
+    {
+      "epoch": 0.6058842782849038,
+      "grad_norm": 0.0,
+      "learning_rate": 7.098843987770439e-06,
+      "loss": 0.9723,
+      "step": 16660
+    },
+    {
+      "epoch": 0.6059206458886424,
+      "grad_norm": 0.0,
+      "learning_rate": 7.09771680416145e-06,
+      "loss": 1.007,
+      "step": 16661
+    },
+    {
+      "epoch": 0.605957013492381,
+      "grad_norm": 0.0,
+      "learning_rate": 7.096589660817526e-06,
+      "loss": 1.1146,
+      "step": 16662
+    },
+    {
+      "epoch": 0.6059933810961196,
+      "grad_norm": 0.0,
+      "learning_rate": 7.095462557754296e-06,
+      "loss": 1.0885,
+      "step": 16663
+    },
+    {
+      "epoch": 0.6060297486998582,
+      "grad_norm": 0.0,
+      "learning_rate": 7.094335494987405e-06,
+      "loss": 1.0714,
+      "step": 16664
+    },
+    {
+      "epoch": 0.6060661163035967,
+      "grad_norm": 0.0,
+      "learning_rate": 7.093208472532484e-06,
+      "loss": 1.0812,
+      "step": 16665
+    },
+    {
+      "epoch": 0.6061024839073353,
+      "grad_norm": 0.0,
+      "learning_rate": 7.092081490405174e-06,
+      "loss": 0.9376,
+      "step": 16666
+    },
+    {
+      "epoch": 0.6061388515110739,
+      "grad_norm": 0.0,
+      "learning_rate": 7.090954548621104e-06,
+      "loss": 0.9971,
+      "step": 16667
+    },
+    {
+      "epoch": 0.6061752191148125,
+      "grad_norm": 0.0,
+      "learning_rate": 7.089827647195917e-06,
+      "loss": 1.0306,
+      "step": 16668
+    },
+    {
+      "epoch": 0.6062115867185511,
+      "grad_norm": 0.0,
+      "learning_rate": 7.088700786145237e-06,
+      "loss": 0.9308,
+      "step": 16669
+    },
+    {
+      "epoch": 0.6062479543222897,
+      "grad_norm": 0.0,
+      "learning_rate": 7.087573965484709e-06,
+      "loss": 0.9971,
+      "step": 16670
+    },
+    {
+      "epoch": 0.6062843219260283,
+      "grad_norm": 0.0,
+      "learning_rate": 7.086447185229954e-06,
+      "loss": 1.1122,
+      "step": 16671
+    },
+    {
+      "epoch": 0.6063206895297669,
+      "grad_norm": 0.0,
+      "learning_rate": 7.085320445396615e-06,
+      "loss": 1.016,
+      "step": 16672
+    },
+    {
+      "epoch": 0.6063570571335055,
+      "grad_norm": 0.0,
+      "learning_rate": 7.084193746000321e-06,
+      "loss": 0.9993,
+      "step": 16673
+    },
+    {
+      "epoch": 0.6063934247372441,
+      "grad_norm": 0.0,
+      "learning_rate": 7.083067087056694e-06,
+      "loss": 0.9185,
+      "step": 16674
+    },
+    {
+      "epoch": 0.6064297923409826,
+      "grad_norm": 0.0,
+      "learning_rate": 7.081940468581381e-06,
+      "loss": 0.9045,
+      "step": 16675
+    },
+    {
+      "epoch": 0.6064661599447212,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0808138905899985e-06,
+      "loss": 1.0779,
+      "step": 16676
+    },
+    {
+      "epoch": 0.6065025275484598,
+      "grad_norm": 0.0,
+      "learning_rate": 7.079687353098183e-06,
+      "loss": 0.9912,
+      "step": 16677
+    },
+    {
+      "epoch": 0.6065388951521984,
+      "grad_norm": 0.0,
+      "learning_rate": 7.078560856121562e-06,
+      "loss": 0.9182,
+      "step": 16678
+    },
+    {
+      "epoch": 0.606575262755937,
+      "grad_norm": 0.0,
+      "learning_rate": 7.077434399675765e-06,
+      "loss": 0.9738,
+      "step": 16679
+    },
+    {
+      "epoch": 0.6066116303596756,
+      "grad_norm": 0.0,
+      "learning_rate": 7.076307983776416e-06,
+      "loss": 0.9021,
+      "step": 16680
+    },
+    {
+      "epoch": 0.6066479979634142,
+      "grad_norm": 0.0,
+      "learning_rate": 7.07518160843915e-06,
+      "loss": 0.8754,
+      "step": 16681
+    },
+    {
+      "epoch": 0.6066843655671528,
+      "grad_norm": 0.0,
+      "learning_rate": 7.074055273679584e-06,
+      "loss": 1.0522,
+      "step": 16682
+    },
+    {
+      "epoch": 0.6067207331708914,
+      "grad_norm": 0.0,
+      "learning_rate": 7.072928979513356e-06,
+      "loss": 1.0111,
+      "step": 16683
+    },
+    {
+      "epoch": 0.60675710077463,
+      "grad_norm": 0.0,
+      "learning_rate": 7.071802725956081e-06,
+      "loss": 0.8631,
+      "step": 16684
+    },
+    {
+      "epoch": 0.6067934683783686,
+      "grad_norm": 0.0,
+      "learning_rate": 7.070676513023391e-06,
+      "loss": 0.8597,
+      "step": 16685
+    },
+    {
+      "epoch": 0.6068298359821072,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0695503407309055e-06,
+      "loss": 0.9973,
+      "step": 16686
+    },
+    {
+      "epoch": 0.6068662035858458,
+      "grad_norm": 0.0,
+      "learning_rate": 7.068424209094254e-06,
+      "loss": 0.8866,
+      "step": 16687
+    },
+    {
+      "epoch": 0.6069025711895844,
+      "grad_norm": 0.0,
+      "learning_rate": 7.067298118129054e-06,
+      "loss": 0.8457,
+      "step": 16688
+    },
+    {
+      "epoch": 0.606938938793323,
+      "grad_norm": 0.0,
+      "learning_rate": 7.066172067850937e-06,
+      "loss": 0.9971,
+      "step": 16689
+    },
+    {
+      "epoch": 0.6069753063970615,
+      "grad_norm": 0.0,
+      "learning_rate": 7.065046058275515e-06,
+      "loss": 1.0363,
+      "step": 16690
+    },
+    {
+      "epoch": 0.6070116740008001,
+      "grad_norm": 0.0,
+      "learning_rate": 7.063920089418418e-06,
+      "loss": 0.943,
+      "step": 16691
+    },
+    {
+      "epoch": 0.6070480416045386,
+      "grad_norm": 0.0,
+      "learning_rate": 7.062794161295264e-06,
+      "loss": 1.0563,
+      "step": 16692
+    },
+    {
+      "epoch": 0.6070844092082772,
+      "grad_norm": 0.0,
+      "learning_rate": 7.06166827392167e-06,
+      "loss": 0.9803,
+      "step": 16693
+    },
+    {
+      "epoch": 0.6071207768120158,
+      "grad_norm": 0.0,
+      "learning_rate": 7.060542427313266e-06,
+      "loss": 1.0428,
+      "step": 16694
+    },
+    {
+      "epoch": 0.6071571444157544,
+      "grad_norm": 0.0,
+      "learning_rate": 7.05941662148566e-06,
+      "loss": 0.9346,
+      "step": 16695
+    },
+    {
+      "epoch": 0.607193512019493,
+      "grad_norm": 0.0,
+      "learning_rate": 7.058290856454477e-06,
+      "loss": 0.9609,
+      "step": 16696
+    },
+    {
+      "epoch": 0.6072298796232316,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0571651322353355e-06,
+      "loss": 1.0458,
+      "step": 16697
+    },
+    {
+      "epoch": 0.6072662472269702,
+      "grad_norm": 0.0,
+      "learning_rate": 7.056039448843852e-06,
+      "loss": 0.893,
+      "step": 16698
+    },
+    {
+      "epoch": 0.6073026148307088,
+      "grad_norm": 0.0,
+      "learning_rate": 7.054913806295643e-06,
+      "loss": 0.9842,
+      "step": 16699
+    },
+    {
+      "epoch": 0.6073389824344474,
+      "grad_norm": 0.0,
+      "learning_rate": 7.053788204606329e-06,
+      "loss": 0.9965,
+      "step": 16700
+    },
+    {
+      "epoch": 0.607375350038186,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0526626437915205e-06,
+      "loss": 0.9781,
+      "step": 16701
+    },
+    {
+      "epoch": 0.6074117176419246,
+      "grad_norm": 0.0,
+      "learning_rate": 7.05153712386684e-06,
+      "loss": 0.9489,
+      "step": 16702
+    },
+    {
+      "epoch": 0.6074480852456632,
+      "grad_norm": 0.0,
+      "learning_rate": 7.050411644847897e-06,
+      "loss": 0.9506,
+      "step": 16703
+    },
+    {
+      "epoch": 0.6074844528494018,
+      "grad_norm": 0.0,
+      "learning_rate": 7.049286206750307e-06,
+      "loss": 1.0054,
+      "step": 16704
+    },
+    {
+      "epoch": 0.6075208204531404,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0481608095896825e-06,
+      "loss": 1.0421,
+      "step": 16705
+    },
+    {
+      "epoch": 0.607557188056879,
+      "grad_norm": 0.0,
+      "learning_rate": 7.047035453381641e-06,
+      "loss": 0.9417,
+      "step": 16706
+    },
+    {
+      "epoch": 0.6075935556606176,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0459101381417915e-06,
+      "loss": 0.8834,
+      "step": 16707
+    },
+    {
+      "epoch": 0.6076299232643562,
+      "grad_norm": 0.0,
+      "learning_rate": 7.044784863885752e-06,
+      "loss": 0.9719,
+      "step": 16708
+    },
+    {
+      "epoch": 0.6076662908680946,
+      "grad_norm": 0.0,
+      "learning_rate": 7.043659630629121e-06,
+      "loss": 0.8849,
+      "step": 16709
+    },
+    {
+      "epoch": 0.6077026584718332,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0425344383875274e-06,
+      "loss": 0.9455,
+      "step": 16710
+    },
+    {
+      "epoch": 0.6077390260755718,
+      "grad_norm": 0.0,
+      "learning_rate": 7.041409287176569e-06,
+      "loss": 1.0249,
+      "step": 16711
+    },
+    {
+      "epoch": 0.6077753936793104,
+      "grad_norm": 0.0,
+      "learning_rate": 7.040284177011856e-06,
+      "loss": 0.9337,
+      "step": 16712
+    },
+    {
+      "epoch": 0.607811761283049,
+      "grad_norm": 0.0,
+      "learning_rate": 7.039159107909006e-06,
+      "loss": 1.0466,
+      "step": 16713
+    },
+    {
+      "epoch": 0.6078481288867876,
+      "grad_norm": 0.0,
+      "learning_rate": 7.038034079883615e-06,
+      "loss": 0.9219,
+      "step": 16714
+    },
+    {
+      "epoch": 0.6078844964905262,
+      "grad_norm": 0.0,
+      "learning_rate": 7.036909092951308e-06,
+      "loss": 0.8537,
+      "step": 16715
+    },
+    {
+      "epoch": 0.6079208640942648,
+      "grad_norm": 0.0,
+      "learning_rate": 7.035784147127676e-06,
+      "loss": 0.9754,
+      "step": 16716
+    },
+    {
+      "epoch": 0.6079572316980034,
+      "grad_norm": 0.0,
+      "learning_rate": 7.034659242428337e-06,
+      "loss": 0.8997,
+      "step": 16717
+    },
+    {
+      "epoch": 0.607993599301742,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0335343788688894e-06,
+      "loss": 0.9195,
+      "step": 16718
+    },
+    {
+      "epoch": 0.6080299669054806,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0324095564649475e-06,
+      "loss": 0.9964,
+      "step": 16719
+    },
+    {
+      "epoch": 0.6080663345092192,
+      "grad_norm": 0.0,
+      "learning_rate": 7.03128477523211e-06,
+      "loss": 0.9632,
+      "step": 16720
+    },
+    {
+      "epoch": 0.6081027021129578,
+      "grad_norm": 0.0,
+      "learning_rate": 7.030160035185989e-06,
+      "loss": 0.9247,
+      "step": 16721
+    },
+    {
+      "epoch": 0.6081390697166964,
+      "grad_norm": 0.0,
+      "learning_rate": 7.029035336342177e-06,
+      "loss": 0.9636,
+      "step": 16722
+    },
+    {
+      "epoch": 0.608175437320435,
+      "grad_norm": 0.0,
+      "learning_rate": 7.027910678716291e-06,
+      "loss": 1.063,
+      "step": 16723
+    },
+    {
+      "epoch": 0.6082118049241736,
+      "grad_norm": 0.0,
+      "learning_rate": 7.026786062323922e-06,
+      "loss": 0.9332,
+      "step": 16724
+    },
+    {
+      "epoch": 0.6082481725279122,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0256614871806815e-06,
+      "loss": 0.9356,
+      "step": 16725
+    },
+    {
+      "epoch": 0.6082845401316507,
+      "grad_norm": 0.0,
+      "learning_rate": 7.024536953302165e-06,
+      "loss": 0.9247,
+      "step": 16726
+    },
+    {
+      "epoch": 0.6083209077353893,
+      "grad_norm": 0.0,
+      "learning_rate": 7.02341246070398e-06,
+      "loss": 1.0036,
+      "step": 16727
+    },
+    {
+      "epoch": 0.6083572753391279,
+      "grad_norm": 0.0,
+      "learning_rate": 7.022288009401721e-06,
+      "loss": 1.0242,
+      "step": 16728
+    },
+    {
+      "epoch": 0.6083936429428665,
+      "grad_norm": 0.0,
+      "learning_rate": 7.021163599410995e-06,
+      "loss": 0.9108,
+      "step": 16729
+    },
+    {
+      "epoch": 0.6084300105466051,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0200392307473955e-06,
+      "loss": 1.0513,
+      "step": 16730
+    },
+    {
+      "epoch": 0.6084663781503437,
+      "grad_norm": 0.0,
+      "learning_rate": 7.018914903426521e-06,
+      "loss": 0.9682,
+      "step": 16731
+    },
+    {
+      "epoch": 0.6085027457540823,
+      "grad_norm": 0.0,
+      "learning_rate": 7.017790617463977e-06,
+      "loss": 0.9381,
+      "step": 16732
+    },
+    {
+      "epoch": 0.6085391133578208,
+      "grad_norm": 0.0,
+      "learning_rate": 7.016666372875352e-06,
+      "loss": 1.0273,
+      "step": 16733
+    },
+    {
+      "epoch": 0.6085754809615594,
+      "grad_norm": 0.0,
+      "learning_rate": 7.015542169676255e-06,
+      "loss": 0.9613,
+      "step": 16734
+    },
+    {
+      "epoch": 0.608611848565298,
+      "grad_norm": 0.0,
+      "learning_rate": 7.014418007882271e-06,
+      "loss": 0.9619,
+      "step": 16735
+    },
+    {
+      "epoch": 0.6086482161690366,
+      "grad_norm": 0.0,
+      "learning_rate": 7.013293887509003e-06,
+      "loss": 1.0023,
+      "step": 16736
+    },
+    {
+      "epoch": 0.6086845837727752,
+      "grad_norm": 0.0,
+      "learning_rate": 7.012169808572041e-06,
+      "loss": 0.9739,
+      "step": 16737
+    },
+    {
+      "epoch": 0.6087209513765138,
+      "grad_norm": 0.0,
+      "learning_rate": 7.011045771086988e-06,
+      "loss": 1.021,
+      "step": 16738
+    },
+    {
+      "epoch": 0.6087573189802524,
+      "grad_norm": 0.0,
+      "learning_rate": 7.009921775069432e-06,
+      "loss": 0.8709,
+      "step": 16739
+    },
+    {
+      "epoch": 0.608793686583991,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0087978205349715e-06,
+      "loss": 0.9055,
+      "step": 16740
+    },
+    {
+      "epoch": 0.6088300541877296,
+      "grad_norm": 0.0,
+      "learning_rate": 7.007673907499193e-06,
+      "loss": 1.008,
+      "step": 16741
+    },
+    {
+      "epoch": 0.6088664217914682,
+      "grad_norm": 0.0,
+      "learning_rate": 7.006550035977698e-06,
+      "loss": 0.9522,
+      "step": 16742
+    },
+    {
+      "epoch": 0.6089027893952067,
+      "grad_norm": 0.0,
+      "learning_rate": 7.005426205986071e-06,
+      "loss": 0.9648,
+      "step": 16743
+    },
+    {
+      "epoch": 0.6089391569989453,
+      "grad_norm": 0.0,
+      "learning_rate": 7.004302417539909e-06,
+      "loss": 1.0184,
+      "step": 16744
+    },
+    {
+      "epoch": 0.6089755246026839,
+      "grad_norm": 0.0,
+      "learning_rate": 7.003178670654798e-06,
+      "loss": 0.9424,
+      "step": 16745
+    },
+    {
+      "epoch": 0.6090118922064225,
+      "grad_norm": 0.0,
+      "learning_rate": 7.002054965346332e-06,
+      "loss": 0.8851,
+      "step": 16746
+    },
+    {
+      "epoch": 0.6090482598101611,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0009313016300994e-06,
+      "loss": 0.8848,
+      "step": 16747
+    },
+    {
+      "epoch": 0.6090846274138997,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9998076795216926e-06,
+      "loss": 0.9469,
+      "step": 16748
+    },
+    {
+      "epoch": 0.6091209950176383,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9986840990366965e-06,
+      "loss": 0.9435,
+      "step": 16749
+    },
+    {
+      "epoch": 0.6091573626213769,
+      "grad_norm": 0.0,
+      "learning_rate": 6.997560560190699e-06,
+      "loss": 0.908,
+      "step": 16750
+    },
+    {
+      "epoch": 0.6091937302251155,
+      "grad_norm": 0.0,
+      "learning_rate": 6.996437062999291e-06,
+      "loss": 0.8089,
+      "step": 16751
+    },
+    {
+      "epoch": 0.6092300978288541,
+      "grad_norm": 0.0,
+      "learning_rate": 6.995313607478054e-06,
+      "loss": 0.8867,
+      "step": 16752
+    },
+    {
+      "epoch": 0.6092664654325927,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9941901936425806e-06,
+      "loss": 0.943,
+      "step": 16753
+    },
+    {
+      "epoch": 0.6093028330363313,
+      "grad_norm": 0.0,
+      "learning_rate": 6.993066821508448e-06,
+      "loss": 0.922,
+      "step": 16754
+    },
+    {
+      "epoch": 0.6093392006400699,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9919434910912545e-06,
+      "loss": 1.0115,
+      "step": 16755
+    },
+    {
+      "epoch": 0.6093755682438085,
+      "grad_norm": 0.0,
+      "learning_rate": 6.990820202406572e-06,
+      "loss": 1.0179,
+      "step": 16756
+    },
+    {
+      "epoch": 0.609411935847547,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9896969554699926e-06,
+      "loss": 1.0695,
+      "step": 16757
+    },
+    {
+      "epoch": 0.6094483034512856,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9885737502970944e-06,
+      "loss": 0.9729,
+      "step": 16758
+    },
+    {
+      "epoch": 0.6094846710550241,
+      "grad_norm": 0.0,
+      "learning_rate": 6.987450586903464e-06,
+      "loss": 0.9316,
+      "step": 16759
+    },
+    {
+      "epoch": 0.6095210386587627,
+      "grad_norm": 0.0,
+      "learning_rate": 6.986327465304681e-06,
+      "loss": 0.8095,
+      "step": 16760
+    },
+    {
+      "epoch": 0.6095574062625013,
+      "grad_norm": 0.0,
+      "learning_rate": 6.985204385516333e-06,
+      "loss": 0.8635,
+      "step": 16761
+    },
+    {
+      "epoch": 0.6095937738662399,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9840813475539925e-06,
+      "loss": 0.9605,
+      "step": 16762
+    },
+    {
+      "epoch": 0.6096301414699785,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9829583514332464e-06,
+      "loss": 0.9876,
+      "step": 16763
+    },
+    {
+      "epoch": 0.6096665090737171,
+      "grad_norm": 0.0,
+      "learning_rate": 6.98183539716967e-06,
+      "loss": 0.9707,
+      "step": 16764
+    },
+    {
+      "epoch": 0.6097028766774557,
+      "grad_norm": 0.0,
+      "learning_rate": 6.980712484778848e-06,
+      "loss": 1.0214,
+      "step": 16765
+    },
+    {
+      "epoch": 0.6097392442811943,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9795896142763545e-06,
+      "loss": 0.9863,
+      "step": 16766
+    },
+    {
+      "epoch": 0.6097756118849329,
+      "grad_norm": 0.0,
+      "learning_rate": 6.978466785677773e-06,
+      "loss": 0.8949,
+      "step": 16767
+    },
+    {
+      "epoch": 0.6098119794886715,
+      "grad_norm": 0.0,
+      "learning_rate": 6.977343998998675e-06,
+      "loss": 0.9316,
+      "step": 16768
+    },
+    {
+      "epoch": 0.6098483470924101,
+      "grad_norm": 0.0,
+      "learning_rate": 6.976221254254641e-06,
+      "loss": 0.9225,
+      "step": 16769
+    },
+    {
+      "epoch": 0.6098847146961487,
+      "grad_norm": 0.0,
+      "learning_rate": 6.975098551461248e-06,
+      "loss": 0.9888,
+      "step": 16770
+    },
+    {
+      "epoch": 0.6099210822998873,
+      "grad_norm": 0.0,
+      "learning_rate": 6.973975890634069e-06,
+      "loss": 0.951,
+      "step": 16771
+    },
+    {
+      "epoch": 0.6099574499036259,
+      "grad_norm": 0.0,
+      "learning_rate": 6.972853271788681e-06,
+      "loss": 0.94,
+      "step": 16772
+    },
+    {
+      "epoch": 0.6099938175073645,
+      "grad_norm": 0.0,
+      "learning_rate": 6.971730694940659e-06,
+      "loss": 0.9036,
+      "step": 16773
+    },
+    {
+      "epoch": 0.6100301851111031,
+      "grad_norm": 0.0,
+      "learning_rate": 6.970608160105581e-06,
+      "loss": 1.0209,
+      "step": 16774
+    },
+    {
+      "epoch": 0.6100665527148417,
+      "grad_norm": 0.0,
+      "learning_rate": 6.969485667299012e-06,
+      "loss": 0.9926,
+      "step": 16775
+    },
+    {
+      "epoch": 0.6101029203185802,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9683632165365315e-06,
+      "loss": 0.975,
+      "step": 16776
+    },
+    {
+      "epoch": 0.6101392879223188,
+      "grad_norm": 0.0,
+      "learning_rate": 6.967240807833707e-06,
+      "loss": 1.0984,
+      "step": 16777
+    },
+    {
+      "epoch": 0.6101756555260573,
+      "grad_norm": 0.0,
+      "learning_rate": 6.966118441206117e-06,
+      "loss": 1.0137,
+      "step": 16778
+    },
+    {
+      "epoch": 0.6102120231297959,
+      "grad_norm": 0.0,
+      "learning_rate": 6.964996116669327e-06,
+      "loss": 1.0095,
+      "step": 16779
+    },
+    {
+      "epoch": 0.6102483907335345,
+      "grad_norm": 0.0,
+      "learning_rate": 6.963873834238913e-06,
+      "loss": 0.9463,
+      "step": 16780
+    },
+    {
+      "epoch": 0.6102847583372731,
+      "grad_norm": 0.0,
+      "learning_rate": 6.962751593930437e-06,
+      "loss": 0.9596,
+      "step": 16781
+    },
+    {
+      "epoch": 0.6103211259410117,
+      "grad_norm": 0.0,
+      "learning_rate": 6.961629395759478e-06,
+      "loss": 0.9283,
+      "step": 16782
+    },
+    {
+      "epoch": 0.6103574935447503,
+      "grad_norm": 0.0,
+      "learning_rate": 6.960507239741597e-06,
+      "loss": 1.0129,
+      "step": 16783
+    },
+    {
+      "epoch": 0.6103938611484889,
+      "grad_norm": 0.0,
+      "learning_rate": 6.959385125892369e-06,
+      "loss": 0.9197,
+      "step": 16784
+    },
+    {
+      "epoch": 0.6104302287522275,
+      "grad_norm": 0.0,
+      "learning_rate": 6.958263054227353e-06,
+      "loss": 1.0219,
+      "step": 16785
+    },
+    {
+      "epoch": 0.6104665963559661,
+      "grad_norm": 0.0,
+      "learning_rate": 6.957141024762126e-06,
+      "loss": 0.9725,
+      "step": 16786
+    },
+    {
+      "epoch": 0.6105029639597047,
+      "grad_norm": 0.0,
+      "learning_rate": 6.956019037512252e-06,
+      "loss": 0.9713,
+      "step": 16787
+    },
+    {
+      "epoch": 0.6105393315634433,
+      "grad_norm": 0.0,
+      "learning_rate": 6.954897092493289e-06,
+      "loss": 1.0161,
+      "step": 16788
+    },
+    {
+      "epoch": 0.6105756991671819,
+      "grad_norm": 0.0,
+      "learning_rate": 6.953775189720813e-06,
+      "loss": 0.8974,
+      "step": 16789
+    },
+    {
+      "epoch": 0.6106120667709205,
+      "grad_norm": 0.0,
+      "learning_rate": 6.95265332921038e-06,
+      "loss": 0.9832,
+      "step": 16790
+    },
+    {
+      "epoch": 0.6106484343746591,
+      "grad_norm": 0.0,
+      "learning_rate": 6.951531510977562e-06,
+      "loss": 1.0146,
+      "step": 16791
+    },
+    {
+      "epoch": 0.6106848019783977,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9504097350379156e-06,
+      "loss": 1.0239,
+      "step": 16792
+    },
+    {
+      "epoch": 0.6107211695821362,
+      "grad_norm": 0.0,
+      "learning_rate": 6.949288001407013e-06,
+      "loss": 0.8766,
+      "step": 16793
+    },
+    {
+      "epoch": 0.6107575371858748,
+      "grad_norm": 0.0,
+      "learning_rate": 6.948166310100404e-06,
+      "loss": 1.017,
+      "step": 16794
+    },
+    {
+      "epoch": 0.6107939047896134,
+      "grad_norm": 0.0,
+      "learning_rate": 6.947044661133664e-06,
+      "loss": 0.9825,
+      "step": 16795
+    },
+    {
+      "epoch": 0.610830272393352,
+      "grad_norm": 0.0,
+      "learning_rate": 6.945923054522343e-06,
+      "loss": 0.9337,
+      "step": 16796
+    },
+    {
+      "epoch": 0.6108666399970906,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9448014902820095e-06,
+      "loss": 1.0208,
+      "step": 16797
+    },
+    {
+      "epoch": 0.6109030076008292,
+      "grad_norm": 0.0,
+      "learning_rate": 6.943679968428216e-06,
+      "loss": 0.9684,
+      "step": 16798
+    },
+    {
+      "epoch": 0.6109393752045678,
+      "grad_norm": 0.0,
+      "learning_rate": 6.942558488976532e-06,
+      "loss": 1.0383,
+      "step": 16799
+    },
+    {
+      "epoch": 0.6109757428083064,
+      "grad_norm": 0.0,
+      "learning_rate": 6.941437051942507e-06,
+      "loss": 0.8495,
+      "step": 16800
+    },
+    {
+      "epoch": 0.611012110412045,
+      "grad_norm": 0.0,
+      "learning_rate": 6.940315657341707e-06,
+      "loss": 0.9241,
+      "step": 16801
+    },
+    {
+      "epoch": 0.6110484780157835,
+      "grad_norm": 0.0,
+      "learning_rate": 6.939194305189684e-06,
+      "loss": 0.9463,
+      "step": 16802
+    },
+    {
+      "epoch": 0.6110848456195221,
+      "grad_norm": 0.0,
+      "learning_rate": 6.938072995501998e-06,
+      "loss": 0.8329,
+      "step": 16803
+    },
+    {
+      "epoch": 0.6111212132232607,
+      "grad_norm": 0.0,
+      "learning_rate": 6.936951728294203e-06,
+      "loss": 0.9589,
+      "step": 16804
+    },
+    {
+      "epoch": 0.6111575808269993,
+      "grad_norm": 0.0,
+      "learning_rate": 6.93583050358186e-06,
+      "loss": 0.9758,
+      "step": 16805
+    },
+    {
+      "epoch": 0.6111939484307379,
+      "grad_norm": 0.0,
+      "learning_rate": 6.934709321380522e-06,
+      "loss": 1.0395,
+      "step": 16806
+    },
+    {
+      "epoch": 0.6112303160344765,
+      "grad_norm": 0.0,
+      "learning_rate": 6.93358818170574e-06,
+      "loss": 0.9678,
+      "step": 16807
+    },
+    {
+      "epoch": 0.6112666836382151,
+      "grad_norm": 0.0,
+      "learning_rate": 6.932467084573071e-06,
+      "loss": 0.9133,
+      "step": 16808
+    },
+    {
+      "epoch": 0.6113030512419537,
+      "grad_norm": 0.0,
+      "learning_rate": 6.931346029998069e-06,
+      "loss": 0.9613,
+      "step": 16809
+    },
+    {
+      "epoch": 0.6113394188456922,
+      "grad_norm": 0.0,
+      "learning_rate": 6.930225017996289e-06,
+      "loss": 0.9938,
+      "step": 16810
+    },
+    {
+      "epoch": 0.6113757864494308,
+      "grad_norm": 0.0,
+      "learning_rate": 6.929104048583279e-06,
+      "loss": 0.9251,
+      "step": 16811
+    },
+    {
+      "epoch": 0.6114121540531694,
+      "grad_norm": 0.0,
+      "learning_rate": 6.927983121774598e-06,
+      "loss": 0.8966,
+      "step": 16812
+    },
+    {
+      "epoch": 0.611448521656908,
+      "grad_norm": 0.0,
+      "learning_rate": 6.926862237585785e-06,
+      "loss": 1.062,
+      "step": 16813
+    },
+    {
+      "epoch": 0.6114848892606466,
+      "grad_norm": 0.0,
+      "learning_rate": 6.925741396032406e-06,
+      "loss": 1.0367,
+      "step": 16814
+    },
+    {
+      "epoch": 0.6115212568643852,
+      "grad_norm": 0.0,
+      "learning_rate": 6.924620597129998e-06,
+      "loss": 0.7682,
+      "step": 16815
+    },
+    {
+      "epoch": 0.6115576244681238,
+      "grad_norm": 0.0,
+      "learning_rate": 6.923499840894119e-06,
+      "loss": 0.9414,
+      "step": 16816
+    },
+    {
+      "epoch": 0.6115939920718624,
+      "grad_norm": 0.0,
+      "learning_rate": 6.922379127340311e-06,
+      "loss": 0.9508,
+      "step": 16817
+    },
+    {
+      "epoch": 0.611630359675601,
+      "grad_norm": 0.0,
+      "learning_rate": 6.92125845648413e-06,
+      "loss": 0.8883,
+      "step": 16818
+    },
+    {
+      "epoch": 0.6116667272793396,
+      "grad_norm": 0.0,
+      "learning_rate": 6.920137828341116e-06,
+      "loss": 0.9833,
+      "step": 16819
+    },
+    {
+      "epoch": 0.6117030948830782,
+      "grad_norm": 0.0,
+      "learning_rate": 6.919017242926826e-06,
+      "loss": 0.9519,
+      "step": 16820
+    },
+    {
+      "epoch": 0.6117394624868168,
+      "grad_norm": 0.0,
+      "learning_rate": 6.917896700256792e-06,
+      "loss": 0.95,
+      "step": 16821
+    },
+    {
+      "epoch": 0.6117758300905554,
+      "grad_norm": 0.0,
+      "learning_rate": 6.916776200346576e-06,
+      "loss": 0.8976,
+      "step": 16822
+    },
+    {
+      "epoch": 0.611812197694294,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9156557432117114e-06,
+      "loss": 1.0803,
+      "step": 16823
+    },
+    {
+      "epoch": 0.6118485652980326,
+      "grad_norm": 0.0,
+      "learning_rate": 6.914535328867749e-06,
+      "loss": 0.9822,
+      "step": 16824
+    },
+    {
+      "epoch": 0.6118849329017712,
+      "grad_norm": 0.0,
+      "learning_rate": 6.913414957330233e-06,
+      "loss": 0.8213,
+      "step": 16825
+    },
+    {
+      "epoch": 0.6119213005055097,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9122946286147e-06,
+      "loss": 0.8986,
+      "step": 16826
+    },
+    {
+      "epoch": 0.6119576681092482,
+      "grad_norm": 0.0,
+      "learning_rate": 6.911174342736703e-06,
+      "loss": 0.9745,
+      "step": 16827
+    },
+    {
+      "epoch": 0.6119940357129868,
+      "grad_norm": 0.0,
+      "learning_rate": 6.910054099711776e-06,
+      "loss": 0.9299,
+      "step": 16828
+    },
+    {
+      "epoch": 0.6120304033167254,
+      "grad_norm": 0.0,
+      "learning_rate": 6.908933899555466e-06,
+      "loss": 0.9984,
+      "step": 16829
+    },
+    {
+      "epoch": 0.612066770920464,
+      "grad_norm": 0.0,
+      "learning_rate": 6.907813742283312e-06,
+      "loss": 0.8938,
+      "step": 16830
+    },
+    {
+      "epoch": 0.6121031385242026,
+      "grad_norm": 0.0,
+      "learning_rate": 6.906693627910855e-06,
+      "loss": 0.9904,
+      "step": 16831
+    },
+    {
+      "epoch": 0.6121395061279412,
+      "grad_norm": 0.0,
+      "learning_rate": 6.905573556453634e-06,
+      "loss": 0.996,
+      "step": 16832
+    },
+    {
+      "epoch": 0.6121758737316798,
+      "grad_norm": 0.0,
+      "learning_rate": 6.904453527927194e-06,
+      "loss": 1.0264,
+      "step": 16833
+    },
+    {
+      "epoch": 0.6122122413354184,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9033335423470645e-06,
+      "loss": 1.0084,
+      "step": 16834
+    },
+    {
+      "epoch": 0.612248608939157,
+      "grad_norm": 0.0,
+      "learning_rate": 6.902213599728791e-06,
+      "loss": 0.9366,
+      "step": 16835
+    },
+    {
+      "epoch": 0.6122849765428956,
+      "grad_norm": 0.0,
+      "learning_rate": 6.901093700087906e-06,
+      "loss": 0.8741,
+      "step": 16836
+    },
+    {
+      "epoch": 0.6123213441466342,
+      "grad_norm": 0.0,
+      "learning_rate": 6.899973843439952e-06,
+      "loss": 1.0733,
+      "step": 16837
+    },
+    {
+      "epoch": 0.6123577117503728,
+      "grad_norm": 0.0,
+      "learning_rate": 6.898854029800461e-06,
+      "loss": 1.0542,
+      "step": 16838
+    },
+    {
+      "epoch": 0.6123940793541114,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8977342591849724e-06,
+      "loss": 0.9549,
+      "step": 16839
+    },
+    {
+      "epoch": 0.61243044695785,
+      "grad_norm": 0.0,
+      "learning_rate": 6.896614531609015e-06,
+      "loss": 0.9263,
+      "step": 16840
+    },
+    {
+      "epoch": 0.6124668145615886,
+      "grad_norm": 0.0,
+      "learning_rate": 6.895494847088134e-06,
+      "loss": 0.959,
+      "step": 16841
+    },
+    {
+      "epoch": 0.6125031821653272,
+      "grad_norm": 0.0,
+      "learning_rate": 6.894375205637852e-06,
+      "loss": 0.9046,
+      "step": 16842
+    },
+    {
+      "epoch": 0.6125395497690657,
+      "grad_norm": 0.0,
+      "learning_rate": 6.89325560727371e-06,
+      "loss": 1.0227,
+      "step": 16843
+    },
+    {
+      "epoch": 0.6125759173728043,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8921360520112405e-06,
+      "loss": 0.828,
+      "step": 16844
+    },
+    {
+      "epoch": 0.6126122849765429,
+      "grad_norm": 0.0,
+      "learning_rate": 6.891016539865969e-06,
+      "loss": 1.0262,
+      "step": 16845
+    },
+    {
+      "epoch": 0.6126486525802814,
+      "grad_norm": 0.0,
+      "learning_rate": 6.889897070853437e-06,
+      "loss": 0.8639,
+      "step": 16846
+    },
+    {
+      "epoch": 0.61268502018402,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8887776449891666e-06,
+      "loss": 0.9555,
+      "step": 16847
+    },
+    {
+      "epoch": 0.6127213877877586,
+      "grad_norm": 0.0,
+      "learning_rate": 6.887658262288695e-06,
+      "loss": 1.0283,
+      "step": 16848
+    },
+    {
+      "epoch": 0.6127577553914972,
+      "grad_norm": 0.0,
+      "learning_rate": 6.886538922767546e-06,
+      "loss": 1.0143,
+      "step": 16849
+    },
+    {
+      "epoch": 0.6127941229952358,
+      "grad_norm": 0.0,
+      "learning_rate": 6.885419626441254e-06,
+      "loss": 0.9268,
+      "step": 16850
+    },
+    {
+      "epoch": 0.6128304905989744,
+      "grad_norm": 0.0,
+      "learning_rate": 6.884300373325344e-06,
+      "loss": 1.0173,
+      "step": 16851
+    },
+    {
+      "epoch": 0.612866858202713,
+      "grad_norm": 0.0,
+      "learning_rate": 6.88318116343535e-06,
+      "loss": 1.1483,
+      "step": 16852
+    },
+    {
+      "epoch": 0.6129032258064516,
+      "grad_norm": 0.0,
+      "learning_rate": 6.882061996786789e-06,
+      "loss": 0.8783,
+      "step": 16853
+    },
+    {
+      "epoch": 0.6129395934101902,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8809428733952e-06,
+      "loss": 0.9294,
+      "step": 16854
+    },
+    {
+      "epoch": 0.6129759610139288,
+      "grad_norm": 0.0,
+      "learning_rate": 6.879823793276099e-06,
+      "loss": 0.9703,
+      "step": 16855
+    },
+    {
+      "epoch": 0.6130123286176674,
+      "grad_norm": 0.0,
+      "learning_rate": 6.878704756445018e-06,
+      "loss": 1.0767,
+      "step": 16856
+    },
+    {
+      "epoch": 0.613048696221406,
+      "grad_norm": 0.0,
+      "learning_rate": 6.877585762917478e-06,
+      "loss": 0.868,
+      "step": 16857
+    },
+    {
+      "epoch": 0.6130850638251446,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8764668127090085e-06,
+      "loss": 0.8548,
+      "step": 16858
+    },
+    {
+      "epoch": 0.6131214314288832,
+      "grad_norm": 0.0,
+      "learning_rate": 6.875347905835127e-06,
+      "loss": 0.9771,
+      "step": 16859
+    },
+    {
+      "epoch": 0.6131577990326217,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8742290423113645e-06,
+      "loss": 0.9502,
+      "step": 16860
+    },
+    {
+      "epoch": 0.6131941666363603,
+      "grad_norm": 0.0,
+      "learning_rate": 6.873110222153235e-06,
+      "loss": 0.991,
+      "step": 16861
+    },
+    {
+      "epoch": 0.6132305342400989,
+      "grad_norm": 0.0,
+      "learning_rate": 6.871991445376267e-06,
+      "loss": 1.038,
+      "step": 16862
+    },
+    {
+      "epoch": 0.6132669018438375,
+      "grad_norm": 0.0,
+      "learning_rate": 6.87087271199598e-06,
+      "loss": 0.9305,
+      "step": 16863
+    },
+    {
+      "epoch": 0.6133032694475761,
+      "grad_norm": 0.0,
+      "learning_rate": 6.86975402202789e-06,
+      "loss": 0.906,
+      "step": 16864
+    },
+    {
+      "epoch": 0.6133396370513147,
+      "grad_norm": 0.0,
+      "learning_rate": 6.868635375487529e-06,
+      "loss": 1.0627,
+      "step": 16865
+    },
+    {
+      "epoch": 0.6133760046550533,
+      "grad_norm": 0.0,
+      "learning_rate": 6.867516772390401e-06,
+      "loss": 0.9197,
+      "step": 16866
+    },
+    {
+      "epoch": 0.6134123722587919,
+      "grad_norm": 0.0,
+      "learning_rate": 6.86639821275204e-06,
+      "loss": 1.0451,
+      "step": 16867
+    },
+    {
+      "epoch": 0.6134487398625305,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8652796965879546e-06,
+      "loss": 0.9391,
+      "step": 16868
+    },
+    {
+      "epoch": 0.613485107466269,
+      "grad_norm": 0.0,
+      "learning_rate": 6.864161223913666e-06,
+      "loss": 1.1119,
+      "step": 16869
+    },
+    {
+      "epoch": 0.6135214750700076,
+      "grad_norm": 0.0,
+      "learning_rate": 6.863042794744692e-06,
+      "loss": 1.0047,
+      "step": 16870
+    },
+    {
+      "epoch": 0.6135578426737462,
+      "grad_norm": 0.0,
+      "learning_rate": 6.861924409096548e-06,
+      "loss": 1.0033,
+      "step": 16871
+    },
+    {
+      "epoch": 0.6135942102774848,
+      "grad_norm": 0.0,
+      "learning_rate": 6.860806066984749e-06,
+      "loss": 0.9562,
+      "step": 16872
+    },
+    {
+      "epoch": 0.6136305778812234,
+      "grad_norm": 0.0,
+      "learning_rate": 6.859687768424816e-06,
+      "loss": 1.0214,
+      "step": 16873
+    },
+    {
+      "epoch": 0.613666945484962,
+      "grad_norm": 0.0,
+      "learning_rate": 6.858569513432256e-06,
+      "loss": 0.8815,
+      "step": 16874
+    },
+    {
+      "epoch": 0.6137033130887006,
+      "grad_norm": 0.0,
+      "learning_rate": 6.85745130202259e-06,
+      "loss": 0.9811,
+      "step": 16875
+    },
+    {
+      "epoch": 0.6137396806924392,
+      "grad_norm": 0.0,
+      "learning_rate": 6.856333134211325e-06,
+      "loss": 0.8792,
+      "step": 16876
+    },
+    {
+      "epoch": 0.6137760482961777,
+      "grad_norm": 0.0,
+      "learning_rate": 6.85521501001398e-06,
+      "loss": 1.0641,
+      "step": 16877
+    },
+    {
+      "epoch": 0.6138124158999163,
+      "grad_norm": 0.0,
+      "learning_rate": 6.854096929446062e-06,
+      "loss": 1.0573,
+      "step": 16878
+    },
+    {
+      "epoch": 0.6138487835036549,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8529788925230915e-06,
+      "loss": 0.9399,
+      "step": 16879
+    },
+    {
+      "epoch": 0.6138851511073935,
+      "grad_norm": 0.0,
+      "learning_rate": 6.851860899260566e-06,
+      "loss": 0.9871,
+      "step": 16880
+    },
+    {
+      "epoch": 0.6139215187111321,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8507429496740105e-06,
+      "loss": 1.092,
+      "step": 16881
+    },
+    {
+      "epoch": 0.6139578863148707,
+      "grad_norm": 0.0,
+      "learning_rate": 6.849625043778927e-06,
+      "loss": 0.876,
+      "step": 16882
+    },
+    {
+      "epoch": 0.6139942539186093,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8485071815908235e-06,
+      "loss": 0.7928,
+      "step": 16883
+    },
+    {
+      "epoch": 0.6140306215223479,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8473893631252155e-06,
+      "loss": 0.8902,
+      "step": 16884
+    },
+    {
+      "epoch": 0.6140669891260865,
+      "grad_norm": 0.0,
+      "learning_rate": 6.846271588397601e-06,
+      "loss": 1.0553,
+      "step": 16885
+    },
+    {
+      "epoch": 0.6141033567298251,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8451538574235e-06,
+      "loss": 1.055,
+      "step": 16886
+    },
+    {
+      "epoch": 0.6141397243335637,
+      "grad_norm": 0.0,
+      "learning_rate": 6.844036170218409e-06,
+      "loss": 0.8797,
+      "step": 16887
+    },
+    {
+      "epoch": 0.6141760919373023,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8429185267978405e-06,
+      "loss": 0.9999,
+      "step": 16888
+    },
+    {
+      "epoch": 0.6142124595410409,
+      "grad_norm": 0.0,
+      "learning_rate": 6.841800927177296e-06,
+      "loss": 0.8888,
+      "step": 16889
+    },
+    {
+      "epoch": 0.6142488271447795,
+      "grad_norm": 0.0,
+      "learning_rate": 6.840683371372286e-06,
+      "loss": 0.8642,
+      "step": 16890
+    },
+    {
+      "epoch": 0.6142851947485181,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8395658593983076e-06,
+      "loss": 0.9987,
+      "step": 16891
+    },
+    {
+      "epoch": 0.6143215623522567,
+      "grad_norm": 0.0,
+      "learning_rate": 6.838448391270873e-06,
+      "loss": 0.9635,
+      "step": 16892
+    },
+    {
+      "epoch": 0.6143579299559953,
+      "grad_norm": 0.0,
+      "learning_rate": 6.837330967005477e-06,
+      "loss": 0.8774,
+      "step": 16893
+    },
+    {
+      "epoch": 0.6143942975597337,
+      "grad_norm": 0.0,
+      "learning_rate": 6.836213586617631e-06,
+      "loss": 1.041,
+      "step": 16894
+    },
+    {
+      "epoch": 0.6144306651634723,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8350962501228305e-06,
+      "loss": 0.8387,
+      "step": 16895
+    },
+    {
+      "epoch": 0.6144670327672109,
+      "grad_norm": 0.0,
+      "learning_rate": 6.833978957536579e-06,
+      "loss": 1.0805,
+      "step": 16896
+    },
+    {
+      "epoch": 0.6145034003709495,
+      "grad_norm": 0.0,
+      "learning_rate": 6.832861708874377e-06,
+      "loss": 1.0639,
+      "step": 16897
+    },
+    {
+      "epoch": 0.6145397679746881,
+      "grad_norm": 0.0,
+      "learning_rate": 6.831744504151725e-06,
+      "loss": 1.0185,
+      "step": 16898
+    },
+    {
+      "epoch": 0.6145761355784267,
+      "grad_norm": 0.0,
+      "learning_rate": 6.830627343384123e-06,
+      "loss": 0.9084,
+      "step": 16899
+    },
+    {
+      "epoch": 0.6146125031821653,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8295102265870735e-06,
+      "loss": 1.031,
+      "step": 16900
+    },
+    {
+      "epoch": 0.6146488707859039,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8283931537760674e-06,
+      "loss": 0.9706,
+      "step": 16901
+    },
+    {
+      "epoch": 0.6146852383896425,
+      "grad_norm": 0.0,
+      "learning_rate": 6.827276124966608e-06,
+      "loss": 0.9537,
+      "step": 16902
+    },
+    {
+      "epoch": 0.6147216059933811,
+      "grad_norm": 0.0,
+      "learning_rate": 6.826159140174191e-06,
+      "loss": 1.004,
+      "step": 16903
+    },
+    {
+      "epoch": 0.6147579735971197,
+      "grad_norm": 0.0,
+      "learning_rate": 6.82504219941431e-06,
+      "loss": 0.9673,
+      "step": 16904
+    },
+    {
+      "epoch": 0.6147943412008583,
+      "grad_norm": 0.0,
+      "learning_rate": 6.823925302702469e-06,
+      "loss": 1.0527,
+      "step": 16905
+    },
+    {
+      "epoch": 0.6148307088045969,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8228084500541534e-06,
+      "loss": 0.9291,
+      "step": 16906
+    },
+    {
+      "epoch": 0.6148670764083355,
+      "grad_norm": 0.0,
+      "learning_rate": 6.821691641484865e-06,
+      "loss": 1.045,
+      "step": 16907
+    },
+    {
+      "epoch": 0.6149034440120741,
+      "grad_norm": 0.0,
+      "learning_rate": 6.820574877010093e-06,
+      "loss": 0.92,
+      "step": 16908
+    },
+    {
+      "epoch": 0.6149398116158127,
+      "grad_norm": 0.0,
+      "learning_rate": 6.819458156645337e-06,
+      "loss": 0.9907,
+      "step": 16909
+    },
+    {
+      "epoch": 0.6149761792195513,
+      "grad_norm": 0.0,
+      "learning_rate": 6.818341480406083e-06,
+      "loss": 0.9736,
+      "step": 16910
+    },
+    {
+      "epoch": 0.6150125468232898,
+      "grad_norm": 0.0,
+      "learning_rate": 6.817224848307833e-06,
+      "loss": 1.0487,
+      "step": 16911
+    },
+    {
+      "epoch": 0.6150489144270284,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8161082603660645e-06,
+      "loss": 0.8394,
+      "step": 16912
+    },
+    {
+      "epoch": 0.615085282030767,
+      "grad_norm": 0.0,
+      "learning_rate": 6.814991716596283e-06,
+      "loss": 0.9209,
+      "step": 16913
+    },
+    {
+      "epoch": 0.6151216496345056,
+      "grad_norm": 0.0,
+      "learning_rate": 6.813875217013968e-06,
+      "loss": 1.0557,
+      "step": 16914
+    },
+    {
+      "epoch": 0.6151580172382441,
+      "grad_norm": 0.0,
+      "learning_rate": 6.812758761634617e-06,
+      "loss": 0.9752,
+      "step": 16915
+    },
+    {
+      "epoch": 0.6151943848419827,
+      "grad_norm": 0.0,
+      "learning_rate": 6.811642350473713e-06,
+      "loss": 0.8515,
+      "step": 16916
+    },
+    {
+      "epoch": 0.6152307524457213,
+      "grad_norm": 0.0,
+      "learning_rate": 6.81052598354675e-06,
+      "loss": 0.8853,
+      "step": 16917
+    },
+    {
+      "epoch": 0.6152671200494599,
+      "grad_norm": 0.0,
+      "learning_rate": 6.809409660869212e-06,
+      "loss": 0.8472,
+      "step": 16918
+    },
+    {
+      "epoch": 0.6153034876531985,
+      "grad_norm": 0.0,
+      "learning_rate": 6.808293382456592e-06,
+      "loss": 0.9441,
+      "step": 16919
+    },
+    {
+      "epoch": 0.6153398552569371,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8071771483243665e-06,
+      "loss": 0.9836,
+      "step": 16920
+    },
+    {
+      "epoch": 0.6153762228606757,
+      "grad_norm": 0.0,
+      "learning_rate": 6.806060958488034e-06,
+      "loss": 0.949,
+      "step": 16921
+    },
+    {
+      "epoch": 0.6154125904644143,
+      "grad_norm": 0.0,
+      "learning_rate": 6.804944812963072e-06,
+      "loss": 0.9303,
+      "step": 16922
+    },
+    {
+      "epoch": 0.6154489580681529,
+      "grad_norm": 0.0,
+      "learning_rate": 6.803828711764966e-06,
+      "loss": 0.9016,
+      "step": 16923
+    },
+    {
+      "epoch": 0.6154853256718915,
+      "grad_norm": 0.0,
+      "learning_rate": 6.802712654909207e-06,
+      "loss": 0.9128,
+      "step": 16924
+    },
+    {
+      "epoch": 0.6155216932756301,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8015966424112655e-06,
+      "loss": 1.0238,
+      "step": 16925
+    },
+    {
+      "epoch": 0.6155580608793687,
+      "grad_norm": 0.0,
+      "learning_rate": 6.8004806742866405e-06,
+      "loss": 1.0394,
+      "step": 16926
+    },
+    {
+      "epoch": 0.6155944284831073,
+      "grad_norm": 0.0,
+      "learning_rate": 6.799364750550801e-06,
+      "loss": 1.0892,
+      "step": 16927
+    },
+    {
+      "epoch": 0.6156307960868458,
+      "grad_norm": 0.0,
+      "learning_rate": 6.798248871219237e-06,
+      "loss": 0.965,
+      "step": 16928
+    },
+    {
+      "epoch": 0.6156671636905844,
+      "grad_norm": 0.0,
+      "learning_rate": 6.797133036307426e-06,
+      "loss": 0.9613,
+      "step": 16929
+    },
+    {
+      "epoch": 0.615703531294323,
+      "grad_norm": 0.0,
+      "learning_rate": 6.796017245830851e-06,
+      "loss": 1.0204,
+      "step": 16930
+    },
+    {
+      "epoch": 0.6157398988980616,
+      "grad_norm": 0.0,
+      "learning_rate": 6.794901499804989e-06,
+      "loss": 1.0186,
+      "step": 16931
+    },
+    {
+      "epoch": 0.6157762665018002,
+      "grad_norm": 0.0,
+      "learning_rate": 6.793785798245325e-06,
+      "loss": 0.971,
+      "step": 16932
+    },
+    {
+      "epoch": 0.6158126341055388,
+      "grad_norm": 0.0,
+      "learning_rate": 6.79267014116733e-06,
+      "loss": 0.9662,
+      "step": 16933
+    },
+    {
+      "epoch": 0.6158490017092774,
+      "grad_norm": 0.0,
+      "learning_rate": 6.791554528586488e-06,
+      "loss": 1.0146,
+      "step": 16934
+    },
+    {
+      "epoch": 0.615885369313016,
+      "grad_norm": 0.0,
+      "learning_rate": 6.790438960518274e-06,
+      "loss": 0.9735,
+      "step": 16935
+    },
+    {
+      "epoch": 0.6159217369167546,
+      "grad_norm": 0.0,
+      "learning_rate": 6.789323436978167e-06,
+      "loss": 1.0261,
+      "step": 16936
+    },
+    {
+      "epoch": 0.6159581045204932,
+      "grad_norm": 0.0,
+      "learning_rate": 6.78820795798164e-06,
+      "loss": 0.9544,
+      "step": 16937
+    },
+    {
+      "epoch": 0.6159944721242318,
+      "grad_norm": 0.0,
+      "learning_rate": 6.787092523544172e-06,
+      "loss": 0.9514,
+      "step": 16938
+    },
+    {
+      "epoch": 0.6160308397279703,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7859771336812354e-06,
+      "loss": 1.0319,
+      "step": 16939
+    },
+    {
+      "epoch": 0.6160672073317089,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7848617884083105e-06,
+      "loss": 0.9279,
+      "step": 16940
+    },
+    {
+      "epoch": 0.6161035749354475,
+      "grad_norm": 0.0,
+      "learning_rate": 6.783746487740863e-06,
+      "loss": 0.9391,
+      "step": 16941
+    },
+    {
+      "epoch": 0.6161399425391861,
+      "grad_norm": 0.0,
+      "learning_rate": 6.782631231694368e-06,
+      "loss": 1.0822,
+      "step": 16942
+    },
+    {
+      "epoch": 0.6161763101429247,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7815160202843035e-06,
+      "loss": 1.0235,
+      "step": 16943
+    },
+    {
+      "epoch": 0.6162126777466632,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7804008535261345e-06,
+      "loss": 1.0144,
+      "step": 16944
+    },
+    {
+      "epoch": 0.6162490453504018,
+      "grad_norm": 0.0,
+      "learning_rate": 6.779285731435339e-06,
+      "loss": 0.9914,
+      "step": 16945
+    },
+    {
+      "epoch": 0.6162854129541404,
+      "grad_norm": 0.0,
+      "learning_rate": 6.778170654027381e-06,
+      "loss": 1.0304,
+      "step": 16946
+    },
+    {
+      "epoch": 0.616321780557879,
+      "grad_norm": 0.0,
+      "learning_rate": 6.777055621317736e-06,
+      "loss": 1.0841,
+      "step": 16947
+    },
+    {
+      "epoch": 0.6163581481616176,
+      "grad_norm": 0.0,
+      "learning_rate": 6.775940633321871e-06,
+      "loss": 1.0011,
+      "step": 16948
+    },
+    {
+      "epoch": 0.6163945157653562,
+      "grad_norm": 0.0,
+      "learning_rate": 6.774825690055256e-06,
+      "loss": 1.0356,
+      "step": 16949
+    },
+    {
+      "epoch": 0.6164308833690948,
+      "grad_norm": 0.0,
+      "learning_rate": 6.773710791533356e-06,
+      "loss": 1.0386,
+      "step": 16950
+    },
+    {
+      "epoch": 0.6164672509728334,
+      "grad_norm": 0.0,
+      "learning_rate": 6.772595937771646e-06,
+      "loss": 0.9373,
+      "step": 16951
+    },
+    {
+      "epoch": 0.616503618576572,
+      "grad_norm": 0.0,
+      "learning_rate": 6.771481128785582e-06,
+      "loss": 1.1245,
+      "step": 16952
+    },
+    {
+      "epoch": 0.6165399861803106,
+      "grad_norm": 0.0,
+      "learning_rate": 6.770366364590644e-06,
+      "loss": 1.0759,
+      "step": 16953
+    },
+    {
+      "epoch": 0.6165763537840492,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7692516452022846e-06,
+      "loss": 0.7814,
+      "step": 16954
+    },
+    {
+      "epoch": 0.6166127213877878,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7681369706359774e-06,
+      "loss": 0.9449,
+      "step": 16955
+    },
+    {
+      "epoch": 0.6166490889915264,
+      "grad_norm": 0.0,
+      "learning_rate": 6.767022340907183e-06,
+      "loss": 0.9483,
+      "step": 16956
+    },
+    {
+      "epoch": 0.616685456595265,
+      "grad_norm": 0.0,
+      "learning_rate": 6.765907756031368e-06,
+      "loss": 0.9448,
+      "step": 16957
+    },
+    {
+      "epoch": 0.6167218241990036,
+      "grad_norm": 0.0,
+      "learning_rate": 6.764793216023993e-06,
+      "loss": 1.0218,
+      "step": 16958
+    },
+    {
+      "epoch": 0.6167581918027422,
+      "grad_norm": 0.0,
+      "learning_rate": 6.763678720900524e-06,
+      "loss": 0.9061,
+      "step": 16959
+    },
+    {
+      "epoch": 0.6167945594064808,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7625642706764215e-06,
+      "loss": 0.8899,
+      "step": 16960
+    },
+    {
+      "epoch": 0.6168309270102192,
+      "grad_norm": 0.0,
+      "learning_rate": 6.761449865367142e-06,
+      "loss": 0.9628,
+      "step": 16961
+    },
+    {
+      "epoch": 0.6168672946139578,
+      "grad_norm": 0.0,
+      "learning_rate": 6.760335504988155e-06,
+      "loss": 0.854,
+      "step": 16962
+    },
+    {
+      "epoch": 0.6169036622176964,
+      "grad_norm": 0.0,
+      "learning_rate": 6.759221189554913e-06,
+      "loss": 0.9775,
+      "step": 16963
+    },
+    {
+      "epoch": 0.616940029821435,
+      "grad_norm": 0.0,
+      "learning_rate": 6.758106919082883e-06,
+      "loss": 0.993,
+      "step": 16964
+    },
+    {
+      "epoch": 0.6169763974251736,
+      "grad_norm": 0.0,
+      "learning_rate": 6.756992693587514e-06,
+      "loss": 0.9367,
+      "step": 16965
+    },
+    {
+      "epoch": 0.6170127650289122,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7558785130842775e-06,
+      "loss": 0.7731,
+      "step": 16966
+    },
+    {
+      "epoch": 0.6170491326326508,
+      "grad_norm": 0.0,
+      "learning_rate": 6.754764377588618e-06,
+      "loss": 1.0461,
+      "step": 16967
+    },
+    {
+      "epoch": 0.6170855002363894,
+      "grad_norm": 0.0,
+      "learning_rate": 6.753650287116e-06,
+      "loss": 0.9178,
+      "step": 16968
+    },
+    {
+      "epoch": 0.617121867840128,
+      "grad_norm": 0.0,
+      "learning_rate": 6.752536241681878e-06,
+      "loss": 0.9934,
+      "step": 16969
+    },
+    {
+      "epoch": 0.6171582354438666,
+      "grad_norm": 0.0,
+      "learning_rate": 6.75142224130171e-06,
+      "loss": 1.0159,
+      "step": 16970
+    },
+    {
+      "epoch": 0.6171946030476052,
+      "grad_norm": 0.0,
+      "learning_rate": 6.750308285990946e-06,
+      "loss": 0.9535,
+      "step": 16971
+    },
+    {
+      "epoch": 0.6172309706513438,
+      "grad_norm": 0.0,
+      "learning_rate": 6.749194375765049e-06,
+      "loss": 0.9422,
+      "step": 16972
+    },
+    {
+      "epoch": 0.6172673382550824,
+      "grad_norm": 0.0,
+      "learning_rate": 6.748080510639463e-06,
+      "loss": 0.9679,
+      "step": 16973
+    },
+    {
+      "epoch": 0.617303705858821,
+      "grad_norm": 0.0,
+      "learning_rate": 6.746966690629648e-06,
+      "loss": 0.8473,
+      "step": 16974
+    },
+    {
+      "epoch": 0.6173400734625596,
+      "grad_norm": 0.0,
+      "learning_rate": 6.745852915751054e-06,
+      "loss": 0.8686,
+      "step": 16975
+    },
+    {
+      "epoch": 0.6173764410662982,
+      "grad_norm": 0.0,
+      "learning_rate": 6.744739186019133e-06,
+      "loss": 1.1074,
+      "step": 16976
+    },
+    {
+      "epoch": 0.6174128086700368,
+      "grad_norm": 0.0,
+      "learning_rate": 6.743625501449338e-06,
+      "loss": 0.9464,
+      "step": 16977
+    },
+    {
+      "epoch": 0.6174491762737753,
+      "grad_norm": 0.0,
+      "learning_rate": 6.742511862057122e-06,
+      "loss": 0.8755,
+      "step": 16978
+    },
+    {
+      "epoch": 0.6174855438775139,
+      "grad_norm": 0.0,
+      "learning_rate": 6.74139826785793e-06,
+      "loss": 0.9033,
+      "step": 16979
+    },
+    {
+      "epoch": 0.6175219114812525,
+      "grad_norm": 0.0,
+      "learning_rate": 6.740284718867211e-06,
+      "loss": 0.8933,
+      "step": 16980
+    },
+    {
+      "epoch": 0.6175582790849911,
+      "grad_norm": 0.0,
+      "learning_rate": 6.739171215100419e-06,
+      "loss": 0.8483,
+      "step": 16981
+    },
+    {
+      "epoch": 0.6175946466887297,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7380577565729996e-06,
+      "loss": 0.9605,
+      "step": 16982
+    },
+    {
+      "epoch": 0.6176310142924683,
+      "grad_norm": 0.0,
+      "learning_rate": 6.736944343300401e-06,
+      "loss": 0.9605,
+      "step": 16983
+    },
+    {
+      "epoch": 0.6176673818962068,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7358309752980685e-06,
+      "loss": 1.0332,
+      "step": 16984
+    },
+    {
+      "epoch": 0.6177037494999454,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7347176525814536e-06,
+      "loss": 0.952,
+      "step": 16985
+    },
+    {
+      "epoch": 0.617740117103684,
+      "grad_norm": 0.0,
+      "learning_rate": 6.733604375165995e-06,
+      "loss": 1.1222,
+      "step": 16986
+    },
+    {
+      "epoch": 0.6177764847074226,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7324911430671435e-06,
+      "loss": 0.8454,
+      "step": 16987
+    },
+    {
+      "epoch": 0.6178128523111612,
+      "grad_norm": 0.0,
+      "learning_rate": 6.731377956300338e-06,
+      "loss": 0.8688,
+      "step": 16988
+    },
+    {
+      "epoch": 0.6178492199148998,
+      "grad_norm": 0.0,
+      "learning_rate": 6.730264814881029e-06,
+      "loss": 0.9604,
+      "step": 16989
+    },
+    {
+      "epoch": 0.6178855875186384,
+      "grad_norm": 0.0,
+      "learning_rate": 6.729151718824654e-06,
+      "loss": 0.9158,
+      "step": 16990
+    },
+    {
+      "epoch": 0.617921955122377,
+      "grad_norm": 0.0,
+      "learning_rate": 6.728038668146663e-06,
+      "loss": 0.92,
+      "step": 16991
+    },
+    {
+      "epoch": 0.6179583227261156,
+      "grad_norm": 0.0,
+      "learning_rate": 6.726925662862485e-06,
+      "loss": 0.9614,
+      "step": 16992
+    },
+    {
+      "epoch": 0.6179946903298542,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7258127029875775e-06,
+      "loss": 0.9606,
+      "step": 16993
+    },
+    {
+      "epoch": 0.6180310579335928,
+      "grad_norm": 0.0,
+      "learning_rate": 6.72469978853737e-06,
+      "loss": 0.9515,
+      "step": 16994
+    },
+    {
+      "epoch": 0.6180674255373313,
+      "grad_norm": 0.0,
+      "learning_rate": 6.723586919527306e-06,
+      "loss": 1.0859,
+      "step": 16995
+    },
+    {
+      "epoch": 0.6181037931410699,
+      "grad_norm": 0.0,
+      "learning_rate": 6.722474095972824e-06,
+      "loss": 0.9579,
+      "step": 16996
+    },
+    {
+      "epoch": 0.6181401607448085,
+      "grad_norm": 0.0,
+      "learning_rate": 6.721361317889366e-06,
+      "loss": 0.9519,
+      "step": 16997
+    },
+    {
+      "epoch": 0.6181765283485471,
+      "grad_norm": 0.0,
+      "learning_rate": 6.720248585292368e-06,
+      "loss": 0.8667,
+      "step": 16998
+    },
+    {
+      "epoch": 0.6182128959522857,
+      "grad_norm": 0.0,
+      "learning_rate": 6.719135898197264e-06,
+      "loss": 1.0229,
+      "step": 16999
+    },
+    {
+      "epoch": 0.6182492635560243,
+      "grad_norm": 0.0,
+      "learning_rate": 6.718023256619498e-06,
+      "loss": 1.0202,
+      "step": 17000
+    },
+    {
+      "epoch": 0.6182856311597629,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7169106605745e-06,
+      "loss": 0.993,
+      "step": 17001
+    },
+    {
+      "epoch": 0.6183219987635015,
+      "grad_norm": 0.0,
+      "learning_rate": 6.715798110077711e-06,
+      "loss": 0.9327,
+      "step": 17002
+    },
+    {
+      "epoch": 0.6183583663672401,
+      "grad_norm": 0.0,
+      "learning_rate": 6.714685605144561e-06,
+      "loss": 0.9923,
+      "step": 17003
+    },
+    {
+      "epoch": 0.6183947339709787,
+      "grad_norm": 0.0,
+      "learning_rate": 6.713573145790491e-06,
+      "loss": 0.9509,
+      "step": 17004
+    },
+    {
+      "epoch": 0.6184311015747173,
+      "grad_norm": 0.0,
+      "learning_rate": 6.712460732030928e-06,
+      "loss": 1.0345,
+      "step": 17005
+    },
+    {
+      "epoch": 0.6184674691784559,
+      "grad_norm": 0.0,
+      "learning_rate": 6.711348363881307e-06,
+      "loss": 0.9567,
+      "step": 17006
+    },
+    {
+      "epoch": 0.6185038367821944,
+      "grad_norm": 0.0,
+      "learning_rate": 6.710236041357061e-06,
+      "loss": 0.9966,
+      "step": 17007
+    },
+    {
+      "epoch": 0.618540204385933,
+      "grad_norm": 0.0,
+      "learning_rate": 6.709123764473625e-06,
+      "loss": 0.9638,
+      "step": 17008
+    },
+    {
+      "epoch": 0.6185765719896716,
+      "grad_norm": 0.0,
+      "learning_rate": 6.708011533246426e-06,
+      "loss": 0.8147,
+      "step": 17009
+    },
+    {
+      "epoch": 0.6186129395934102,
+      "grad_norm": 0.0,
+      "learning_rate": 6.706899347690897e-06,
+      "loss": 0.9453,
+      "step": 17010
+    },
+    {
+      "epoch": 0.6186493071971488,
+      "grad_norm": 0.0,
+      "learning_rate": 6.705787207822465e-06,
+      "loss": 1.1127,
+      "step": 17011
+    },
+    {
+      "epoch": 0.6186856748008873,
+      "grad_norm": 0.0,
+      "learning_rate": 6.704675113656567e-06,
+      "loss": 0.9319,
+      "step": 17012
+    },
+    {
+      "epoch": 0.6187220424046259,
+      "grad_norm": 0.0,
+      "learning_rate": 6.703563065208622e-06,
+      "loss": 1.0074,
+      "step": 17013
+    },
+    {
+      "epoch": 0.6187584100083645,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7024510624940645e-06,
+      "loss": 0.8174,
+      "step": 17014
+    },
+    {
+      "epoch": 0.6187947776121031,
+      "grad_norm": 0.0,
+      "learning_rate": 6.701339105528317e-06,
+      "loss": 1.0217,
+      "step": 17015
+    },
+    {
+      "epoch": 0.6188311452158417,
+      "grad_norm": 0.0,
+      "learning_rate": 6.7002271943268116e-06,
+      "loss": 1.0089,
+      "step": 17016
+    },
+    {
+      "epoch": 0.6188675128195803,
+      "grad_norm": 0.0,
+      "learning_rate": 6.699115328904975e-06,
+      "loss": 1.0398,
+      "step": 17017
+    },
+    {
+      "epoch": 0.6189038804233189,
+      "grad_norm": 0.0,
+      "learning_rate": 6.698003509278225e-06,
+      "loss": 0.8926,
+      "step": 17018
+    },
+    {
+      "epoch": 0.6189402480270575,
+      "grad_norm": 0.0,
+      "learning_rate": 6.696891735461993e-06,
+      "loss": 0.9246,
+      "step": 17019
+    },
+    {
+      "epoch": 0.6189766156307961,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6957800074716994e-06,
+      "loss": 0.8438,
+      "step": 17020
+    },
+    {
+      "epoch": 0.6190129832345347,
+      "grad_norm": 0.0,
+      "learning_rate": 6.694668325322774e-06,
+      "loss": 0.9742,
+      "step": 17021
+    },
+    {
+      "epoch": 0.6190493508382733,
+      "grad_norm": 0.0,
+      "learning_rate": 6.693556689030633e-06,
+      "loss": 1.1076,
+      "step": 17022
+    },
+    {
+      "epoch": 0.6190857184420119,
+      "grad_norm": 0.0,
+      "learning_rate": 6.692445098610704e-06,
+      "loss": 0.9101,
+      "step": 17023
+    },
+    {
+      "epoch": 0.6191220860457505,
+      "grad_norm": 0.0,
+      "learning_rate": 6.691333554078403e-06,
+      "loss": 1.0848,
+      "step": 17024
+    },
+    {
+      "epoch": 0.6191584536494891,
+      "grad_norm": 0.0,
+      "learning_rate": 6.69022205544916e-06,
+      "loss": 1.1309,
+      "step": 17025
+    },
+    {
+      "epoch": 0.6191948212532277,
+      "grad_norm": 0.0,
+      "learning_rate": 6.689110602738384e-06,
+      "loss": 0.8851,
+      "step": 17026
+    },
+    {
+      "epoch": 0.6192311888569663,
+      "grad_norm": 0.0,
+      "learning_rate": 6.687999195961505e-06,
+      "loss": 0.9136,
+      "step": 17027
+    },
+    {
+      "epoch": 0.6192675564607047,
+      "grad_norm": 0.0,
+      "learning_rate": 6.686887835133936e-06,
+      "loss": 0.9788,
+      "step": 17028
+    },
+    {
+      "epoch": 0.6193039240644433,
+      "grad_norm": 0.0,
+      "learning_rate": 6.685776520271099e-06,
+      "loss": 0.9513,
+      "step": 17029
+    },
+    {
+      "epoch": 0.6193402916681819,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6846652513884095e-06,
+      "loss": 0.8995,
+      "step": 17030
+    },
+    {
+      "epoch": 0.6193766592719205,
+      "grad_norm": 0.0,
+      "learning_rate": 6.683554028501289e-06,
+      "loss": 0.8719,
+      "step": 17031
+    },
+    {
+      "epoch": 0.6194130268756591,
+      "grad_norm": 0.0,
+      "learning_rate": 6.682442851625144e-06,
+      "loss": 0.9009,
+      "step": 17032
+    },
+    {
+      "epoch": 0.6194493944793977,
+      "grad_norm": 0.0,
+      "learning_rate": 6.681331720775403e-06,
+      "loss": 0.9659,
+      "step": 17033
+    },
+    {
+      "epoch": 0.6194857620831363,
+      "grad_norm": 0.0,
+      "learning_rate": 6.680220635967474e-06,
+      "loss": 0.9435,
+      "step": 17034
+    },
+    {
+      "epoch": 0.6195221296868749,
+      "grad_norm": 0.0,
+      "learning_rate": 6.679109597216775e-06,
+      "loss": 0.9287,
+      "step": 17035
+    },
+    {
+      "epoch": 0.6195584972906135,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6779986045387204e-06,
+      "loss": 1.0534,
+      "step": 17036
+    },
+    {
+      "epoch": 0.6195948648943521,
+      "grad_norm": 0.0,
+      "learning_rate": 6.676887657948715e-06,
+      "loss": 1.0502,
+      "step": 17037
+    },
+    {
+      "epoch": 0.6196312324980907,
+      "grad_norm": 0.0,
+      "learning_rate": 6.675776757462184e-06,
+      "loss": 1.0136,
+      "step": 17038
+    },
+    {
+      "epoch": 0.6196676001018293,
+      "grad_norm": 0.0,
+      "learning_rate": 6.674665903094533e-06,
+      "loss": 0.821,
+      "step": 17039
+    },
+    {
+      "epoch": 0.6197039677055679,
+      "grad_norm": 0.0,
+      "learning_rate": 6.673555094861175e-06,
+      "loss": 0.9986,
+      "step": 17040
+    },
+    {
+      "epoch": 0.6197403353093065,
+      "grad_norm": 0.0,
+      "learning_rate": 6.672444332777517e-06,
+      "loss": 0.9374,
+      "step": 17041
+    },
+    {
+      "epoch": 0.6197767029130451,
+      "grad_norm": 0.0,
+      "learning_rate": 6.671333616858976e-06,
+      "loss": 0.9852,
+      "step": 17042
+    },
+    {
+      "epoch": 0.6198130705167837,
+      "grad_norm": 0.0,
+      "learning_rate": 6.670222947120957e-06,
+      "loss": 0.9784,
+      "step": 17043
+    },
+    {
+      "epoch": 0.6198494381205223,
+      "grad_norm": 0.0,
+      "learning_rate": 6.669112323578873e-06,
+      "loss": 1.0125,
+      "step": 17044
+    },
+    {
+      "epoch": 0.6198858057242608,
+      "grad_norm": 0.0,
+      "learning_rate": 6.668001746248125e-06,
+      "loss": 0.8918,
+      "step": 17045
+    },
+    {
+      "epoch": 0.6199221733279994,
+      "grad_norm": 0.0,
+      "learning_rate": 6.666891215144129e-06,
+      "loss": 0.9599,
+      "step": 17046
+    },
+    {
+      "epoch": 0.619958540931738,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6657807302822856e-06,
+      "loss": 1.0168,
+      "step": 17047
+    },
+    {
+      "epoch": 0.6199949085354766,
+      "grad_norm": 0.0,
+      "learning_rate": 6.664670291678006e-06,
+      "loss": 0.906,
+      "step": 17048
+    },
+    {
+      "epoch": 0.6200312761392152,
+      "grad_norm": 0.0,
+      "learning_rate": 6.663559899346692e-06,
+      "loss": 0.9627,
+      "step": 17049
+    },
+    {
+      "epoch": 0.6200676437429538,
+      "grad_norm": 0.0,
+      "learning_rate": 6.662449553303753e-06,
+      "loss": 1.0203,
+      "step": 17050
+    },
+    {
+      "epoch": 0.6201040113466924,
+      "grad_norm": 0.0,
+      "learning_rate": 6.661339253564587e-06,
+      "loss": 0.9889,
+      "step": 17051
+    },
+    {
+      "epoch": 0.620140378950431,
+      "grad_norm": 0.0,
+      "learning_rate": 6.660229000144609e-06,
+      "loss": 1.0169,
+      "step": 17052
+    },
+    {
+      "epoch": 0.6201767465541695,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6591187930592095e-06,
+      "loss": 0.9905,
+      "step": 17053
+    },
+    {
+      "epoch": 0.6202131141579081,
+      "grad_norm": 0.0,
+      "learning_rate": 6.658008632323798e-06,
+      "loss": 0.9962,
+      "step": 17054
+    },
+    {
+      "epoch": 0.6202494817616467,
+      "grad_norm": 0.0,
+      "learning_rate": 6.656898517953776e-06,
+      "loss": 0.9948,
+      "step": 17055
+    },
+    {
+      "epoch": 0.6202858493653853,
+      "grad_norm": 0.0,
+      "learning_rate": 6.655788449964542e-06,
+      "loss": 0.9714,
+      "step": 17056
+    },
+    {
+      "epoch": 0.6203222169691239,
+      "grad_norm": 0.0,
+      "learning_rate": 6.654678428371502e-06,
+      "loss": 0.9032,
+      "step": 17057
+    },
+    {
+      "epoch": 0.6203585845728625,
+      "grad_norm": 0.0,
+      "learning_rate": 6.653568453190048e-06,
+      "loss": 0.9901,
+      "step": 17058
+    },
+    {
+      "epoch": 0.6203949521766011,
+      "grad_norm": 0.0,
+      "learning_rate": 6.652458524435587e-06,
+      "loss": 0.9955,
+      "step": 17059
+    },
+    {
+      "epoch": 0.6204313197803397,
+      "grad_norm": 0.0,
+      "learning_rate": 6.65134864212351e-06,
+      "loss": 0.9614,
+      "step": 17060
+    },
+    {
+      "epoch": 0.6204676873840783,
+      "grad_norm": 0.0,
+      "learning_rate": 6.650238806269225e-06,
+      "loss": 0.9224,
+      "step": 17061
+    },
+    {
+      "epoch": 0.6205040549878168,
+      "grad_norm": 0.0,
+      "learning_rate": 6.649129016888118e-06,
+      "loss": 1.086,
+      "step": 17062
+    },
+    {
+      "epoch": 0.6205404225915554,
+      "grad_norm": 0.0,
+      "learning_rate": 6.648019273995599e-06,
+      "loss": 0.8577,
+      "step": 17063
+    },
+    {
+      "epoch": 0.620576790195294,
+      "grad_norm": 0.0,
+      "learning_rate": 6.646909577607049e-06,
+      "loss": 0.8249,
+      "step": 17064
+    },
+    {
+      "epoch": 0.6206131577990326,
+      "grad_norm": 0.0,
+      "learning_rate": 6.645799927737877e-06,
+      "loss": 0.8998,
+      "step": 17065
+    },
+    {
+      "epoch": 0.6206495254027712,
+      "grad_norm": 0.0,
+      "learning_rate": 6.644690324403469e-06,
+      "loss": 0.9305,
+      "step": 17066
+    },
+    {
+      "epoch": 0.6206858930065098,
+      "grad_norm": 0.0,
+      "learning_rate": 6.643580767619225e-06,
+      "loss": 0.8498,
+      "step": 17067
+    },
+    {
+      "epoch": 0.6207222606102484,
+      "grad_norm": 0.0,
+      "learning_rate": 6.642471257400531e-06,
+      "loss": 0.9422,
+      "step": 17068
+    },
+    {
+      "epoch": 0.620758628213987,
+      "grad_norm": 0.0,
+      "learning_rate": 6.641361793762788e-06,
+      "loss": 0.9896,
+      "step": 17069
+    },
+    {
+      "epoch": 0.6207949958177256,
+      "grad_norm": 0.0,
+      "learning_rate": 6.640252376721383e-06,
+      "loss": 0.8342,
+      "step": 17070
+    },
+    {
+      "epoch": 0.6208313634214642,
+      "grad_norm": 0.0,
+      "learning_rate": 6.639143006291713e-06,
+      "loss": 1.0483,
+      "step": 17071
+    },
+    {
+      "epoch": 0.6208677310252028,
+      "grad_norm": 0.0,
+      "learning_rate": 6.638033682489162e-06,
+      "loss": 0.9652,
+      "step": 17072
+    },
+    {
+      "epoch": 0.6209040986289414,
+      "grad_norm": 0.0,
+      "learning_rate": 6.636924405329126e-06,
+      "loss": 0.9365,
+      "step": 17073
+    },
+    {
+      "epoch": 0.62094046623268,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6358151748269916e-06,
+      "loss": 0.8891,
+      "step": 17074
+    },
+    {
+      "epoch": 0.6209768338364186,
+      "grad_norm": 0.0,
+      "learning_rate": 6.634705990998146e-06,
+      "loss": 0.9633,
+      "step": 17075
+    },
+    {
+      "epoch": 0.6210132014401571,
+      "grad_norm": 0.0,
+      "learning_rate": 6.633596853857984e-06,
+      "loss": 1.0083,
+      "step": 17076
+    },
+    {
+      "epoch": 0.6210495690438957,
+      "grad_norm": 0.0,
+      "learning_rate": 6.632487763421886e-06,
+      "loss": 1.0452,
+      "step": 17077
+    },
+    {
+      "epoch": 0.6210859366476343,
+      "grad_norm": 0.0,
+      "learning_rate": 6.631378719705245e-06,
+      "loss": 0.9319,
+      "step": 17078
+    },
+    {
+      "epoch": 0.6211223042513728,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6302697227234415e-06,
+      "loss": 0.9758,
+      "step": 17079
+    },
+    {
+      "epoch": 0.6211586718551114,
+      "grad_norm": 0.0,
+      "learning_rate": 6.629160772491867e-06,
+      "loss": 1.0198,
+      "step": 17080
+    },
+    {
+      "epoch": 0.62119503945885,
+      "grad_norm": 0.0,
+      "learning_rate": 6.628051869025902e-06,
+      "loss": 0.9302,
+      "step": 17081
+    },
+    {
+      "epoch": 0.6212314070625886,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6269430123409365e-06,
+      "loss": 0.8751,
+      "step": 17082
+    },
+    {
+      "epoch": 0.6212677746663272,
+      "grad_norm": 0.0,
+      "learning_rate": 6.625834202452348e-06,
+      "loss": 1.0125,
+      "step": 17083
+    },
+    {
+      "epoch": 0.6213041422700658,
+      "grad_norm": 0.0,
+      "learning_rate": 6.624725439375527e-06,
+      "loss": 0.8928,
+      "step": 17084
+    },
+    {
+      "epoch": 0.6213405098738044,
+      "grad_norm": 0.0,
+      "learning_rate": 6.623616723125847e-06,
+      "loss": 0.9622,
+      "step": 17085
+    },
+    {
+      "epoch": 0.621376877477543,
+      "grad_norm": 0.0,
+      "learning_rate": 6.622508053718697e-06,
+      "loss": 1.0018,
+      "step": 17086
+    },
+    {
+      "epoch": 0.6214132450812816,
+      "grad_norm": 0.0,
+      "learning_rate": 6.621399431169453e-06,
+      "loss": 1.0083,
+      "step": 17087
+    },
+    {
+      "epoch": 0.6214496126850202,
+      "grad_norm": 0.0,
+      "learning_rate": 6.620290855493501e-06,
+      "loss": 0.9044,
+      "step": 17088
+    },
+    {
+      "epoch": 0.6214859802887588,
+      "grad_norm": 0.0,
+      "learning_rate": 6.619182326706216e-06,
+      "loss": 0.9835,
+      "step": 17089
+    },
+    {
+      "epoch": 0.6215223478924974,
+      "grad_norm": 0.0,
+      "learning_rate": 6.618073844822983e-06,
+      "loss": 0.8673,
+      "step": 17090
+    },
+    {
+      "epoch": 0.621558715496236,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6169654098591715e-06,
+      "loss": 0.8881,
+      "step": 17091
+    },
+    {
+      "epoch": 0.6215950830999746,
+      "grad_norm": 0.0,
+      "learning_rate": 6.615857021830172e-06,
+      "loss": 0.9545,
+      "step": 17092
+    },
+    {
+      "epoch": 0.6216314507037132,
+      "grad_norm": 0.0,
+      "learning_rate": 6.61474868075135e-06,
+      "loss": 0.9729,
+      "step": 17093
+    },
+    {
+      "epoch": 0.6216678183074518,
+      "grad_norm": 0.0,
+      "learning_rate": 6.613640386638088e-06,
+      "loss": 0.9167,
+      "step": 17094
+    },
+    {
+      "epoch": 0.6217041859111904,
+      "grad_norm": 0.0,
+      "learning_rate": 6.612532139505764e-06,
+      "loss": 1.0412,
+      "step": 17095
+    },
+    {
+      "epoch": 0.6217405535149289,
+      "grad_norm": 0.0,
+      "learning_rate": 6.611423939369745e-06,
+      "loss": 0.9465,
+      "step": 17096
+    },
+    {
+      "epoch": 0.6217769211186674,
+      "grad_norm": 0.0,
+      "learning_rate": 6.610315786245416e-06,
+      "loss": 1.0062,
+      "step": 17097
+    },
+    {
+      "epoch": 0.621813288722406,
+      "grad_norm": 0.0,
+      "learning_rate": 6.609207680148142e-06,
+      "loss": 0.9336,
+      "step": 17098
+    },
+    {
+      "epoch": 0.6218496563261446,
+      "grad_norm": 0.0,
+      "learning_rate": 6.608099621093304e-06,
+      "loss": 0.9931,
+      "step": 17099
+    },
+    {
+      "epoch": 0.6218860239298832,
+      "grad_norm": 0.0,
+      "learning_rate": 6.606991609096267e-06,
+      "loss": 0.9825,
+      "step": 17100
+    },
+    {
+      "epoch": 0.6219223915336218,
+      "grad_norm": 0.0,
+      "learning_rate": 6.605883644172412e-06,
+      "loss": 0.9954,
+      "step": 17101
+    },
+    {
+      "epoch": 0.6219587591373604,
+      "grad_norm": 0.0,
+      "learning_rate": 6.604775726337102e-06,
+      "loss": 1.028,
+      "step": 17102
+    },
+    {
+      "epoch": 0.621995126741099,
+      "grad_norm": 0.0,
+      "learning_rate": 6.603667855605716e-06,
+      "loss": 0.8645,
+      "step": 17103
+    },
+    {
+      "epoch": 0.6220314943448376,
+      "grad_norm": 0.0,
+      "learning_rate": 6.602560031993614e-06,
+      "loss": 1.0109,
+      "step": 17104
+    },
+    {
+      "epoch": 0.6220678619485762,
+      "grad_norm": 0.0,
+      "learning_rate": 6.601452255516176e-06,
+      "loss": 0.8789,
+      "step": 17105
+    },
+    {
+      "epoch": 0.6221042295523148,
+      "grad_norm": 0.0,
+      "learning_rate": 6.600344526188765e-06,
+      "loss": 0.9207,
+      "step": 17106
+    },
+    {
+      "epoch": 0.6221405971560534,
+      "grad_norm": 0.0,
+      "learning_rate": 6.59923684402675e-06,
+      "loss": 0.9918,
+      "step": 17107
+    },
+    {
+      "epoch": 0.622176964759792,
+      "grad_norm": 0.0,
+      "learning_rate": 6.598129209045497e-06,
+      "loss": 0.8925,
+      "step": 17108
+    },
+    {
+      "epoch": 0.6222133323635306,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5970216212603755e-06,
+      "loss": 0.8906,
+      "step": 17109
+    },
+    {
+      "epoch": 0.6222496999672692,
+      "grad_norm": 0.0,
+      "learning_rate": 6.595914080686751e-06,
+      "loss": 0.9411,
+      "step": 17110
+    },
+    {
+      "epoch": 0.6222860675710078,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5948065873399925e-06,
+      "loss": 1.0384,
+      "step": 17111
+    },
+    {
+      "epoch": 0.6223224351747464,
+      "grad_norm": 0.0,
+      "learning_rate": 6.593699141235458e-06,
+      "loss": 0.8545,
+      "step": 17112
+    },
+    {
+      "epoch": 0.6223588027784849,
+      "grad_norm": 0.0,
+      "learning_rate": 6.592591742388513e-06,
+      "loss": 0.9502,
+      "step": 17113
+    },
+    {
+      "epoch": 0.6223951703822235,
+      "grad_norm": 0.0,
+      "learning_rate": 6.591484390814526e-06,
+      "loss": 0.8976,
+      "step": 17114
+    },
+    {
+      "epoch": 0.6224315379859621,
+      "grad_norm": 0.0,
+      "learning_rate": 6.590377086528854e-06,
+      "loss": 0.8957,
+      "step": 17115
+    },
+    {
+      "epoch": 0.6224679055897007,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5892698295468685e-06,
+      "loss": 1.0125,
+      "step": 17116
+    },
+    {
+      "epoch": 0.6225042731934393,
+      "grad_norm": 0.0,
+      "learning_rate": 6.588162619883919e-06,
+      "loss": 0.9239,
+      "step": 17117
+    },
+    {
+      "epoch": 0.6225406407971779,
+      "grad_norm": 0.0,
+      "learning_rate": 6.587055457555375e-06,
+      "loss": 1.0419,
+      "step": 17118
+    },
+    {
+      "epoch": 0.6225770084009165,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5859483425765916e-06,
+      "loss": 0.9193,
+      "step": 17119
+    },
+    {
+      "epoch": 0.622613376004655,
+      "grad_norm": 0.0,
+      "learning_rate": 6.584841274962935e-06,
+      "loss": 0.9903,
+      "step": 17120
+    },
+    {
+      "epoch": 0.6226497436083936,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5837342547297554e-06,
+      "loss": 0.8989,
+      "step": 17121
+    },
+    {
+      "epoch": 0.6226861112121322,
+      "grad_norm": 0.0,
+      "learning_rate": 6.582627281892422e-06,
+      "loss": 1.0032,
+      "step": 17122
+    },
+    {
+      "epoch": 0.6227224788158708,
+      "grad_norm": 0.0,
+      "learning_rate": 6.581520356466279e-06,
+      "loss": 0.858,
+      "step": 17123
+    },
+    {
+      "epoch": 0.6227588464196094,
+      "grad_norm": 0.0,
+      "learning_rate": 6.580413478466699e-06,
+      "loss": 0.9871,
+      "step": 17124
+    },
+    {
+      "epoch": 0.622795214023348,
+      "grad_norm": 0.0,
+      "learning_rate": 6.579306647909023e-06,
+      "loss": 0.9709,
+      "step": 17125
+    },
+    {
+      "epoch": 0.6228315816270866,
+      "grad_norm": 0.0,
+      "learning_rate": 6.578199864808619e-06,
+      "loss": 0.9064,
+      "step": 17126
+    },
+    {
+      "epoch": 0.6228679492308252,
+      "grad_norm": 0.0,
+      "learning_rate": 6.577093129180834e-06,
+      "loss": 0.944,
+      "step": 17127
+    },
+    {
+      "epoch": 0.6229043168345638,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5759864410410265e-06,
+      "loss": 0.9483,
+      "step": 17128
+    },
+    {
+      "epoch": 0.6229406844383023,
+      "grad_norm": 0.0,
+      "learning_rate": 6.574879800404548e-06,
+      "loss": 0.9659,
+      "step": 17129
+    },
+    {
+      "epoch": 0.6229770520420409,
+      "grad_norm": 0.0,
+      "learning_rate": 6.573773207286756e-06,
+      "loss": 0.9626,
+      "step": 17130
+    },
+    {
+      "epoch": 0.6230134196457795,
+      "grad_norm": 0.0,
+      "learning_rate": 6.572666661702997e-06,
+      "loss": 0.9323,
+      "step": 17131
+    },
+    {
+      "epoch": 0.6230497872495181,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5715601636686244e-06,
+      "loss": 0.909,
+      "step": 17132
+    },
+    {
+      "epoch": 0.6230861548532567,
+      "grad_norm": 0.0,
+      "learning_rate": 6.570453713198992e-06,
+      "loss": 1.0128,
+      "step": 17133
+    },
+    {
+      "epoch": 0.6231225224569953,
+      "grad_norm": 0.0,
+      "learning_rate": 6.569347310309446e-06,
+      "loss": 0.9418,
+      "step": 17134
+    },
+    {
+      "epoch": 0.6231588900607339,
+      "grad_norm": 0.0,
+      "learning_rate": 6.568240955015344e-06,
+      "loss": 0.9429,
+      "step": 17135
+    },
+    {
+      "epoch": 0.6231952576644725,
+      "grad_norm": 0.0,
+      "learning_rate": 6.567134647332023e-06,
+      "loss": 1.0152,
+      "step": 17136
+    },
+    {
+      "epoch": 0.6232316252682111,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5660283872748445e-06,
+      "loss": 1.0672,
+      "step": 17137
+    },
+    {
+      "epoch": 0.6232679928719497,
+      "grad_norm": 0.0,
+      "learning_rate": 6.564922174859145e-06,
+      "loss": 0.9716,
+      "step": 17138
+    },
+    {
+      "epoch": 0.6233043604756883,
+      "grad_norm": 0.0,
+      "learning_rate": 6.563816010100279e-06,
+      "loss": 1.0264,
+      "step": 17139
+    },
+    {
+      "epoch": 0.6233407280794269,
+      "grad_norm": 0.0,
+      "learning_rate": 6.562709893013588e-06,
+      "loss": 1.0512,
+      "step": 17140
+    },
+    {
+      "epoch": 0.6233770956831655,
+      "grad_norm": 0.0,
+      "learning_rate": 6.561603823614424e-06,
+      "loss": 0.9268,
+      "step": 17141
+    },
+    {
+      "epoch": 0.6234134632869041,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5604978019181245e-06,
+      "loss": 0.9824,
+      "step": 17142
+    },
+    {
+      "epoch": 0.6234498308906427,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5593918279400434e-06,
+      "loss": 0.9018,
+      "step": 17143
+    },
+    {
+      "epoch": 0.6234861984943812,
+      "grad_norm": 0.0,
+      "learning_rate": 6.558285901695515e-06,
+      "loss": 0.9712,
+      "step": 17144
+    },
+    {
+      "epoch": 0.6235225660981198,
+      "grad_norm": 0.0,
+      "learning_rate": 6.557180023199888e-06,
+      "loss": 0.9982,
+      "step": 17145
+    },
+    {
+      "epoch": 0.6235589337018583,
+      "grad_norm": 0.0,
+      "learning_rate": 6.556074192468503e-06,
+      "loss": 1.034,
+      "step": 17146
+    },
+    {
+      "epoch": 0.6235953013055969,
+      "grad_norm": 0.0,
+      "learning_rate": 6.554968409516702e-06,
+      "loss": 1.0116,
+      "step": 17147
+    },
+    {
+      "epoch": 0.6236316689093355,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5538626743598255e-06,
+      "loss": 0.8942,
+      "step": 17148
+    },
+    {
+      "epoch": 0.6236680365130741,
+      "grad_norm": 0.0,
+      "learning_rate": 6.552756987013218e-06,
+      "loss": 1.0434,
+      "step": 17149
+    },
+    {
+      "epoch": 0.6237044041168127,
+      "grad_norm": 0.0,
+      "learning_rate": 6.551651347492212e-06,
+      "loss": 0.9933,
+      "step": 17150
+    },
+    {
+      "epoch": 0.6237407717205513,
+      "grad_norm": 0.0,
+      "learning_rate": 6.550545755812156e-06,
+      "loss": 0.9451,
+      "step": 17151
+    },
+    {
+      "epoch": 0.6237771393242899,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5494402119883805e-06,
+      "loss": 0.9147,
+      "step": 17152
+    },
+    {
+      "epoch": 0.6238135069280285,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5483347160362265e-06,
+      "loss": 1.053,
+      "step": 17153
+    },
+    {
+      "epoch": 0.6238498745317671,
+      "grad_norm": 0.0,
+      "learning_rate": 6.547229267971031e-06,
+      "loss": 1.0921,
+      "step": 17154
+    },
+    {
+      "epoch": 0.6238862421355057,
+      "grad_norm": 0.0,
+      "learning_rate": 6.54612386780813e-06,
+      "loss": 0.9122,
+      "step": 17155
+    },
+    {
+      "epoch": 0.6239226097392443,
+      "grad_norm": 0.0,
+      "learning_rate": 6.545018515562864e-06,
+      "loss": 0.9132,
+      "step": 17156
+    },
+    {
+      "epoch": 0.6239589773429829,
+      "grad_norm": 0.0,
+      "learning_rate": 6.543913211250559e-06,
+      "loss": 0.9023,
+      "step": 17157
+    },
+    {
+      "epoch": 0.6239953449467215,
+      "grad_norm": 0.0,
+      "learning_rate": 6.542807954886559e-06,
+      "loss": 1.0403,
+      "step": 17158
+    },
+    {
+      "epoch": 0.6240317125504601,
+      "grad_norm": 0.0,
+      "learning_rate": 6.541702746486189e-06,
+      "loss": 0.8687,
+      "step": 17159
+    },
+    {
+      "epoch": 0.6240680801541987,
+      "grad_norm": 0.0,
+      "learning_rate": 6.54059758606479e-06,
+      "loss": 0.8401,
+      "step": 17160
+    },
+    {
+      "epoch": 0.6241044477579373,
+      "grad_norm": 0.0,
+      "learning_rate": 6.539492473637688e-06,
+      "loss": 0.9471,
+      "step": 17161
+    },
+    {
+      "epoch": 0.6241408153616759,
+      "grad_norm": 0.0,
+      "learning_rate": 6.538387409220224e-06,
+      "loss": 0.9423,
+      "step": 17162
+    },
+    {
+      "epoch": 0.6241771829654144,
+      "grad_norm": 0.0,
+      "learning_rate": 6.537282392827715e-06,
+      "loss": 0.9939,
+      "step": 17163
+    },
+    {
+      "epoch": 0.624213550569153,
+      "grad_norm": 0.0,
+      "learning_rate": 6.536177424475507e-06,
+      "loss": 1.0323,
+      "step": 17164
+    },
+    {
+      "epoch": 0.6242499181728915,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5350725041789185e-06,
+      "loss": 0.9771,
+      "step": 17165
+    },
+    {
+      "epoch": 0.6242862857766301,
+      "grad_norm": 0.0,
+      "learning_rate": 6.533967631953286e-06,
+      "loss": 0.8864,
+      "step": 17166
+    },
+    {
+      "epoch": 0.6243226533803687,
+      "grad_norm": 0.0,
+      "learning_rate": 6.532862807813931e-06,
+      "loss": 1.011,
+      "step": 17167
+    },
+    {
+      "epoch": 0.6243590209841073,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5317580317761894e-06,
+      "loss": 0.8904,
+      "step": 17168
+    },
+    {
+      "epoch": 0.6243953885878459,
+      "grad_norm": 0.0,
+      "learning_rate": 6.530653303855381e-06,
+      "loss": 0.8897,
+      "step": 17169
+    },
+    {
+      "epoch": 0.6244317561915845,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5295486240668395e-06,
+      "loss": 1.0709,
+      "step": 17170
+    },
+    {
+      "epoch": 0.6244681237953231,
+      "grad_norm": 0.0,
+      "learning_rate": 6.528443992425886e-06,
+      "loss": 0.9337,
+      "step": 17171
+    },
+    {
+      "epoch": 0.6245044913990617,
+      "grad_norm": 0.0,
+      "learning_rate": 6.527339408947843e-06,
+      "loss": 1.0441,
+      "step": 17172
+    },
+    {
+      "epoch": 0.6245408590028003,
+      "grad_norm": 0.0,
+      "learning_rate": 6.526234873648043e-06,
+      "loss": 0.9644,
+      "step": 17173
+    },
+    {
+      "epoch": 0.6245772266065389,
+      "grad_norm": 0.0,
+      "learning_rate": 6.525130386541802e-06,
+      "loss": 0.9647,
+      "step": 17174
+    },
+    {
+      "epoch": 0.6246135942102775,
+      "grad_norm": 0.0,
+      "learning_rate": 6.52402594764445e-06,
+      "loss": 1.0417,
+      "step": 17175
+    },
+    {
+      "epoch": 0.6246499618140161,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5229215569713025e-06,
+      "loss": 0.9532,
+      "step": 17176
+    },
+    {
+      "epoch": 0.6246863294177547,
+      "grad_norm": 0.0,
+      "learning_rate": 6.521817214537689e-06,
+      "loss": 0.9648,
+      "step": 17177
+    },
+    {
+      "epoch": 0.6247226970214933,
+      "grad_norm": 0.0,
+      "learning_rate": 6.520712920358923e-06,
+      "loss": 0.9555,
+      "step": 17178
+    },
+    {
+      "epoch": 0.6247590646252319,
+      "grad_norm": 0.0,
+      "learning_rate": 6.519608674450332e-06,
+      "loss": 0.9477,
+      "step": 17179
+    },
+    {
+      "epoch": 0.6247954322289704,
+      "grad_norm": 0.0,
+      "learning_rate": 6.51850447682723e-06,
+      "loss": 1.048,
+      "step": 17180
+    },
+    {
+      "epoch": 0.624831799832709,
+      "grad_norm": 0.0,
+      "learning_rate": 6.517400327504939e-06,
+      "loss": 0.9397,
+      "step": 17181
+    },
+    {
+      "epoch": 0.6248681674364476,
+      "grad_norm": 0.0,
+      "learning_rate": 6.516296226498777e-06,
+      "loss": 0.942,
+      "step": 17182
+    },
+    {
+      "epoch": 0.6249045350401862,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5151921738240675e-06,
+      "loss": 1.0055,
+      "step": 17183
+    },
+    {
+      "epoch": 0.6249409026439248,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5140881694961154e-06,
+      "loss": 0.9327,
+      "step": 17184
+    },
+    {
+      "epoch": 0.6249772702476634,
+      "grad_norm": 0.0,
+      "learning_rate": 6.512984213530248e-06,
+      "loss": 0.8722,
+      "step": 17185
+    },
+    {
+      "epoch": 0.625013637851402,
+      "grad_norm": 0.0,
+      "learning_rate": 6.511880305941775e-06,
+      "loss": 0.9919,
+      "step": 17186
+    },
+    {
+      "epoch": 0.6250500054551406,
+      "grad_norm": 0.0,
+      "learning_rate": 6.510776446746014e-06,
+      "loss": 0.9843,
+      "step": 17187
+    },
+    {
+      "epoch": 0.6250863730588792,
+      "grad_norm": 0.0,
+      "learning_rate": 6.509672635958279e-06,
+      "loss": 0.9519,
+      "step": 17188
+    },
+    {
+      "epoch": 0.6251227406626177,
+      "grad_norm": 0.0,
+      "learning_rate": 6.508568873593888e-06,
+      "loss": 0.866,
+      "step": 17189
+    },
+    {
+      "epoch": 0.6251591082663563,
+      "grad_norm": 0.0,
+      "learning_rate": 6.507465159668149e-06,
+      "loss": 0.9935,
+      "step": 17190
+    },
+    {
+      "epoch": 0.6251954758700949,
+      "grad_norm": 0.0,
+      "learning_rate": 6.506361494196371e-06,
+      "loss": 0.8946,
+      "step": 17191
+    },
+    {
+      "epoch": 0.6252318434738335,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5052578771938754e-06,
+      "loss": 0.972,
+      "step": 17192
+    },
+    {
+      "epoch": 0.6252682110775721,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5041543086759655e-06,
+      "loss": 1.0106,
+      "step": 17193
+    },
+    {
+      "epoch": 0.6253045786813107,
+      "grad_norm": 0.0,
+      "learning_rate": 6.503050788657958e-06,
+      "loss": 0.8973,
+      "step": 17194
+    },
+    {
+      "epoch": 0.6253409462850493,
+      "grad_norm": 0.0,
+      "learning_rate": 6.501947317155153e-06,
+      "loss": 0.9964,
+      "step": 17195
+    },
+    {
+      "epoch": 0.6253773138887879,
+      "grad_norm": 0.0,
+      "learning_rate": 6.5008438941828726e-06,
+      "loss": 0.9793,
+      "step": 17196
+    },
+    {
+      "epoch": 0.6254136814925264,
+      "grad_norm": 0.0,
+      "learning_rate": 6.499740519756413e-06,
+      "loss": 1.0241,
+      "step": 17197
+    },
+    {
+      "epoch": 0.625450049096265,
+      "grad_norm": 0.0,
+      "learning_rate": 6.498637193891091e-06,
+      "loss": 0.9963,
+      "step": 17198
+    },
+    {
+      "epoch": 0.6254864167000036,
+      "grad_norm": 0.0,
+      "learning_rate": 6.497533916602208e-06,
+      "loss": 1.0104,
+      "step": 17199
+    },
+    {
+      "epoch": 0.6255227843037422,
+      "grad_norm": 0.0,
+      "learning_rate": 6.496430687905074e-06,
+      "loss": 0.8268,
+      "step": 17200
+    },
+    {
+      "epoch": 0.6255591519074808,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4953275078149925e-06,
+      "loss": 0.9766,
+      "step": 17201
+    },
+    {
+      "epoch": 0.6255955195112194,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4942243763472716e-06,
+      "loss": 0.9492,
+      "step": 17202
+    },
+    {
+      "epoch": 0.625631887114958,
+      "grad_norm": 0.0,
+      "learning_rate": 6.493121293517208e-06,
+      "loss": 0.8187,
+      "step": 17203
+    },
+    {
+      "epoch": 0.6256682547186966,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4920182593401186e-06,
+      "loss": 0.9266,
+      "step": 17204
+    },
+    {
+      "epoch": 0.6257046223224352,
+      "grad_norm": 0.0,
+      "learning_rate": 6.490915273831293e-06,
+      "loss": 0.9586,
+      "step": 17205
+    },
+    {
+      "epoch": 0.6257409899261738,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4898123370060405e-06,
+      "loss": 0.8893,
+      "step": 17206
+    },
+    {
+      "epoch": 0.6257773575299124,
+      "grad_norm": 0.0,
+      "learning_rate": 6.488709448879661e-06,
+      "loss": 0.9229,
+      "step": 17207
+    },
+    {
+      "epoch": 0.625813725133651,
+      "grad_norm": 0.0,
+      "learning_rate": 6.487606609467457e-06,
+      "loss": 0.9704,
+      "step": 17208
+    },
+    {
+      "epoch": 0.6258500927373896,
+      "grad_norm": 0.0,
+      "learning_rate": 6.486503818784731e-06,
+      "loss": 1.0032,
+      "step": 17209
+    },
+    {
+      "epoch": 0.6258864603411282,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4854010768467745e-06,
+      "loss": 1.0128,
+      "step": 17210
+    },
+    {
+      "epoch": 0.6259228279448668,
+      "grad_norm": 0.0,
+      "learning_rate": 6.484298383668895e-06,
+      "loss": 0.92,
+      "step": 17211
+    },
+    {
+      "epoch": 0.6259591955486054,
+      "grad_norm": 0.0,
+      "learning_rate": 6.483195739266383e-06,
+      "loss": 0.9644,
+      "step": 17212
+    },
+    {
+      "epoch": 0.6259955631523438,
+      "grad_norm": 0.0,
+      "learning_rate": 6.482093143654544e-06,
+      "loss": 1.0087,
+      "step": 17213
+    },
+    {
+      "epoch": 0.6260319307560824,
+      "grad_norm": 0.0,
+      "learning_rate": 6.480990596848668e-06,
+      "loss": 0.9607,
+      "step": 17214
+    },
+    {
+      "epoch": 0.626068298359821,
+      "grad_norm": 0.0,
+      "learning_rate": 6.47988809886406e-06,
+      "loss": 0.9481,
+      "step": 17215
+    },
+    {
+      "epoch": 0.6261046659635596,
+      "grad_norm": 0.0,
+      "learning_rate": 6.478785649716006e-06,
+      "loss": 0.8763,
+      "step": 17216
+    },
+    {
+      "epoch": 0.6261410335672982,
+      "grad_norm": 0.0,
+      "learning_rate": 6.477683249419807e-06,
+      "loss": 0.9432,
+      "step": 17217
+    },
+    {
+      "epoch": 0.6261774011710368,
+      "grad_norm": 0.0,
+      "learning_rate": 6.476580897990754e-06,
+      "loss": 0.9265,
+      "step": 17218
+    },
+    {
+      "epoch": 0.6262137687747754,
+      "grad_norm": 0.0,
+      "learning_rate": 6.475478595444144e-06,
+      "loss": 0.9097,
+      "step": 17219
+    },
+    {
+      "epoch": 0.626250136378514,
+      "grad_norm": 0.0,
+      "learning_rate": 6.474376341795265e-06,
+      "loss": 1.0194,
+      "step": 17220
+    },
+    {
+      "epoch": 0.6262865039822526,
+      "grad_norm": 0.0,
+      "learning_rate": 6.473274137059415e-06,
+      "loss": 0.8842,
+      "step": 17221
+    },
+    {
+      "epoch": 0.6263228715859912,
+      "grad_norm": 0.0,
+      "learning_rate": 6.472171981251878e-06,
+      "loss": 1.0032,
+      "step": 17222
+    },
+    {
+      "epoch": 0.6263592391897298,
+      "grad_norm": 0.0,
+      "learning_rate": 6.471069874387957e-06,
+      "loss": 0.8919,
+      "step": 17223
+    },
+    {
+      "epoch": 0.6263956067934684,
+      "grad_norm": 0.0,
+      "learning_rate": 6.469967816482927e-06,
+      "loss": 0.9443,
+      "step": 17224
+    },
+    {
+      "epoch": 0.626431974397207,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4688658075520895e-06,
+      "loss": 1.0361,
+      "step": 17225
+    },
+    {
+      "epoch": 0.6264683420009456,
+      "grad_norm": 0.0,
+      "learning_rate": 6.467763847610726e-06,
+      "loss": 1.0254,
+      "step": 17226
+    },
+    {
+      "epoch": 0.6265047096046842,
+      "grad_norm": 0.0,
+      "learning_rate": 6.46666193667413e-06,
+      "loss": 0.9083,
+      "step": 17227
+    },
+    {
+      "epoch": 0.6265410772084228,
+      "grad_norm": 0.0,
+      "learning_rate": 6.465560074757589e-06,
+      "loss": 0.9035,
+      "step": 17228
+    },
+    {
+      "epoch": 0.6265774448121614,
+      "grad_norm": 0.0,
+      "learning_rate": 6.464458261876383e-06,
+      "loss": 1.0786,
+      "step": 17229
+    },
+    {
+      "epoch": 0.6266138124158999,
+      "grad_norm": 0.0,
+      "learning_rate": 6.463356498045803e-06,
+      "loss": 1.0052,
+      "step": 17230
+    },
+    {
+      "epoch": 0.6266501800196385,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4622547832811335e-06,
+      "loss": 0.9318,
+      "step": 17231
+    },
+    {
+      "epoch": 0.626686547623377,
+      "grad_norm": 0.0,
+      "learning_rate": 6.461153117597661e-06,
+      "loss": 0.8914,
+      "step": 17232
+    },
+    {
+      "epoch": 0.6267229152271157,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4600515010106645e-06,
+      "loss": 0.902,
+      "step": 17233
+    },
+    {
+      "epoch": 0.6267592828308542,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4589499335354365e-06,
+      "loss": 0.9808,
+      "step": 17234
+    },
+    {
+      "epoch": 0.6267956504345928,
+      "grad_norm": 0.0,
+      "learning_rate": 6.457848415187248e-06,
+      "loss": 0.8949,
+      "step": 17235
+    },
+    {
+      "epoch": 0.6268320180383314,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4567469459813934e-06,
+      "loss": 1.0847,
+      "step": 17236
+    },
+    {
+      "epoch": 0.62686838564207,
+      "grad_norm": 0.0,
+      "learning_rate": 6.455645525933143e-06,
+      "loss": 0.9521,
+      "step": 17237
+    },
+    {
+      "epoch": 0.6269047532458086,
+      "grad_norm": 0.0,
+      "learning_rate": 6.454544155057784e-06,
+      "loss": 0.9144,
+      "step": 17238
+    },
+    {
+      "epoch": 0.6269411208495472,
+      "grad_norm": 0.0,
+      "learning_rate": 6.453442833370592e-06,
+      "loss": 0.9614,
+      "step": 17239
+    },
+    {
+      "epoch": 0.6269774884532858,
+      "grad_norm": 0.0,
+      "learning_rate": 6.452341560886852e-06,
+      "loss": 0.9454,
+      "step": 17240
+    },
+    {
+      "epoch": 0.6270138560570244,
+      "grad_norm": 0.0,
+      "learning_rate": 6.451240337621837e-06,
+      "loss": 0.9359,
+      "step": 17241
+    },
+    {
+      "epoch": 0.627050223660763,
+      "grad_norm": 0.0,
+      "learning_rate": 6.450139163590832e-06,
+      "loss": 1.0477,
+      "step": 17242
+    },
+    {
+      "epoch": 0.6270865912645016,
+      "grad_norm": 0.0,
+      "learning_rate": 6.449038038809103e-06,
+      "loss": 0.8974,
+      "step": 17243
+    },
+    {
+      "epoch": 0.6271229588682402,
+      "grad_norm": 0.0,
+      "learning_rate": 6.447936963291938e-06,
+      "loss": 0.9808,
+      "step": 17244
+    },
+    {
+      "epoch": 0.6271593264719788,
+      "grad_norm": 0.0,
+      "learning_rate": 6.446835937054604e-06,
+      "loss": 1.0273,
+      "step": 17245
+    },
+    {
+      "epoch": 0.6271956940757174,
+      "grad_norm": 0.0,
+      "learning_rate": 6.445734960112383e-06,
+      "loss": 1.0097,
+      "step": 17246
+    },
+    {
+      "epoch": 0.6272320616794559,
+      "grad_norm": 0.0,
+      "learning_rate": 6.444634032480548e-06,
+      "loss": 0.899,
+      "step": 17247
+    },
+    {
+      "epoch": 0.6272684292831945,
+      "grad_norm": 0.0,
+      "learning_rate": 6.443533154174366e-06,
+      "loss": 0.9054,
+      "step": 17248
+    },
+    {
+      "epoch": 0.6273047968869331,
+      "grad_norm": 0.0,
+      "learning_rate": 6.442432325209121e-06,
+      "loss": 0.9174,
+      "step": 17249
+    },
+    {
+      "epoch": 0.6273411644906717,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4413315456000754e-06,
+      "loss": 0.9689,
+      "step": 17250
+    },
+    {
+      "epoch": 0.6273775320944103,
+      "grad_norm": 0.0,
+      "learning_rate": 6.440230815362509e-06,
+      "loss": 0.9013,
+      "step": 17251
+    },
+    {
+      "epoch": 0.6274138996981489,
+      "grad_norm": 0.0,
+      "learning_rate": 6.439130134511686e-06,
+      "loss": 1.008,
+      "step": 17252
+    },
+    {
+      "epoch": 0.6274502673018875,
+      "grad_norm": 0.0,
+      "learning_rate": 6.438029503062884e-06,
+      "loss": 0.9325,
+      "step": 17253
+    },
+    {
+      "epoch": 0.6274866349056261,
+      "grad_norm": 0.0,
+      "learning_rate": 6.436928921031366e-06,
+      "loss": 1.0432,
+      "step": 17254
+    },
+    {
+      "epoch": 0.6275230025093647,
+      "grad_norm": 0.0,
+      "learning_rate": 6.435828388432408e-06,
+      "loss": 0.949,
+      "step": 17255
+    },
+    {
+      "epoch": 0.6275593701131033,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4347279052812695e-06,
+      "loss": 0.7985,
+      "step": 17256
+    },
+    {
+      "epoch": 0.6275957377168419,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4336274715932245e-06,
+      "loss": 0.9978,
+      "step": 17257
+    },
+    {
+      "epoch": 0.6276321053205804,
+      "grad_norm": 0.0,
+      "learning_rate": 6.432527087383538e-06,
+      "loss": 0.8974,
+      "step": 17258
+    },
+    {
+      "epoch": 0.627668472924319,
+      "grad_norm": 0.0,
+      "learning_rate": 6.431426752667478e-06,
+      "loss": 0.9088,
+      "step": 17259
+    },
+    {
+      "epoch": 0.6277048405280576,
+      "grad_norm": 0.0,
+      "learning_rate": 6.430326467460306e-06,
+      "loss": 0.8389,
+      "step": 17260
+    },
+    {
+      "epoch": 0.6277412081317962,
+      "grad_norm": 0.0,
+      "learning_rate": 6.429226231777293e-06,
+      "loss": 1.018,
+      "step": 17261
+    },
+    {
+      "epoch": 0.6277775757355348,
+      "grad_norm": 0.0,
+      "learning_rate": 6.428126045633695e-06,
+      "loss": 1.1049,
+      "step": 17262
+    },
+    {
+      "epoch": 0.6278139433392734,
+      "grad_norm": 0.0,
+      "learning_rate": 6.427025909044786e-06,
+      "loss": 0.883,
+      "step": 17263
+    },
+    {
+      "epoch": 0.6278503109430119,
+      "grad_norm": 0.0,
+      "learning_rate": 6.425925822025818e-06,
+      "loss": 1.0291,
+      "step": 17264
+    },
+    {
+      "epoch": 0.6278866785467505,
+      "grad_norm": 0.0,
+      "learning_rate": 6.424825784592061e-06,
+      "loss": 1.1367,
+      "step": 17265
+    },
+    {
+      "epoch": 0.6279230461504891,
+      "grad_norm": 0.0,
+      "learning_rate": 6.423725796758777e-06,
+      "loss": 1.0689,
+      "step": 17266
+    },
+    {
+      "epoch": 0.6279594137542277,
+      "grad_norm": 0.0,
+      "learning_rate": 6.422625858541216e-06,
+      "loss": 0.9822,
+      "step": 17267
+    },
+    {
+      "epoch": 0.6279957813579663,
+      "grad_norm": 0.0,
+      "learning_rate": 6.421525969954653e-06,
+      "loss": 0.9648,
+      "step": 17268
+    },
+    {
+      "epoch": 0.6280321489617049,
+      "grad_norm": 0.0,
+      "learning_rate": 6.420426131014335e-06,
+      "loss": 0.9833,
+      "step": 17269
+    },
+    {
+      "epoch": 0.6280685165654435,
+      "grad_norm": 0.0,
+      "learning_rate": 6.419326341735528e-06,
+      "loss": 0.9716,
+      "step": 17270
+    },
+    {
+      "epoch": 0.6281048841691821,
+      "grad_norm": 0.0,
+      "learning_rate": 6.418226602133485e-06,
+      "loss": 1.0063,
+      "step": 17271
+    },
+    {
+      "epoch": 0.6281412517729207,
+      "grad_norm": 0.0,
+      "learning_rate": 6.417126912223468e-06,
+      "loss": 0.9977,
+      "step": 17272
+    },
+    {
+      "epoch": 0.6281776193766593,
+      "grad_norm": 0.0,
+      "learning_rate": 6.416027272020731e-06,
+      "loss": 0.9989,
+      "step": 17273
+    },
+    {
+      "epoch": 0.6282139869803979,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4149276815405325e-06,
+      "loss": 1.02,
+      "step": 17274
+    },
+    {
+      "epoch": 0.6282503545841365,
+      "grad_norm": 0.0,
+      "learning_rate": 6.413828140798121e-06,
+      "loss": 0.9678,
+      "step": 17275
+    },
+    {
+      "epoch": 0.6282867221878751,
+      "grad_norm": 0.0,
+      "learning_rate": 6.412728649808763e-06,
+      "loss": 0.9448,
+      "step": 17276
+    },
+    {
+      "epoch": 0.6283230897916137,
+      "grad_norm": 0.0,
+      "learning_rate": 6.4116292085877e-06,
+      "loss": 0.9605,
+      "step": 17277
+    },
+    {
+      "epoch": 0.6283594573953523,
+      "grad_norm": 0.0,
+      "learning_rate": 6.410529817150193e-06,
+      "loss": 1.0928,
+      "step": 17278
+    },
+    {
+      "epoch": 0.6283958249990909,
+      "grad_norm": 0.0,
+      "learning_rate": 6.40943047551149e-06,
+      "loss": 0.8877,
+      "step": 17279
+    },
+    {
+      "epoch": 0.6284321926028295,
+      "grad_norm": 0.0,
+      "learning_rate": 6.408331183686847e-06,
+      "loss": 0.9358,
+      "step": 17280
+    },
+    {
+      "epoch": 0.6284685602065679,
+      "grad_norm": 0.0,
+      "learning_rate": 6.40723194169151e-06,
+      "loss": 1.0677,
+      "step": 17281
+    },
+    {
+      "epoch": 0.6285049278103065,
+      "grad_norm": 0.0,
+      "learning_rate": 6.406132749540737e-06,
+      "loss": 1.0015,
+      "step": 17282
+    },
+    {
+      "epoch": 0.6285412954140451,
+      "grad_norm": 0.0,
+      "learning_rate": 6.405033607249769e-06,
+      "loss": 0.8609,
+      "step": 17283
+    },
+    {
+      "epoch": 0.6285776630177837,
+      "grad_norm": 0.0,
+      "learning_rate": 6.40393451483386e-06,
+      "loss": 1.0904,
+      "step": 17284
+    },
+    {
+      "epoch": 0.6286140306215223,
+      "grad_norm": 0.0,
+      "learning_rate": 6.40283547230826e-06,
+      "loss": 0.9674,
+      "step": 17285
+    },
+    {
+      "epoch": 0.6286503982252609,
+      "grad_norm": 0.0,
+      "learning_rate": 6.40173647968821e-06,
+      "loss": 0.8889,
+      "step": 17286
+    },
+    {
+      "epoch": 0.6286867658289995,
+      "grad_norm": 0.0,
+      "learning_rate": 6.400637536988965e-06,
+      "loss": 0.9598,
+      "step": 17287
+    },
+    {
+      "epoch": 0.6287231334327381,
+      "grad_norm": 0.0,
+      "learning_rate": 6.399538644225764e-06,
+      "loss": 0.9192,
+      "step": 17288
+    },
+    {
+      "epoch": 0.6287595010364767,
+      "grad_norm": 0.0,
+      "learning_rate": 6.398439801413858e-06,
+      "loss": 1.0705,
+      "step": 17289
+    },
+    {
+      "epoch": 0.6287958686402153,
+      "grad_norm": 0.0,
+      "learning_rate": 6.397341008568488e-06,
+      "loss": 0.9106,
+      "step": 17290
+    },
+    {
+      "epoch": 0.6288322362439539,
+      "grad_norm": 0.0,
+      "learning_rate": 6.396242265704902e-06,
+      "loss": 0.9396,
+      "step": 17291
+    },
+    {
+      "epoch": 0.6288686038476925,
+      "grad_norm": 0.0,
+      "learning_rate": 6.395143572838341e-06,
+      "loss": 0.9043,
+      "step": 17292
+    },
+    {
+      "epoch": 0.6289049714514311,
+      "grad_norm": 0.0,
+      "learning_rate": 6.394044929984046e-06,
+      "loss": 1.0273,
+      "step": 17293
+    },
+    {
+      "epoch": 0.6289413390551697,
+      "grad_norm": 0.0,
+      "learning_rate": 6.392946337157263e-06,
+      "loss": 1.0791,
+      "step": 17294
+    },
+    {
+      "epoch": 0.6289777066589083,
+      "grad_norm": 0.0,
+      "learning_rate": 6.391847794373236e-06,
+      "loss": 0.9949,
+      "step": 17295
+    },
+    {
+      "epoch": 0.6290140742626469,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3907493016471955e-06,
+      "loss": 1.0505,
+      "step": 17296
+    },
+    {
+      "epoch": 0.6290504418663855,
+      "grad_norm": 0.0,
+      "learning_rate": 6.389650858994391e-06,
+      "loss": 0.8678,
+      "step": 17297
+    },
+    {
+      "epoch": 0.629086809470124,
+      "grad_norm": 0.0,
+      "learning_rate": 6.388552466430056e-06,
+      "loss": 0.912,
+      "step": 17298
+    },
+    {
+      "epoch": 0.6291231770738626,
+      "grad_norm": 0.0,
+      "learning_rate": 6.387454123969433e-06,
+      "loss": 0.9603,
+      "step": 17299
+    },
+    {
+      "epoch": 0.6291595446776012,
+      "grad_norm": 0.0,
+      "learning_rate": 6.386355831627757e-06,
+      "loss": 0.849,
+      "step": 17300
+    },
+    {
+      "epoch": 0.6291959122813398,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3852575894202705e-06,
+      "loss": 0.9473,
+      "step": 17301
+    },
+    {
+      "epoch": 0.6292322798850783,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3841593973622e-06,
+      "loss": 0.9047,
+      "step": 17302
+    },
+    {
+      "epoch": 0.629268647488817,
+      "grad_norm": 0.0,
+      "learning_rate": 6.383061255468794e-06,
+      "loss": 1.0793,
+      "step": 17303
+    },
+    {
+      "epoch": 0.6293050150925555,
+      "grad_norm": 0.0,
+      "learning_rate": 6.38196316375528e-06,
+      "loss": 1.0625,
+      "step": 17304
+    },
+    {
+      "epoch": 0.6293413826962941,
+      "grad_norm": 0.0,
+      "learning_rate": 6.380865122236892e-06,
+      "loss": 1.0945,
+      "step": 17305
+    },
+    {
+      "epoch": 0.6293777503000327,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3797671309288685e-06,
+      "loss": 0.9262,
+      "step": 17306
+    },
+    {
+      "epoch": 0.6294141179037713,
+      "grad_norm": 0.0,
+      "learning_rate": 6.378669189846434e-06,
+      "loss": 0.9868,
+      "step": 17307
+    },
+    {
+      "epoch": 0.6294504855075099,
+      "grad_norm": 0.0,
+      "learning_rate": 6.377571299004832e-06,
+      "loss": 0.8677,
+      "step": 17308
+    },
+    {
+      "epoch": 0.6294868531112485,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3764734584192855e-06,
+      "loss": 0.9732,
+      "step": 17309
+    },
+    {
+      "epoch": 0.6295232207149871,
+      "grad_norm": 0.0,
+      "learning_rate": 6.375375668105031e-06,
+      "loss": 0.8511,
+      "step": 17310
+    },
+    {
+      "epoch": 0.6295595883187257,
+      "grad_norm": 0.0,
+      "learning_rate": 6.374277928077292e-06,
+      "loss": 0.9888,
+      "step": 17311
+    },
+    {
+      "epoch": 0.6295959559224643,
+      "grad_norm": 0.0,
+      "learning_rate": 6.373180238351306e-06,
+      "loss": 1.0826,
+      "step": 17312
+    },
+    {
+      "epoch": 0.6296323235262029,
+      "grad_norm": 0.0,
+      "learning_rate": 6.372082598942297e-06,
+      "loss": 0.9576,
+      "step": 17313
+    },
+    {
+      "epoch": 0.6296686911299414,
+      "grad_norm": 0.0,
+      "learning_rate": 6.370985009865498e-06,
+      "loss": 1.0795,
+      "step": 17314
+    },
+    {
+      "epoch": 0.62970505873368,
+      "grad_norm": 0.0,
+      "learning_rate": 6.369887471136131e-06,
+      "loss": 1.0362,
+      "step": 17315
+    },
+    {
+      "epoch": 0.6297414263374186,
+      "grad_norm": 0.0,
+      "learning_rate": 6.368789982769425e-06,
+      "loss": 1.0666,
+      "step": 17316
+    },
+    {
+      "epoch": 0.6297777939411572,
+      "grad_norm": 0.0,
+      "learning_rate": 6.367692544780604e-06,
+      "loss": 0.964,
+      "step": 17317
+    },
+    {
+      "epoch": 0.6298141615448958,
+      "grad_norm": 0.0,
+      "learning_rate": 6.366595157184896e-06,
+      "loss": 0.922,
+      "step": 17318
+    },
+    {
+      "epoch": 0.6298505291486344,
+      "grad_norm": 0.0,
+      "learning_rate": 6.365497819997526e-06,
+      "loss": 1.0653,
+      "step": 17319
+    },
+    {
+      "epoch": 0.629886896752373,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3644005332337164e-06,
+      "loss": 1.0399,
+      "step": 17320
+    },
+    {
+      "epoch": 0.6299232643561116,
+      "grad_norm": 0.0,
+      "learning_rate": 6.363303296908692e-06,
+      "loss": 0.9256,
+      "step": 17321
+    },
+    {
+      "epoch": 0.6299596319598502,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3622061110376765e-06,
+      "loss": 0.9522,
+      "step": 17322
+    },
+    {
+      "epoch": 0.6299959995635888,
+      "grad_norm": 0.0,
+      "learning_rate": 6.361108975635888e-06,
+      "loss": 1.0116,
+      "step": 17323
+    },
+    {
+      "epoch": 0.6300323671673274,
+      "grad_norm": 0.0,
+      "learning_rate": 6.360011890718547e-06,
+      "loss": 0.9451,
+      "step": 17324
+    },
+    {
+      "epoch": 0.630068734771066,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3589148563008796e-06,
+      "loss": 0.9576,
+      "step": 17325
+    },
+    {
+      "epoch": 0.6301051023748045,
+      "grad_norm": 0.0,
+      "learning_rate": 6.357817872398101e-06,
+      "loss": 0.9279,
+      "step": 17326
+    },
+    {
+      "epoch": 0.6301414699785431,
+      "grad_norm": 0.0,
+      "learning_rate": 6.356720939025436e-06,
+      "loss": 1.0143,
+      "step": 17327
+    },
+    {
+      "epoch": 0.6301778375822817,
+      "grad_norm": 0.0,
+      "learning_rate": 6.355624056198093e-06,
+      "loss": 0.8395,
+      "step": 17328
+    },
+    {
+      "epoch": 0.6302142051860203,
+      "grad_norm": 0.0,
+      "learning_rate": 6.354527223931299e-06,
+      "loss": 0.985,
+      "step": 17329
+    },
+    {
+      "epoch": 0.6302505727897589,
+      "grad_norm": 0.0,
+      "learning_rate": 6.353430442240265e-06,
+      "loss": 1.0019,
+      "step": 17330
+    },
+    {
+      "epoch": 0.6302869403934974,
+      "grad_norm": 0.0,
+      "learning_rate": 6.352333711140212e-06,
+      "loss": 0.9559,
+      "step": 17331
+    },
+    {
+      "epoch": 0.630323307997236,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3512370306463515e-06,
+      "loss": 0.8533,
+      "step": 17332
+    },
+    {
+      "epoch": 0.6303596756009746,
+      "grad_norm": 0.0,
+      "learning_rate": 6.350140400773903e-06,
+      "loss": 0.9139,
+      "step": 17333
+    },
+    {
+      "epoch": 0.6303960432047132,
+      "grad_norm": 0.0,
+      "learning_rate": 6.349043821538074e-06,
+      "loss": 0.9991,
+      "step": 17334
+    },
+    {
+      "epoch": 0.6304324108084518,
+      "grad_norm": 0.0,
+      "learning_rate": 6.347947292954087e-06,
+      "loss": 0.8712,
+      "step": 17335
+    },
+    {
+      "epoch": 0.6304687784121904,
+      "grad_norm": 0.0,
+      "learning_rate": 6.346850815037145e-06,
+      "loss": 1.0843,
+      "step": 17336
+    },
+    {
+      "epoch": 0.630505146015929,
+      "grad_norm": 0.0,
+      "learning_rate": 6.345754387802467e-06,
+      "loss": 1.0586,
+      "step": 17337
+    },
+    {
+      "epoch": 0.6305415136196676,
+      "grad_norm": 0.0,
+      "learning_rate": 6.344658011265261e-06,
+      "loss": 0.9312,
+      "step": 17338
+    },
+    {
+      "epoch": 0.6305778812234062,
+      "grad_norm": 0.0,
+      "learning_rate": 6.343561685440741e-06,
+      "loss": 0.9652,
+      "step": 17339
+    },
+    {
+      "epoch": 0.6306142488271448,
+      "grad_norm": 0.0,
+      "learning_rate": 6.342465410344112e-06,
+      "loss": 0.9909,
+      "step": 17340
+    },
+    {
+      "epoch": 0.6306506164308834,
+      "grad_norm": 0.0,
+      "learning_rate": 6.341369185990591e-06,
+      "loss": 0.9134,
+      "step": 17341
+    },
+    {
+      "epoch": 0.630686984034622,
+      "grad_norm": 0.0,
+      "learning_rate": 6.340273012395379e-06,
+      "loss": 0.9108,
+      "step": 17342
+    },
+    {
+      "epoch": 0.6307233516383606,
+      "grad_norm": 0.0,
+      "learning_rate": 6.339176889573684e-06,
+      "loss": 1.0015,
+      "step": 17343
+    },
+    {
+      "epoch": 0.6307597192420992,
+      "grad_norm": 0.0,
+      "learning_rate": 6.338080817540719e-06,
+      "loss": 0.8942,
+      "step": 17344
+    },
+    {
+      "epoch": 0.6307960868458378,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3369847963116835e-06,
+      "loss": 0.943,
+      "step": 17345
+    },
+    {
+      "epoch": 0.6308324544495764,
+      "grad_norm": 0.0,
+      "learning_rate": 6.335888825901792e-06,
+      "loss": 0.9462,
+      "step": 17346
+    },
+    {
+      "epoch": 0.630868822053315,
+      "grad_norm": 0.0,
+      "learning_rate": 6.334792906326237e-06,
+      "loss": 0.8731,
+      "step": 17347
+    },
+    {
+      "epoch": 0.6309051896570534,
+      "grad_norm": 0.0,
+      "learning_rate": 6.333697037600237e-06,
+      "loss": 0.955,
+      "step": 17348
+    },
+    {
+      "epoch": 0.630941557260792,
+      "grad_norm": 0.0,
+      "learning_rate": 6.332601219738985e-06,
+      "loss": 0.9972,
+      "step": 17349
+    },
+    {
+      "epoch": 0.6309779248645306,
+      "grad_norm": 0.0,
+      "learning_rate": 6.331505452757691e-06,
+      "loss": 0.9912,
+      "step": 17350
+    },
+    {
+      "epoch": 0.6310142924682692,
+      "grad_norm": 0.0,
+      "learning_rate": 6.33040973667155e-06,
+      "loss": 0.9809,
+      "step": 17351
+    },
+    {
+      "epoch": 0.6310506600720078,
+      "grad_norm": 0.0,
+      "learning_rate": 6.329314071495769e-06,
+      "loss": 0.8855,
+      "step": 17352
+    },
+    {
+      "epoch": 0.6310870276757464,
+      "grad_norm": 0.0,
+      "learning_rate": 6.328218457245547e-06,
+      "loss": 1.0372,
+      "step": 17353
+    },
+    {
+      "epoch": 0.631123395279485,
+      "grad_norm": 0.0,
+      "learning_rate": 6.327122893936087e-06,
+      "loss": 1.0279,
+      "step": 17354
+    },
+    {
+      "epoch": 0.6311597628832236,
+      "grad_norm": 0.0,
+      "learning_rate": 6.326027381582583e-06,
+      "loss": 0.9961,
+      "step": 17355
+    },
+    {
+      "epoch": 0.6311961304869622,
+      "grad_norm": 0.0,
+      "learning_rate": 6.324931920200237e-06,
+      "loss": 1.0044,
+      "step": 17356
+    },
+    {
+      "epoch": 0.6312324980907008,
+      "grad_norm": 0.0,
+      "learning_rate": 6.323836509804244e-06,
+      "loss": 1.0159,
+      "step": 17357
+    },
+    {
+      "epoch": 0.6312688656944394,
+      "grad_norm": 0.0,
+      "learning_rate": 6.322741150409808e-06,
+      "loss": 0.8984,
+      "step": 17358
+    },
+    {
+      "epoch": 0.631305233298178,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3216458420321185e-06,
+      "loss": 0.9303,
+      "step": 17359
+    },
+    {
+      "epoch": 0.6313416009019166,
+      "grad_norm": 0.0,
+      "learning_rate": 6.320550584686376e-06,
+      "loss": 0.9672,
+      "step": 17360
+    },
+    {
+      "epoch": 0.6313779685056552,
+      "grad_norm": 0.0,
+      "learning_rate": 6.319455378387773e-06,
+      "loss": 0.9424,
+      "step": 17361
+    },
+    {
+      "epoch": 0.6314143361093938,
+      "grad_norm": 0.0,
+      "learning_rate": 6.318360223151503e-06,
+      "loss": 0.9305,
+      "step": 17362
+    },
+    {
+      "epoch": 0.6314507037131324,
+      "grad_norm": 0.0,
+      "learning_rate": 6.317265118992764e-06,
+      "loss": 0.9767,
+      "step": 17363
+    },
+    {
+      "epoch": 0.631487071316871,
+      "grad_norm": 0.0,
+      "learning_rate": 6.316170065926742e-06,
+      "loss": 0.9033,
+      "step": 17364
+    },
+    {
+      "epoch": 0.6315234389206095,
+      "grad_norm": 0.0,
+      "learning_rate": 6.315075063968636e-06,
+      "loss": 0.8678,
+      "step": 17365
+    },
+    {
+      "epoch": 0.6315598065243481,
+      "grad_norm": 0.0,
+      "learning_rate": 6.313980113133636e-06,
+      "loss": 1.0712,
+      "step": 17366
+    },
+    {
+      "epoch": 0.6315961741280867,
+      "grad_norm": 0.0,
+      "learning_rate": 6.312885213436934e-06,
+      "loss": 1.0096,
+      "step": 17367
+    },
+    {
+      "epoch": 0.6316325417318253,
+      "grad_norm": 0.0,
+      "learning_rate": 6.311790364893715e-06,
+      "loss": 0.9629,
+      "step": 17368
+    },
+    {
+      "epoch": 0.6316689093355639,
+      "grad_norm": 0.0,
+      "learning_rate": 6.310695567519174e-06,
+      "loss": 0.925,
+      "step": 17369
+    },
+    {
+      "epoch": 0.6317052769393025,
+      "grad_norm": 0.0,
+      "learning_rate": 6.309600821328495e-06,
+      "loss": 0.9438,
+      "step": 17370
+    },
+    {
+      "epoch": 0.631741644543041,
+      "grad_norm": 0.0,
+      "learning_rate": 6.308506126336871e-06,
+      "loss": 1.0371,
+      "step": 17371
+    },
+    {
+      "epoch": 0.6317780121467796,
+      "grad_norm": 0.0,
+      "learning_rate": 6.307411482559484e-06,
+      "loss": 0.9347,
+      "step": 17372
+    },
+    {
+      "epoch": 0.6318143797505182,
+      "grad_norm": 0.0,
+      "learning_rate": 6.306316890011528e-06,
+      "loss": 1.0089,
+      "step": 17373
+    },
+    {
+      "epoch": 0.6318507473542568,
+      "grad_norm": 0.0,
+      "learning_rate": 6.305222348708178e-06,
+      "loss": 0.8437,
+      "step": 17374
+    },
+    {
+      "epoch": 0.6318871149579954,
+      "grad_norm": 0.0,
+      "learning_rate": 6.304127858664632e-06,
+      "loss": 1.0337,
+      "step": 17375
+    },
+    {
+      "epoch": 0.631923482561734,
+      "grad_norm": 0.0,
+      "learning_rate": 6.303033419896065e-06,
+      "loss": 1.0045,
+      "step": 17376
+    },
+    {
+      "epoch": 0.6319598501654726,
+      "grad_norm": 0.0,
+      "learning_rate": 6.301939032417667e-06,
+      "loss": 0.9632,
+      "step": 17377
+    },
+    {
+      "epoch": 0.6319962177692112,
+      "grad_norm": 0.0,
+      "learning_rate": 6.300844696244615e-06,
+      "loss": 0.9002,
+      "step": 17378
+    },
+    {
+      "epoch": 0.6320325853729498,
+      "grad_norm": 0.0,
+      "learning_rate": 6.299750411392096e-06,
+      "loss": 0.9164,
+      "step": 17379
+    },
+    {
+      "epoch": 0.6320689529766884,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2986561778752875e-06,
+      "loss": 1.0367,
+      "step": 17380
+    },
+    {
+      "epoch": 0.632105320580427,
+      "grad_norm": 0.0,
+      "learning_rate": 6.297561995709378e-06,
+      "loss": 0.9103,
+      "step": 17381
+    },
+    {
+      "epoch": 0.6321416881841655,
+      "grad_norm": 0.0,
+      "learning_rate": 6.29646786490954e-06,
+      "loss": 0.882,
+      "step": 17382
+    },
+    {
+      "epoch": 0.6321780557879041,
+      "grad_norm": 0.0,
+      "learning_rate": 6.295373785490955e-06,
+      "loss": 1.0617,
+      "step": 17383
+    },
+    {
+      "epoch": 0.6322144233916427,
+      "grad_norm": 0.0,
+      "learning_rate": 6.294279757468803e-06,
+      "loss": 1.0528,
+      "step": 17384
+    },
+    {
+      "epoch": 0.6322507909953813,
+      "grad_norm": 0.0,
+      "learning_rate": 6.293185780858261e-06,
+      "loss": 0.9112,
+      "step": 17385
+    },
+    {
+      "epoch": 0.6322871585991199,
+      "grad_norm": 0.0,
+      "learning_rate": 6.29209185567451e-06,
+      "loss": 0.964,
+      "step": 17386
+    },
+    {
+      "epoch": 0.6323235262028585,
+      "grad_norm": 0.0,
+      "learning_rate": 6.29099798193272e-06,
+      "loss": 0.9935,
+      "step": 17387
+    },
+    {
+      "epoch": 0.6323598938065971,
+      "grad_norm": 0.0,
+      "learning_rate": 6.289904159648073e-06,
+      "loss": 1.0018,
+      "step": 17388
+    },
+    {
+      "epoch": 0.6323962614103357,
+      "grad_norm": 0.0,
+      "learning_rate": 6.28881038883574e-06,
+      "loss": 0.8843,
+      "step": 17389
+    },
+    {
+      "epoch": 0.6324326290140743,
+      "grad_norm": 0.0,
+      "learning_rate": 6.287716669510898e-06,
+      "loss": 0.9571,
+      "step": 17390
+    },
+    {
+      "epoch": 0.6324689966178129,
+      "grad_norm": 0.0,
+      "learning_rate": 6.286623001688718e-06,
+      "loss": 0.9352,
+      "step": 17391
+    },
+    {
+      "epoch": 0.6325053642215515,
+      "grad_norm": 0.0,
+      "learning_rate": 6.285529385384379e-06,
+      "loss": 0.9479,
+      "step": 17392
+    },
+    {
+      "epoch": 0.63254173182529,
+      "grad_norm": 0.0,
+      "learning_rate": 6.284435820613047e-06,
+      "loss": 1.0308,
+      "step": 17393
+    },
+    {
+      "epoch": 0.6325780994290287,
+      "grad_norm": 0.0,
+      "learning_rate": 6.283342307389901e-06,
+      "loss": 0.9583,
+      "step": 17394
+    },
+    {
+      "epoch": 0.6326144670327672,
+      "grad_norm": 0.0,
+      "learning_rate": 6.282248845730102e-06,
+      "loss": 0.9507,
+      "step": 17395
+    },
+    {
+      "epoch": 0.6326508346365058,
+      "grad_norm": 0.0,
+      "learning_rate": 6.281155435648829e-06,
+      "loss": 0.8601,
+      "step": 17396
+    },
+    {
+      "epoch": 0.6326872022402444,
+      "grad_norm": 0.0,
+      "learning_rate": 6.280062077161244e-06,
+      "loss": 1.0041,
+      "step": 17397
+    },
+    {
+      "epoch": 0.6327235698439829,
+      "grad_norm": 0.0,
+      "learning_rate": 6.278968770282523e-06,
+      "loss": 0.9511,
+      "step": 17398
+    },
+    {
+      "epoch": 0.6327599374477215,
+      "grad_norm": 0.0,
+      "learning_rate": 6.277875515027829e-06,
+      "loss": 0.9602,
+      "step": 17399
+    },
+    {
+      "epoch": 0.6327963050514601,
+      "grad_norm": 0.0,
+      "learning_rate": 6.276782311412335e-06,
+      "loss": 0.9561,
+      "step": 17400
+    },
+    {
+      "epoch": 0.6328326726551987,
+      "grad_norm": 0.0,
+      "learning_rate": 6.275689159451202e-06,
+      "loss": 0.8879,
+      "step": 17401
+    },
+    {
+      "epoch": 0.6328690402589373,
+      "grad_norm": 0.0,
+      "learning_rate": 6.274596059159595e-06,
+      "loss": 0.8591,
+      "step": 17402
+    },
+    {
+      "epoch": 0.6329054078626759,
+      "grad_norm": 0.0,
+      "learning_rate": 6.273503010552684e-06,
+      "loss": 0.9959,
+      "step": 17403
+    },
+    {
+      "epoch": 0.6329417754664145,
+      "grad_norm": 0.0,
+      "learning_rate": 6.27241001364563e-06,
+      "loss": 1.0535,
+      "step": 17404
+    },
+    {
+      "epoch": 0.6329781430701531,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2713170684536015e-06,
+      "loss": 1.0491,
+      "step": 17405
+    },
+    {
+      "epoch": 0.6330145106738917,
+      "grad_norm": 0.0,
+      "learning_rate": 6.270224174991755e-06,
+      "loss": 0.9257,
+      "step": 17406
+    },
+    {
+      "epoch": 0.6330508782776303,
+      "grad_norm": 0.0,
+      "learning_rate": 6.26913133327526e-06,
+      "loss": 1.0511,
+      "step": 17407
+    },
+    {
+      "epoch": 0.6330872458813689,
+      "grad_norm": 0.0,
+      "learning_rate": 6.26803854331927e-06,
+      "loss": 1.0237,
+      "step": 17408
+    },
+    {
+      "epoch": 0.6331236134851075,
+      "grad_norm": 0.0,
+      "learning_rate": 6.266945805138953e-06,
+      "loss": 0.8889,
+      "step": 17409
+    },
+    {
+      "epoch": 0.6331599810888461,
+      "grad_norm": 0.0,
+      "learning_rate": 6.265853118749466e-06,
+      "loss": 1.0418,
+      "step": 17410
+    },
+    {
+      "epoch": 0.6331963486925847,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2647604841659684e-06,
+      "loss": 0.9641,
+      "step": 17411
+    },
+    {
+      "epoch": 0.6332327162963233,
+      "grad_norm": 0.0,
+      "learning_rate": 6.263667901403618e-06,
+      "loss": 0.917,
+      "step": 17412
+    },
+    {
+      "epoch": 0.6332690839000619,
+      "grad_norm": 0.0,
+      "learning_rate": 6.262575370477579e-06,
+      "loss": 0.9728,
+      "step": 17413
+    },
+    {
+      "epoch": 0.6333054515038005,
+      "grad_norm": 0.0,
+      "learning_rate": 6.261482891402997e-06,
+      "loss": 1.0227,
+      "step": 17414
+    },
+    {
+      "epoch": 0.633341819107539,
+      "grad_norm": 0.0,
+      "learning_rate": 6.260390464195043e-06,
+      "loss": 1.0669,
+      "step": 17415
+    },
+    {
+      "epoch": 0.6333781867112775,
+      "grad_norm": 0.0,
+      "learning_rate": 6.259298088868861e-06,
+      "loss": 0.9408,
+      "step": 17416
+    },
+    {
+      "epoch": 0.6334145543150161,
+      "grad_norm": 0.0,
+      "learning_rate": 6.258205765439612e-06,
+      "loss": 1.0698,
+      "step": 17417
+    },
+    {
+      "epoch": 0.6334509219187547,
+      "grad_norm": 0.0,
+      "learning_rate": 6.257113493922449e-06,
+      "loss": 0.8771,
+      "step": 17418
+    },
+    {
+      "epoch": 0.6334872895224933,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2560212743325254e-06,
+      "loss": 0.8949,
+      "step": 17419
+    },
+    {
+      "epoch": 0.6335236571262319,
+      "grad_norm": 0.0,
+      "learning_rate": 6.254929106684997e-06,
+      "loss": 1.0055,
+      "step": 17420
+    },
+    {
+      "epoch": 0.6335600247299705,
+      "grad_norm": 0.0,
+      "learning_rate": 6.25383699099501e-06,
+      "loss": 0.8587,
+      "step": 17421
+    },
+    {
+      "epoch": 0.6335963923337091,
+      "grad_norm": 0.0,
+      "learning_rate": 6.252744927277723e-06,
+      "loss": 0.9258,
+      "step": 17422
+    },
+    {
+      "epoch": 0.6336327599374477,
+      "grad_norm": 0.0,
+      "learning_rate": 6.25165291554828e-06,
+      "loss": 1.0622,
+      "step": 17423
+    },
+    {
+      "epoch": 0.6336691275411863,
+      "grad_norm": 0.0,
+      "learning_rate": 6.250560955821837e-06,
+      "loss": 0.9612,
+      "step": 17424
+    },
+    {
+      "epoch": 0.6337054951449249,
+      "grad_norm": 0.0,
+      "learning_rate": 6.249469048113539e-06,
+      "loss": 1.006,
+      "step": 17425
+    },
+    {
+      "epoch": 0.6337418627486635,
+      "grad_norm": 0.0,
+      "learning_rate": 6.248377192438542e-06,
+      "loss": 1.0305,
+      "step": 17426
+    },
+    {
+      "epoch": 0.6337782303524021,
+      "grad_norm": 0.0,
+      "learning_rate": 6.247285388811982e-06,
+      "loss": 0.9538,
+      "step": 17427
+    },
+    {
+      "epoch": 0.6338145979561407,
+      "grad_norm": 0.0,
+      "learning_rate": 6.246193637249016e-06,
+      "loss": 0.8577,
+      "step": 17428
+    },
+    {
+      "epoch": 0.6338509655598793,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2451019377647855e-06,
+      "loss": 0.8889,
+      "step": 17429
+    },
+    {
+      "epoch": 0.6338873331636179,
+      "grad_norm": 0.0,
+      "learning_rate": 6.244010290374441e-06,
+      "loss": 1.014,
+      "step": 17430
+    },
+    {
+      "epoch": 0.6339237007673565,
+      "grad_norm": 0.0,
+      "learning_rate": 6.242918695093121e-06,
+      "loss": 0.955,
+      "step": 17431
+    },
+    {
+      "epoch": 0.633960068371095,
+      "grad_norm": 0.0,
+      "learning_rate": 6.241827151935977e-06,
+      "loss": 0.9733,
+      "step": 17432
+    },
+    {
+      "epoch": 0.6339964359748336,
+      "grad_norm": 0.0,
+      "learning_rate": 6.240735660918144e-06,
+      "loss": 1.0466,
+      "step": 17433
+    },
+    {
+      "epoch": 0.6340328035785722,
+      "grad_norm": 0.0,
+      "learning_rate": 6.239644222054777e-06,
+      "loss": 0.9034,
+      "step": 17434
+    },
+    {
+      "epoch": 0.6340691711823108,
+      "grad_norm": 0.0,
+      "learning_rate": 6.238552835361007e-06,
+      "loss": 0.9717,
+      "step": 17435
+    },
+    {
+      "epoch": 0.6341055387860494,
+      "grad_norm": 0.0,
+      "learning_rate": 6.237461500851981e-06,
+      "loss": 0.9459,
+      "step": 17436
+    },
+    {
+      "epoch": 0.634141906389788,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2363702185428375e-06,
+      "loss": 0.9067,
+      "step": 17437
+    },
+    {
+      "epoch": 0.6341782739935266,
+      "grad_norm": 0.0,
+      "learning_rate": 6.235278988448717e-06,
+      "loss": 0.959,
+      "step": 17438
+    },
+    {
+      "epoch": 0.6342146415972651,
+      "grad_norm": 0.0,
+      "learning_rate": 6.234187810584763e-06,
+      "loss": 1.0176,
+      "step": 17439
+    },
+    {
+      "epoch": 0.6342510092010037,
+      "grad_norm": 0.0,
+      "learning_rate": 6.233096684966105e-06,
+      "loss": 0.9197,
+      "step": 17440
+    },
+    {
+      "epoch": 0.6342873768047423,
+      "grad_norm": 0.0,
+      "learning_rate": 6.23200561160789e-06,
+      "loss": 1.0113,
+      "step": 17441
+    },
+    {
+      "epoch": 0.6343237444084809,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2309145905252465e-06,
+      "loss": 0.9523,
+      "step": 17442
+    },
+    {
+      "epoch": 0.6343601120122195,
+      "grad_norm": 0.0,
+      "learning_rate": 6.229823621733319e-06,
+      "loss": 0.9785,
+      "step": 17443
+    },
+    {
+      "epoch": 0.6343964796159581,
+      "grad_norm": 0.0,
+      "learning_rate": 6.228732705247236e-06,
+      "loss": 1.0118,
+      "step": 17444
+    },
+    {
+      "epoch": 0.6344328472196967,
+      "grad_norm": 0.0,
+      "learning_rate": 6.227641841082142e-06,
+      "loss": 0.9354,
+      "step": 17445
+    },
+    {
+      "epoch": 0.6344692148234353,
+      "grad_norm": 0.0,
+      "learning_rate": 6.226551029253157e-06,
+      "loss": 1.0322,
+      "step": 17446
+    },
+    {
+      "epoch": 0.6345055824271739,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2254602697754296e-06,
+      "loss": 0.9433,
+      "step": 17447
+    },
+    {
+      "epoch": 0.6345419500309125,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2243695626640836e-06,
+      "loss": 1.076,
+      "step": 17448
+    },
+    {
+      "epoch": 0.634578317634651,
+      "grad_norm": 0.0,
+      "learning_rate": 6.223278907934251e-06,
+      "loss": 1.0148,
+      "step": 17449
+    },
+    {
+      "epoch": 0.6346146852383896,
+      "grad_norm": 0.0,
+      "learning_rate": 6.222188305601067e-06,
+      "loss": 0.967,
+      "step": 17450
+    },
+    {
+      "epoch": 0.6346510528421282,
+      "grad_norm": 0.0,
+      "learning_rate": 6.221097755679659e-06,
+      "loss": 0.9733,
+      "step": 17451
+    },
+    {
+      "epoch": 0.6346874204458668,
+      "grad_norm": 0.0,
+      "learning_rate": 6.220007258185159e-06,
+      "loss": 0.9821,
+      "step": 17452
+    },
+    {
+      "epoch": 0.6347237880496054,
+      "grad_norm": 0.0,
+      "learning_rate": 6.218916813132697e-06,
+      "loss": 1.0022,
+      "step": 17453
+    },
+    {
+      "epoch": 0.634760155653344,
+      "grad_norm": 0.0,
+      "learning_rate": 6.217826420537396e-06,
+      "loss": 1.0726,
+      "step": 17454
+    },
+    {
+      "epoch": 0.6347965232570826,
+      "grad_norm": 0.0,
+      "learning_rate": 6.216736080414389e-06,
+      "loss": 0.9337,
+      "step": 17455
+    },
+    {
+      "epoch": 0.6348328908608212,
+      "grad_norm": 0.0,
+      "learning_rate": 6.215645792778801e-06,
+      "loss": 0.868,
+      "step": 17456
+    },
+    {
+      "epoch": 0.6348692584645598,
+      "grad_norm": 0.0,
+      "learning_rate": 6.214555557645759e-06,
+      "loss": 1.0072,
+      "step": 17457
+    },
+    {
+      "epoch": 0.6349056260682984,
+      "grad_norm": 0.0,
+      "learning_rate": 6.21346537503039e-06,
+      "loss": 1.0503,
+      "step": 17458
+    },
+    {
+      "epoch": 0.634941993672037,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2123752449478125e-06,
+      "loss": 1.0515,
+      "step": 17459
+    },
+    {
+      "epoch": 0.6349783612757756,
+      "grad_norm": 0.0,
+      "learning_rate": 6.211285167413156e-06,
+      "loss": 0.873,
+      "step": 17460
+    },
+    {
+      "epoch": 0.6350147288795142,
+      "grad_norm": 0.0,
+      "learning_rate": 6.210195142441541e-06,
+      "loss": 0.9633,
+      "step": 17461
+    },
+    {
+      "epoch": 0.6350510964832528,
+      "grad_norm": 0.0,
+      "learning_rate": 6.209105170048093e-06,
+      "loss": 0.9902,
+      "step": 17462
+    },
+    {
+      "epoch": 0.6350874640869913,
+      "grad_norm": 0.0,
+      "learning_rate": 6.20801525024793e-06,
+      "loss": 1.0775,
+      "step": 17463
+    },
+    {
+      "epoch": 0.63512383169073,
+      "grad_norm": 0.0,
+      "learning_rate": 6.206925383056177e-06,
+      "loss": 0.8759,
+      "step": 17464
+    },
+    {
+      "epoch": 0.6351601992944685,
+      "grad_norm": 0.0,
+      "learning_rate": 6.205835568487952e-06,
+      "loss": 1.023,
+      "step": 17465
+    },
+    {
+      "epoch": 0.635196566898207,
+      "grad_norm": 0.0,
+      "learning_rate": 6.204745806558379e-06,
+      "loss": 0.9883,
+      "step": 17466
+    },
+    {
+      "epoch": 0.6352329345019456,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2036560972825675e-06,
+      "loss": 0.9465,
+      "step": 17467
+    },
+    {
+      "epoch": 0.6352693021056842,
+      "grad_norm": 0.0,
+      "learning_rate": 6.202566440675645e-06,
+      "loss": 1.0066,
+      "step": 17468
+    },
+    {
+      "epoch": 0.6353056697094228,
+      "grad_norm": 0.0,
+      "learning_rate": 6.201476836752723e-06,
+      "loss": 0.8738,
+      "step": 17469
+    },
+    {
+      "epoch": 0.6353420373131614,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2003872855289216e-06,
+      "loss": 0.9439,
+      "step": 17470
+    },
+    {
+      "epoch": 0.6353784049169,
+      "grad_norm": 0.0,
+      "learning_rate": 6.199297787019354e-06,
+      "loss": 1.003,
+      "step": 17471
+    },
+    {
+      "epoch": 0.6354147725206386,
+      "grad_norm": 0.0,
+      "learning_rate": 6.198208341239143e-06,
+      "loss": 0.9357,
+      "step": 17472
+    },
+    {
+      "epoch": 0.6354511401243772,
+      "grad_norm": 0.0,
+      "learning_rate": 6.197118948203389e-06,
+      "loss": 0.9879,
+      "step": 17473
+    },
+    {
+      "epoch": 0.6354875077281158,
+      "grad_norm": 0.0,
+      "learning_rate": 6.196029607927221e-06,
+      "loss": 1.0905,
+      "step": 17474
+    },
+    {
+      "epoch": 0.6355238753318544,
+      "grad_norm": 0.0,
+      "learning_rate": 6.194940320425739e-06,
+      "loss": 0.8775,
+      "step": 17475
+    },
+    {
+      "epoch": 0.635560242935593,
+      "grad_norm": 0.0,
+      "learning_rate": 6.193851085714066e-06,
+      "loss": 0.9906,
+      "step": 17476
+    },
+    {
+      "epoch": 0.6355966105393316,
+      "grad_norm": 0.0,
+      "learning_rate": 6.19276190380731e-06,
+      "loss": 0.8064,
+      "step": 17477
+    },
+    {
+      "epoch": 0.6356329781430702,
+      "grad_norm": 0.0,
+      "learning_rate": 6.191672774720575e-06,
+      "loss": 0.9087,
+      "step": 17478
+    },
+    {
+      "epoch": 0.6356693457468088,
+      "grad_norm": 0.0,
+      "learning_rate": 6.190583698468984e-06,
+      "loss": 1.0145,
+      "step": 17479
+    },
+    {
+      "epoch": 0.6357057133505474,
+      "grad_norm": 0.0,
+      "learning_rate": 6.189494675067633e-06,
+      "loss": 1.0855,
+      "step": 17480
+    },
+    {
+      "epoch": 0.635742080954286,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1884057045316405e-06,
+      "loss": 0.912,
+      "step": 17481
+    },
+    {
+      "epoch": 0.6357784485580246,
+      "grad_norm": 0.0,
+      "learning_rate": 6.187316786876109e-06,
+      "loss": 0.9948,
+      "step": 17482
+    },
+    {
+      "epoch": 0.635814816161763,
+      "grad_norm": 0.0,
+      "learning_rate": 6.186227922116148e-06,
+      "loss": 0.8789,
+      "step": 17483
+    },
+    {
+      "epoch": 0.6358511837655016,
+      "grad_norm": 0.0,
+      "learning_rate": 6.185139110266864e-06,
+      "loss": 0.9784,
+      "step": 17484
+    },
+    {
+      "epoch": 0.6358875513692402,
+      "grad_norm": 0.0,
+      "learning_rate": 6.184050351343365e-06,
+      "loss": 0.9408,
+      "step": 17485
+    },
+    {
+      "epoch": 0.6359239189729788,
+      "grad_norm": 0.0,
+      "learning_rate": 6.182961645360747e-06,
+      "loss": 0.9771,
+      "step": 17486
+    },
+    {
+      "epoch": 0.6359602865767174,
+      "grad_norm": 0.0,
+      "learning_rate": 6.181872992334126e-06,
+      "loss": 0.9842,
+      "step": 17487
+    },
+    {
+      "epoch": 0.635996654180456,
+      "grad_norm": 0.0,
+      "learning_rate": 6.180784392278597e-06,
+      "loss": 0.9849,
+      "step": 17488
+    },
+    {
+      "epoch": 0.6360330217841946,
+      "grad_norm": 0.0,
+      "learning_rate": 6.179695845209268e-06,
+      "loss": 1.0112,
+      "step": 17489
+    },
+    {
+      "epoch": 0.6360693893879332,
+      "grad_norm": 0.0,
+      "learning_rate": 6.178607351141236e-06,
+      "loss": 0.8991,
+      "step": 17490
+    },
+    {
+      "epoch": 0.6361057569916718,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1775189100896075e-06,
+      "loss": 1.0743,
+      "step": 17491
+    },
+    {
+      "epoch": 0.6361421245954104,
+      "grad_norm": 0.0,
+      "learning_rate": 6.176430522069479e-06,
+      "loss": 0.8459,
+      "step": 17492
+    },
+    {
+      "epoch": 0.636178492199149,
+      "grad_norm": 0.0,
+      "learning_rate": 6.175342187095956e-06,
+      "loss": 0.9921,
+      "step": 17493
+    },
+    {
+      "epoch": 0.6362148598028876,
+      "grad_norm": 0.0,
+      "learning_rate": 6.174253905184129e-06,
+      "loss": 0.9913,
+      "step": 17494
+    },
+    {
+      "epoch": 0.6362512274066262,
+      "grad_norm": 0.0,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.9789,
+      "step": 17495
+    },
+    {
+      "epoch": 0.6362875950103648,
+      "grad_norm": 0.0,
+      "learning_rate": 6.172077500605975e-06,
+      "loss": 0.9968,
+      "step": 17496
+    },
+    {
+      "epoch": 0.6363239626141034,
+      "grad_norm": 0.0,
+      "learning_rate": 6.170989377969838e-06,
+      "loss": 0.9523,
+      "step": 17497
+    },
+    {
+      "epoch": 0.636360330217842,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1699013084557944e-06,
+      "loss": 0.8535,
+      "step": 17498
+    },
+    {
+      "epoch": 0.6363966978215805,
+      "grad_norm": 0.0,
+      "learning_rate": 6.168813292078932e-06,
+      "loss": 0.9359,
+      "step": 17499
+    },
+    {
+      "epoch": 0.6364330654253191,
+      "grad_norm": 0.0,
+      "learning_rate": 6.16772532885435e-06,
+      "loss": 0.9657,
+      "step": 17500
+    },
+    {
+      "epoch": 0.6364694330290577,
+      "grad_norm": 0.0,
+      "learning_rate": 6.166637418797142e-06,
+      "loss": 0.9966,
+      "step": 17501
+    },
+    {
+      "epoch": 0.6365058006327963,
+      "grad_norm": 0.0,
+      "learning_rate": 6.165549561922403e-06,
+      "loss": 1.066,
+      "step": 17502
+    },
+    {
+      "epoch": 0.6365421682365349,
+      "grad_norm": 0.0,
+      "learning_rate": 6.164461758245221e-06,
+      "loss": 0.9439,
+      "step": 17503
+    },
+    {
+      "epoch": 0.6365785358402735,
+      "grad_norm": 0.0,
+      "learning_rate": 6.163374007780694e-06,
+      "loss": 1.0154,
+      "step": 17504
+    },
+    {
+      "epoch": 0.6366149034440121,
+      "grad_norm": 0.0,
+      "learning_rate": 6.162286310543903e-06,
+      "loss": 0.8831,
+      "step": 17505
+    },
+    {
+      "epoch": 0.6366512710477507,
+      "grad_norm": 0.0,
+      "learning_rate": 6.16119866654995e-06,
+      "loss": 1.0059,
+      "step": 17506
+    },
+    {
+      "epoch": 0.6366876386514893,
+      "grad_norm": 0.0,
+      "learning_rate": 6.160111075813915e-06,
+      "loss": 1.1025,
+      "step": 17507
+    },
+    {
+      "epoch": 0.6367240062552278,
+      "grad_norm": 0.0,
+      "learning_rate": 6.159023538350893e-06,
+      "loss": 0.9578,
+      "step": 17508
+    },
+    {
+      "epoch": 0.6367603738589664,
+      "grad_norm": 0.0,
+      "learning_rate": 6.157936054175968e-06,
+      "loss": 0.8996,
+      "step": 17509
+    },
+    {
+      "epoch": 0.636796741462705,
+      "grad_norm": 0.0,
+      "learning_rate": 6.156848623304232e-06,
+      "loss": 1.0645,
+      "step": 17510
+    },
+    {
+      "epoch": 0.6368331090664436,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1557612457507655e-06,
+      "loss": 0.9161,
+      "step": 17511
+    },
+    {
+      "epoch": 0.6368694766701822,
+      "grad_norm": 0.0,
+      "learning_rate": 6.154673921530661e-06,
+      "loss": 0.9939,
+      "step": 17512
+    },
+    {
+      "epoch": 0.6369058442739208,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1535866506589945e-06,
+      "loss": 0.9983,
+      "step": 17513
+    },
+    {
+      "epoch": 0.6369422118776594,
+      "grad_norm": 0.0,
+      "learning_rate": 6.152499433150862e-06,
+      "loss": 0.9923,
+      "step": 17514
+    },
+    {
+      "epoch": 0.636978579481398,
+      "grad_norm": 0.0,
+      "learning_rate": 6.15141226902134e-06,
+      "loss": 0.9709,
+      "step": 17515
+    },
+    {
+      "epoch": 0.6370149470851365,
+      "grad_norm": 0.0,
+      "learning_rate": 6.150325158285509e-06,
+      "loss": 0.8762,
+      "step": 17516
+    },
+    {
+      "epoch": 0.6370513146888751,
+      "grad_norm": 0.0,
+      "learning_rate": 6.149238100958459e-06,
+      "loss": 1.018,
+      "step": 17517
+    },
+    {
+      "epoch": 0.6370876822926137,
+      "grad_norm": 0.0,
+      "learning_rate": 6.14815109705526e-06,
+      "loss": 0.9802,
+      "step": 17518
+    },
+    {
+      "epoch": 0.6371240498963523,
+      "grad_norm": 0.0,
+      "learning_rate": 6.147064146591006e-06,
+      "loss": 1.0114,
+      "step": 17519
+    },
+    {
+      "epoch": 0.6371604175000909,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1459772495807665e-06,
+      "loss": 0.9275,
+      "step": 17520
+    },
+    {
+      "epoch": 0.6371967851038295,
+      "grad_norm": 0.0,
+      "learning_rate": 6.144890406039627e-06,
+      "loss": 0.9603,
+      "step": 17521
+    },
+    {
+      "epoch": 0.6372331527075681,
+      "grad_norm": 0.0,
+      "learning_rate": 6.143803615982661e-06,
+      "loss": 0.9716,
+      "step": 17522
+    },
+    {
+      "epoch": 0.6372695203113067,
+      "grad_norm": 0.0,
+      "learning_rate": 6.14271687942495e-06,
+      "loss": 0.8421,
+      "step": 17523
+    },
+    {
+      "epoch": 0.6373058879150453,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1416301963815685e-06,
+      "loss": 0.9417,
+      "step": 17524
+    },
+    {
+      "epoch": 0.6373422555187839,
+      "grad_norm": 0.0,
+      "learning_rate": 6.140543566867597e-06,
+      "loss": 1.0171,
+      "step": 17525
+    },
+    {
+      "epoch": 0.6373786231225225,
+      "grad_norm": 0.0,
+      "learning_rate": 6.139456990898105e-06,
+      "loss": 1.0123,
+      "step": 17526
+    },
+    {
+      "epoch": 0.6374149907262611,
+      "grad_norm": 0.0,
+      "learning_rate": 6.138370468488171e-06,
+      "loss": 0.8404,
+      "step": 17527
+    },
+    {
+      "epoch": 0.6374513583299997,
+      "grad_norm": 0.0,
+      "learning_rate": 6.137283999652864e-06,
+      "loss": 1.0029,
+      "step": 17528
+    },
+    {
+      "epoch": 0.6374877259337383,
+      "grad_norm": 0.0,
+      "learning_rate": 6.136197584407265e-06,
+      "loss": 0.8706,
+      "step": 17529
+    },
+    {
+      "epoch": 0.6375240935374769,
+      "grad_norm": 0.0,
+      "learning_rate": 6.135111222766441e-06,
+      "loss": 0.9962,
+      "step": 17530
+    },
+    {
+      "epoch": 0.6375604611412155,
+      "grad_norm": 0.0,
+      "learning_rate": 6.134024914745466e-06,
+      "loss": 1.0288,
+      "step": 17531
+    },
+    {
+      "epoch": 0.637596828744954,
+      "grad_norm": 0.0,
+      "learning_rate": 6.132938660359409e-06,
+      "loss": 0.9366,
+      "step": 17532
+    },
+    {
+      "epoch": 0.6376331963486925,
+      "grad_norm": 0.0,
+      "learning_rate": 6.131852459623344e-06,
+      "loss": 0.9715,
+      "step": 17533
+    },
+    {
+      "epoch": 0.6376695639524311,
+      "grad_norm": 0.0,
+      "learning_rate": 6.130766312552336e-06,
+      "loss": 1.029,
+      "step": 17534
+    },
+    {
+      "epoch": 0.6377059315561697,
+      "grad_norm": 0.0,
+      "learning_rate": 6.129680219161456e-06,
+      "loss": 0.9802,
+      "step": 17535
+    },
+    {
+      "epoch": 0.6377422991599083,
+      "grad_norm": 0.0,
+      "learning_rate": 6.128594179465772e-06,
+      "loss": 1.0518,
+      "step": 17536
+    },
+    {
+      "epoch": 0.6377786667636469,
+      "grad_norm": 0.0,
+      "learning_rate": 6.12750819348035e-06,
+      "loss": 0.9574,
+      "step": 17537
+    },
+    {
+      "epoch": 0.6378150343673855,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1264222612202596e-06,
+      "loss": 0.9569,
+      "step": 17538
+    },
+    {
+      "epoch": 0.6378514019711241,
+      "grad_norm": 0.0,
+      "learning_rate": 6.125336382700562e-06,
+      "loss": 0.9521,
+      "step": 17539
+    },
+    {
+      "epoch": 0.6378877695748627,
+      "grad_norm": 0.0,
+      "learning_rate": 6.124250557936326e-06,
+      "loss": 1.0386,
+      "step": 17540
+    },
+    {
+      "epoch": 0.6379241371786013,
+      "grad_norm": 0.0,
+      "learning_rate": 6.123164786942612e-06,
+      "loss": 0.9948,
+      "step": 17541
+    },
+    {
+      "epoch": 0.6379605047823399,
+      "grad_norm": 0.0,
+      "learning_rate": 6.12207906973449e-06,
+      "loss": 0.9909,
+      "step": 17542
+    },
+    {
+      "epoch": 0.6379968723860785,
+      "grad_norm": 0.0,
+      "learning_rate": 6.120993406327016e-06,
+      "loss": 0.9993,
+      "step": 17543
+    },
+    {
+      "epoch": 0.6380332399898171,
+      "grad_norm": 0.0,
+      "learning_rate": 6.119907796735258e-06,
+      "loss": 0.966,
+      "step": 17544
+    },
+    {
+      "epoch": 0.6380696075935557,
+      "grad_norm": 0.0,
+      "learning_rate": 6.118822240974269e-06,
+      "loss": 0.9575,
+      "step": 17545
+    },
+    {
+      "epoch": 0.6381059751972943,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1177367390591215e-06,
+      "loss": 0.9728,
+      "step": 17546
+    },
+    {
+      "epoch": 0.6381423428010329,
+      "grad_norm": 0.0,
+      "learning_rate": 6.116651291004863e-06,
+      "loss": 0.9804,
+      "step": 17547
+    },
+    {
+      "epoch": 0.6381787104047715,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1155658968265605e-06,
+      "loss": 0.9425,
+      "step": 17548
+    },
+    {
+      "epoch": 0.6382150780085101,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1144805565392685e-06,
+      "loss": 0.9982,
+      "step": 17549
+    },
+    {
+      "epoch": 0.6382514456122486,
+      "grad_norm": 0.0,
+      "learning_rate": 6.113395270158047e-06,
+      "loss": 0.9872,
+      "step": 17550
+    },
+    {
+      "epoch": 0.6382878132159872,
+      "grad_norm": 0.0,
+      "learning_rate": 6.112310037697951e-06,
+      "loss": 0.7751,
+      "step": 17551
+    },
+    {
+      "epoch": 0.6383241808197258,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1112248591740395e-06,
+      "loss": 0.9986,
+      "step": 17552
+    },
+    {
+      "epoch": 0.6383605484234643,
+      "grad_norm": 0.0,
+      "learning_rate": 6.110139734601364e-06,
+      "loss": 1.0312,
+      "step": 17553
+    },
+    {
+      "epoch": 0.6383969160272029,
+      "grad_norm": 0.0,
+      "learning_rate": 6.10905466399498e-06,
+      "loss": 0.9079,
+      "step": 17554
+    },
+    {
+      "epoch": 0.6384332836309415,
+      "grad_norm": 0.0,
+      "learning_rate": 6.107969647369944e-06,
+      "loss": 0.8896,
+      "step": 17555
+    },
+    {
+      "epoch": 0.6384696512346801,
+      "grad_norm": 0.0,
+      "learning_rate": 6.1068846847413044e-06,
+      "loss": 1.131,
+      "step": 17556
+    },
+    {
+      "epoch": 0.6385060188384187,
+      "grad_norm": 0.0,
+      "learning_rate": 6.105799776124119e-06,
+      "loss": 1.0843,
+      "step": 17557
+    },
+    {
+      "epoch": 0.6385423864421573,
+      "grad_norm": 0.0,
+      "learning_rate": 6.104714921533432e-06,
+      "loss": 0.9383,
+      "step": 17558
+    },
+    {
+      "epoch": 0.6385787540458959,
+      "grad_norm": 0.0,
+      "learning_rate": 6.103630120984305e-06,
+      "loss": 1.0467,
+      "step": 17559
+    },
+    {
+      "epoch": 0.6386151216496345,
+      "grad_norm": 0.0,
+      "learning_rate": 6.102545374491777e-06,
+      "loss": 0.9084,
+      "step": 17560
+    },
+    {
+      "epoch": 0.6386514892533731,
+      "grad_norm": 0.0,
+      "learning_rate": 6.101460682070905e-06,
+      "loss": 0.9033,
+      "step": 17561
+    },
+    {
+      "epoch": 0.6386878568571117,
+      "grad_norm": 0.0,
+      "learning_rate": 6.100376043736733e-06,
+      "loss": 1.0267,
+      "step": 17562
+    },
+    {
+      "epoch": 0.6387242244608503,
+      "grad_norm": 0.0,
+      "learning_rate": 6.099291459504311e-06,
+      "loss": 0.9474,
+      "step": 17563
+    },
+    {
+      "epoch": 0.6387605920645889,
+      "grad_norm": 0.0,
+      "learning_rate": 6.098206929388686e-06,
+      "loss": 0.9825,
+      "step": 17564
+    },
+    {
+      "epoch": 0.6387969596683275,
+      "grad_norm": 0.0,
+      "learning_rate": 6.097122453404906e-06,
+      "loss": 1.0461,
+      "step": 17565
+    },
+    {
+      "epoch": 0.6388333272720661,
+      "grad_norm": 0.0,
+      "learning_rate": 6.09603803156801e-06,
+      "loss": 0.9353,
+      "step": 17566
+    },
+    {
+      "epoch": 0.6388696948758046,
+      "grad_norm": 0.0,
+      "learning_rate": 6.094953663893051e-06,
+      "loss": 0.8699,
+      "step": 17567
+    },
+    {
+      "epoch": 0.6389060624795432,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0938693503950665e-06,
+      "loss": 0.9264,
+      "step": 17568
+    },
+    {
+      "epoch": 0.6389424300832818,
+      "grad_norm": 0.0,
+      "learning_rate": 6.092785091089104e-06,
+      "loss": 0.9377,
+      "step": 17569
+    },
+    {
+      "epoch": 0.6389787976870204,
+      "grad_norm": 0.0,
+      "learning_rate": 6.091700885990204e-06,
+      "loss": 0.8392,
+      "step": 17570
+    },
+    {
+      "epoch": 0.639015165290759,
+      "grad_norm": 0.0,
+      "learning_rate": 6.090616735113413e-06,
+      "loss": 0.9086,
+      "step": 17571
+    },
+    {
+      "epoch": 0.6390515328944976,
+      "grad_norm": 0.0,
+      "learning_rate": 6.089532638473765e-06,
+      "loss": 0.8855,
+      "step": 17572
+    },
+    {
+      "epoch": 0.6390879004982362,
+      "grad_norm": 0.0,
+      "learning_rate": 6.088448596086303e-06,
+      "loss": 1.078,
+      "step": 17573
+    },
+    {
+      "epoch": 0.6391242681019748,
+      "grad_norm": 0.0,
+      "learning_rate": 6.087364607966069e-06,
+      "loss": 0.9429,
+      "step": 17574
+    },
+    {
+      "epoch": 0.6391606357057134,
+      "grad_norm": 0.0,
+      "learning_rate": 6.086280674128096e-06,
+      "loss": 1.002,
+      "step": 17575
+    },
+    {
+      "epoch": 0.639197003309452,
+      "grad_norm": 0.0,
+      "learning_rate": 6.08519679458743e-06,
+      "loss": 0.8343,
+      "step": 17576
+    },
+    {
+      "epoch": 0.6392333709131905,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0841129693590995e-06,
+      "loss": 0.8915,
+      "step": 17577
+    },
+    {
+      "epoch": 0.6392697385169291,
+      "grad_norm": 0.0,
+      "learning_rate": 6.08302919845815e-06,
+      "loss": 0.8457,
+      "step": 17578
+    },
+    {
+      "epoch": 0.6393061061206677,
+      "grad_norm": 0.0,
+      "learning_rate": 6.08194548189961e-06,
+      "loss": 0.9486,
+      "step": 17579
+    },
+    {
+      "epoch": 0.6393424737244063,
+      "grad_norm": 0.0,
+      "learning_rate": 6.08086181969852e-06,
+      "loss": 0.9348,
+      "step": 17580
+    },
+    {
+      "epoch": 0.6393788413281449,
+      "grad_norm": 0.0,
+      "learning_rate": 6.079778211869909e-06,
+      "loss": 0.8175,
+      "step": 17581
+    },
+    {
+      "epoch": 0.6394152089318835,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0786946584288145e-06,
+      "loss": 0.9879,
+      "step": 17582
+    },
+    {
+      "epoch": 0.639451576535622,
+      "grad_norm": 0.0,
+      "learning_rate": 6.077611159390267e-06,
+      "loss": 0.9113,
+      "step": 17583
+    },
+    {
+      "epoch": 0.6394879441393606,
+      "grad_norm": 0.0,
+      "learning_rate": 6.076527714769304e-06,
+      "loss": 0.8608,
+      "step": 17584
+    },
+    {
+      "epoch": 0.6395243117430992,
+      "grad_norm": 0.0,
+      "learning_rate": 6.075444324580945e-06,
+      "loss": 1.0969,
+      "step": 17585
+    },
+    {
+      "epoch": 0.6395606793468378,
+      "grad_norm": 0.0,
+      "learning_rate": 6.074360988840234e-06,
+      "loss": 0.9238,
+      "step": 17586
+    },
+    {
+      "epoch": 0.6395970469505764,
+      "grad_norm": 0.0,
+      "learning_rate": 6.073277707562191e-06,
+      "loss": 1.0058,
+      "step": 17587
+    },
+    {
+      "epoch": 0.639633414554315,
+      "grad_norm": 0.0,
+      "learning_rate": 6.072194480761851e-06,
+      "loss": 0.9554,
+      "step": 17588
+    },
+    {
+      "epoch": 0.6396697821580536,
+      "grad_norm": 0.0,
+      "learning_rate": 6.071111308454236e-06,
+      "loss": 1.0094,
+      "step": 17589
+    },
+    {
+      "epoch": 0.6397061497617922,
+      "grad_norm": 0.0,
+      "learning_rate": 6.07002819065438e-06,
+      "loss": 1.004,
+      "step": 17590
+    },
+    {
+      "epoch": 0.6397425173655308,
+      "grad_norm": 0.0,
+      "learning_rate": 6.06894512737731e-06,
+      "loss": 0.8663,
+      "step": 17591
+    },
+    {
+      "epoch": 0.6397788849692694,
+      "grad_norm": 0.0,
+      "learning_rate": 6.067862118638044e-06,
+      "loss": 1.087,
+      "step": 17592
+    },
+    {
+      "epoch": 0.639815252573008,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0667791644516146e-06,
+      "loss": 0.9684,
+      "step": 17593
+    },
+    {
+      "epoch": 0.6398516201767466,
+      "grad_norm": 0.0,
+      "learning_rate": 6.065696264833042e-06,
+      "loss": 0.8003,
+      "step": 17594
+    },
+    {
+      "epoch": 0.6398879877804852,
+      "grad_norm": 0.0,
+      "learning_rate": 6.064613419797353e-06,
+      "loss": 1.0175,
+      "step": 17595
+    },
+    {
+      "epoch": 0.6399243553842238,
+      "grad_norm": 0.0,
+      "learning_rate": 6.063530629359568e-06,
+      "loss": 1.0963,
+      "step": 17596
+    },
+    {
+      "epoch": 0.6399607229879624,
+      "grad_norm": 0.0,
+      "learning_rate": 6.062447893534713e-06,
+      "loss": 0.8752,
+      "step": 17597
+    },
+    {
+      "epoch": 0.639997090591701,
+      "grad_norm": 0.0,
+      "learning_rate": 6.061365212337804e-06,
+      "loss": 0.9196,
+      "step": 17598
+    },
+    {
+      "epoch": 0.6400334581954396,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0602825857838655e-06,
+      "loss": 1.0896,
+      "step": 17599
+    },
+    {
+      "epoch": 0.640069825799178,
+      "grad_norm": 0.0,
+      "learning_rate": 6.059200013887916e-06,
+      "loss": 1.0337,
+      "step": 17600
+    },
+    {
+      "epoch": 0.6401061934029166,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0581174966649755e-06,
+      "loss": 0.8937,
+      "step": 17601
+    },
+    {
+      "epoch": 0.6401425610066552,
+      "grad_norm": 0.0,
+      "learning_rate": 6.057035034130061e-06,
+      "loss": 1.0276,
+      "step": 17602
+    },
+    {
+      "epoch": 0.6401789286103938,
+      "grad_norm": 0.0,
+      "learning_rate": 6.055952626298191e-06,
+      "loss": 0.9307,
+      "step": 17603
+    },
+    {
+      "epoch": 0.6402152962141324,
+      "grad_norm": 0.0,
+      "learning_rate": 6.054870273184383e-06,
+      "loss": 0.871,
+      "step": 17604
+    },
+    {
+      "epoch": 0.640251663817871,
+      "grad_norm": 0.0,
+      "learning_rate": 6.053787974803656e-06,
+      "loss": 0.9469,
+      "step": 17605
+    },
+    {
+      "epoch": 0.6402880314216096,
+      "grad_norm": 0.0,
+      "learning_rate": 6.052705731171018e-06,
+      "loss": 0.9153,
+      "step": 17606
+    },
+    {
+      "epoch": 0.6403243990253482,
+      "grad_norm": 0.0,
+      "learning_rate": 6.051623542301488e-06,
+      "loss": 0.9625,
+      "step": 17607
+    },
+    {
+      "epoch": 0.6403607666290868,
+      "grad_norm": 0.0,
+      "learning_rate": 6.050541408210078e-06,
+      "loss": 0.7726,
+      "step": 17608
+    },
+    {
+      "epoch": 0.6403971342328254,
+      "grad_norm": 0.0,
+      "learning_rate": 6.049459328911806e-06,
+      "loss": 0.8832,
+      "step": 17609
+    },
+    {
+      "epoch": 0.640433501836564,
+      "grad_norm": 0.0,
+      "learning_rate": 6.048377304421681e-06,
+      "loss": 0.9355,
+      "step": 17610
+    },
+    {
+      "epoch": 0.6404698694403026,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0472953347547105e-06,
+      "loss": 0.8539,
+      "step": 17611
+    },
+    {
+      "epoch": 0.6405062370440412,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0462134199259115e-06,
+      "loss": 1.0006,
+      "step": 17612
+    },
+    {
+      "epoch": 0.6405426046477798,
+      "grad_norm": 0.0,
+      "learning_rate": 6.045131559950289e-06,
+      "loss": 0.8634,
+      "step": 17613
+    },
+    {
+      "epoch": 0.6405789722515184,
+      "grad_norm": 0.0,
+      "learning_rate": 6.044049754842858e-06,
+      "loss": 0.9413,
+      "step": 17614
+    },
+    {
+      "epoch": 0.640615339855257,
+      "grad_norm": 0.0,
+      "learning_rate": 6.042968004618621e-06,
+      "loss": 0.9553,
+      "step": 17615
+    },
+    {
+      "epoch": 0.6406517074589956,
+      "grad_norm": 0.0,
+      "learning_rate": 6.041886309292593e-06,
+      "loss": 1.001,
+      "step": 17616
+    },
+    {
+      "epoch": 0.6406880750627341,
+      "grad_norm": 0.0,
+      "learning_rate": 6.040804668879769e-06,
+      "loss": 0.8206,
+      "step": 17617
+    },
+    {
+      "epoch": 0.6407244426664727,
+      "grad_norm": 0.0,
+      "learning_rate": 6.039723083395171e-06,
+      "loss": 1.0342,
+      "step": 17618
+    },
+    {
+      "epoch": 0.6407608102702113,
+      "grad_norm": 0.0,
+      "learning_rate": 6.038641552853792e-06,
+      "loss": 0.9461,
+      "step": 17619
+    },
+    {
+      "epoch": 0.6407971778739499,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0375600772706435e-06,
+      "loss": 0.9205,
+      "step": 17620
+    },
+    {
+      "epoch": 0.6408335454776884,
+      "grad_norm": 0.0,
+      "learning_rate": 6.036478656660722e-06,
+      "loss": 0.9173,
+      "step": 17621
+    },
+    {
+      "epoch": 0.640869913081427,
+      "grad_norm": 0.0,
+      "learning_rate": 6.03539729103904e-06,
+      "loss": 0.9009,
+      "step": 17622
+    },
+    {
+      "epoch": 0.6409062806851656,
+      "grad_norm": 0.0,
+      "learning_rate": 6.034315980420594e-06,
+      "loss": 0.8265,
+      "step": 17623
+    },
+    {
+      "epoch": 0.6409426482889042,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0332347248203915e-06,
+      "loss": 1.0492,
+      "step": 17624
+    },
+    {
+      "epoch": 0.6409790158926428,
+      "grad_norm": 0.0,
+      "learning_rate": 6.032153524253424e-06,
+      "loss": 0.9369,
+      "step": 17625
+    },
+    {
+      "epoch": 0.6410153834963814,
+      "grad_norm": 0.0,
+      "learning_rate": 6.031072378734698e-06,
+      "loss": 1.0706,
+      "step": 17626
+    },
+    {
+      "epoch": 0.64105175110012,
+      "grad_norm": 0.0,
+      "learning_rate": 6.029991288279211e-06,
+      "loss": 0.9349,
+      "step": 17627
+    },
+    {
+      "epoch": 0.6410881187038586,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0289102529019625e-06,
+      "loss": 0.8656,
+      "step": 17628
+    },
+    {
+      "epoch": 0.6411244863075972,
+      "grad_norm": 0.0,
+      "learning_rate": 6.027829272617949e-06,
+      "loss": 0.9347,
+      "step": 17629
+    },
+    {
+      "epoch": 0.6411608539113358,
+      "grad_norm": 0.0,
+      "learning_rate": 6.02674834744217e-06,
+      "loss": 0.9702,
+      "step": 17630
+    },
+    {
+      "epoch": 0.6411972215150744,
+      "grad_norm": 0.0,
+      "learning_rate": 6.025667477389625e-06,
+      "loss": 1.0686,
+      "step": 17631
+    },
+    {
+      "epoch": 0.641233589118813,
+      "grad_norm": 0.0,
+      "learning_rate": 6.024586662475298e-06,
+      "loss": 0.9123,
+      "step": 17632
+    },
+    {
+      "epoch": 0.6412699567225516,
+      "grad_norm": 0.0,
+      "learning_rate": 6.023505902714194e-06,
+      "loss": 1.0572,
+      "step": 17633
+    },
+    {
+      "epoch": 0.6413063243262901,
+      "grad_norm": 0.0,
+      "learning_rate": 6.022425198121301e-06,
+      "loss": 1.1136,
+      "step": 17634
+    },
+    {
+      "epoch": 0.6413426919300287,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0213445487116175e-06,
+      "loss": 0.8301,
+      "step": 17635
+    },
+    {
+      "epoch": 0.6413790595337673,
+      "grad_norm": 0.0,
+      "learning_rate": 6.020263954500131e-06,
+      "loss": 0.8888,
+      "step": 17636
+    },
+    {
+      "epoch": 0.6414154271375059,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0191834155018395e-06,
+      "loss": 1.0404,
+      "step": 17637
+    },
+    {
+      "epoch": 0.6414517947412445,
+      "grad_norm": 0.0,
+      "learning_rate": 6.018102931731726e-06,
+      "loss": 0.9912,
+      "step": 17638
+    },
+    {
+      "epoch": 0.6414881623449831,
+      "grad_norm": 0.0,
+      "learning_rate": 6.017022503204788e-06,
+      "loss": 0.8835,
+      "step": 17639
+    },
+    {
+      "epoch": 0.6415245299487217,
+      "grad_norm": 0.0,
+      "learning_rate": 6.015942129936008e-06,
+      "loss": 0.797,
+      "step": 17640
+    },
+    {
+      "epoch": 0.6415608975524603,
+      "grad_norm": 0.0,
+      "learning_rate": 6.014861811940381e-06,
+      "loss": 0.8593,
+      "step": 17641
+    },
+    {
+      "epoch": 0.6415972651561989,
+      "grad_norm": 0.0,
+      "learning_rate": 6.013781549232891e-06,
+      "loss": 0.9959,
+      "step": 17642
+    },
+    {
+      "epoch": 0.6416336327599375,
+      "grad_norm": 0.0,
+      "learning_rate": 6.012701341828528e-06,
+      "loss": 1.0052,
+      "step": 17643
+    },
+    {
+      "epoch": 0.641670000363676,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0116211897422735e-06,
+      "loss": 0.9676,
+      "step": 17644
+    },
+    {
+      "epoch": 0.6417063679674146,
+      "grad_norm": 0.0,
+      "learning_rate": 6.010541092989121e-06,
+      "loss": 0.9245,
+      "step": 17645
+    },
+    {
+      "epoch": 0.6417427355711532,
+      "grad_norm": 0.0,
+      "learning_rate": 6.009461051584047e-06,
+      "loss": 0.9835,
+      "step": 17646
+    },
+    {
+      "epoch": 0.6417791031748918,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0083810655420394e-06,
+      "loss": 0.9417,
+      "step": 17647
+    },
+    {
+      "epoch": 0.6418154707786304,
+      "grad_norm": 0.0,
+      "learning_rate": 6.007301134878081e-06,
+      "loss": 0.9428,
+      "step": 17648
+    },
+    {
+      "epoch": 0.641851838382369,
+      "grad_norm": 0.0,
+      "learning_rate": 6.006221259607154e-06,
+      "loss": 0.9289,
+      "step": 17649
+    },
+    {
+      "epoch": 0.6418882059861076,
+      "grad_norm": 0.0,
+      "learning_rate": 6.005141439744247e-06,
+      "loss": 0.9656,
+      "step": 17650
+    },
+    {
+      "epoch": 0.6419245735898461,
+      "grad_norm": 0.0,
+      "learning_rate": 6.004061675304327e-06,
+      "loss": 0.9795,
+      "step": 17651
+    },
+    {
+      "epoch": 0.6419609411935847,
+      "grad_norm": 0.0,
+      "learning_rate": 6.002981966302384e-06,
+      "loss": 0.9265,
+      "step": 17652
+    },
+    {
+      "epoch": 0.6419973087973233,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0019023127533935e-06,
+      "loss": 1.027,
+      "step": 17653
+    },
+    {
+      "epoch": 0.6420336764010619,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0008227146723386e-06,
+      "loss": 0.9189,
+      "step": 17654
+    },
+    {
+      "epoch": 0.6420700440048005,
+      "grad_norm": 0.0,
+      "learning_rate": 5.999743172074193e-06,
+      "loss": 0.8196,
+      "step": 17655
+    },
+    {
+      "epoch": 0.6421064116085391,
+      "grad_norm": 0.0,
+      "learning_rate": 5.998663684973938e-06,
+      "loss": 0.8068,
+      "step": 17656
+    },
+    {
+      "epoch": 0.6421427792122777,
+      "grad_norm": 0.0,
+      "learning_rate": 5.997584253386541e-06,
+      "loss": 1.0078,
+      "step": 17657
+    },
+    {
+      "epoch": 0.6421791468160163,
+      "grad_norm": 0.0,
+      "learning_rate": 5.996504877326989e-06,
+      "loss": 0.8727,
+      "step": 17658
+    },
+    {
+      "epoch": 0.6422155144197549,
+      "grad_norm": 0.0,
+      "learning_rate": 5.995425556810251e-06,
+      "loss": 0.9859,
+      "step": 17659
+    },
+    {
+      "epoch": 0.6422518820234935,
+      "grad_norm": 0.0,
+      "learning_rate": 5.994346291851301e-06,
+      "loss": 0.9444,
+      "step": 17660
+    },
+    {
+      "epoch": 0.6422882496272321,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9932670824651105e-06,
+      "loss": 0.82,
+      "step": 17661
+    },
+    {
+      "epoch": 0.6423246172309707,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9921879286666575e-06,
+      "loss": 0.9916,
+      "step": 17662
+    },
+    {
+      "epoch": 0.6423609848347093,
+      "grad_norm": 0.0,
+      "learning_rate": 5.991108830470909e-06,
+      "loss": 0.8134,
+      "step": 17663
+    },
+    {
+      "epoch": 0.6423973524384479,
+      "grad_norm": 0.0,
+      "learning_rate": 5.990029787892841e-06,
+      "loss": 0.9096,
+      "step": 17664
+    },
+    {
+      "epoch": 0.6424337200421865,
+      "grad_norm": 0.0,
+      "learning_rate": 5.988950800947417e-06,
+      "loss": 0.9103,
+      "step": 17665
+    },
+    {
+      "epoch": 0.6424700876459251,
+      "grad_norm": 0.0,
+      "learning_rate": 5.987871869649611e-06,
+      "loss": 0.8512,
+      "step": 17666
+    },
+    {
+      "epoch": 0.6425064552496637,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9867929940143875e-06,
+      "loss": 1.0399,
+      "step": 17667
+    },
+    {
+      "epoch": 0.6425428228534021,
+      "grad_norm": 0.0,
+      "learning_rate": 5.985714174056721e-06,
+      "loss": 0.9922,
+      "step": 17668
+    },
+    {
+      "epoch": 0.6425791904571407,
+      "grad_norm": 0.0,
+      "learning_rate": 5.984635409791576e-06,
+      "loss": 1.0163,
+      "step": 17669
+    },
+    {
+      "epoch": 0.6426155580608793,
+      "grad_norm": 0.0,
+      "learning_rate": 5.983556701233913e-06,
+      "loss": 0.9932,
+      "step": 17670
+    },
+    {
+      "epoch": 0.6426519256646179,
+      "grad_norm": 0.0,
+      "learning_rate": 5.982478048398706e-06,
+      "loss": 0.9568,
+      "step": 17671
+    },
+    {
+      "epoch": 0.6426882932683565,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9813994513009125e-06,
+      "loss": 0.8369,
+      "step": 17672
+    },
+    {
+      "epoch": 0.6427246608720951,
+      "grad_norm": 0.0,
+      "learning_rate": 5.980320909955501e-06,
+      "loss": 0.7276,
+      "step": 17673
+    },
+    {
+      "epoch": 0.6427610284758337,
+      "grad_norm": 0.0,
+      "learning_rate": 5.979242424377434e-06,
+      "loss": 0.9449,
+      "step": 17674
+    },
+    {
+      "epoch": 0.6427973960795723,
+      "grad_norm": 0.0,
+      "learning_rate": 5.978163994581674e-06,
+      "loss": 0.8851,
+      "step": 17675
+    },
+    {
+      "epoch": 0.6428337636833109,
+      "grad_norm": 0.0,
+      "learning_rate": 5.97708562058318e-06,
+      "loss": 1.0547,
+      "step": 17676
+    },
+    {
+      "epoch": 0.6428701312870495,
+      "grad_norm": 0.0,
+      "learning_rate": 5.976007302396923e-06,
+      "loss": 1.0714,
+      "step": 17677
+    },
+    {
+      "epoch": 0.6429064988907881,
+      "grad_norm": 0.0,
+      "learning_rate": 5.974929040037847e-06,
+      "loss": 1.0257,
+      "step": 17678
+    },
+    {
+      "epoch": 0.6429428664945267,
+      "grad_norm": 0.0,
+      "learning_rate": 5.973850833520923e-06,
+      "loss": 1.0355,
+      "step": 17679
+    },
+    {
+      "epoch": 0.6429792340982653,
+      "grad_norm": 0.0,
+      "learning_rate": 5.972772682861105e-06,
+      "loss": 0.9503,
+      "step": 17680
+    },
+    {
+      "epoch": 0.6430156017020039,
+      "grad_norm": 0.0,
+      "learning_rate": 5.971694588073353e-06,
+      "loss": 0.9362,
+      "step": 17681
+    },
+    {
+      "epoch": 0.6430519693057425,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9706165491726235e-06,
+      "loss": 0.8284,
+      "step": 17682
+    },
+    {
+      "epoch": 0.6430883369094811,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9695385661738735e-06,
+      "loss": 0.9401,
+      "step": 17683
+    },
+    {
+      "epoch": 0.6431247045132196,
+      "grad_norm": 0.0,
+      "learning_rate": 5.968460639092053e-06,
+      "loss": 0.877,
+      "step": 17684
+    },
+    {
+      "epoch": 0.6431610721169582,
+      "grad_norm": 0.0,
+      "learning_rate": 5.967382767942127e-06,
+      "loss": 0.8824,
+      "step": 17685
+    },
+    {
+      "epoch": 0.6431974397206968,
+      "grad_norm": 0.0,
+      "learning_rate": 5.966304952739039e-06,
+      "loss": 0.8558,
+      "step": 17686
+    },
+    {
+      "epoch": 0.6432338073244354,
+      "grad_norm": 0.0,
+      "learning_rate": 5.96522719349775e-06,
+      "loss": 0.9809,
+      "step": 17687
+    },
+    {
+      "epoch": 0.643270174928174,
+      "grad_norm": 0.0,
+      "learning_rate": 5.964149490233211e-06,
+      "loss": 0.7968,
+      "step": 17688
+    },
+    {
+      "epoch": 0.6433065425319126,
+      "grad_norm": 0.0,
+      "learning_rate": 5.963071842960364e-06,
+      "loss": 0.908,
+      "step": 17689
+    },
+    {
+      "epoch": 0.6433429101356511,
+      "grad_norm": 0.0,
+      "learning_rate": 5.961994251694174e-06,
+      "loss": 1.048,
+      "step": 17690
+    },
+    {
+      "epoch": 0.6433792777393897,
+      "grad_norm": 0.0,
+      "learning_rate": 5.960916716449581e-06,
+      "loss": 1.0454,
+      "step": 17691
+    },
+    {
+      "epoch": 0.6434156453431283,
+      "grad_norm": 0.0,
+      "learning_rate": 5.95983923724154e-06,
+      "loss": 0.9185,
+      "step": 17692
+    },
+    {
+      "epoch": 0.6434520129468669,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9587618140849966e-06,
+      "loss": 0.9982,
+      "step": 17693
+    },
+    {
+      "epoch": 0.6434883805506055,
+      "grad_norm": 0.0,
+      "learning_rate": 5.957684446994901e-06,
+      "loss": 0.8936,
+      "step": 17694
+    },
+    {
+      "epoch": 0.6435247481543441,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9566071359861945e-06,
+      "loss": 0.8278,
+      "step": 17695
+    },
+    {
+      "epoch": 0.6435611157580827,
+      "grad_norm": 0.0,
+      "learning_rate": 5.955529881073832e-06,
+      "loss": 0.8707,
+      "step": 17696
+    },
+    {
+      "epoch": 0.6435974833618213,
+      "grad_norm": 0.0,
+      "learning_rate": 5.954452682272749e-06,
+      "loss": 0.9483,
+      "step": 17697
+    },
+    {
+      "epoch": 0.6436338509655599,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9533755395978985e-06,
+      "loss": 0.8911,
+      "step": 17698
+    },
+    {
+      "epoch": 0.6436702185692985,
+      "grad_norm": 0.0,
+      "learning_rate": 5.952298453064218e-06,
+      "loss": 0.9905,
+      "step": 17699
+    },
+    {
+      "epoch": 0.6437065861730371,
+      "grad_norm": 0.0,
+      "learning_rate": 5.951221422686657e-06,
+      "loss": 1.0132,
+      "step": 17700
+    },
+    {
+      "epoch": 0.6437429537767756,
+      "grad_norm": 0.0,
+      "learning_rate": 5.950144448480152e-06,
+      "loss": 0.844,
+      "step": 17701
+    },
+    {
+      "epoch": 0.6437793213805142,
+      "grad_norm": 0.0,
+      "learning_rate": 5.949067530459648e-06,
+      "loss": 0.9494,
+      "step": 17702
+    },
+    {
+      "epoch": 0.6438156889842528,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9479906686400826e-06,
+      "loss": 1.1018,
+      "step": 17703
+    },
+    {
+      "epoch": 0.6438520565879914,
+      "grad_norm": 0.0,
+      "learning_rate": 5.946913863036402e-06,
+      "loss": 0.828,
+      "step": 17704
+    },
+    {
+      "epoch": 0.64388842419173,
+      "grad_norm": 0.0,
+      "learning_rate": 5.945837113663537e-06,
+      "loss": 0.8634,
+      "step": 17705
+    },
+    {
+      "epoch": 0.6439247917954686,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9447604205364304e-06,
+      "loss": 0.9338,
+      "step": 17706
+    },
+    {
+      "epoch": 0.6439611593992072,
+      "grad_norm": 0.0,
+      "learning_rate": 5.943683783670022e-06,
+      "loss": 0.9773,
+      "step": 17707
+    },
+    {
+      "epoch": 0.6439975270029458,
+      "grad_norm": 0.0,
+      "learning_rate": 5.942607203079243e-06,
+      "loss": 0.926,
+      "step": 17708
+    },
+    {
+      "epoch": 0.6440338946066844,
+      "grad_norm": 0.0,
+      "learning_rate": 5.941530678779035e-06,
+      "loss": 0.962,
+      "step": 17709
+    },
+    {
+      "epoch": 0.644070262210423,
+      "grad_norm": 0.0,
+      "learning_rate": 5.940454210784328e-06,
+      "loss": 1.0704,
+      "step": 17710
+    },
+    {
+      "epoch": 0.6441066298141616,
+      "grad_norm": 0.0,
+      "learning_rate": 5.93937779911006e-06,
+      "loss": 0.9212,
+      "step": 17711
+    },
+    {
+      "epoch": 0.6441429974179002,
+      "grad_norm": 0.0,
+      "learning_rate": 5.938301443771163e-06,
+      "loss": 1.0039,
+      "step": 17712
+    },
+    {
+      "epoch": 0.6441793650216388,
+      "grad_norm": 0.0,
+      "learning_rate": 5.937225144782572e-06,
+      "loss": 0.9778,
+      "step": 17713
+    },
+    {
+      "epoch": 0.6442157326253773,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9361489021592155e-06,
+      "loss": 0.9274,
+      "step": 17714
+    },
+    {
+      "epoch": 0.6442521002291159,
+      "grad_norm": 0.0,
+      "learning_rate": 5.935072715916033e-06,
+      "loss": 0.9619,
+      "step": 17715
+    },
+    {
+      "epoch": 0.6442884678328545,
+      "grad_norm": 0.0,
+      "learning_rate": 5.933996586067941e-06,
+      "loss": 0.8561,
+      "step": 17716
+    },
+    {
+      "epoch": 0.6443248354365931,
+      "grad_norm": 0.0,
+      "learning_rate": 5.932920512629884e-06,
+      "loss": 0.9615,
+      "step": 17717
+    },
+    {
+      "epoch": 0.6443612030403316,
+      "grad_norm": 0.0,
+      "learning_rate": 5.931844495616781e-06,
+      "loss": 0.9866,
+      "step": 17718
+    },
+    {
+      "epoch": 0.6443975706440702,
+      "grad_norm": 0.0,
+      "learning_rate": 5.930768535043565e-06,
+      "loss": 0.9075,
+      "step": 17719
+    },
+    {
+      "epoch": 0.6444339382478088,
+      "grad_norm": 0.0,
+      "learning_rate": 5.92969263092516e-06,
+      "loss": 0.9171,
+      "step": 17720
+    },
+    {
+      "epoch": 0.6444703058515474,
+      "grad_norm": 0.0,
+      "learning_rate": 5.928616783276498e-06,
+      "loss": 1.0411,
+      "step": 17721
+    },
+    {
+      "epoch": 0.644506673455286,
+      "grad_norm": 0.0,
+      "learning_rate": 5.927540992112498e-06,
+      "loss": 0.9653,
+      "step": 17722
+    },
+    {
+      "epoch": 0.6445430410590246,
+      "grad_norm": 0.0,
+      "learning_rate": 5.926465257448093e-06,
+      "loss": 0.9723,
+      "step": 17723
+    },
+    {
+      "epoch": 0.6445794086627632,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9253895792981976e-06,
+      "loss": 0.8749,
+      "step": 17724
+    },
+    {
+      "epoch": 0.6446157762665018,
+      "grad_norm": 0.0,
+      "learning_rate": 5.924313957677747e-06,
+      "loss": 1.0431,
+      "step": 17725
+    },
+    {
+      "epoch": 0.6446521438702404,
+      "grad_norm": 0.0,
+      "learning_rate": 5.923238392601655e-06,
+      "loss": 1.0367,
+      "step": 17726
+    },
+    {
+      "epoch": 0.644688511473979,
+      "grad_norm": 0.0,
+      "learning_rate": 5.922162884084843e-06,
+      "loss": 1.0129,
+      "step": 17727
+    },
+    {
+      "epoch": 0.6447248790777176,
+      "grad_norm": 0.0,
+      "learning_rate": 5.921087432142241e-06,
+      "loss": 0.9761,
+      "step": 17728
+    },
+    {
+      "epoch": 0.6447612466814562,
+      "grad_norm": 0.0,
+      "learning_rate": 5.920012036788758e-06,
+      "loss": 0.8849,
+      "step": 17729
+    },
+    {
+      "epoch": 0.6447976142851948,
+      "grad_norm": 0.0,
+      "learning_rate": 5.918936698039324e-06,
+      "loss": 0.9237,
+      "step": 17730
+    },
+    {
+      "epoch": 0.6448339818889334,
+      "grad_norm": 0.0,
+      "learning_rate": 5.91786141590885e-06,
+      "loss": 0.9882,
+      "step": 17731
+    },
+    {
+      "epoch": 0.644870349492672,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9167861904122585e-06,
+      "loss": 0.9218,
+      "step": 17732
+    },
+    {
+      "epoch": 0.6449067170964106,
+      "grad_norm": 0.0,
+      "learning_rate": 5.915711021564463e-06,
+      "loss": 1.0387,
+      "step": 17733
+    },
+    {
+      "epoch": 0.6449430847001492,
+      "grad_norm": 0.0,
+      "learning_rate": 5.914635909380384e-06,
+      "loss": 0.9198,
+      "step": 17734
+    },
+    {
+      "epoch": 0.6449794523038876,
+      "grad_norm": 0.0,
+      "learning_rate": 5.913560853874933e-06,
+      "loss": 0.8593,
+      "step": 17735
+    },
+    {
+      "epoch": 0.6450158199076262,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9124858550630326e-06,
+      "loss": 0.9897,
+      "step": 17736
+    },
+    {
+      "epoch": 0.6450521875113648,
+      "grad_norm": 0.0,
+      "learning_rate": 5.911410912959588e-06,
+      "loss": 1.1077,
+      "step": 17737
+    },
+    {
+      "epoch": 0.6450885551151034,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9103360275795166e-06,
+      "loss": 0.9615,
+      "step": 17738
+    },
+    {
+      "epoch": 0.645124922718842,
+      "grad_norm": 0.0,
+      "learning_rate": 5.909261198937728e-06,
+      "loss": 0.8931,
+      "step": 17739
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 0.0,
+      "learning_rate": 5.908186427049139e-06,
+      "loss": 0.8858,
+      "step": 17740
+    },
+    {
+      "epoch": 0.6451976579263192,
+      "grad_norm": 0.0,
+      "learning_rate": 5.907111711928657e-06,
+      "loss": 0.969,
+      "step": 17741
+    },
+    {
+      "epoch": 0.6452340255300578,
+      "grad_norm": 0.0,
+      "learning_rate": 5.906037053591196e-06,
+      "loss": 0.925,
+      "step": 17742
+    },
+    {
+      "epoch": 0.6452703931337964,
+      "grad_norm": 0.0,
+      "learning_rate": 5.904962452051655e-06,
+      "loss": 0.9012,
+      "step": 17743
+    },
+    {
+      "epoch": 0.645306760737535,
+      "grad_norm": 0.0,
+      "learning_rate": 5.903887907324958e-06,
+      "loss": 0.8615,
+      "step": 17744
+    },
+    {
+      "epoch": 0.6453431283412736,
+      "grad_norm": 0.0,
+      "learning_rate": 5.902813419426001e-06,
+      "loss": 0.8161,
+      "step": 17745
+    },
+    {
+      "epoch": 0.6453794959450122,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9017389883696945e-06,
+      "loss": 0.8963,
+      "step": 17746
+    },
+    {
+      "epoch": 0.6454158635487508,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9006646141709455e-06,
+      "loss": 0.8809,
+      "step": 17747
+    },
+    {
+      "epoch": 0.6454522311524894,
+      "grad_norm": 0.0,
+      "learning_rate": 5.899590296844658e-06,
+      "loss": 0.932,
+      "step": 17748
+    },
+    {
+      "epoch": 0.645488598756228,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8985160364057395e-06,
+      "loss": 1.0051,
+      "step": 17749
+    },
+    {
+      "epoch": 0.6455249663599666,
+      "grad_norm": 0.0,
+      "learning_rate": 5.897441832869089e-06,
+      "loss": 1.1366,
+      "step": 17750
+    },
+    {
+      "epoch": 0.6455613339637052,
+      "grad_norm": 0.0,
+      "learning_rate": 5.896367686249614e-06,
+      "loss": 0.8951,
+      "step": 17751
+    },
+    {
+      "epoch": 0.6455977015674437,
+      "grad_norm": 0.0,
+      "learning_rate": 5.895293596562212e-06,
+      "loss": 1.0024,
+      "step": 17752
+    },
+    {
+      "epoch": 0.6456340691711823,
+      "grad_norm": 0.0,
+      "learning_rate": 5.894219563821791e-06,
+      "loss": 0.8622,
+      "step": 17753
+    },
+    {
+      "epoch": 0.6456704367749209,
+      "grad_norm": 0.0,
+      "learning_rate": 5.893145588043245e-06,
+      "loss": 0.8797,
+      "step": 17754
+    },
+    {
+      "epoch": 0.6457068043786595,
+      "grad_norm": 0.0,
+      "learning_rate": 5.892071669241482e-06,
+      "loss": 0.9149,
+      "step": 17755
+    },
+    {
+      "epoch": 0.6457431719823981,
+      "grad_norm": 0.0,
+      "learning_rate": 5.89099780743139e-06,
+      "loss": 0.9891,
+      "step": 17756
+    },
+    {
+      "epoch": 0.6457795395861367,
+      "grad_norm": 0.0,
+      "learning_rate": 5.889924002627881e-06,
+      "loss": 0.8833,
+      "step": 17757
+    },
+    {
+      "epoch": 0.6458159071898752,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8888502548458384e-06,
+      "loss": 0.7282,
+      "step": 17758
+    },
+    {
+      "epoch": 0.6458522747936138,
+      "grad_norm": 0.0,
+      "learning_rate": 5.887776564100169e-06,
+      "loss": 0.9406,
+      "step": 17759
+    },
+    {
+      "epoch": 0.6458886423973524,
+      "grad_norm": 0.0,
+      "learning_rate": 5.886702930405764e-06,
+      "loss": 0.8255,
+      "step": 17760
+    },
+    {
+      "epoch": 0.645925010001091,
+      "grad_norm": 0.0,
+      "learning_rate": 5.88562935377752e-06,
+      "loss": 0.9628,
+      "step": 17761
+    },
+    {
+      "epoch": 0.6459613776048296,
+      "grad_norm": 0.0,
+      "learning_rate": 5.88455583423033e-06,
+      "loss": 1.0216,
+      "step": 17762
+    },
+    {
+      "epoch": 0.6459977452085682,
+      "grad_norm": 0.0,
+      "learning_rate": 5.883482371779093e-06,
+      "loss": 0.9386,
+      "step": 17763
+    },
+    {
+      "epoch": 0.6460341128123068,
+      "grad_norm": 0.0,
+      "learning_rate": 5.882408966438695e-06,
+      "loss": 0.9668,
+      "step": 17764
+    },
+    {
+      "epoch": 0.6460704804160454,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8813356182240285e-06,
+      "loss": 0.9724,
+      "step": 17765
+    },
+    {
+      "epoch": 0.646106848019784,
+      "grad_norm": 0.0,
+      "learning_rate": 5.880262327149989e-06,
+      "loss": 0.9633,
+      "step": 17766
+    },
+    {
+      "epoch": 0.6461432156235226,
+      "grad_norm": 0.0,
+      "learning_rate": 5.879189093231463e-06,
+      "loss": 0.8773,
+      "step": 17767
+    },
+    {
+      "epoch": 0.6461795832272611,
+      "grad_norm": 0.0,
+      "learning_rate": 5.878115916483344e-06,
+      "loss": 1.07,
+      "step": 17768
+    },
+    {
+      "epoch": 0.6462159508309997,
+      "grad_norm": 0.0,
+      "learning_rate": 5.877042796920514e-06,
+      "loss": 1.0526,
+      "step": 17769
+    },
+    {
+      "epoch": 0.6462523184347383,
+      "grad_norm": 0.0,
+      "learning_rate": 5.875969734557872e-06,
+      "loss": 1.0222,
+      "step": 17770
+    },
+    {
+      "epoch": 0.6462886860384769,
+      "grad_norm": 0.0,
+      "learning_rate": 5.874896729410293e-06,
+      "loss": 1.0456,
+      "step": 17771
+    },
+    {
+      "epoch": 0.6463250536422155,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8738237814926735e-06,
+      "loss": 0.9208,
+      "step": 17772
+    },
+    {
+      "epoch": 0.6463614212459541,
+      "grad_norm": 0.0,
+      "learning_rate": 5.87275089081989e-06,
+      "loss": 0.9093,
+      "step": 17773
+    },
+    {
+      "epoch": 0.6463977888496927,
+      "grad_norm": 0.0,
+      "learning_rate": 5.871678057406835e-06,
+      "loss": 1.0008,
+      "step": 17774
+    },
+    {
+      "epoch": 0.6464341564534313,
+      "grad_norm": 0.0,
+      "learning_rate": 5.870605281268389e-06,
+      "loss": 1.0188,
+      "step": 17775
+    },
+    {
+      "epoch": 0.6464705240571699,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8695325624194385e-06,
+      "loss": 0.8963,
+      "step": 17776
+    },
+    {
+      "epoch": 0.6465068916609085,
+      "grad_norm": 0.0,
+      "learning_rate": 5.86845990087486e-06,
+      "loss": 0.9014,
+      "step": 17777
+    },
+    {
+      "epoch": 0.6465432592646471,
+      "grad_norm": 0.0,
+      "learning_rate": 5.86738729664954e-06,
+      "loss": 0.92,
+      "step": 17778
+    },
+    {
+      "epoch": 0.6465796268683857,
+      "grad_norm": 0.0,
+      "learning_rate": 5.866314749758356e-06,
+      "loss": 0.9777,
+      "step": 17779
+    },
+    {
+      "epoch": 0.6466159944721243,
+      "grad_norm": 0.0,
+      "learning_rate": 5.865242260216193e-06,
+      "loss": 1.0318,
+      "step": 17780
+    },
+    {
+      "epoch": 0.6466523620758629,
+      "grad_norm": 0.0,
+      "learning_rate": 5.864169828037923e-06,
+      "loss": 0.9698,
+      "step": 17781
+    },
+    {
+      "epoch": 0.6466887296796014,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8630974532384345e-06,
+      "loss": 0.8974,
+      "step": 17782
+    },
+    {
+      "epoch": 0.64672509728334,
+      "grad_norm": 0.0,
+      "learning_rate": 5.862025135832596e-06,
+      "loss": 0.9533,
+      "step": 17783
+    },
+    {
+      "epoch": 0.6467614648870786,
+      "grad_norm": 0.0,
+      "learning_rate": 5.860952875835286e-06,
+      "loss": 1.0655,
+      "step": 17784
+    },
+    {
+      "epoch": 0.6467978324908171,
+      "grad_norm": 0.0,
+      "learning_rate": 5.859880673261384e-06,
+      "loss": 0.9746,
+      "step": 17785
+    },
+    {
+      "epoch": 0.6468342000945557,
+      "grad_norm": 0.0,
+      "learning_rate": 5.858808528125763e-06,
+      "loss": 0.9073,
+      "step": 17786
+    },
+    {
+      "epoch": 0.6468705676982943,
+      "grad_norm": 0.0,
+      "learning_rate": 5.857736440443302e-06,
+      "loss": 0.8388,
+      "step": 17787
+    },
+    {
+      "epoch": 0.6469069353020329,
+      "grad_norm": 0.0,
+      "learning_rate": 5.856664410228864e-06,
+      "loss": 0.9651,
+      "step": 17788
+    },
+    {
+      "epoch": 0.6469433029057715,
+      "grad_norm": 0.0,
+      "learning_rate": 5.855592437497335e-06,
+      "loss": 0.9112,
+      "step": 17789
+    },
+    {
+      "epoch": 0.6469796705095101,
+      "grad_norm": 0.0,
+      "learning_rate": 5.854520522263576e-06,
+      "loss": 0.9137,
+      "step": 17790
+    },
+    {
+      "epoch": 0.6470160381132487,
+      "grad_norm": 0.0,
+      "learning_rate": 5.853448664542466e-06,
+      "loss": 0.937,
+      "step": 17791
+    },
+    {
+      "epoch": 0.6470524057169873,
+      "grad_norm": 0.0,
+      "learning_rate": 5.85237686434887e-06,
+      "loss": 1.0372,
+      "step": 17792
+    },
+    {
+      "epoch": 0.6470887733207259,
+      "grad_norm": 0.0,
+      "learning_rate": 5.851305121697663e-06,
+      "loss": 0.8871,
+      "step": 17793
+    },
+    {
+      "epoch": 0.6471251409244645,
+      "grad_norm": 0.0,
+      "learning_rate": 5.850233436603709e-06,
+      "loss": 0.8867,
+      "step": 17794
+    },
+    {
+      "epoch": 0.6471615085282031,
+      "grad_norm": 0.0,
+      "learning_rate": 5.849161809081882e-06,
+      "loss": 0.9116,
+      "step": 17795
+    },
+    {
+      "epoch": 0.6471978761319417,
+      "grad_norm": 0.0,
+      "learning_rate": 5.84809023914704e-06,
+      "loss": 0.8127,
+      "step": 17796
+    },
+    {
+      "epoch": 0.6472342437356803,
+      "grad_norm": 0.0,
+      "learning_rate": 5.84701872681406e-06,
+      "loss": 0.9578,
+      "step": 17797
+    },
+    {
+      "epoch": 0.6472706113394189,
+      "grad_norm": 0.0,
+      "learning_rate": 5.845947272097801e-06,
+      "loss": 1.0783,
+      "step": 17798
+    },
+    {
+      "epoch": 0.6473069789431575,
+      "grad_norm": 0.0,
+      "learning_rate": 5.844875875013129e-06,
+      "loss": 0.969,
+      "step": 17799
+    },
+    {
+      "epoch": 0.6473433465468961,
+      "grad_norm": 0.0,
+      "learning_rate": 5.843804535574908e-06,
+      "loss": 1.0454,
+      "step": 17800
+    },
+    {
+      "epoch": 0.6473797141506347,
+      "grad_norm": 0.0,
+      "learning_rate": 5.842733253798005e-06,
+      "loss": 0.9837,
+      "step": 17801
+    },
+    {
+      "epoch": 0.6474160817543732,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8416620296972815e-06,
+      "loss": 0.9718,
+      "step": 17802
+    },
+    {
+      "epoch": 0.6474524493581117,
+      "grad_norm": 0.0,
+      "learning_rate": 5.840590863287594e-06,
+      "loss": 0.9922,
+      "step": 17803
+    },
+    {
+      "epoch": 0.6474888169618503,
+      "grad_norm": 0.0,
+      "learning_rate": 5.839519754583808e-06,
+      "loss": 1.004,
+      "step": 17804
+    },
+    {
+      "epoch": 0.6475251845655889,
+      "grad_norm": 0.0,
+      "learning_rate": 5.838448703600781e-06,
+      "loss": 0.889,
+      "step": 17805
+    },
+    {
+      "epoch": 0.6475615521693275,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8373777103533746e-06,
+      "loss": 0.9522,
+      "step": 17806
+    },
+    {
+      "epoch": 0.6475979197730661,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8363067748564474e-06,
+      "loss": 1.0034,
+      "step": 17807
+    },
+    {
+      "epoch": 0.6476342873768047,
+      "grad_norm": 0.0,
+      "learning_rate": 5.835235897124856e-06,
+      "loss": 0.8723,
+      "step": 17808
+    },
+    {
+      "epoch": 0.6476706549805433,
+      "grad_norm": 0.0,
+      "learning_rate": 5.834165077173452e-06,
+      "loss": 1.021,
+      "step": 17809
+    },
+    {
+      "epoch": 0.6477070225842819,
+      "grad_norm": 0.0,
+      "learning_rate": 5.833094315017104e-06,
+      "loss": 1.0161,
+      "step": 17810
+    },
+    {
+      "epoch": 0.6477433901880205,
+      "grad_norm": 0.0,
+      "learning_rate": 5.832023610670659e-06,
+      "loss": 1.0097,
+      "step": 17811
+    },
+    {
+      "epoch": 0.6477797577917591,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8309529641489726e-06,
+      "loss": 0.9908,
+      "step": 17812
+    },
+    {
+      "epoch": 0.6478161253954977,
+      "grad_norm": 0.0,
+      "learning_rate": 5.829882375466897e-06,
+      "loss": 0.9778,
+      "step": 17813
+    },
+    {
+      "epoch": 0.6478524929992363,
+      "grad_norm": 0.0,
+      "learning_rate": 5.828811844639291e-06,
+      "loss": 0.981,
+      "step": 17814
+    },
+    {
+      "epoch": 0.6478888606029749,
+      "grad_norm": 0.0,
+      "learning_rate": 5.827741371680995e-06,
+      "loss": 0.9391,
+      "step": 17815
+    },
+    {
+      "epoch": 0.6479252282067135,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8266709566068726e-06,
+      "loss": 0.9415,
+      "step": 17816
+    },
+    {
+      "epoch": 0.6479615958104521,
+      "grad_norm": 0.0,
+      "learning_rate": 5.82560059943177e-06,
+      "loss": 0.939,
+      "step": 17817
+    },
+    {
+      "epoch": 0.6479979634141907,
+      "grad_norm": 0.0,
+      "learning_rate": 5.824530300170537e-06,
+      "loss": 0.9489,
+      "step": 17818
+    },
+    {
+      "epoch": 0.6480343310179292,
+      "grad_norm": 0.0,
+      "learning_rate": 5.823460058838015e-06,
+      "loss": 0.938,
+      "step": 17819
+    },
+    {
+      "epoch": 0.6480706986216678,
+      "grad_norm": 0.0,
+      "learning_rate": 5.822389875449071e-06,
+      "loss": 1.0657,
+      "step": 17820
+    },
+    {
+      "epoch": 0.6481070662254064,
+      "grad_norm": 0.0,
+      "learning_rate": 5.821319750018533e-06,
+      "loss": 0.9394,
+      "step": 17821
+    },
+    {
+      "epoch": 0.648143433829145,
+      "grad_norm": 0.0,
+      "learning_rate": 5.820249682561252e-06,
+      "loss": 0.9758,
+      "step": 17822
+    },
+    {
+      "epoch": 0.6481798014328836,
+      "grad_norm": 0.0,
+      "learning_rate": 5.819179673092084e-06,
+      "loss": 1.0556,
+      "step": 17823
+    },
+    {
+      "epoch": 0.6482161690366222,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8181097216258585e-06,
+      "loss": 0.9384,
+      "step": 17824
+    },
+    {
+      "epoch": 0.6482525366403608,
+      "grad_norm": 0.0,
+      "learning_rate": 5.817039828177431e-06,
+      "loss": 0.906,
+      "step": 17825
+    },
+    {
+      "epoch": 0.6482889042440994,
+      "grad_norm": 0.0,
+      "learning_rate": 5.815969992761642e-06,
+      "loss": 0.999,
+      "step": 17826
+    },
+    {
+      "epoch": 0.648325271847838,
+      "grad_norm": 0.0,
+      "learning_rate": 5.814900215393332e-06,
+      "loss": 0.8552,
+      "step": 17827
+    },
+    {
+      "epoch": 0.6483616394515765,
+      "grad_norm": 0.0,
+      "learning_rate": 5.813830496087341e-06,
+      "loss": 0.8723,
+      "step": 17828
+    },
+    {
+      "epoch": 0.6483980070553151,
+      "grad_norm": 0.0,
+      "learning_rate": 5.812760834858514e-06,
+      "loss": 1.0375,
+      "step": 17829
+    },
+    {
+      "epoch": 0.6484343746590537,
+      "grad_norm": 0.0,
+      "learning_rate": 5.811691231721691e-06,
+      "loss": 0.986,
+      "step": 17830
+    },
+    {
+      "epoch": 0.6484707422627923,
+      "grad_norm": 0.0,
+      "learning_rate": 5.81062168669171e-06,
+      "loss": 0.9231,
+      "step": 17831
+    },
+    {
+      "epoch": 0.6485071098665309,
+      "grad_norm": 0.0,
+      "learning_rate": 5.809552199783405e-06,
+      "loss": 0.8902,
+      "step": 17832
+    },
+    {
+      "epoch": 0.6485434774702695,
+      "grad_norm": 0.0,
+      "learning_rate": 5.808482771011625e-06,
+      "loss": 0.845,
+      "step": 17833
+    },
+    {
+      "epoch": 0.6485798450740081,
+      "grad_norm": 0.0,
+      "learning_rate": 5.807413400391192e-06,
+      "loss": 0.8813,
+      "step": 17834
+    },
+    {
+      "epoch": 0.6486162126777467,
+      "grad_norm": 0.0,
+      "learning_rate": 5.806344087936954e-06,
+      "loss": 0.9626,
+      "step": 17835
+    },
+    {
+      "epoch": 0.6486525802814852,
+      "grad_norm": 0.0,
+      "learning_rate": 5.805274833663741e-06,
+      "loss": 1.0792,
+      "step": 17836
+    },
+    {
+      "epoch": 0.6486889478852238,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8042056375863885e-06,
+      "loss": 0.9323,
+      "step": 17837
+    },
+    {
+      "epoch": 0.6487253154889624,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8031364997197255e-06,
+      "loss": 0.9345,
+      "step": 17838
+    },
+    {
+      "epoch": 0.648761683092701,
+      "grad_norm": 0.0,
+      "learning_rate": 5.802067420078599e-06,
+      "loss": 1.0099,
+      "step": 17839
+    },
+    {
+      "epoch": 0.6487980506964396,
+      "grad_norm": 0.0,
+      "learning_rate": 5.800998398677824e-06,
+      "loss": 0.9147,
+      "step": 17840
+    },
+    {
+      "epoch": 0.6488344183001782,
+      "grad_norm": 0.0,
+      "learning_rate": 5.799929435532237e-06,
+      "loss": 0.8978,
+      "step": 17841
+    },
+    {
+      "epoch": 0.6488707859039168,
+      "grad_norm": 0.0,
+      "learning_rate": 5.798860530656671e-06,
+      "loss": 0.8164,
+      "step": 17842
+    },
+    {
+      "epoch": 0.6489071535076554,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7977916840659564e-06,
+      "loss": 0.8819,
+      "step": 17843
+    },
+    {
+      "epoch": 0.648943521111394,
+      "grad_norm": 0.0,
+      "learning_rate": 5.79672289577492e-06,
+      "loss": 1.0322,
+      "step": 17844
+    },
+    {
+      "epoch": 0.6489798887151326,
+      "grad_norm": 0.0,
+      "learning_rate": 5.79565416579839e-06,
+      "loss": 0.8231,
+      "step": 17845
+    },
+    {
+      "epoch": 0.6490162563188712,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7945854941511925e-06,
+      "loss": 0.8892,
+      "step": 17846
+    },
+    {
+      "epoch": 0.6490526239226098,
+      "grad_norm": 0.0,
+      "learning_rate": 5.793516880848151e-06,
+      "loss": 0.9197,
+      "step": 17847
+    },
+    {
+      "epoch": 0.6490889915263484,
+      "grad_norm": 0.0,
+      "learning_rate": 5.792448325904099e-06,
+      "loss": 0.8976,
+      "step": 17848
+    },
+    {
+      "epoch": 0.649125359130087,
+      "grad_norm": 0.0,
+      "learning_rate": 5.791379829333856e-06,
+      "loss": 1.0049,
+      "step": 17849
+    },
+    {
+      "epoch": 0.6491617267338256,
+      "grad_norm": 0.0,
+      "learning_rate": 5.790311391152248e-06,
+      "loss": 0.9435,
+      "step": 17850
+    },
+    {
+      "epoch": 0.6491980943375641,
+      "grad_norm": 0.0,
+      "learning_rate": 5.78924301137409e-06,
+      "loss": 0.9337,
+      "step": 17851
+    },
+    {
+      "epoch": 0.6492344619413027,
+      "grad_norm": 0.0,
+      "learning_rate": 5.788174690014222e-06,
+      "loss": 0.9282,
+      "step": 17852
+    },
+    {
+      "epoch": 0.6492708295450412,
+      "grad_norm": 0.0,
+      "learning_rate": 5.787106427087446e-06,
+      "loss": 0.9539,
+      "step": 17853
+    },
+    {
+      "epoch": 0.6493071971487798,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7860382226085924e-06,
+      "loss": 0.8743,
+      "step": 17854
+    },
+    {
+      "epoch": 0.6493435647525184,
+      "grad_norm": 0.0,
+      "learning_rate": 5.784970076592482e-06,
+      "loss": 0.9068,
+      "step": 17855
+    },
+    {
+      "epoch": 0.649379932356257,
+      "grad_norm": 0.0,
+      "learning_rate": 5.783901989053931e-06,
+      "loss": 0.9261,
+      "step": 17856
+    },
+    {
+      "epoch": 0.6494162999599956,
+      "grad_norm": 0.0,
+      "learning_rate": 5.782833960007752e-06,
+      "loss": 0.9825,
+      "step": 17857
+    },
+    {
+      "epoch": 0.6494526675637342,
+      "grad_norm": 0.0,
+      "learning_rate": 5.78176598946878e-06,
+      "loss": 0.9596,
+      "step": 17858
+    },
+    {
+      "epoch": 0.6494890351674728,
+      "grad_norm": 0.0,
+      "learning_rate": 5.780698077451809e-06,
+      "loss": 0.9264,
+      "step": 17859
+    },
+    {
+      "epoch": 0.6495254027712114,
+      "grad_norm": 0.0,
+      "learning_rate": 5.779630223971671e-06,
+      "loss": 0.9491,
+      "step": 17860
+    },
+    {
+      "epoch": 0.64956177037495,
+      "grad_norm": 0.0,
+      "learning_rate": 5.778562429043175e-06,
+      "loss": 0.963,
+      "step": 17861
+    },
+    {
+      "epoch": 0.6495981379786886,
+      "grad_norm": 0.0,
+      "learning_rate": 5.777494692681137e-06,
+      "loss": 1.0693,
+      "step": 17862
+    },
+    {
+      "epoch": 0.6496345055824272,
+      "grad_norm": 0.0,
+      "learning_rate": 5.776427014900368e-06,
+      "loss": 1.005,
+      "step": 17863
+    },
+    {
+      "epoch": 0.6496708731861658,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7753593957156785e-06,
+      "loss": 0.8582,
+      "step": 17864
+    },
+    {
+      "epoch": 0.6497072407899044,
+      "grad_norm": 0.0,
+      "learning_rate": 5.774291835141892e-06,
+      "loss": 0.9983,
+      "step": 17865
+    },
+    {
+      "epoch": 0.649743608393643,
+      "grad_norm": 0.0,
+      "learning_rate": 5.773224333193801e-06,
+      "loss": 1.0545,
+      "step": 17866
+    },
+    {
+      "epoch": 0.6497799759973816,
+      "grad_norm": 0.0,
+      "learning_rate": 5.772156889886231e-06,
+      "loss": 0.9001,
+      "step": 17867
+    },
+    {
+      "epoch": 0.6498163436011202,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7710895052339844e-06,
+      "loss": 0.8677,
+      "step": 17868
+    },
+    {
+      "epoch": 0.6498527112048587,
+      "grad_norm": 0.0,
+      "learning_rate": 5.770022179251873e-06,
+      "loss": 0.9947,
+      "step": 17869
+    },
+    {
+      "epoch": 0.6498890788085973,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7689549119546965e-06,
+      "loss": 0.8948,
+      "step": 17870
+    },
+    {
+      "epoch": 0.6499254464123358,
+      "grad_norm": 0.0,
+      "learning_rate": 5.767887703357275e-06,
+      "loss": 0.9653,
+      "step": 17871
+    },
+    {
+      "epoch": 0.6499618140160744,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7668205534744e-06,
+      "loss": 1.0081,
+      "step": 17872
+    },
+    {
+      "epoch": 0.649998181619813,
+      "grad_norm": 0.0,
+      "learning_rate": 5.765753462320888e-06,
+      "loss": 0.9446,
+      "step": 17873
+    },
+    {
+      "epoch": 0.6500345492235516,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7646864299115394e-06,
+      "loss": 0.9004,
+      "step": 17874
+    },
+    {
+      "epoch": 0.6500709168272902,
+      "grad_norm": 0.0,
+      "learning_rate": 5.763619456261156e-06,
+      "loss": 1.1575,
+      "step": 17875
+    },
+    {
+      "epoch": 0.6501072844310288,
+      "grad_norm": 0.0,
+      "learning_rate": 5.762552541384539e-06,
+      "loss": 0.8764,
+      "step": 17876
+    },
+    {
+      "epoch": 0.6501436520347674,
+      "grad_norm": 0.0,
+      "learning_rate": 5.761485685296497e-06,
+      "loss": 0.9851,
+      "step": 17877
+    },
+    {
+      "epoch": 0.650180019638506,
+      "grad_norm": 0.0,
+      "learning_rate": 5.760418888011827e-06,
+      "loss": 0.9104,
+      "step": 17878
+    },
+    {
+      "epoch": 0.6502163872422446,
+      "grad_norm": 0.0,
+      "learning_rate": 5.759352149545331e-06,
+      "loss": 0.9641,
+      "step": 17879
+    },
+    {
+      "epoch": 0.6502527548459832,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7582854699118065e-06,
+      "loss": 1.1014,
+      "step": 17880
+    },
+    {
+      "epoch": 0.6502891224497218,
+      "grad_norm": 0.0,
+      "learning_rate": 5.757218849126053e-06,
+      "loss": 1.0132,
+      "step": 17881
+    },
+    {
+      "epoch": 0.6503254900534604,
+      "grad_norm": 0.0,
+      "learning_rate": 5.756152287202867e-06,
+      "loss": 1.0483,
+      "step": 17882
+    },
+    {
+      "epoch": 0.650361857657199,
+      "grad_norm": 0.0,
+      "learning_rate": 5.755085784157044e-06,
+      "loss": 0.8965,
+      "step": 17883
+    },
+    {
+      "epoch": 0.6503982252609376,
+      "grad_norm": 0.0,
+      "learning_rate": 5.754019340003392e-06,
+      "loss": 1.0017,
+      "step": 17884
+    },
+    {
+      "epoch": 0.6504345928646762,
+      "grad_norm": 0.0,
+      "learning_rate": 5.752952954756688e-06,
+      "loss": 0.8121,
+      "step": 17885
+    },
+    {
+      "epoch": 0.6504709604684147,
+      "grad_norm": 0.0,
+      "learning_rate": 5.75188662843174e-06,
+      "loss": 0.9983,
+      "step": 17886
+    },
+    {
+      "epoch": 0.6505073280721533,
+      "grad_norm": 0.0,
+      "learning_rate": 5.750820361043339e-06,
+      "loss": 0.9608,
+      "step": 17887
+    },
+    {
+      "epoch": 0.6505436956758919,
+      "grad_norm": 0.0,
+      "learning_rate": 5.749754152606275e-06,
+      "loss": 0.9692,
+      "step": 17888
+    },
+    {
+      "epoch": 0.6505800632796305,
+      "grad_norm": 0.0,
+      "learning_rate": 5.748688003135338e-06,
+      "loss": 0.8894,
+      "step": 17889
+    },
+    {
+      "epoch": 0.6506164308833691,
+      "grad_norm": 0.0,
+      "learning_rate": 5.747621912645333e-06,
+      "loss": 1.0399,
+      "step": 17890
+    },
+    {
+      "epoch": 0.6506527984871077,
+      "grad_norm": 0.0,
+      "learning_rate": 5.74655588115103e-06,
+      "loss": 0.9589,
+      "step": 17891
+    },
+    {
+      "epoch": 0.6506891660908463,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7454899086672355e-06,
+      "loss": 0.904,
+      "step": 17892
+    },
+    {
+      "epoch": 0.6507255336945849,
+      "grad_norm": 0.0,
+      "learning_rate": 5.744423995208732e-06,
+      "loss": 0.8596,
+      "step": 17893
+    },
+    {
+      "epoch": 0.6507619012983235,
+      "grad_norm": 0.0,
+      "learning_rate": 5.743358140790307e-06,
+      "loss": 0.8842,
+      "step": 17894
+    },
+    {
+      "epoch": 0.650798268902062,
+      "grad_norm": 0.0,
+      "learning_rate": 5.742292345426743e-06,
+      "loss": 0.9399,
+      "step": 17895
+    },
+    {
+      "epoch": 0.6508346365058006,
+      "grad_norm": 0.0,
+      "learning_rate": 5.741226609132838e-06,
+      "loss": 0.9917,
+      "step": 17896
+    },
+    {
+      "epoch": 0.6508710041095392,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7401609319233705e-06,
+      "loss": 1.0173,
+      "step": 17897
+    },
+    {
+      "epoch": 0.6509073717132778,
+      "grad_norm": 0.0,
+      "learning_rate": 5.739095313813126e-06,
+      "loss": 0.9203,
+      "step": 17898
+    },
+    {
+      "epoch": 0.6509437393170164,
+      "grad_norm": 0.0,
+      "learning_rate": 5.73802975481689e-06,
+      "loss": 0.9095,
+      "step": 17899
+    },
+    {
+      "epoch": 0.650980106920755,
+      "grad_norm": 0.0,
+      "learning_rate": 5.736964254949442e-06,
+      "loss": 1.0033,
+      "step": 17900
+    },
+    {
+      "epoch": 0.6510164745244936,
+      "grad_norm": 0.0,
+      "learning_rate": 5.735898814225569e-06,
+      "loss": 0.918,
+      "step": 17901
+    },
+    {
+      "epoch": 0.6510528421282322,
+      "grad_norm": 0.0,
+      "learning_rate": 5.734833432660043e-06,
+      "loss": 0.9498,
+      "step": 17902
+    },
+    {
+      "epoch": 0.6510892097319707,
+      "grad_norm": 0.0,
+      "learning_rate": 5.733768110267661e-06,
+      "loss": 0.983,
+      "step": 17903
+    },
+    {
+      "epoch": 0.6511255773357093,
+      "grad_norm": 0.0,
+      "learning_rate": 5.732702847063187e-06,
+      "loss": 0.8381,
+      "step": 17904
+    },
+    {
+      "epoch": 0.6511619449394479,
+      "grad_norm": 0.0,
+      "learning_rate": 5.73163764306141e-06,
+      "loss": 0.9605,
+      "step": 17905
+    },
+    {
+      "epoch": 0.6511983125431865,
+      "grad_norm": 0.0,
+      "learning_rate": 5.730572498277102e-06,
+      "loss": 0.8375,
+      "step": 17906
+    },
+    {
+      "epoch": 0.6512346801469251,
+      "grad_norm": 0.0,
+      "learning_rate": 5.729507412725046e-06,
+      "loss": 1.0592,
+      "step": 17907
+    },
+    {
+      "epoch": 0.6512710477506637,
+      "grad_norm": 0.0,
+      "learning_rate": 5.72844238642001e-06,
+      "loss": 1.0536,
+      "step": 17908
+    },
+    {
+      "epoch": 0.6513074153544023,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7273774193767805e-06,
+      "loss": 0.9331,
+      "step": 17909
+    },
+    {
+      "epoch": 0.6513437829581409,
+      "grad_norm": 0.0,
+      "learning_rate": 5.726312511610126e-06,
+      "loss": 0.9874,
+      "step": 17910
+    },
+    {
+      "epoch": 0.6513801505618795,
+      "grad_norm": 0.0,
+      "learning_rate": 5.725247663134823e-06,
+      "loss": 0.8391,
+      "step": 17911
+    },
+    {
+      "epoch": 0.6514165181656181,
+      "grad_norm": 0.0,
+      "learning_rate": 5.724182873965642e-06,
+      "loss": 0.7692,
+      "step": 17912
+    },
+    {
+      "epoch": 0.6514528857693567,
+      "grad_norm": 0.0,
+      "learning_rate": 5.723118144117359e-06,
+      "loss": 0.9605,
+      "step": 17913
+    },
+    {
+      "epoch": 0.6514892533730953,
+      "grad_norm": 0.0,
+      "learning_rate": 5.722053473604738e-06,
+      "loss": 0.9505,
+      "step": 17914
+    },
+    {
+      "epoch": 0.6515256209768339,
+      "grad_norm": 0.0,
+      "learning_rate": 5.72098886244256e-06,
+      "loss": 0.861,
+      "step": 17915
+    },
+    {
+      "epoch": 0.6515619885805725,
+      "grad_norm": 0.0,
+      "learning_rate": 5.719924310645591e-06,
+      "loss": 1.0248,
+      "step": 17916
+    },
+    {
+      "epoch": 0.6515983561843111,
+      "grad_norm": 0.0,
+      "learning_rate": 5.718859818228599e-06,
+      "loss": 0.9317,
+      "step": 17917
+    },
+    {
+      "epoch": 0.6516347237880497,
+      "grad_norm": 0.0,
+      "learning_rate": 5.717795385206353e-06,
+      "loss": 0.9136,
+      "step": 17918
+    },
+    {
+      "epoch": 0.6516710913917882,
+      "grad_norm": 0.0,
+      "learning_rate": 5.716731011593622e-06,
+      "loss": 0.8989,
+      "step": 17919
+    },
+    {
+      "epoch": 0.6517074589955267,
+      "grad_norm": 0.0,
+      "learning_rate": 5.715666697405169e-06,
+      "loss": 0.8552,
+      "step": 17920
+    },
+    {
+      "epoch": 0.6517438265992653,
+      "grad_norm": 0.0,
+      "learning_rate": 5.714602442655758e-06,
+      "loss": 1.1162,
+      "step": 17921
+    },
+    {
+      "epoch": 0.6517801942030039,
+      "grad_norm": 0.0,
+      "learning_rate": 5.713538247360166e-06,
+      "loss": 0.9786,
+      "step": 17922
+    },
+    {
+      "epoch": 0.6518165618067425,
+      "grad_norm": 0.0,
+      "learning_rate": 5.712474111533142e-06,
+      "loss": 0.9904,
+      "step": 17923
+    },
+    {
+      "epoch": 0.6518529294104811,
+      "grad_norm": 0.0,
+      "learning_rate": 5.711410035189459e-06,
+      "loss": 0.9337,
+      "step": 17924
+    },
+    {
+      "epoch": 0.6518892970142197,
+      "grad_norm": 0.0,
+      "learning_rate": 5.710346018343878e-06,
+      "loss": 0.8862,
+      "step": 17925
+    },
+    {
+      "epoch": 0.6519256646179583,
+      "grad_norm": 0.0,
+      "learning_rate": 5.70928206101116e-06,
+      "loss": 1.065,
+      "step": 17926
+    },
+    {
+      "epoch": 0.6519620322216969,
+      "grad_norm": 0.0,
+      "learning_rate": 5.708218163206061e-06,
+      "loss": 1.0478,
+      "step": 17927
+    },
+    {
+      "epoch": 0.6519983998254355,
+      "grad_norm": 0.0,
+      "learning_rate": 5.70715432494335e-06,
+      "loss": 0.9001,
+      "step": 17928
+    },
+    {
+      "epoch": 0.6520347674291741,
+      "grad_norm": 0.0,
+      "learning_rate": 5.706090546237782e-06,
+      "loss": 0.9622,
+      "step": 17929
+    },
+    {
+      "epoch": 0.6520711350329127,
+      "grad_norm": 0.0,
+      "learning_rate": 5.705026827104117e-06,
+      "loss": 0.9053,
+      "step": 17930
+    },
+    {
+      "epoch": 0.6521075026366513,
+      "grad_norm": 0.0,
+      "learning_rate": 5.703963167557105e-06,
+      "loss": 1.0313,
+      "step": 17931
+    },
+    {
+      "epoch": 0.6521438702403899,
+      "grad_norm": 0.0,
+      "learning_rate": 5.702899567611517e-06,
+      "loss": 0.9435,
+      "step": 17932
+    },
+    {
+      "epoch": 0.6521802378441285,
+      "grad_norm": 0.0,
+      "learning_rate": 5.701836027282093e-06,
+      "loss": 0.9673,
+      "step": 17933
+    },
+    {
+      "epoch": 0.6522166054478671,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7007725465836e-06,
+      "loss": 0.9574,
+      "step": 17934
+    },
+    {
+      "epoch": 0.6522529730516057,
+      "grad_norm": 0.0,
+      "learning_rate": 5.699709125530788e-06,
+      "loss": 0.9356,
+      "step": 17935
+    },
+    {
+      "epoch": 0.6522893406553443,
+      "grad_norm": 0.0,
+      "learning_rate": 5.698645764138411e-06,
+      "loss": 0.8132,
+      "step": 17936
+    },
+    {
+      "epoch": 0.6523257082590828,
+      "grad_norm": 0.0,
+      "learning_rate": 5.697582462421221e-06,
+      "loss": 0.9897,
+      "step": 17937
+    },
+    {
+      "epoch": 0.6523620758628214,
+      "grad_norm": 0.0,
+      "learning_rate": 5.696519220393969e-06,
+      "loss": 0.929,
+      "step": 17938
+    },
+    {
+      "epoch": 0.65239844346656,
+      "grad_norm": 0.0,
+      "learning_rate": 5.695456038071407e-06,
+      "loss": 1.0482,
+      "step": 17939
+    },
+    {
+      "epoch": 0.6524348110702985,
+      "grad_norm": 0.0,
+      "learning_rate": 5.694392915468283e-06,
+      "loss": 1.0648,
+      "step": 17940
+    },
+    {
+      "epoch": 0.6524711786740371,
+      "grad_norm": 0.0,
+      "learning_rate": 5.693329852599351e-06,
+      "loss": 0.857,
+      "step": 17941
+    },
+    {
+      "epoch": 0.6525075462777757,
+      "grad_norm": 0.0,
+      "learning_rate": 5.692266849479356e-06,
+      "loss": 0.9111,
+      "step": 17942
+    },
+    {
+      "epoch": 0.6525439138815143,
+      "grad_norm": 0.0,
+      "learning_rate": 5.691203906123049e-06,
+      "loss": 0.9971,
+      "step": 17943
+    },
+    {
+      "epoch": 0.6525802814852529,
+      "grad_norm": 0.0,
+      "learning_rate": 5.690141022545172e-06,
+      "loss": 0.9142,
+      "step": 17944
+    },
+    {
+      "epoch": 0.6526166490889915,
+      "grad_norm": 0.0,
+      "learning_rate": 5.689078198760476e-06,
+      "loss": 1.0323,
+      "step": 17945
+    },
+    {
+      "epoch": 0.6526530166927301,
+      "grad_norm": 0.0,
+      "learning_rate": 5.688015434783698e-06,
+      "loss": 1.0465,
+      "step": 17946
+    },
+    {
+      "epoch": 0.6526893842964687,
+      "grad_norm": 0.0,
+      "learning_rate": 5.686952730629591e-06,
+      "loss": 1.0225,
+      "step": 17947
+    },
+    {
+      "epoch": 0.6527257519002073,
+      "grad_norm": 0.0,
+      "learning_rate": 5.685890086312897e-06,
+      "loss": 0.8903,
+      "step": 17948
+    },
+    {
+      "epoch": 0.6527621195039459,
+      "grad_norm": 0.0,
+      "learning_rate": 5.684827501848358e-06,
+      "loss": 0.9596,
+      "step": 17949
+    },
+    {
+      "epoch": 0.6527984871076845,
+      "grad_norm": 0.0,
+      "learning_rate": 5.683764977250708e-06,
+      "loss": 1.0501,
+      "step": 17950
+    },
+    {
+      "epoch": 0.6528348547114231,
+      "grad_norm": 0.0,
+      "learning_rate": 5.682702512534706e-06,
+      "loss": 0.8467,
+      "step": 17951
+    },
+    {
+      "epoch": 0.6528712223151617,
+      "grad_norm": 0.0,
+      "learning_rate": 5.681640107715072e-06,
+      "loss": 0.9633,
+      "step": 17952
+    },
+    {
+      "epoch": 0.6529075899189002,
+      "grad_norm": 0.0,
+      "learning_rate": 5.680577762806558e-06,
+      "loss": 0.963,
+      "step": 17953
+    },
+    {
+      "epoch": 0.6529439575226388,
+      "grad_norm": 0.0,
+      "learning_rate": 5.679515477823899e-06,
+      "loss": 0.9617,
+      "step": 17954
+    },
+    {
+      "epoch": 0.6529803251263774,
+      "grad_norm": 0.0,
+      "learning_rate": 5.678453252781834e-06,
+      "loss": 0.8293,
+      "step": 17955
+    },
+    {
+      "epoch": 0.653016692730116,
+      "grad_norm": 0.0,
+      "learning_rate": 5.677391087695098e-06,
+      "loss": 1.0187,
+      "step": 17956
+    },
+    {
+      "epoch": 0.6530530603338546,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6763289825784274e-06,
+      "loss": 0.8761,
+      "step": 17957
+    },
+    {
+      "epoch": 0.6530894279375932,
+      "grad_norm": 0.0,
+      "learning_rate": 5.675266937446558e-06,
+      "loss": 0.9923,
+      "step": 17958
+    },
+    {
+      "epoch": 0.6531257955413318,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6742049523142196e-06,
+      "loss": 0.9575,
+      "step": 17959
+    },
+    {
+      "epoch": 0.6531621631450704,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6731430271961545e-06,
+      "loss": 0.9798,
+      "step": 17960
+    },
+    {
+      "epoch": 0.653198530748809,
+      "grad_norm": 0.0,
+      "learning_rate": 5.672081162107091e-06,
+      "loss": 0.9164,
+      "step": 17961
+    },
+    {
+      "epoch": 0.6532348983525476,
+      "grad_norm": 0.0,
+      "learning_rate": 5.67101935706176e-06,
+      "loss": 1.0665,
+      "step": 17962
+    },
+    {
+      "epoch": 0.6532712659562862,
+      "grad_norm": 0.0,
+      "learning_rate": 5.66995761207489e-06,
+      "loss": 1.0351,
+      "step": 17963
+    },
+    {
+      "epoch": 0.6533076335600247,
+      "grad_norm": 0.0,
+      "learning_rate": 5.668895927161222e-06,
+      "loss": 0.972,
+      "step": 17964
+    },
+    {
+      "epoch": 0.6533440011637633,
+      "grad_norm": 0.0,
+      "learning_rate": 5.667834302335471e-06,
+      "loss": 1.0139,
+      "step": 17965
+    },
+    {
+      "epoch": 0.6533803687675019,
+      "grad_norm": 0.0,
+      "learning_rate": 5.666772737612376e-06,
+      "loss": 0.8958,
+      "step": 17966
+    },
+    {
+      "epoch": 0.6534167363712405,
+      "grad_norm": 0.0,
+      "learning_rate": 5.665711233006661e-06,
+      "loss": 0.8288,
+      "step": 17967
+    },
+    {
+      "epoch": 0.6534531039749791,
+      "grad_norm": 0.0,
+      "learning_rate": 5.664649788533052e-06,
+      "loss": 0.9478,
+      "step": 17968
+    },
+    {
+      "epoch": 0.6534894715787177,
+      "grad_norm": 0.0,
+      "learning_rate": 5.663588404206274e-06,
+      "loss": 0.9552,
+      "step": 17969
+    },
+    {
+      "epoch": 0.6535258391824562,
+      "grad_norm": 0.0,
+      "learning_rate": 5.662527080041063e-06,
+      "loss": 0.9391,
+      "step": 17970
+    },
+    {
+      "epoch": 0.6535622067861948,
+      "grad_norm": 0.0,
+      "learning_rate": 5.661465816052124e-06,
+      "loss": 0.981,
+      "step": 17971
+    },
+    {
+      "epoch": 0.6535985743899334,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6604046122541965e-06,
+      "loss": 0.8994,
+      "step": 17972
+    },
+    {
+      "epoch": 0.653634941993672,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6593434686619985e-06,
+      "loss": 1.0006,
+      "step": 17973
+    },
+    {
+      "epoch": 0.6536713095974106,
+      "grad_norm": 0.0,
+      "learning_rate": 5.65828238529025e-06,
+      "loss": 0.97,
+      "step": 17974
+    },
+    {
+      "epoch": 0.6537076772011492,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6572213621536755e-06,
+      "loss": 0.9528,
+      "step": 17975
+    },
+    {
+      "epoch": 0.6537440448048878,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6561603992669875e-06,
+      "loss": 0.9388,
+      "step": 17976
+    },
+    {
+      "epoch": 0.6537804124086264,
+      "grad_norm": 0.0,
+      "learning_rate": 5.655099496644918e-06,
+      "loss": 0.9418,
+      "step": 17977
+    },
+    {
+      "epoch": 0.653816780012365,
+      "grad_norm": 0.0,
+      "learning_rate": 5.654038654302172e-06,
+      "loss": 0.8974,
+      "step": 17978
+    },
+    {
+      "epoch": 0.6538531476161036,
+      "grad_norm": 0.0,
+      "learning_rate": 5.652977872253476e-06,
+      "loss": 0.8537,
+      "step": 17979
+    },
+    {
+      "epoch": 0.6538895152198422,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6519171505135465e-06,
+      "loss": 0.9039,
+      "step": 17980
+    },
+    {
+      "epoch": 0.6539258828235808,
+      "grad_norm": 0.0,
+      "learning_rate": 5.650856489097097e-06,
+      "loss": 0.9347,
+      "step": 17981
+    },
+    {
+      "epoch": 0.6539622504273194,
+      "grad_norm": 0.0,
+      "learning_rate": 5.649795888018836e-06,
+      "loss": 0.8495,
+      "step": 17982
+    },
+    {
+      "epoch": 0.653998618031058,
+      "grad_norm": 0.0,
+      "learning_rate": 5.648735347293497e-06,
+      "loss": 0.9878,
+      "step": 17983
+    },
+    {
+      "epoch": 0.6540349856347966,
+      "grad_norm": 0.0,
+      "learning_rate": 5.647674866935771e-06,
+      "loss": 0.9854,
+      "step": 17984
+    },
+    {
+      "epoch": 0.6540713532385352,
+      "grad_norm": 0.0,
+      "learning_rate": 5.646614446960386e-06,
+      "loss": 0.8314,
+      "step": 17985
+    },
+    {
+      "epoch": 0.6541077208422738,
+      "grad_norm": 0.0,
+      "learning_rate": 5.645554087382048e-06,
+      "loss": 0.8811,
+      "step": 17986
+    },
+    {
+      "epoch": 0.6541440884460122,
+      "grad_norm": 0.0,
+      "learning_rate": 5.644493788215471e-06,
+      "loss": 1.1765,
+      "step": 17987
+    },
+    {
+      "epoch": 0.6541804560497508,
+      "grad_norm": 0.0,
+      "learning_rate": 5.643433549475357e-06,
+      "loss": 0.9301,
+      "step": 17988
+    },
+    {
+      "epoch": 0.6542168236534894,
+      "grad_norm": 0.0,
+      "learning_rate": 5.64237337117643e-06,
+      "loss": 1.0803,
+      "step": 17989
+    },
+    {
+      "epoch": 0.654253191257228,
+      "grad_norm": 0.0,
+      "learning_rate": 5.641313253333381e-06,
+      "loss": 0.9693,
+      "step": 17990
+    },
+    {
+      "epoch": 0.6542895588609666,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6402531959609315e-06,
+      "loss": 0.9635,
+      "step": 17991
+    },
+    {
+      "epoch": 0.6543259264647052,
+      "grad_norm": 0.0,
+      "learning_rate": 5.639193199073782e-06,
+      "loss": 0.9848,
+      "step": 17992
+    },
+    {
+      "epoch": 0.6543622940684438,
+      "grad_norm": 0.0,
+      "learning_rate": 5.638133262686641e-06,
+      "loss": 0.8973,
+      "step": 17993
+    },
+    {
+      "epoch": 0.6543986616721824,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6370733868142126e-06,
+      "loss": 0.8892,
+      "step": 17994
+    },
+    {
+      "epoch": 0.654435029275921,
+      "grad_norm": 0.0,
+      "learning_rate": 5.636013571471196e-06,
+      "loss": 1.0612,
+      "step": 17995
+    },
+    {
+      "epoch": 0.6544713968796596,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6349538166723074e-06,
+      "loss": 0.9541,
+      "step": 17996
+    },
+    {
+      "epoch": 0.6545077644833982,
+      "grad_norm": 0.0,
+      "learning_rate": 5.633894122432233e-06,
+      "loss": 0.9336,
+      "step": 17997
+    },
+    {
+      "epoch": 0.6545441320871368,
+      "grad_norm": 0.0,
+      "learning_rate": 5.632834488765688e-06,
+      "loss": 1.0153,
+      "step": 17998
+    },
+    {
+      "epoch": 0.6545804996908754,
+      "grad_norm": 0.0,
+      "learning_rate": 5.631774915687367e-06,
+      "loss": 1.0354,
+      "step": 17999
+    },
+    {
+      "epoch": 0.654616867294614,
+      "grad_norm": 0.0,
+      "learning_rate": 5.630715403211973e-06,
+      "loss": 0.9814,
+      "step": 18000
+    },
+    {
+      "epoch": 0.6546532348983526,
+      "grad_norm": 0.0,
+      "learning_rate": 5.629655951354198e-06,
+      "loss": 0.9892,
+      "step": 18001
+    },
+    {
+      "epoch": 0.6546896025020912,
+      "grad_norm": 0.0,
+      "learning_rate": 5.628596560128755e-06,
+      "loss": 0.8809,
+      "step": 18002
+    },
+    {
+      "epoch": 0.6547259701058298,
+      "grad_norm": 0.0,
+      "learning_rate": 5.627537229550324e-06,
+      "loss": 1.0011,
+      "step": 18003
+    },
+    {
+      "epoch": 0.6547623377095683,
+      "grad_norm": 0.0,
+      "learning_rate": 5.626477959633614e-06,
+      "loss": 0.914,
+      "step": 18004
+    },
+    {
+      "epoch": 0.6547987053133069,
+      "grad_norm": 0.0,
+      "learning_rate": 5.625418750393318e-06,
+      "loss": 0.9625,
+      "step": 18005
+    },
+    {
+      "epoch": 0.6548350729170455,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6243596018441295e-06,
+      "loss": 1.0866,
+      "step": 18006
+    },
+    {
+      "epoch": 0.654871440520784,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6233005140007394e-06,
+      "loss": 0.9715,
+      "step": 18007
+    },
+    {
+      "epoch": 0.6549078081245227,
+      "grad_norm": 0.0,
+      "learning_rate": 5.622241486877849e-06,
+      "loss": 0.8763,
+      "step": 18008
+    },
+    {
+      "epoch": 0.6549441757282612,
+      "grad_norm": 0.0,
+      "learning_rate": 5.621182520490147e-06,
+      "loss": 0.993,
+      "step": 18009
+    },
+    {
+      "epoch": 0.6549805433319998,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6201236148523245e-06,
+      "loss": 0.8691,
+      "step": 18010
+    },
+    {
+      "epoch": 0.6550169109357384,
+      "grad_norm": 0.0,
+      "learning_rate": 5.619064769979072e-06,
+      "loss": 0.9048,
+      "step": 18011
+    },
+    {
+      "epoch": 0.655053278539477,
+      "grad_norm": 0.0,
+      "learning_rate": 5.61800598588508e-06,
+      "loss": 0.9058,
+      "step": 18012
+    },
+    {
+      "epoch": 0.6550896461432156,
+      "grad_norm": 0.0,
+      "learning_rate": 5.61694726258504e-06,
+      "loss": 0.9474,
+      "step": 18013
+    },
+    {
+      "epoch": 0.6551260137469542,
+      "grad_norm": 0.0,
+      "learning_rate": 5.615888600093631e-06,
+      "loss": 0.9743,
+      "step": 18014
+    },
+    {
+      "epoch": 0.6551623813506928,
+      "grad_norm": 0.0,
+      "learning_rate": 5.614829998425558e-06,
+      "loss": 0.9148,
+      "step": 18015
+    },
+    {
+      "epoch": 0.6551987489544314,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6137714575954875e-06,
+      "loss": 0.8828,
+      "step": 18016
+    },
+    {
+      "epoch": 0.65523511655817,
+      "grad_norm": 0.0,
+      "learning_rate": 5.612712977618118e-06,
+      "loss": 1.0354,
+      "step": 18017
+    },
+    {
+      "epoch": 0.6552714841619086,
+      "grad_norm": 0.0,
+      "learning_rate": 5.611654558508133e-06,
+      "loss": 0.9828,
+      "step": 18018
+    },
+    {
+      "epoch": 0.6553078517656472,
+      "grad_norm": 0.0,
+      "learning_rate": 5.610596200280214e-06,
+      "loss": 0.9088,
+      "step": 18019
+    },
+    {
+      "epoch": 0.6553442193693858,
+      "grad_norm": 0.0,
+      "learning_rate": 5.60953790294904e-06,
+      "loss": 0.972,
+      "step": 18020
+    },
+    {
+      "epoch": 0.6553805869731243,
+      "grad_norm": 0.0,
+      "learning_rate": 5.608479666529304e-06,
+      "loss": 1.0402,
+      "step": 18021
+    },
+    {
+      "epoch": 0.6554169545768629,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6074214910356804e-06,
+      "loss": 0.9279,
+      "step": 18022
+    },
+    {
+      "epoch": 0.6554533221806015,
+      "grad_norm": 0.0,
+      "learning_rate": 5.606363376482853e-06,
+      "loss": 0.8931,
+      "step": 18023
+    },
+    {
+      "epoch": 0.6554896897843401,
+      "grad_norm": 0.0,
+      "learning_rate": 5.605305322885499e-06,
+      "loss": 0.8273,
+      "step": 18024
+    },
+    {
+      "epoch": 0.6555260573880787,
+      "grad_norm": 0.0,
+      "learning_rate": 5.604247330258299e-06,
+      "loss": 0.9832,
+      "step": 18025
+    },
+    {
+      "epoch": 0.6555624249918173,
+      "grad_norm": 0.0,
+      "learning_rate": 5.603189398615926e-06,
+      "loss": 0.9393,
+      "step": 18026
+    },
+    {
+      "epoch": 0.6555987925955559,
+      "grad_norm": 0.0,
+      "learning_rate": 5.602131527973066e-06,
+      "loss": 0.9242,
+      "step": 18027
+    },
+    {
+      "epoch": 0.6556351601992945,
+      "grad_norm": 0.0,
+      "learning_rate": 5.601073718344391e-06,
+      "loss": 0.8535,
+      "step": 18028
+    },
+    {
+      "epoch": 0.6556715278030331,
+      "grad_norm": 0.0,
+      "learning_rate": 5.600015969744576e-06,
+      "loss": 0.7834,
+      "step": 18029
+    },
+    {
+      "epoch": 0.6557078954067717,
+      "grad_norm": 0.0,
+      "learning_rate": 5.598958282188295e-06,
+      "loss": 0.8923,
+      "step": 18030
+    },
+    {
+      "epoch": 0.6557442630105103,
+      "grad_norm": 0.0,
+      "learning_rate": 5.597900655690231e-06,
+      "loss": 1.0803,
+      "step": 18031
+    },
+    {
+      "epoch": 0.6557806306142488,
+      "grad_norm": 0.0,
+      "learning_rate": 5.596843090265045e-06,
+      "loss": 0.8993,
+      "step": 18032
+    },
+    {
+      "epoch": 0.6558169982179874,
+      "grad_norm": 0.0,
+      "learning_rate": 5.595785585927409e-06,
+      "loss": 0.8237,
+      "step": 18033
+    },
+    {
+      "epoch": 0.655853365821726,
+      "grad_norm": 0.0,
+      "learning_rate": 5.594728142692007e-06,
+      "loss": 0.9436,
+      "step": 18034
+    },
+    {
+      "epoch": 0.6558897334254646,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5936707605734965e-06,
+      "loss": 0.9885,
+      "step": 18035
+    },
+    {
+      "epoch": 0.6559261010292032,
+      "grad_norm": 0.0,
+      "learning_rate": 5.592613439586554e-06,
+      "loss": 0.9947,
+      "step": 18036
+    },
+    {
+      "epoch": 0.6559624686329418,
+      "grad_norm": 0.0,
+      "learning_rate": 5.591556179745848e-06,
+      "loss": 0.9354,
+      "step": 18037
+    },
+    {
+      "epoch": 0.6559988362366803,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5904989810660445e-06,
+      "loss": 0.9204,
+      "step": 18038
+    },
+    {
+      "epoch": 0.6560352038404189,
+      "grad_norm": 0.0,
+      "learning_rate": 5.589441843561808e-06,
+      "loss": 1.0329,
+      "step": 18039
+    },
+    {
+      "epoch": 0.6560715714441575,
+      "grad_norm": 0.0,
+      "learning_rate": 5.588384767247812e-06,
+      "loss": 1.0072,
+      "step": 18040
+    },
+    {
+      "epoch": 0.6561079390478961,
+      "grad_norm": 0.0,
+      "learning_rate": 5.587327752138718e-06,
+      "loss": 0.9449,
+      "step": 18041
+    },
+    {
+      "epoch": 0.6561443066516347,
+      "grad_norm": 0.0,
+      "learning_rate": 5.586270798249191e-06,
+      "loss": 0.8369,
+      "step": 18042
+    },
+    {
+      "epoch": 0.6561806742553733,
+      "grad_norm": 0.0,
+      "learning_rate": 5.585213905593889e-06,
+      "loss": 0.9686,
+      "step": 18043
+    },
+    {
+      "epoch": 0.6562170418591119,
+      "grad_norm": 0.0,
+      "learning_rate": 5.58415707418749e-06,
+      "loss": 0.9119,
+      "step": 18044
+    },
+    {
+      "epoch": 0.6562534094628505,
+      "grad_norm": 0.0,
+      "learning_rate": 5.583100304044637e-06,
+      "loss": 0.9177,
+      "step": 18045
+    },
+    {
+      "epoch": 0.6562897770665891,
+      "grad_norm": 0.0,
+      "learning_rate": 5.582043595180005e-06,
+      "loss": 1.0328,
+      "step": 18046
+    },
+    {
+      "epoch": 0.6563261446703277,
+      "grad_norm": 0.0,
+      "learning_rate": 5.58098694760825e-06,
+      "loss": 0.9246,
+      "step": 18047
+    },
+    {
+      "epoch": 0.6563625122740663,
+      "grad_norm": 0.0,
+      "learning_rate": 5.579930361344032e-06,
+      "loss": 1.0581,
+      "step": 18048
+    },
+    {
+      "epoch": 0.6563988798778049,
+      "grad_norm": 0.0,
+      "learning_rate": 5.578873836402003e-06,
+      "loss": 1.0211,
+      "step": 18049
+    },
+    {
+      "epoch": 0.6564352474815435,
+      "grad_norm": 0.0,
+      "learning_rate": 5.577817372796836e-06,
+      "loss": 0.9312,
+      "step": 18050
+    },
+    {
+      "epoch": 0.6564716150852821,
+      "grad_norm": 0.0,
+      "learning_rate": 5.576760970543176e-06,
+      "loss": 0.9408,
+      "step": 18051
+    },
+    {
+      "epoch": 0.6565079826890207,
+      "grad_norm": 0.0,
+      "learning_rate": 5.575704629655676e-06,
+      "loss": 1.0319,
+      "step": 18052
+    },
+    {
+      "epoch": 0.6565443502927593,
+      "grad_norm": 0.0,
+      "learning_rate": 5.574648350149001e-06,
+      "loss": 1.0061,
+      "step": 18053
+    },
+    {
+      "epoch": 0.6565807178964977,
+      "grad_norm": 0.0,
+      "learning_rate": 5.573592132037803e-06,
+      "loss": 0.9282,
+      "step": 18054
+    },
+    {
+      "epoch": 0.6566170855002363,
+      "grad_norm": 0.0,
+      "learning_rate": 5.572535975336732e-06,
+      "loss": 0.9544,
+      "step": 18055
+    },
+    {
+      "epoch": 0.6566534531039749,
+      "grad_norm": 0.0,
+      "learning_rate": 5.571479880060444e-06,
+      "loss": 0.9701,
+      "step": 18056
+    },
+    {
+      "epoch": 0.6566898207077135,
+      "grad_norm": 0.0,
+      "learning_rate": 5.570423846223589e-06,
+      "loss": 1.0724,
+      "step": 18057
+    },
+    {
+      "epoch": 0.6567261883114521,
+      "grad_norm": 0.0,
+      "learning_rate": 5.569367873840814e-06,
+      "loss": 1.0269,
+      "step": 18058
+    },
+    {
+      "epoch": 0.6567625559151907,
+      "grad_norm": 0.0,
+      "learning_rate": 5.568311962926778e-06,
+      "loss": 0.9453,
+      "step": 18059
+    },
+    {
+      "epoch": 0.6567989235189293,
+      "grad_norm": 0.0,
+      "learning_rate": 5.567256113496125e-06,
+      "loss": 0.8857,
+      "step": 18060
+    },
+    {
+      "epoch": 0.6568352911226679,
+      "grad_norm": 0.0,
+      "learning_rate": 5.566200325563505e-06,
+      "loss": 0.9453,
+      "step": 18061
+    },
+    {
+      "epoch": 0.6568716587264065,
+      "grad_norm": 0.0,
+      "learning_rate": 5.565144599143562e-06,
+      "loss": 0.9632,
+      "step": 18062
+    },
+    {
+      "epoch": 0.6569080263301451,
+      "grad_norm": 0.0,
+      "learning_rate": 5.564088934250951e-06,
+      "loss": 0.7656,
+      "step": 18063
+    },
+    {
+      "epoch": 0.6569443939338837,
+      "grad_norm": 0.0,
+      "learning_rate": 5.563033330900305e-06,
+      "loss": 1.0475,
+      "step": 18064
+    },
+    {
+      "epoch": 0.6569807615376223,
+      "grad_norm": 0.0,
+      "learning_rate": 5.561977789106281e-06,
+      "loss": 0.9036,
+      "step": 18065
+    },
+    {
+      "epoch": 0.6570171291413609,
+      "grad_norm": 0.0,
+      "learning_rate": 5.560922308883518e-06,
+      "loss": 0.8937,
+      "step": 18066
+    },
+    {
+      "epoch": 0.6570534967450995,
+      "grad_norm": 0.0,
+      "learning_rate": 5.559866890246659e-06,
+      "loss": 0.9387,
+      "step": 18067
+    },
+    {
+      "epoch": 0.6570898643488381,
+      "grad_norm": 0.0,
+      "learning_rate": 5.558811533210344e-06,
+      "loss": 1.0233,
+      "step": 18068
+    },
+    {
+      "epoch": 0.6571262319525767,
+      "grad_norm": 0.0,
+      "learning_rate": 5.557756237789225e-06,
+      "loss": 0.9327,
+      "step": 18069
+    },
+    {
+      "epoch": 0.6571625995563153,
+      "grad_norm": 0.0,
+      "learning_rate": 5.556701003997933e-06,
+      "loss": 1.0134,
+      "step": 18070
+    },
+    {
+      "epoch": 0.6571989671600538,
+      "grad_norm": 0.0,
+      "learning_rate": 5.555645831851105e-06,
+      "loss": 0.8937,
+      "step": 18071
+    },
+    {
+      "epoch": 0.6572353347637924,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5545907213633905e-06,
+      "loss": 1.0234,
+      "step": 18072
+    },
+    {
+      "epoch": 0.657271702367531,
+      "grad_norm": 0.0,
+      "learning_rate": 5.553535672549421e-06,
+      "loss": 0.9057,
+      "step": 18073
+    },
+    {
+      "epoch": 0.6573080699712696,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5524806854238366e-06,
+      "loss": 0.9578,
+      "step": 18074
+    },
+    {
+      "epoch": 0.6573444375750082,
+      "grad_norm": 0.0,
+      "learning_rate": 5.551425760001267e-06,
+      "loss": 0.9952,
+      "step": 18075
+    },
+    {
+      "epoch": 0.6573808051787468,
+      "grad_norm": 0.0,
+      "learning_rate": 5.550370896296365e-06,
+      "loss": 1.0123,
+      "step": 18076
+    },
+    {
+      "epoch": 0.6574171727824853,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5493160943237425e-06,
+      "loss": 0.9875,
+      "step": 18077
+    },
+    {
+      "epoch": 0.657453540386224,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5482613540980475e-06,
+      "loss": 0.9509,
+      "step": 18078
+    },
+    {
+      "epoch": 0.6574899079899625,
+      "grad_norm": 0.0,
+      "learning_rate": 5.547206675633912e-06,
+      "loss": 0.8408,
+      "step": 18079
+    },
+    {
+      "epoch": 0.6575262755937011,
+      "grad_norm": 0.0,
+      "learning_rate": 5.546152058945966e-06,
+      "loss": 0.9269,
+      "step": 18080
+    },
+    {
+      "epoch": 0.6575626431974397,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5450975040488364e-06,
+      "loss": 0.9462,
+      "step": 18081
+    },
+    {
+      "epoch": 0.6575990108011783,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5440430109571664e-06,
+      "loss": 0.8802,
+      "step": 18082
+    },
+    {
+      "epoch": 0.6576353784049169,
+      "grad_norm": 0.0,
+      "learning_rate": 5.54298857968557e-06,
+      "loss": 0.9736,
+      "step": 18083
+    },
+    {
+      "epoch": 0.6576717460086555,
+      "grad_norm": 0.0,
+      "learning_rate": 5.541934210248688e-06,
+      "loss": 0.893,
+      "step": 18084
+    },
+    {
+      "epoch": 0.6577081136123941,
+      "grad_norm": 0.0,
+      "learning_rate": 5.540879902661141e-06,
+      "loss": 0.9239,
+      "step": 18085
+    },
+    {
+      "epoch": 0.6577444812161327,
+      "grad_norm": 0.0,
+      "learning_rate": 5.539825656937561e-06,
+      "loss": 0.8633,
+      "step": 18086
+    },
+    {
+      "epoch": 0.6577808488198713,
+      "grad_norm": 0.0,
+      "learning_rate": 5.538771473092567e-06,
+      "loss": 1.0238,
+      "step": 18087
+    },
+    {
+      "epoch": 0.6578172164236098,
+      "grad_norm": 0.0,
+      "learning_rate": 5.537717351140797e-06,
+      "loss": 1.0187,
+      "step": 18088
+    },
+    {
+      "epoch": 0.6578535840273484,
+      "grad_norm": 0.0,
+      "learning_rate": 5.536663291096864e-06,
+      "loss": 1.0136,
+      "step": 18089
+    },
+    {
+      "epoch": 0.657889951631087,
+      "grad_norm": 0.0,
+      "learning_rate": 5.535609292975392e-06,
+      "loss": 0.8937,
+      "step": 18090
+    },
+    {
+      "epoch": 0.6579263192348256,
+      "grad_norm": 0.0,
+      "learning_rate": 5.534555356791009e-06,
+      "loss": 1.0679,
+      "step": 18091
+    },
+    {
+      "epoch": 0.6579626868385642,
+      "grad_norm": 0.0,
+      "learning_rate": 5.533501482558337e-06,
+      "loss": 0.9524,
+      "step": 18092
+    },
+    {
+      "epoch": 0.6579990544423028,
+      "grad_norm": 0.0,
+      "learning_rate": 5.532447670291993e-06,
+      "loss": 0.9436,
+      "step": 18093
+    },
+    {
+      "epoch": 0.6580354220460414,
+      "grad_norm": 0.0,
+      "learning_rate": 5.531393920006596e-06,
+      "loss": 0.9399,
+      "step": 18094
+    },
+    {
+      "epoch": 0.65807178964978,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5303402317167755e-06,
+      "loss": 0.7769,
+      "step": 18095
+    },
+    {
+      "epoch": 0.6581081572535186,
+      "grad_norm": 0.0,
+      "learning_rate": 5.529286605437134e-06,
+      "loss": 0.9025,
+      "step": 18096
+    },
+    {
+      "epoch": 0.6581445248572572,
+      "grad_norm": 0.0,
+      "learning_rate": 5.528233041182302e-06,
+      "loss": 0.9516,
+      "step": 18097
+    },
+    {
+      "epoch": 0.6581808924609958,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5271795389668935e-06,
+      "loss": 1.0715,
+      "step": 18098
+    },
+    {
+      "epoch": 0.6582172600647344,
+      "grad_norm": 0.0,
+      "learning_rate": 5.526126098805522e-06,
+      "loss": 1.0863,
+      "step": 18099
+    },
+    {
+      "epoch": 0.658253627668473,
+      "grad_norm": 0.0,
+      "learning_rate": 5.525072720712799e-06,
+      "loss": 0.9338,
+      "step": 18100
+    },
+    {
+      "epoch": 0.6582899952722115,
+      "grad_norm": 0.0,
+      "learning_rate": 5.524019404703351e-06,
+      "loss": 0.9218,
+      "step": 18101
+    },
+    {
+      "epoch": 0.6583263628759501,
+      "grad_norm": 0.0,
+      "learning_rate": 5.522966150791776e-06,
+      "loss": 0.9662,
+      "step": 18102
+    },
+    {
+      "epoch": 0.6583627304796887,
+      "grad_norm": 0.0,
+      "learning_rate": 5.521912958992697e-06,
+      "loss": 0.9329,
+      "step": 18103
+    },
+    {
+      "epoch": 0.6583990980834273,
+      "grad_norm": 0.0,
+      "learning_rate": 5.520859829320721e-06,
+      "loss": 0.9887,
+      "step": 18104
+    },
+    {
+      "epoch": 0.6584354656871658,
+      "grad_norm": 0.0,
+      "learning_rate": 5.519806761790462e-06,
+      "loss": 0.9047,
+      "step": 18105
+    },
+    {
+      "epoch": 0.6584718332909044,
+      "grad_norm": 0.0,
+      "learning_rate": 5.518753756416522e-06,
+      "loss": 1.0058,
+      "step": 18106
+    },
+    {
+      "epoch": 0.658508200894643,
+      "grad_norm": 0.0,
+      "learning_rate": 5.51770081321352e-06,
+      "loss": 1.0422,
+      "step": 18107
+    },
+    {
+      "epoch": 0.6585445684983816,
+      "grad_norm": 0.0,
+      "learning_rate": 5.51664793219606e-06,
+      "loss": 0.8258,
+      "step": 18108
+    },
+    {
+      "epoch": 0.6585809361021202,
+      "grad_norm": 0.0,
+      "learning_rate": 5.515595113378749e-06,
+      "loss": 0.939,
+      "step": 18109
+    },
+    {
+      "epoch": 0.6586173037058588,
+      "grad_norm": 0.0,
+      "learning_rate": 5.514542356776192e-06,
+      "loss": 0.8436,
+      "step": 18110
+    },
+    {
+      "epoch": 0.6586536713095974,
+      "grad_norm": 0.0,
+      "learning_rate": 5.513489662402998e-06,
+      "loss": 0.9167,
+      "step": 18111
+    },
+    {
+      "epoch": 0.658690038913336,
+      "grad_norm": 0.0,
+      "learning_rate": 5.512437030273768e-06,
+      "loss": 0.9409,
+      "step": 18112
+    },
+    {
+      "epoch": 0.6587264065170746,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5113844604031015e-06,
+      "loss": 1.0542,
+      "step": 18113
+    },
+    {
+      "epoch": 0.6587627741208132,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5103319528056176e-06,
+      "loss": 0.8534,
+      "step": 18114
+    },
+    {
+      "epoch": 0.6587991417245518,
+      "grad_norm": 0.0,
+      "learning_rate": 5.509279507495899e-06,
+      "loss": 0.9469,
+      "step": 18115
+    },
+    {
+      "epoch": 0.6588355093282904,
+      "grad_norm": 0.0,
+      "learning_rate": 5.50822712448856e-06,
+      "loss": 0.8954,
+      "step": 18116
+    },
+    {
+      "epoch": 0.658871876932029,
+      "grad_norm": 0.0,
+      "learning_rate": 5.507174803798196e-06,
+      "loss": 0.906,
+      "step": 18117
+    },
+    {
+      "epoch": 0.6589082445357676,
+      "grad_norm": 0.0,
+      "learning_rate": 5.506122545439407e-06,
+      "loss": 0.9658,
+      "step": 18118
+    },
+    {
+      "epoch": 0.6589446121395062,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5050703494267885e-06,
+      "loss": 0.9789,
+      "step": 18119
+    },
+    {
+      "epoch": 0.6589809797432448,
+      "grad_norm": 0.0,
+      "learning_rate": 5.504018215774945e-06,
+      "loss": 0.8692,
+      "step": 18120
+    },
+    {
+      "epoch": 0.6590173473469834,
+      "grad_norm": 0.0,
+      "learning_rate": 5.502966144498471e-06,
+      "loss": 0.9798,
+      "step": 18121
+    },
+    {
+      "epoch": 0.6590537149507218,
+      "grad_norm": 0.0,
+      "learning_rate": 5.50191413561196e-06,
+      "loss": 0.9419,
+      "step": 18122
+    },
+    {
+      "epoch": 0.6590900825544604,
+      "grad_norm": 0.0,
+      "learning_rate": 5.500862189130009e-06,
+      "loss": 1.055,
+      "step": 18123
+    },
+    {
+      "epoch": 0.659126450158199,
+      "grad_norm": 0.0,
+      "learning_rate": 5.49981030506721e-06,
+      "loss": 0.9929,
+      "step": 18124
+    },
+    {
+      "epoch": 0.6591628177619376,
+      "grad_norm": 0.0,
+      "learning_rate": 5.498758483438156e-06,
+      "loss": 0.9722,
+      "step": 18125
+    },
+    {
+      "epoch": 0.6591991853656762,
+      "grad_norm": 0.0,
+      "learning_rate": 5.497706724257446e-06,
+      "loss": 0.944,
+      "step": 18126
+    },
+    {
+      "epoch": 0.6592355529694148,
+      "grad_norm": 0.0,
+      "learning_rate": 5.496655027539666e-06,
+      "loss": 0.903,
+      "step": 18127
+    },
+    {
+      "epoch": 0.6592719205731534,
+      "grad_norm": 0.0,
+      "learning_rate": 5.495603393299408e-06,
+      "loss": 0.9212,
+      "step": 18128
+    },
+    {
+      "epoch": 0.659308288176892,
+      "grad_norm": 0.0,
+      "learning_rate": 5.494551821551263e-06,
+      "loss": 0.925,
+      "step": 18129
+    },
+    {
+      "epoch": 0.6593446557806306,
+      "grad_norm": 0.0,
+      "learning_rate": 5.493500312309819e-06,
+      "loss": 0.8194,
+      "step": 18130
+    },
+    {
+      "epoch": 0.6593810233843692,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4924488655896635e-06,
+      "loss": 1.0193,
+      "step": 18131
+    },
+    {
+      "epoch": 0.6594173909881078,
+      "grad_norm": 0.0,
+      "learning_rate": 5.491397481405381e-06,
+      "loss": 0.9983,
+      "step": 18132
+    },
+    {
+      "epoch": 0.6594537585918464,
+      "grad_norm": 0.0,
+      "learning_rate": 5.490346159771569e-06,
+      "loss": 0.974,
+      "step": 18133
+    },
+    {
+      "epoch": 0.659490126195585,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4892949007027975e-06,
+      "loss": 0.7489,
+      "step": 18134
+    },
+    {
+      "epoch": 0.6595264937993236,
+      "grad_norm": 0.0,
+      "learning_rate": 5.488243704213665e-06,
+      "loss": 0.9914,
+      "step": 18135
+    },
+    {
+      "epoch": 0.6595628614030622,
+      "grad_norm": 0.0,
+      "learning_rate": 5.487192570318749e-06,
+      "loss": 0.9427,
+      "step": 18136
+    },
+    {
+      "epoch": 0.6595992290068008,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4861414990326325e-06,
+      "loss": 0.9248,
+      "step": 18137
+    },
+    {
+      "epoch": 0.6596355966105393,
+      "grad_norm": 0.0,
+      "learning_rate": 5.485090490369895e-06,
+      "loss": 0.9844,
+      "step": 18138
+    },
+    {
+      "epoch": 0.6596719642142779,
+      "grad_norm": 0.0,
+      "learning_rate": 5.484039544345126e-06,
+      "loss": 0.9925,
+      "step": 18139
+    },
+    {
+      "epoch": 0.6597083318180165,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4829886609728985e-06,
+      "loss": 0.9647,
+      "step": 18140
+    },
+    {
+      "epoch": 0.6597446994217551,
+      "grad_norm": 0.0,
+      "learning_rate": 5.481937840267797e-06,
+      "loss": 1.0249,
+      "step": 18141
+    },
+    {
+      "epoch": 0.6597810670254937,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4808870822443926e-06,
+      "loss": 0.9822,
+      "step": 18142
+    },
+    {
+      "epoch": 0.6598174346292323,
+      "grad_norm": 0.0,
+      "learning_rate": 5.479836386917277e-06,
+      "loss": 0.8777,
+      "step": 18143
+    },
+    {
+      "epoch": 0.6598538022329709,
+      "grad_norm": 0.0,
+      "learning_rate": 5.478785754301008e-06,
+      "loss": 0.9693,
+      "step": 18144
+    },
+    {
+      "epoch": 0.6598901698367095,
+      "grad_norm": 0.0,
+      "learning_rate": 5.477735184410178e-06,
+      "loss": 1.0388,
+      "step": 18145
+    },
+    {
+      "epoch": 0.659926537440448,
+      "grad_norm": 0.0,
+      "learning_rate": 5.476684677259356e-06,
+      "loss": 0.8641,
+      "step": 18146
+    },
+    {
+      "epoch": 0.6599629050441866,
+      "grad_norm": 0.0,
+      "learning_rate": 5.475634232863117e-06,
+      "loss": 0.8736,
+      "step": 18147
+    },
+    {
+      "epoch": 0.6599992726479252,
+      "grad_norm": 0.0,
+      "learning_rate": 5.474583851236031e-06,
+      "loss": 1.0359,
+      "step": 18148
+    },
+    {
+      "epoch": 0.6600356402516638,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4735335323926754e-06,
+      "loss": 0.8316,
+      "step": 18149
+    },
+    {
+      "epoch": 0.6600720078554024,
+      "grad_norm": 0.0,
+      "learning_rate": 5.47248327634762e-06,
+      "loss": 1.0291,
+      "step": 18150
+    },
+    {
+      "epoch": 0.660108375459141,
+      "grad_norm": 0.0,
+      "learning_rate": 5.471433083115431e-06,
+      "loss": 0.8461,
+      "step": 18151
+    },
+    {
+      "epoch": 0.6601447430628796,
+      "grad_norm": 0.0,
+      "learning_rate": 5.470382952710687e-06,
+      "loss": 0.9809,
+      "step": 18152
+    },
+    {
+      "epoch": 0.6601811106666182,
+      "grad_norm": 0.0,
+      "learning_rate": 5.469332885147951e-06,
+      "loss": 1.0438,
+      "step": 18153
+    },
+    {
+      "epoch": 0.6602174782703568,
+      "grad_norm": 0.0,
+      "learning_rate": 5.468282880441795e-06,
+      "loss": 0.9535,
+      "step": 18154
+    },
+    {
+      "epoch": 0.6602538458740953,
+      "grad_norm": 0.0,
+      "learning_rate": 5.467232938606784e-06,
+      "loss": 0.845,
+      "step": 18155
+    },
+    {
+      "epoch": 0.6602902134778339,
+      "grad_norm": 0.0,
+      "learning_rate": 5.466183059657484e-06,
+      "loss": 1.0594,
+      "step": 18156
+    },
+    {
+      "epoch": 0.6603265810815725,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4651332436084585e-06,
+      "loss": 0.9556,
+      "step": 18157
+    },
+    {
+      "epoch": 0.6603629486853111,
+      "grad_norm": 0.0,
+      "learning_rate": 5.464083490474279e-06,
+      "loss": 0.9516,
+      "step": 18158
+    },
+    {
+      "epoch": 0.6603993162890497,
+      "grad_norm": 0.0,
+      "learning_rate": 5.463033800269506e-06,
+      "loss": 1.0007,
+      "step": 18159
+    },
+    {
+      "epoch": 0.6604356838927883,
+      "grad_norm": 0.0,
+      "learning_rate": 5.461984173008701e-06,
+      "loss": 0.9548,
+      "step": 18160
+    },
+    {
+      "epoch": 0.6604720514965269,
+      "grad_norm": 0.0,
+      "learning_rate": 5.460934608706423e-06,
+      "loss": 1.0072,
+      "step": 18161
+    },
+    {
+      "epoch": 0.6605084191002655,
+      "grad_norm": 0.0,
+      "learning_rate": 5.459885107377245e-06,
+      "loss": 0.9521,
+      "step": 18162
+    },
+    {
+      "epoch": 0.6605447867040041,
+      "grad_norm": 0.0,
+      "learning_rate": 5.458835669035712e-06,
+      "loss": 0.9651,
+      "step": 18163
+    },
+    {
+      "epoch": 0.6605811543077427,
+      "grad_norm": 0.0,
+      "learning_rate": 5.457786293696395e-06,
+      "loss": 0.9032,
+      "step": 18164
+    },
+    {
+      "epoch": 0.6606175219114813,
+      "grad_norm": 0.0,
+      "learning_rate": 5.456736981373848e-06,
+      "loss": 0.946,
+      "step": 18165
+    },
+    {
+      "epoch": 0.6606538895152199,
+      "grad_norm": 0.0,
+      "learning_rate": 5.455687732082631e-06,
+      "loss": 0.8868,
+      "step": 18166
+    },
+    {
+      "epoch": 0.6606902571189585,
+      "grad_norm": 0.0,
+      "learning_rate": 5.454638545837296e-06,
+      "loss": 0.8785,
+      "step": 18167
+    },
+    {
+      "epoch": 0.660726624722697,
+      "grad_norm": 0.0,
+      "learning_rate": 5.453589422652406e-06,
+      "loss": 0.9524,
+      "step": 18168
+    },
+    {
+      "epoch": 0.6607629923264356,
+      "grad_norm": 0.0,
+      "learning_rate": 5.452540362542509e-06,
+      "loss": 0.9466,
+      "step": 18169
+    },
+    {
+      "epoch": 0.6607993599301742,
+      "grad_norm": 0.0,
+      "learning_rate": 5.45149136552216e-06,
+      "loss": 0.9485,
+      "step": 18170
+    },
+    {
+      "epoch": 0.6608357275339128,
+      "grad_norm": 0.0,
+      "learning_rate": 5.450442431605919e-06,
+      "loss": 0.9703,
+      "step": 18171
+    },
+    {
+      "epoch": 0.6608720951376513,
+      "grad_norm": 0.0,
+      "learning_rate": 5.449393560808334e-06,
+      "loss": 0.9256,
+      "step": 18172
+    },
+    {
+      "epoch": 0.6609084627413899,
+      "grad_norm": 0.0,
+      "learning_rate": 5.448344753143957e-06,
+      "loss": 1.1337,
+      "step": 18173
+    },
+    {
+      "epoch": 0.6609448303451285,
+      "grad_norm": 0.0,
+      "learning_rate": 5.447296008627334e-06,
+      "loss": 0.9115,
+      "step": 18174
+    },
+    {
+      "epoch": 0.6609811979488671,
+      "grad_norm": 0.0,
+      "learning_rate": 5.446247327273029e-06,
+      "loss": 1.0013,
+      "step": 18175
+    },
+    {
+      "epoch": 0.6610175655526057,
+      "grad_norm": 0.0,
+      "learning_rate": 5.445198709095572e-06,
+      "loss": 1.0223,
+      "step": 18176
+    },
+    {
+      "epoch": 0.6610539331563443,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4441501541095245e-06,
+      "loss": 0.9638,
+      "step": 18177
+    },
+    {
+      "epoch": 0.6610903007600829,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4431016623294305e-06,
+      "loss": 0.9491,
+      "step": 18178
+    },
+    {
+      "epoch": 0.6611266683638215,
+      "grad_norm": 0.0,
+      "learning_rate": 5.442053233769835e-06,
+      "loss": 0.9277,
+      "step": 18179
+    },
+    {
+      "epoch": 0.6611630359675601,
+      "grad_norm": 0.0,
+      "learning_rate": 5.44100486844528e-06,
+      "loss": 0.9174,
+      "step": 18180
+    },
+    {
+      "epoch": 0.6611994035712987,
+      "grad_norm": 0.0,
+      "learning_rate": 5.439956566370322e-06,
+      "loss": 0.9478,
+      "step": 18181
+    },
+    {
+      "epoch": 0.6612357711750373,
+      "grad_norm": 0.0,
+      "learning_rate": 5.438908327559489e-06,
+      "loss": 0.9176,
+      "step": 18182
+    },
+    {
+      "epoch": 0.6612721387787759,
+      "grad_norm": 0.0,
+      "learning_rate": 5.437860152027334e-06,
+      "loss": 1.0925,
+      "step": 18183
+    },
+    {
+      "epoch": 0.6613085063825145,
+      "grad_norm": 0.0,
+      "learning_rate": 5.436812039788399e-06,
+      "loss": 1.0009,
+      "step": 18184
+    },
+    {
+      "epoch": 0.6613448739862531,
+      "grad_norm": 0.0,
+      "learning_rate": 5.435763990857222e-06,
+      "loss": 0.951,
+      "step": 18185
+    },
+    {
+      "epoch": 0.6613812415899917,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4347160052483435e-06,
+      "loss": 1.0314,
+      "step": 18186
+    },
+    {
+      "epoch": 0.6614176091937303,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4336680829763e-06,
+      "loss": 0.9276,
+      "step": 18187
+    },
+    {
+      "epoch": 0.6614539767974689,
+      "grad_norm": 0.0,
+      "learning_rate": 5.43262022405564e-06,
+      "loss": 1.0321,
+      "step": 18188
+    },
+    {
+      "epoch": 0.6614903444012074,
+      "grad_norm": 0.0,
+      "learning_rate": 5.431572428500886e-06,
+      "loss": 0.9836,
+      "step": 18189
+    },
+    {
+      "epoch": 0.661526712004946,
+      "grad_norm": 0.0,
+      "learning_rate": 5.430524696326587e-06,
+      "loss": 0.9193,
+      "step": 18190
+    },
+    {
+      "epoch": 0.6615630796086845,
+      "grad_norm": 0.0,
+      "learning_rate": 5.429477027547274e-06,
+      "loss": 0.9069,
+      "step": 18191
+    },
+    {
+      "epoch": 0.6615994472124231,
+      "grad_norm": 0.0,
+      "learning_rate": 5.428429422177484e-06,
+      "loss": 0.9373,
+      "step": 18192
+    },
+    {
+      "epoch": 0.6616358148161617,
+      "grad_norm": 0.0,
+      "learning_rate": 5.427381880231746e-06,
+      "loss": 0.991,
+      "step": 18193
+    },
+    {
+      "epoch": 0.6616721824199003,
+      "grad_norm": 0.0,
+      "learning_rate": 5.426334401724603e-06,
+      "loss": 0.8448,
+      "step": 18194
+    },
+    {
+      "epoch": 0.6617085500236389,
+      "grad_norm": 0.0,
+      "learning_rate": 5.425286986670573e-06,
+      "loss": 0.8899,
+      "step": 18195
+    },
+    {
+      "epoch": 0.6617449176273775,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4242396350842006e-06,
+      "loss": 0.9608,
+      "step": 18196
+    },
+    {
+      "epoch": 0.6617812852311161,
+      "grad_norm": 0.0,
+      "learning_rate": 5.42319234698001e-06,
+      "loss": 0.9866,
+      "step": 18197
+    },
+    {
+      "epoch": 0.6618176528348547,
+      "grad_norm": 0.0,
+      "learning_rate": 5.422145122372533e-06,
+      "loss": 0.8672,
+      "step": 18198
+    },
+    {
+      "epoch": 0.6618540204385933,
+      "grad_norm": 0.0,
+      "learning_rate": 5.421097961276291e-06,
+      "loss": 0.9725,
+      "step": 18199
+    },
+    {
+      "epoch": 0.6618903880423319,
+      "grad_norm": 0.0,
+      "learning_rate": 5.420050863705829e-06,
+      "loss": 0.8067,
+      "step": 18200
+    },
+    {
+      "epoch": 0.6619267556460705,
+      "grad_norm": 0.0,
+      "learning_rate": 5.419003829675654e-06,
+      "loss": 0.9954,
+      "step": 18201
+    },
+    {
+      "epoch": 0.6619631232498091,
+      "grad_norm": 0.0,
+      "learning_rate": 5.417956859200306e-06,
+      "loss": 0.9513,
+      "step": 18202
+    },
+    {
+      "epoch": 0.6619994908535477,
+      "grad_norm": 0.0,
+      "learning_rate": 5.416909952294304e-06,
+      "loss": 0.9834,
+      "step": 18203
+    },
+    {
+      "epoch": 0.6620358584572863,
+      "grad_norm": 0.0,
+      "learning_rate": 5.415863108972175e-06,
+      "loss": 0.8671,
+      "step": 18204
+    },
+    {
+      "epoch": 0.6620722260610249,
+      "grad_norm": 0.0,
+      "learning_rate": 5.414816329248441e-06,
+      "loss": 0.8443,
+      "step": 18205
+    },
+    {
+      "epoch": 0.6621085936647634,
+      "grad_norm": 0.0,
+      "learning_rate": 5.41376961313762e-06,
+      "loss": 0.9748,
+      "step": 18206
+    },
+    {
+      "epoch": 0.662144961268502,
+      "grad_norm": 0.0,
+      "learning_rate": 5.412722960654248e-06,
+      "loss": 0.9796,
+      "step": 18207
+    },
+    {
+      "epoch": 0.6621813288722406,
+      "grad_norm": 0.0,
+      "learning_rate": 5.411676371812828e-06,
+      "loss": 0.9184,
+      "step": 18208
+    },
+    {
+      "epoch": 0.6622176964759792,
+      "grad_norm": 0.0,
+      "learning_rate": 5.410629846627893e-06,
+      "loss": 0.9224,
+      "step": 18209
+    },
+    {
+      "epoch": 0.6622540640797178,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4095833851139565e-06,
+      "loss": 0.9223,
+      "step": 18210
+    },
+    {
+      "epoch": 0.6622904316834564,
+      "grad_norm": 0.0,
+      "learning_rate": 5.408536987285538e-06,
+      "loss": 0.8849,
+      "step": 18211
+    },
+    {
+      "epoch": 0.662326799287195,
+      "grad_norm": 0.0,
+      "learning_rate": 5.40749065315715e-06,
+      "loss": 0.8834,
+      "step": 18212
+    },
+    {
+      "epoch": 0.6623631668909336,
+      "grad_norm": 0.0,
+      "learning_rate": 5.406444382743321e-06,
+      "loss": 0.8046,
+      "step": 18213
+    },
+    {
+      "epoch": 0.6623995344946721,
+      "grad_norm": 0.0,
+      "learning_rate": 5.405398176058551e-06,
+      "loss": 1.0058,
+      "step": 18214
+    },
+    {
+      "epoch": 0.6624359020984107,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4043520331173685e-06,
+      "loss": 0.9981,
+      "step": 18215
+    },
+    {
+      "epoch": 0.6624722697021493,
+      "grad_norm": 0.0,
+      "learning_rate": 5.403305953934278e-06,
+      "loss": 1.0751,
+      "step": 18216
+    },
+    {
+      "epoch": 0.6625086373058879,
+      "grad_norm": 0.0,
+      "learning_rate": 5.402259938523798e-06,
+      "loss": 0.8809,
+      "step": 18217
+    },
+    {
+      "epoch": 0.6625450049096265,
+      "grad_norm": 0.0,
+      "learning_rate": 5.401213986900432e-06,
+      "loss": 0.9221,
+      "step": 18218
+    },
+    {
+      "epoch": 0.6625813725133651,
+      "grad_norm": 0.0,
+      "learning_rate": 5.400168099078702e-06,
+      "loss": 0.9612,
+      "step": 18219
+    },
+    {
+      "epoch": 0.6626177401171037,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3991222750731135e-06,
+      "loss": 0.9938,
+      "step": 18220
+    },
+    {
+      "epoch": 0.6626541077208423,
+      "grad_norm": 0.0,
+      "learning_rate": 5.398076514898175e-06,
+      "loss": 0.9015,
+      "step": 18221
+    },
+    {
+      "epoch": 0.6626904753245809,
+      "grad_norm": 0.0,
+      "learning_rate": 5.397030818568396e-06,
+      "loss": 0.9788,
+      "step": 18222
+    },
+    {
+      "epoch": 0.6627268429283194,
+      "grad_norm": 0.0,
+      "learning_rate": 5.395985186098284e-06,
+      "loss": 1.0886,
+      "step": 18223
+    },
+    {
+      "epoch": 0.662763210532058,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3949396175023446e-06,
+      "loss": 0.9275,
+      "step": 18224
+    },
+    {
+      "epoch": 0.6627995781357966,
+      "grad_norm": 0.0,
+      "learning_rate": 5.393894112795082e-06,
+      "loss": 1.0076,
+      "step": 18225
+    },
+    {
+      "epoch": 0.6628359457395352,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3928486719910085e-06,
+      "loss": 0.9134,
+      "step": 18226
+    },
+    {
+      "epoch": 0.6628723133432738,
+      "grad_norm": 0.0,
+      "learning_rate": 5.391803295104617e-06,
+      "loss": 0.9245,
+      "step": 18227
+    },
+    {
+      "epoch": 0.6629086809470124,
+      "grad_norm": 0.0,
+      "learning_rate": 5.390757982150421e-06,
+      "loss": 0.8603,
+      "step": 18228
+    },
+    {
+      "epoch": 0.662945048550751,
+      "grad_norm": 0.0,
+      "learning_rate": 5.389712733142917e-06,
+      "loss": 0.9114,
+      "step": 18229
+    },
+    {
+      "epoch": 0.6629814161544896,
+      "grad_norm": 0.0,
+      "learning_rate": 5.388667548096608e-06,
+      "loss": 1.0304,
+      "step": 18230
+    },
+    {
+      "epoch": 0.6630177837582282,
+      "grad_norm": 0.0,
+      "learning_rate": 5.38762242702599e-06,
+      "loss": 0.9309,
+      "step": 18231
+    },
+    {
+      "epoch": 0.6630541513619668,
+      "grad_norm": 0.0,
+      "learning_rate": 5.386577369945575e-06,
+      "loss": 0.8574,
+      "step": 18232
+    },
+    {
+      "epoch": 0.6630905189657054,
+      "grad_norm": 0.0,
+      "learning_rate": 5.385532376869845e-06,
+      "loss": 0.95,
+      "step": 18233
+    },
+    {
+      "epoch": 0.663126886569444,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3844874478133115e-06,
+      "loss": 0.9861,
+      "step": 18234
+    },
+    {
+      "epoch": 0.6631632541731826,
+      "grad_norm": 0.0,
+      "learning_rate": 5.383442582790464e-06,
+      "loss": 1.0517,
+      "step": 18235
+    },
+    {
+      "epoch": 0.6631996217769212,
+      "grad_norm": 0.0,
+      "learning_rate": 5.382397781815803e-06,
+      "loss": 0.8941,
+      "step": 18236
+    },
+    {
+      "epoch": 0.6632359893806598,
+      "grad_norm": 0.0,
+      "learning_rate": 5.381353044903816e-06,
+      "loss": 0.972,
+      "step": 18237
+    },
+    {
+      "epoch": 0.6632723569843983,
+      "grad_norm": 0.0,
+      "learning_rate": 5.380308372069006e-06,
+      "loss": 1.0111,
+      "step": 18238
+    },
+    {
+      "epoch": 0.6633087245881368,
+      "grad_norm": 0.0,
+      "learning_rate": 5.379263763325864e-06,
+      "loss": 0.9041,
+      "step": 18239
+    },
+    {
+      "epoch": 0.6633450921918754,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3782192186888805e-06,
+      "loss": 0.8834,
+      "step": 18240
+    },
+    {
+      "epoch": 0.663381459795614,
+      "grad_norm": 0.0,
+      "learning_rate": 5.377174738172544e-06,
+      "loss": 0.9089,
+      "step": 18241
+    },
+    {
+      "epoch": 0.6634178273993526,
+      "grad_norm": 0.0,
+      "learning_rate": 5.376130321791358e-06,
+      "loss": 0.9446,
+      "step": 18242
+    },
+    {
+      "epoch": 0.6634541950030912,
+      "grad_norm": 0.0,
+      "learning_rate": 5.375085969559798e-06,
+      "loss": 0.9312,
+      "step": 18243
+    },
+    {
+      "epoch": 0.6634905626068298,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3740416814923544e-06,
+      "loss": 1.0248,
+      "step": 18244
+    },
+    {
+      "epoch": 0.6635269302105684,
+      "grad_norm": 0.0,
+      "learning_rate": 5.372997457603528e-06,
+      "loss": 0.8916,
+      "step": 18245
+    },
+    {
+      "epoch": 0.663563297814307,
+      "grad_norm": 0.0,
+      "learning_rate": 5.371953297907788e-06,
+      "loss": 0.9125,
+      "step": 18246
+    },
+    {
+      "epoch": 0.6635996654180456,
+      "grad_norm": 0.0,
+      "learning_rate": 5.370909202419634e-06,
+      "loss": 1.0043,
+      "step": 18247
+    },
+    {
+      "epoch": 0.6636360330217842,
+      "grad_norm": 0.0,
+      "learning_rate": 5.369865171153548e-06,
+      "loss": 0.9337,
+      "step": 18248
+    },
+    {
+      "epoch": 0.6636724006255228,
+      "grad_norm": 0.0,
+      "learning_rate": 5.368821204124013e-06,
+      "loss": 1.0264,
+      "step": 18249
+    },
+    {
+      "epoch": 0.6637087682292614,
+      "grad_norm": 0.0,
+      "learning_rate": 5.367777301345508e-06,
+      "loss": 1.0694,
+      "step": 18250
+    },
+    {
+      "epoch": 0.663745135833,
+      "grad_norm": 0.0,
+      "learning_rate": 5.366733462832525e-06,
+      "loss": 0.9051,
+      "step": 18251
+    },
+    {
+      "epoch": 0.6637815034367386,
+      "grad_norm": 0.0,
+      "learning_rate": 5.365689688599543e-06,
+      "loss": 0.96,
+      "step": 18252
+    },
+    {
+      "epoch": 0.6638178710404772,
+      "grad_norm": 0.0,
+      "learning_rate": 5.36464597866104e-06,
+      "loss": 0.9823,
+      "step": 18253
+    },
+    {
+      "epoch": 0.6638542386442158,
+      "grad_norm": 0.0,
+      "learning_rate": 5.363602333031497e-06,
+      "loss": 0.9065,
+      "step": 18254
+    },
+    {
+      "epoch": 0.6638906062479544,
+      "grad_norm": 0.0,
+      "learning_rate": 5.362558751725394e-06,
+      "loss": 0.9386,
+      "step": 18255
+    },
+    {
+      "epoch": 0.6639269738516929,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3615152347572045e-06,
+      "loss": 1.028,
+      "step": 18256
+    },
+    {
+      "epoch": 0.6639633414554315,
+      "grad_norm": 0.0,
+      "learning_rate": 5.360471782141414e-06,
+      "loss": 0.9832,
+      "step": 18257
+    },
+    {
+      "epoch": 0.66399970905917,
+      "grad_norm": 0.0,
+      "learning_rate": 5.359428393892495e-06,
+      "loss": 0.9203,
+      "step": 18258
+    },
+    {
+      "epoch": 0.6640360766629086,
+      "grad_norm": 0.0,
+      "learning_rate": 5.358385070024923e-06,
+      "loss": 0.8689,
+      "step": 18259
+    },
+    {
+      "epoch": 0.6640724442666472,
+      "grad_norm": 0.0,
+      "learning_rate": 5.357341810553169e-06,
+      "loss": 1.0268,
+      "step": 18260
+    },
+    {
+      "epoch": 0.6641088118703858,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3562986154917175e-06,
+      "loss": 0.9821,
+      "step": 18261
+    },
+    {
+      "epoch": 0.6641451794741244,
+      "grad_norm": 0.0,
+      "learning_rate": 5.355255484855031e-06,
+      "loss": 0.9156,
+      "step": 18262
+    },
+    {
+      "epoch": 0.664181547077863,
+      "grad_norm": 0.0,
+      "learning_rate": 5.354212418657579e-06,
+      "loss": 0.9426,
+      "step": 18263
+    },
+    {
+      "epoch": 0.6642179146816016,
+      "grad_norm": 0.0,
+      "learning_rate": 5.353169416913843e-06,
+      "loss": 0.9779,
+      "step": 18264
+    },
+    {
+      "epoch": 0.6642542822853402,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3521264796382884e-06,
+      "loss": 0.9466,
+      "step": 18265
+    },
+    {
+      "epoch": 0.6642906498890788,
+      "grad_norm": 0.0,
+      "learning_rate": 5.351083606845385e-06,
+      "loss": 1.0571,
+      "step": 18266
+    },
+    {
+      "epoch": 0.6643270174928174,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3500407985496e-06,
+      "loss": 0.9539,
+      "step": 18267
+    },
+    {
+      "epoch": 0.664363385096556,
+      "grad_norm": 0.0,
+      "learning_rate": 5.348998054765402e-06,
+      "loss": 1.0117,
+      "step": 18268
+    },
+    {
+      "epoch": 0.6643997527002946,
+      "grad_norm": 0.0,
+      "learning_rate": 5.347955375507252e-06,
+      "loss": 0.9522,
+      "step": 18269
+    },
+    {
+      "epoch": 0.6644361203040332,
+      "grad_norm": 0.0,
+      "learning_rate": 5.346912760789626e-06,
+      "loss": 1.0166,
+      "step": 18270
+    },
+    {
+      "epoch": 0.6644724879077718,
+      "grad_norm": 0.0,
+      "learning_rate": 5.345870210626983e-06,
+      "loss": 1.0508,
+      "step": 18271
+    },
+    {
+      "epoch": 0.6645088555115104,
+      "grad_norm": 0.0,
+      "learning_rate": 5.344827725033788e-06,
+      "loss": 0.9808,
+      "step": 18272
+    },
+    {
+      "epoch": 0.6645452231152489,
+      "grad_norm": 0.0,
+      "learning_rate": 5.343785304024499e-06,
+      "loss": 1.069,
+      "step": 18273
+    },
+    {
+      "epoch": 0.6645815907189875,
+      "grad_norm": 0.0,
+      "learning_rate": 5.34274294761359e-06,
+      "loss": 1.017,
+      "step": 18274
+    },
+    {
+      "epoch": 0.6646179583227261,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3417006558155085e-06,
+      "loss": 0.9365,
+      "step": 18275
+    },
+    {
+      "epoch": 0.6646543259264647,
+      "grad_norm": 0.0,
+      "learning_rate": 5.340658428644724e-06,
+      "loss": 0.9081,
+      "step": 18276
+    },
+    {
+      "epoch": 0.6646906935302033,
+      "grad_norm": 0.0,
+      "learning_rate": 5.339616266115693e-06,
+      "loss": 0.9482,
+      "step": 18277
+    },
+    {
+      "epoch": 0.6647270611339419,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3385741682428735e-06,
+      "loss": 0.9556,
+      "step": 18278
+    },
+    {
+      "epoch": 0.6647634287376805,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3375321350407215e-06,
+      "loss": 0.8841,
+      "step": 18279
+    },
+    {
+      "epoch": 0.6647997963414191,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3364901665237024e-06,
+      "loss": 0.9435,
+      "step": 18280
+    },
+    {
+      "epoch": 0.6648361639451577,
+      "grad_norm": 0.0,
+      "learning_rate": 5.335448262706262e-06,
+      "loss": 1.0264,
+      "step": 18281
+    },
+    {
+      "epoch": 0.6648725315488963,
+      "grad_norm": 0.0,
+      "learning_rate": 5.334406423602856e-06,
+      "loss": 0.9916,
+      "step": 18282
+    },
+    {
+      "epoch": 0.6649088991526348,
+      "grad_norm": 0.0,
+      "learning_rate": 5.333364649227944e-06,
+      "loss": 1.0083,
+      "step": 18283
+    },
+    {
+      "epoch": 0.6649452667563734,
+      "grad_norm": 0.0,
+      "learning_rate": 5.332322939595977e-06,
+      "loss": 0.8649,
+      "step": 18284
+    },
+    {
+      "epoch": 0.664981634360112,
+      "grad_norm": 0.0,
+      "learning_rate": 5.331281294721406e-06,
+      "loss": 1.0236,
+      "step": 18285
+    },
+    {
+      "epoch": 0.6650180019638506,
+      "grad_norm": 0.0,
+      "learning_rate": 5.330239714618679e-06,
+      "loss": 0.9564,
+      "step": 18286
+    },
+    {
+      "epoch": 0.6650543695675892,
+      "grad_norm": 0.0,
+      "learning_rate": 5.329198199302259e-06,
+      "loss": 1.0502,
+      "step": 18287
+    },
+    {
+      "epoch": 0.6650907371713278,
+      "grad_norm": 0.0,
+      "learning_rate": 5.32815674878658e-06,
+      "loss": 0.925,
+      "step": 18288
+    },
+    {
+      "epoch": 0.6651271047750664,
+      "grad_norm": 0.0,
+      "learning_rate": 5.327115363086099e-06,
+      "loss": 0.8768,
+      "step": 18289
+    },
+    {
+      "epoch": 0.6651634723788049,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3260740422152635e-06,
+      "loss": 0.9067,
+      "step": 18290
+    },
+    {
+      "epoch": 0.6651998399825435,
+      "grad_norm": 0.0,
+      "learning_rate": 5.32503278618852e-06,
+      "loss": 0.9222,
+      "step": 18291
+    },
+    {
+      "epoch": 0.6652362075862821,
+      "grad_norm": 0.0,
+      "learning_rate": 5.323991595020308e-06,
+      "loss": 0.8516,
+      "step": 18292
+    },
+    {
+      "epoch": 0.6652725751900207,
+      "grad_norm": 0.0,
+      "learning_rate": 5.322950468725087e-06,
+      "loss": 0.9276,
+      "step": 18293
+    },
+    {
+      "epoch": 0.6653089427937593,
+      "grad_norm": 0.0,
+      "learning_rate": 5.321909407317285e-06,
+      "loss": 0.8699,
+      "step": 18294
+    },
+    {
+      "epoch": 0.6653453103974979,
+      "grad_norm": 0.0,
+      "learning_rate": 5.320868410811357e-06,
+      "loss": 1.0076,
+      "step": 18295
+    },
+    {
+      "epoch": 0.6653816780012365,
+      "grad_norm": 0.0,
+      "learning_rate": 5.319827479221739e-06,
+      "loss": 0.9196,
+      "step": 18296
+    },
+    {
+      "epoch": 0.6654180456049751,
+      "grad_norm": 0.0,
+      "learning_rate": 5.318786612562877e-06,
+      "loss": 1.0671,
+      "step": 18297
+    },
+    {
+      "epoch": 0.6654544132087137,
+      "grad_norm": 0.0,
+      "learning_rate": 5.317745810849204e-06,
+      "loss": 1.0344,
+      "step": 18298
+    },
+    {
+      "epoch": 0.6654907808124523,
+      "grad_norm": 0.0,
+      "learning_rate": 5.316705074095172e-06,
+      "loss": 0.9988,
+      "step": 18299
+    },
+    {
+      "epoch": 0.6655271484161909,
+      "grad_norm": 0.0,
+      "learning_rate": 5.315664402315208e-06,
+      "loss": 0.9009,
+      "step": 18300
+    },
+    {
+      "epoch": 0.6655635160199295,
+      "grad_norm": 0.0,
+      "learning_rate": 5.314623795523753e-06,
+      "loss": 0.9132,
+      "step": 18301
+    },
+    {
+      "epoch": 0.6655998836236681,
+      "grad_norm": 0.0,
+      "learning_rate": 5.313583253735246e-06,
+      "loss": 1.0183,
+      "step": 18302
+    },
+    {
+      "epoch": 0.6656362512274067,
+      "grad_norm": 0.0,
+      "learning_rate": 5.312542776964124e-06,
+      "loss": 0.8454,
+      "step": 18303
+    },
+    {
+      "epoch": 0.6656726188311453,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3115023652248185e-06,
+      "loss": 0.9818,
+      "step": 18304
+    },
+    {
+      "epoch": 0.6657089864348839,
+      "grad_norm": 0.0,
+      "learning_rate": 5.310462018531763e-06,
+      "loss": 0.9619,
+      "step": 18305
+    },
+    {
+      "epoch": 0.6657453540386224,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3094217368994e-06,
+      "loss": 1.0344,
+      "step": 18306
+    },
+    {
+      "epoch": 0.6657817216423609,
+      "grad_norm": 0.0,
+      "learning_rate": 5.308381520342149e-06,
+      "loss": 0.9906,
+      "step": 18307
+    },
+    {
+      "epoch": 0.6658180892460995,
+      "grad_norm": 0.0,
+      "learning_rate": 5.307341368874452e-06,
+      "loss": 0.938,
+      "step": 18308
+    },
+    {
+      "epoch": 0.6658544568498381,
+      "grad_norm": 0.0,
+      "learning_rate": 5.306301282510735e-06,
+      "loss": 1.0143,
+      "step": 18309
+    },
+    {
+      "epoch": 0.6658908244535767,
+      "grad_norm": 0.0,
+      "learning_rate": 5.305261261265429e-06,
+      "loss": 0.8967,
+      "step": 18310
+    },
+    {
+      "epoch": 0.6659271920573153,
+      "grad_norm": 0.0,
+      "learning_rate": 5.304221305152957e-06,
+      "loss": 0.9305,
+      "step": 18311
+    },
+    {
+      "epoch": 0.6659635596610539,
+      "grad_norm": 0.0,
+      "learning_rate": 5.303181414187759e-06,
+      "loss": 0.9845,
+      "step": 18312
+    },
+    {
+      "epoch": 0.6659999272647925,
+      "grad_norm": 0.0,
+      "learning_rate": 5.302141588384249e-06,
+      "loss": 0.9329,
+      "step": 18313
+    },
+    {
+      "epoch": 0.6660362948685311,
+      "grad_norm": 0.0,
+      "learning_rate": 5.30110182775686e-06,
+      "loss": 1.0273,
+      "step": 18314
+    },
+    {
+      "epoch": 0.6660726624722697,
+      "grad_norm": 0.0,
+      "learning_rate": 5.30006213232002e-06,
+      "loss": 0.9517,
+      "step": 18315
+    },
+    {
+      "epoch": 0.6661090300760083,
+      "grad_norm": 0.0,
+      "learning_rate": 5.299022502088147e-06,
+      "loss": 0.8737,
+      "step": 18316
+    },
+    {
+      "epoch": 0.6661453976797469,
+      "grad_norm": 0.0,
+      "learning_rate": 5.297982937075666e-06,
+      "loss": 0.8805,
+      "step": 18317
+    },
+    {
+      "epoch": 0.6661817652834855,
+      "grad_norm": 0.0,
+      "learning_rate": 5.296943437297001e-06,
+      "loss": 0.9308,
+      "step": 18318
+    },
+    {
+      "epoch": 0.6662181328872241,
+      "grad_norm": 0.0,
+      "learning_rate": 5.295904002766579e-06,
+      "loss": 1.1406,
+      "step": 18319
+    },
+    {
+      "epoch": 0.6662545004909627,
+      "grad_norm": 0.0,
+      "learning_rate": 5.294864633498805e-06,
+      "loss": 0.9139,
+      "step": 18320
+    },
+    {
+      "epoch": 0.6662908680947013,
+      "grad_norm": 0.0,
+      "learning_rate": 5.293825329508113e-06,
+      "loss": 1.002,
+      "step": 18321
+    },
+    {
+      "epoch": 0.6663272356984399,
+      "grad_norm": 0.0,
+      "learning_rate": 5.292786090808919e-06,
+      "loss": 1.0523,
+      "step": 18322
+    },
+    {
+      "epoch": 0.6663636033021784,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2917469174156386e-06,
+      "loss": 0.8301,
+      "step": 18323
+    },
+    {
+      "epoch": 0.666399970905917,
+      "grad_norm": 0.0,
+      "learning_rate": 5.290707809342686e-06,
+      "loss": 1.0654,
+      "step": 18324
+    },
+    {
+      "epoch": 0.6664363385096556,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2896687666044875e-06,
+      "loss": 0.9713,
+      "step": 18325
+    },
+    {
+      "epoch": 0.6664727061133942,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2886297892154445e-06,
+      "loss": 0.977,
+      "step": 18326
+    },
+    {
+      "epoch": 0.6665090737171327,
+      "grad_norm": 0.0,
+      "learning_rate": 5.287590877189982e-06,
+      "loss": 1.0405,
+      "step": 18327
+    },
+    {
+      "epoch": 0.6665454413208713,
+      "grad_norm": 0.0,
+      "learning_rate": 5.286552030542511e-06,
+      "loss": 0.9117,
+      "step": 18328
+    },
+    {
+      "epoch": 0.6665818089246099,
+      "grad_norm": 0.0,
+      "learning_rate": 5.285513249287443e-06,
+      "loss": 0.9967,
+      "step": 18329
+    },
+    {
+      "epoch": 0.6666181765283485,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2844745334391855e-06,
+      "loss": 0.879,
+      "step": 18330
+    },
+    {
+      "epoch": 0.6666545441320871,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2834358830121565e-06,
+      "loss": 0.9625,
+      "step": 18331
+    },
+    {
+      "epoch": 0.6666909117358257,
+      "grad_norm": 0.0,
+      "learning_rate": 5.282397298020764e-06,
+      "loss": 0.9262,
+      "step": 18332
+    },
+    {
+      "epoch": 0.6667272793395643,
+      "grad_norm": 0.0,
+      "learning_rate": 5.281358778479415e-06,
+      "loss": 0.8985,
+      "step": 18333
+    },
+    {
+      "epoch": 0.6667636469433029,
+      "grad_norm": 0.0,
+      "learning_rate": 5.280320324402518e-06,
+      "loss": 0.7709,
+      "step": 18334
+    },
+    {
+      "epoch": 0.6668000145470415,
+      "grad_norm": 0.0,
+      "learning_rate": 5.279281935804481e-06,
+      "loss": 1.0076,
+      "step": 18335
+    },
+    {
+      "epoch": 0.6668363821507801,
+      "grad_norm": 0.0,
+      "learning_rate": 5.278243612699704e-06,
+      "loss": 0.9643,
+      "step": 18336
+    },
+    {
+      "epoch": 0.6668727497545187,
+      "grad_norm": 0.0,
+      "learning_rate": 5.277205355102603e-06,
+      "loss": 0.8727,
+      "step": 18337
+    },
+    {
+      "epoch": 0.6669091173582573,
+      "grad_norm": 0.0,
+      "learning_rate": 5.276167163027576e-06,
+      "loss": 0.9594,
+      "step": 18338
+    },
+    {
+      "epoch": 0.6669454849619959,
+      "grad_norm": 0.0,
+      "learning_rate": 5.275129036489026e-06,
+      "loss": 0.9464,
+      "step": 18339
+    },
+    {
+      "epoch": 0.6669818525657344,
+      "grad_norm": 0.0,
+      "learning_rate": 5.274090975501359e-06,
+      "loss": 0.9493,
+      "step": 18340
+    },
+    {
+      "epoch": 0.667018220169473,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2730529800789745e-06,
+      "loss": 0.977,
+      "step": 18341
+    },
+    {
+      "epoch": 0.6670545877732116,
+      "grad_norm": 0.0,
+      "learning_rate": 5.272015050236272e-06,
+      "loss": 0.9551,
+      "step": 18342
+    },
+    {
+      "epoch": 0.6670909553769502,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2709771859876495e-06,
+      "loss": 0.9981,
+      "step": 18343
+    },
+    {
+      "epoch": 0.6671273229806888,
+      "grad_norm": 0.0,
+      "learning_rate": 5.269939387347517e-06,
+      "loss": 1.0558,
+      "step": 18344
+    },
+    {
+      "epoch": 0.6671636905844274,
+      "grad_norm": 0.0,
+      "learning_rate": 5.268901654330256e-06,
+      "loss": 0.9661,
+      "step": 18345
+    },
+    {
+      "epoch": 0.667200058188166,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2678639869502765e-06,
+      "loss": 0.9335,
+      "step": 18346
+    },
+    {
+      "epoch": 0.6672364257919046,
+      "grad_norm": 0.0,
+      "learning_rate": 5.266826385221969e-06,
+      "loss": 0.9691,
+      "step": 18347
+    },
+    {
+      "epoch": 0.6672727933956432,
+      "grad_norm": 0.0,
+      "learning_rate": 5.265788849159731e-06,
+      "loss": 0.9826,
+      "step": 18348
+    },
+    {
+      "epoch": 0.6673091609993818,
+      "grad_norm": 0.0,
+      "learning_rate": 5.264751378777951e-06,
+      "loss": 1.0476,
+      "step": 18349
+    },
+    {
+      "epoch": 0.6673455286031204,
+      "grad_norm": 0.0,
+      "learning_rate": 5.263713974091032e-06,
+      "loss": 0.955,
+      "step": 18350
+    },
+    {
+      "epoch": 0.667381896206859,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2626766351133615e-06,
+      "loss": 1.1002,
+      "step": 18351
+    },
+    {
+      "epoch": 0.6674182638105975,
+      "grad_norm": 0.0,
+      "learning_rate": 5.261639361859331e-06,
+      "loss": 1.0013,
+      "step": 18352
+    },
+    {
+      "epoch": 0.6674546314143361,
+      "grad_norm": 0.0,
+      "learning_rate": 5.260602154343327e-06,
+      "loss": 1.1527,
+      "step": 18353
+    },
+    {
+      "epoch": 0.6674909990180747,
+      "grad_norm": 0.0,
+      "learning_rate": 5.259565012579753e-06,
+      "loss": 0.9635,
+      "step": 18354
+    },
+    {
+      "epoch": 0.6675273666218133,
+      "grad_norm": 0.0,
+      "learning_rate": 5.25852793658298e-06,
+      "loss": 1.0774,
+      "step": 18355
+    },
+    {
+      "epoch": 0.6675637342255519,
+      "grad_norm": 0.0,
+      "learning_rate": 5.257490926367408e-06,
+      "loss": 0.9577,
+      "step": 18356
+    },
+    {
+      "epoch": 0.6676001018292904,
+      "grad_norm": 0.0,
+      "learning_rate": 5.256453981947421e-06,
+      "loss": 0.8942,
+      "step": 18357
+    },
+    {
+      "epoch": 0.667636469433029,
+      "grad_norm": 0.0,
+      "learning_rate": 5.255417103337406e-06,
+      "loss": 0.9167,
+      "step": 18358
+    },
+    {
+      "epoch": 0.6676728370367676,
+      "grad_norm": 0.0,
+      "learning_rate": 5.254380290551746e-06,
+      "loss": 0.9979,
+      "step": 18359
+    },
+    {
+      "epoch": 0.6677092046405062,
+      "grad_norm": 0.0,
+      "learning_rate": 5.253343543604827e-06,
+      "loss": 1.048,
+      "step": 18360
+    },
+    {
+      "epoch": 0.6677455722442448,
+      "grad_norm": 0.0,
+      "learning_rate": 5.25230686251103e-06,
+      "loss": 0.8742,
+      "step": 18361
+    },
+    {
+      "epoch": 0.6677819398479834,
+      "grad_norm": 0.0,
+      "learning_rate": 5.251270247284736e-06,
+      "loss": 0.8992,
+      "step": 18362
+    },
+    {
+      "epoch": 0.667818307451722,
+      "grad_norm": 0.0,
+      "learning_rate": 5.250233697940334e-06,
+      "loss": 0.8906,
+      "step": 18363
+    },
+    {
+      "epoch": 0.6678546750554606,
+      "grad_norm": 0.0,
+      "learning_rate": 5.249197214492201e-06,
+      "loss": 0.9703,
+      "step": 18364
+    },
+    {
+      "epoch": 0.6678910426591992,
+      "grad_norm": 0.0,
+      "learning_rate": 5.248160796954715e-06,
+      "loss": 0.9378,
+      "step": 18365
+    },
+    {
+      "epoch": 0.6679274102629378,
+      "grad_norm": 0.0,
+      "learning_rate": 5.247124445342258e-06,
+      "loss": 0.9652,
+      "step": 18366
+    },
+    {
+      "epoch": 0.6679637778666764,
+      "grad_norm": 0.0,
+      "learning_rate": 5.246088159669204e-06,
+      "loss": 0.9189,
+      "step": 18367
+    },
+    {
+      "epoch": 0.668000145470415,
+      "grad_norm": 0.0,
+      "learning_rate": 5.245051939949928e-06,
+      "loss": 0.8711,
+      "step": 18368
+    },
+    {
+      "epoch": 0.6680365130741536,
+      "grad_norm": 0.0,
+      "learning_rate": 5.244015786198814e-06,
+      "loss": 0.9194,
+      "step": 18369
+    },
+    {
+      "epoch": 0.6680728806778922,
+      "grad_norm": 0.0,
+      "learning_rate": 5.242979698430233e-06,
+      "loss": 0.8959,
+      "step": 18370
+    },
+    {
+      "epoch": 0.6681092482816308,
+      "grad_norm": 0.0,
+      "learning_rate": 5.241943676658561e-06,
+      "loss": 0.9628,
+      "step": 18371
+    },
+    {
+      "epoch": 0.6681456158853694,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2409077208981625e-06,
+      "loss": 0.9287,
+      "step": 18372
+    },
+    {
+      "epoch": 0.668181983489108,
+      "grad_norm": 0.0,
+      "learning_rate": 5.239871831163428e-06,
+      "loss": 0.9174,
+      "step": 18373
+    },
+    {
+      "epoch": 0.6682183510928464,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2388360074687095e-06,
+      "loss": 1.0174,
+      "step": 18374
+    },
+    {
+      "epoch": 0.668254718696585,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2378002498283885e-06,
+      "loss": 1.0639,
+      "step": 18375
+    },
+    {
+      "epoch": 0.6682910863003236,
+      "grad_norm": 0.0,
+      "learning_rate": 5.236764558256835e-06,
+      "loss": 1.0757,
+      "step": 18376
+    },
+    {
+      "epoch": 0.6683274539040622,
+      "grad_norm": 0.0,
+      "learning_rate": 5.235728932768414e-06,
+      "loss": 0.952,
+      "step": 18377
+    },
+    {
+      "epoch": 0.6683638215078008,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2346933733774926e-06,
+      "loss": 0.9506,
+      "step": 18378
+    },
+    {
+      "epoch": 0.6684001891115394,
+      "grad_norm": 0.0,
+      "learning_rate": 5.233657880098441e-06,
+      "loss": 0.9369,
+      "step": 18379
+    },
+    {
+      "epoch": 0.668436556715278,
+      "grad_norm": 0.0,
+      "learning_rate": 5.232622452945625e-06,
+      "loss": 0.8675,
+      "step": 18380
+    },
+    {
+      "epoch": 0.6684729243190166,
+      "grad_norm": 0.0,
+      "learning_rate": 5.231587091933403e-06,
+      "loss": 0.993,
+      "step": 18381
+    },
+    {
+      "epoch": 0.6685092919227552,
+      "grad_norm": 0.0,
+      "learning_rate": 5.230551797076149e-06,
+      "loss": 1.0362,
+      "step": 18382
+    },
+    {
+      "epoch": 0.6685456595264938,
+      "grad_norm": 0.0,
+      "learning_rate": 5.229516568388222e-06,
+      "loss": 1.052,
+      "step": 18383
+    },
+    {
+      "epoch": 0.6685820271302324,
+      "grad_norm": 0.0,
+      "learning_rate": 5.228481405883984e-06,
+      "loss": 1.0788,
+      "step": 18384
+    },
+    {
+      "epoch": 0.668618394733971,
+      "grad_norm": 0.0,
+      "learning_rate": 5.227446309577793e-06,
+      "loss": 1.0504,
+      "step": 18385
+    },
+    {
+      "epoch": 0.6686547623377096,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2264112794840205e-06,
+      "loss": 0.9141,
+      "step": 18386
+    },
+    {
+      "epoch": 0.6686911299414482,
+      "grad_norm": 0.0,
+      "learning_rate": 5.22537631561701e-06,
+      "loss": 0.9432,
+      "step": 18387
+    },
+    {
+      "epoch": 0.6687274975451868,
+      "grad_norm": 0.0,
+      "learning_rate": 5.224341417991134e-06,
+      "loss": 0.9984,
+      "step": 18388
+    },
+    {
+      "epoch": 0.6687638651489254,
+      "grad_norm": 0.0,
+      "learning_rate": 5.223306586620744e-06,
+      "loss": 1.0253,
+      "step": 18389
+    },
+    {
+      "epoch": 0.668800232752664,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2222718215201995e-06,
+      "loss": 1.1168,
+      "step": 18390
+    },
+    {
+      "epoch": 0.6688366003564025,
+      "grad_norm": 0.0,
+      "learning_rate": 5.22123712270385e-06,
+      "loss": 1.0981,
+      "step": 18391
+    },
+    {
+      "epoch": 0.6688729679601411,
+      "grad_norm": 0.0,
+      "learning_rate": 5.220202490186063e-06,
+      "loss": 1.0168,
+      "step": 18392
+    },
+    {
+      "epoch": 0.6689093355638797,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2191679239811764e-06,
+      "loss": 0.9631,
+      "step": 18393
+    },
+    {
+      "epoch": 0.6689457031676183,
+      "grad_norm": 0.0,
+      "learning_rate": 5.218133424103558e-06,
+      "loss": 0.9874,
+      "step": 18394
+    },
+    {
+      "epoch": 0.6689820707713569,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2170989905675505e-06,
+      "loss": 0.9429,
+      "step": 18395
+    },
+    {
+      "epoch": 0.6690184383750954,
+      "grad_norm": 0.0,
+      "learning_rate": 5.21606462338751e-06,
+      "loss": 0.9222,
+      "step": 18396
+    },
+    {
+      "epoch": 0.669054805978834,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2150303225777854e-06,
+      "loss": 0.9259,
+      "step": 18397
+    },
+    {
+      "epoch": 0.6690911735825726,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2139960881527266e-06,
+      "loss": 0.8269,
+      "step": 18398
+    },
+    {
+      "epoch": 0.6691275411863112,
+      "grad_norm": 0.0,
+      "learning_rate": 5.21296192012668e-06,
+      "loss": 0.9972,
+      "step": 18399
+    },
+    {
+      "epoch": 0.6691639087900498,
+      "grad_norm": 0.0,
+      "learning_rate": 5.211927818513992e-06,
+      "loss": 0.9201,
+      "step": 18400
+    },
+    {
+      "epoch": 0.6692002763937884,
+      "grad_norm": 0.0,
+      "learning_rate": 5.210893783329016e-06,
+      "loss": 1.0266,
+      "step": 18401
+    },
+    {
+      "epoch": 0.669236643997527,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2098598145860934e-06,
+      "loss": 1.0187,
+      "step": 18402
+    },
+    {
+      "epoch": 0.6692730116012656,
+      "grad_norm": 0.0,
+      "learning_rate": 5.208825912299571e-06,
+      "loss": 0.9552,
+      "step": 18403
+    },
+    {
+      "epoch": 0.6693093792050042,
+      "grad_norm": 0.0,
+      "learning_rate": 5.207792076483786e-06,
+      "loss": 0.9935,
+      "step": 18404
+    },
+    {
+      "epoch": 0.6693457468087428,
+      "grad_norm": 0.0,
+      "learning_rate": 5.206758307153095e-06,
+      "loss": 0.8907,
+      "step": 18405
+    },
+    {
+      "epoch": 0.6693821144124814,
+      "grad_norm": 0.0,
+      "learning_rate": 5.205724604321823e-06,
+      "loss": 1.0504,
+      "step": 18406
+    },
+    {
+      "epoch": 0.66941848201622,
+      "grad_norm": 0.0,
+      "learning_rate": 5.204690968004325e-06,
+      "loss": 0.937,
+      "step": 18407
+    },
+    {
+      "epoch": 0.6694548496199585,
+      "grad_norm": 0.0,
+      "learning_rate": 5.203657398214936e-06,
+      "loss": 0.8328,
+      "step": 18408
+    },
+    {
+      "epoch": 0.6694912172236971,
+      "grad_norm": 0.0,
+      "learning_rate": 5.202623894967996e-06,
+      "loss": 0.9554,
+      "step": 18409
+    },
+    {
+      "epoch": 0.6695275848274357,
+      "grad_norm": 0.0,
+      "learning_rate": 5.201590458277837e-06,
+      "loss": 0.9964,
+      "step": 18410
+    },
+    {
+      "epoch": 0.6695639524311743,
+      "grad_norm": 0.0,
+      "learning_rate": 5.200557088158812e-06,
+      "loss": 1.0162,
+      "step": 18411
+    },
+    {
+      "epoch": 0.6696003200349129,
+      "grad_norm": 0.0,
+      "learning_rate": 5.19952378462524e-06,
+      "loss": 0.8796,
+      "step": 18412
+    },
+    {
+      "epoch": 0.6696366876386515,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1984905476914686e-06,
+      "loss": 0.9783,
+      "step": 18413
+    },
+    {
+      "epoch": 0.6696730552423901,
+      "grad_norm": 0.0,
+      "learning_rate": 5.197457377371826e-06,
+      "loss": 0.941,
+      "step": 18414
+    },
+    {
+      "epoch": 0.6697094228461287,
+      "grad_norm": 0.0,
+      "learning_rate": 5.196424273680651e-06,
+      "loss": 0.9201,
+      "step": 18415
+    },
+    {
+      "epoch": 0.6697457904498673,
+      "grad_norm": 0.0,
+      "learning_rate": 5.195391236632272e-06,
+      "loss": 0.928,
+      "step": 18416
+    },
+    {
+      "epoch": 0.6697821580536059,
+      "grad_norm": 0.0,
+      "learning_rate": 5.19435826624102e-06,
+      "loss": 1.0122,
+      "step": 18417
+    },
+    {
+      "epoch": 0.6698185256573445,
+      "grad_norm": 0.0,
+      "learning_rate": 5.193325362521236e-06,
+      "loss": 0.9571,
+      "step": 18418
+    },
+    {
+      "epoch": 0.669854893261083,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1922925254872355e-06,
+      "loss": 1.0933,
+      "step": 18419
+    },
+    {
+      "epoch": 0.6698912608648216,
+      "grad_norm": 0.0,
+      "learning_rate": 5.191259755153359e-06,
+      "loss": 0.9994,
+      "step": 18420
+    },
+    {
+      "epoch": 0.6699276284685602,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1902270515339305e-06,
+      "loss": 0.9399,
+      "step": 18421
+    },
+    {
+      "epoch": 0.6699639960722988,
+      "grad_norm": 0.0,
+      "learning_rate": 5.189194414643277e-06,
+      "loss": 0.9413,
+      "step": 18422
+    },
+    {
+      "epoch": 0.6700003636760374,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1881618444957224e-06,
+      "loss": 0.8956,
+      "step": 18423
+    },
+    {
+      "epoch": 0.6700367312797759,
+      "grad_norm": 0.0,
+      "learning_rate": 5.187129341105602e-06,
+      "loss": 0.9706,
+      "step": 18424
+    },
+    {
+      "epoch": 0.6700730988835145,
+      "grad_norm": 0.0,
+      "learning_rate": 5.186096904487225e-06,
+      "loss": 0.9645,
+      "step": 18425
+    },
+    {
+      "epoch": 0.6701094664872531,
+      "grad_norm": 0.0,
+      "learning_rate": 5.185064534654929e-06,
+      "loss": 0.9095,
+      "step": 18426
+    },
+    {
+      "epoch": 0.6701458340909917,
+      "grad_norm": 0.0,
+      "learning_rate": 5.18403223162303e-06,
+      "loss": 0.9659,
+      "step": 18427
+    },
+    {
+      "epoch": 0.6701822016947303,
+      "grad_norm": 0.0,
+      "learning_rate": 5.182999995405851e-06,
+      "loss": 0.8933,
+      "step": 18428
+    },
+    {
+      "epoch": 0.6702185692984689,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1819678260177086e-06,
+      "loss": 0.8983,
+      "step": 18429
+    },
+    {
+      "epoch": 0.6702549369022075,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1809357234729314e-06,
+      "loss": 1.0589,
+      "step": 18430
+    },
+    {
+      "epoch": 0.6702913045059461,
+      "grad_norm": 0.0,
+      "learning_rate": 5.179903687785834e-06,
+      "loss": 0.9209,
+      "step": 18431
+    },
+    {
+      "epoch": 0.6703276721096847,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1788717189707326e-06,
+      "loss": 0.8877,
+      "step": 18432
+    },
+    {
+      "epoch": 0.6703640397134233,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1778398170419465e-06,
+      "loss": 0.932,
+      "step": 18433
+    },
+    {
+      "epoch": 0.6704004073171619,
+      "grad_norm": 0.0,
+      "learning_rate": 5.176807982013792e-06,
+      "loss": 1.061,
+      "step": 18434
+    },
+    {
+      "epoch": 0.6704367749209005,
+      "grad_norm": 0.0,
+      "learning_rate": 5.175776213900582e-06,
+      "loss": 0.9597,
+      "step": 18435
+    },
+    {
+      "epoch": 0.6704731425246391,
+      "grad_norm": 0.0,
+      "learning_rate": 5.17474451271663e-06,
+      "loss": 1.0138,
+      "step": 18436
+    },
+    {
+      "epoch": 0.6705095101283777,
+      "grad_norm": 0.0,
+      "learning_rate": 5.173712878476258e-06,
+      "loss": 1.0272,
+      "step": 18437
+    },
+    {
+      "epoch": 0.6705458777321163,
+      "grad_norm": 0.0,
+      "learning_rate": 5.172681311193765e-06,
+      "loss": 0.9886,
+      "step": 18438
+    },
+    {
+      "epoch": 0.6705822453358549,
+      "grad_norm": 0.0,
+      "learning_rate": 5.171649810883474e-06,
+      "loss": 0.9638,
+      "step": 18439
+    },
+    {
+      "epoch": 0.6706186129395935,
+      "grad_norm": 0.0,
+      "learning_rate": 5.170618377559691e-06,
+      "loss": 0.8943,
+      "step": 18440
+    },
+    {
+      "epoch": 0.670654980543332,
+      "grad_norm": 0.0,
+      "learning_rate": 5.169587011236725e-06,
+      "loss": 0.9519,
+      "step": 18441
+    },
+    {
+      "epoch": 0.6706913481470705,
+      "grad_norm": 0.0,
+      "learning_rate": 5.168555711928884e-06,
+      "loss": 0.8076,
+      "step": 18442
+    },
+    {
+      "epoch": 0.6707277157508091,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1675244796504845e-06,
+      "loss": 0.8951,
+      "step": 18443
+    },
+    {
+      "epoch": 0.6707640833545477,
+      "grad_norm": 0.0,
+      "learning_rate": 5.166493314415818e-06,
+      "loss": 0.8878,
+      "step": 18444
+    },
+    {
+      "epoch": 0.6708004509582863,
+      "grad_norm": 0.0,
+      "learning_rate": 5.165462216239203e-06,
+      "loss": 0.9231,
+      "step": 18445
+    },
+    {
+      "epoch": 0.6708368185620249,
+      "grad_norm": 0.0,
+      "learning_rate": 5.164431185134942e-06,
+      "loss": 0.8465,
+      "step": 18446
+    },
+    {
+      "epoch": 0.6708731861657635,
+      "grad_norm": 0.0,
+      "learning_rate": 5.163400221117336e-06,
+      "loss": 0.8757,
+      "step": 18447
+    },
+    {
+      "epoch": 0.6709095537695021,
+      "grad_norm": 0.0,
+      "learning_rate": 5.162369324200686e-06,
+      "loss": 0.9593,
+      "step": 18448
+    },
+    {
+      "epoch": 0.6709459213732407,
+      "grad_norm": 0.0,
+      "learning_rate": 5.161338494399302e-06,
+      "loss": 0.9811,
+      "step": 18449
+    },
+    {
+      "epoch": 0.6709822889769793,
+      "grad_norm": 0.0,
+      "learning_rate": 5.160307731727481e-06,
+      "loss": 0.9303,
+      "step": 18450
+    },
+    {
+      "epoch": 0.6710186565807179,
+      "grad_norm": 0.0,
+      "learning_rate": 5.159277036199524e-06,
+      "loss": 0.9424,
+      "step": 18451
+    },
+    {
+      "epoch": 0.6710550241844565,
+      "grad_norm": 0.0,
+      "learning_rate": 5.158246407829726e-06,
+      "loss": 0.9958,
+      "step": 18452
+    },
+    {
+      "epoch": 0.6710913917881951,
+      "grad_norm": 0.0,
+      "learning_rate": 5.157215846632396e-06,
+      "loss": 0.9696,
+      "step": 18453
+    },
+    {
+      "epoch": 0.6711277593919337,
+      "grad_norm": 0.0,
+      "learning_rate": 5.156185352621823e-06,
+      "loss": 1.0385,
+      "step": 18454
+    },
+    {
+      "epoch": 0.6711641269956723,
+      "grad_norm": 0.0,
+      "learning_rate": 5.155154925812301e-06,
+      "loss": 0.9159,
+      "step": 18455
+    },
+    {
+      "epoch": 0.6712004945994109,
+      "grad_norm": 0.0,
+      "learning_rate": 5.15412456621814e-06,
+      "loss": 1.0359,
+      "step": 18456
+    },
+    {
+      "epoch": 0.6712368622031495,
+      "grad_norm": 0.0,
+      "learning_rate": 5.153094273853614e-06,
+      "loss": 0.9312,
+      "step": 18457
+    },
+    {
+      "epoch": 0.671273229806888,
+      "grad_norm": 0.0,
+      "learning_rate": 5.152064048733034e-06,
+      "loss": 0.8607,
+      "step": 18458
+    },
+    {
+      "epoch": 0.6713095974106266,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1510338908706866e-06,
+      "loss": 0.9351,
+      "step": 18459
+    },
+    {
+      "epoch": 0.6713459650143652,
+      "grad_norm": 0.0,
+      "learning_rate": 5.150003800280865e-06,
+      "loss": 0.9677,
+      "step": 18460
+    },
+    {
+      "epoch": 0.6713823326181038,
+      "grad_norm": 0.0,
+      "learning_rate": 5.148973776977856e-06,
+      "loss": 1.0188,
+      "step": 18461
+    },
+    {
+      "epoch": 0.6714187002218424,
+      "grad_norm": 0.0,
+      "learning_rate": 5.147943820975956e-06,
+      "loss": 0.9993,
+      "step": 18462
+    },
+    {
+      "epoch": 0.671455067825581,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1469139322894514e-06,
+      "loss": 0.8803,
+      "step": 18463
+    },
+    {
+      "epoch": 0.6714914354293195,
+      "grad_norm": 0.0,
+      "learning_rate": 5.145884110932631e-06,
+      "loss": 0.9572,
+      "step": 18464
+    },
+    {
+      "epoch": 0.6715278030330581,
+      "grad_norm": 0.0,
+      "learning_rate": 5.144854356919779e-06,
+      "loss": 0.9468,
+      "step": 18465
+    },
+    {
+      "epoch": 0.6715641706367967,
+      "grad_norm": 0.0,
+      "learning_rate": 5.143824670265187e-06,
+      "loss": 0.8867,
+      "step": 18466
+    },
+    {
+      "epoch": 0.6716005382405353,
+      "grad_norm": 0.0,
+      "learning_rate": 5.142795050983134e-06,
+      "loss": 0.9214,
+      "step": 18467
+    },
+    {
+      "epoch": 0.6716369058442739,
+      "grad_norm": 0.0,
+      "learning_rate": 5.141765499087909e-06,
+      "loss": 0.9348,
+      "step": 18468
+    },
+    {
+      "epoch": 0.6716732734480125,
+      "grad_norm": 0.0,
+      "learning_rate": 5.140736014593798e-06,
+      "loss": 0.9074,
+      "step": 18469
+    },
+    {
+      "epoch": 0.6717096410517511,
+      "grad_norm": 0.0,
+      "learning_rate": 5.139706597515079e-06,
+      "loss": 0.9784,
+      "step": 18470
+    },
+    {
+      "epoch": 0.6717460086554897,
+      "grad_norm": 0.0,
+      "learning_rate": 5.138677247866031e-06,
+      "loss": 0.927,
+      "step": 18471
+    },
+    {
+      "epoch": 0.6717823762592283,
+      "grad_norm": 0.0,
+      "learning_rate": 5.137647965660948e-06,
+      "loss": 0.9894,
+      "step": 18472
+    },
+    {
+      "epoch": 0.6718187438629669,
+      "grad_norm": 0.0,
+      "learning_rate": 5.136618750914096e-06,
+      "loss": 0.9223,
+      "step": 18473
+    },
+    {
+      "epoch": 0.6718551114667055,
+      "grad_norm": 0.0,
+      "learning_rate": 5.135589603639755e-06,
+      "loss": 0.9066,
+      "step": 18474
+    },
+    {
+      "epoch": 0.671891479070444,
+      "grad_norm": 0.0,
+      "learning_rate": 5.134560523852209e-06,
+      "loss": 0.8189,
+      "step": 18475
+    },
+    {
+      "epoch": 0.6719278466741826,
+      "grad_norm": 0.0,
+      "learning_rate": 5.133531511565734e-06,
+      "loss": 1.0184,
+      "step": 18476
+    },
+    {
+      "epoch": 0.6719642142779212,
+      "grad_norm": 0.0,
+      "learning_rate": 5.132502566794604e-06,
+      "loss": 0.8892,
+      "step": 18477
+    },
+    {
+      "epoch": 0.6720005818816598,
+      "grad_norm": 0.0,
+      "learning_rate": 5.131473689553094e-06,
+      "loss": 0.9719,
+      "step": 18478
+    },
+    {
+      "epoch": 0.6720369494853984,
+      "grad_norm": 0.0,
+      "learning_rate": 5.13044487985548e-06,
+      "loss": 0.9788,
+      "step": 18479
+    },
+    {
+      "epoch": 0.672073317089137,
+      "grad_norm": 0.0,
+      "learning_rate": 5.129416137716031e-06,
+      "loss": 0.9264,
+      "step": 18480
+    },
+    {
+      "epoch": 0.6721096846928756,
+      "grad_norm": 0.0,
+      "learning_rate": 5.128387463149027e-06,
+      "loss": 0.8626,
+      "step": 18481
+    },
+    {
+      "epoch": 0.6721460522966142,
+      "grad_norm": 0.0,
+      "learning_rate": 5.127358856168734e-06,
+      "loss": 0.8941,
+      "step": 18482
+    },
+    {
+      "epoch": 0.6721824199003528,
+      "grad_norm": 0.0,
+      "learning_rate": 5.126330316789423e-06,
+      "loss": 1.1304,
+      "step": 18483
+    },
+    {
+      "epoch": 0.6722187875040914,
+      "grad_norm": 0.0,
+      "learning_rate": 5.125301845025361e-06,
+      "loss": 0.8997,
+      "step": 18484
+    },
+    {
+      "epoch": 0.67225515510783,
+      "grad_norm": 0.0,
+      "learning_rate": 5.124273440890827e-06,
+      "loss": 1.0573,
+      "step": 18485
+    },
+    {
+      "epoch": 0.6722915227115686,
+      "grad_norm": 0.0,
+      "learning_rate": 5.123245104400073e-06,
+      "loss": 0.9479,
+      "step": 18486
+    },
+    {
+      "epoch": 0.6723278903153072,
+      "grad_norm": 0.0,
+      "learning_rate": 5.122216835567378e-06,
+      "loss": 0.9525,
+      "step": 18487
+    },
+    {
+      "epoch": 0.6723642579190457,
+      "grad_norm": 0.0,
+      "learning_rate": 5.121188634407005e-06,
+      "loss": 0.9962,
+      "step": 18488
+    },
+    {
+      "epoch": 0.6724006255227843,
+      "grad_norm": 0.0,
+      "learning_rate": 5.120160500933216e-06,
+      "loss": 0.8628,
+      "step": 18489
+    },
+    {
+      "epoch": 0.6724369931265229,
+      "grad_norm": 0.0,
+      "learning_rate": 5.119132435160275e-06,
+      "loss": 0.9221,
+      "step": 18490
+    },
+    {
+      "epoch": 0.6724733607302615,
+      "grad_norm": 0.0,
+      "learning_rate": 5.118104437102451e-06,
+      "loss": 0.8607,
+      "step": 18491
+    },
+    {
+      "epoch": 0.672509728334,
+      "grad_norm": 0.0,
+      "learning_rate": 5.117076506773999e-06,
+      "loss": 0.9554,
+      "step": 18492
+    },
+    {
+      "epoch": 0.6725460959377386,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1160486441891785e-06,
+      "loss": 0.9894,
+      "step": 18493
+    },
+    {
+      "epoch": 0.6725824635414772,
+      "grad_norm": 0.0,
+      "learning_rate": 5.115020849362258e-06,
+      "loss": 1.0246,
+      "step": 18494
+    },
+    {
+      "epoch": 0.6726188311452158,
+      "grad_norm": 0.0,
+      "learning_rate": 5.113993122307492e-06,
+      "loss": 0.9911,
+      "step": 18495
+    },
+    {
+      "epoch": 0.6726551987489544,
+      "grad_norm": 0.0,
+      "learning_rate": 5.112965463039139e-06,
+      "loss": 1.0266,
+      "step": 18496
+    },
+    {
+      "epoch": 0.672691566352693,
+      "grad_norm": 0.0,
+      "learning_rate": 5.111937871571453e-06,
+      "loss": 0.9784,
+      "step": 18497
+    },
+    {
+      "epoch": 0.6727279339564316,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1109103479187e-06,
+      "loss": 1.0997,
+      "step": 18498
+    },
+    {
+      "epoch": 0.6727643015601702,
+      "grad_norm": 0.0,
+      "learning_rate": 5.109882892095124e-06,
+      "loss": 0.9568,
+      "step": 18499
+    },
+    {
+      "epoch": 0.6728006691639088,
+      "grad_norm": 0.0,
+      "learning_rate": 5.108855504114987e-06,
+      "loss": 0.9315,
+      "step": 18500
+    },
+    {
+      "epoch": 0.6728370367676474,
+      "grad_norm": 0.0,
+      "learning_rate": 5.10782818399254e-06,
+      "loss": 0.8073,
+      "step": 18501
+    },
+    {
+      "epoch": 0.672873404371386,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1068009317420365e-06,
+      "loss": 0.8527,
+      "step": 18502
+    },
+    {
+      "epoch": 0.6729097719751246,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1057737473777225e-06,
+      "loss": 0.9026,
+      "step": 18503
+    },
+    {
+      "epoch": 0.6729461395788632,
+      "grad_norm": 0.0,
+      "learning_rate": 5.104746630913863e-06,
+      "loss": 0.9359,
+      "step": 18504
+    },
+    {
+      "epoch": 0.6729825071826018,
+      "grad_norm": 0.0,
+      "learning_rate": 5.10371958236469e-06,
+      "loss": 0.9708,
+      "step": 18505
+    },
+    {
+      "epoch": 0.6730188747863404,
+      "grad_norm": 0.0,
+      "learning_rate": 5.102692601744466e-06,
+      "loss": 1.0881,
+      "step": 18506
+    },
+    {
+      "epoch": 0.673055242390079,
+      "grad_norm": 0.0,
+      "learning_rate": 5.1016656890674325e-06,
+      "loss": 0.8424,
+      "step": 18507
+    },
+    {
+      "epoch": 0.6730916099938175,
+      "grad_norm": 0.0,
+      "learning_rate": 5.100638844347838e-06,
+      "loss": 0.9545,
+      "step": 18508
+    },
+    {
+      "epoch": 0.673127977597556,
+      "grad_norm": 0.0,
+      "learning_rate": 5.099612067599924e-06,
+      "loss": 0.8866,
+      "step": 18509
+    },
+    {
+      "epoch": 0.6731643452012946,
+      "grad_norm": 0.0,
+      "learning_rate": 5.098585358837949e-06,
+      "loss": 1.0054,
+      "step": 18510
+    },
+    {
+      "epoch": 0.6732007128050332,
+      "grad_norm": 0.0,
+      "learning_rate": 5.097558718076142e-06,
+      "loss": 0.8448,
+      "step": 18511
+    },
+    {
+      "epoch": 0.6732370804087718,
+      "grad_norm": 0.0,
+      "learning_rate": 5.09653214532875e-06,
+      "loss": 1.0007,
+      "step": 18512
+    },
+    {
+      "epoch": 0.6732734480125104,
+      "grad_norm": 0.0,
+      "learning_rate": 5.09550564061002e-06,
+      "loss": 0.8602,
+      "step": 18513
+    },
+    {
+      "epoch": 0.673309815616249,
+      "grad_norm": 0.0,
+      "learning_rate": 5.09447920393419e-06,
+      "loss": 0.8857,
+      "step": 18514
+    },
+    {
+      "epoch": 0.6733461832199876,
+      "grad_norm": 0.0,
+      "learning_rate": 5.093452835315503e-06,
+      "loss": 0.9662,
+      "step": 18515
+    },
+    {
+      "epoch": 0.6733825508237262,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0924265347681924e-06,
+      "loss": 0.9805,
+      "step": 18516
+    },
+    {
+      "epoch": 0.6734189184274648,
+      "grad_norm": 0.0,
+      "learning_rate": 5.091400302306509e-06,
+      "loss": 0.944,
+      "step": 18517
+    },
+    {
+      "epoch": 0.6734552860312034,
+      "grad_norm": 0.0,
+      "learning_rate": 5.090374137944671e-06,
+      "loss": 0.9592,
+      "step": 18518
+    },
+    {
+      "epoch": 0.673491653634942,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0893480416969334e-06,
+      "loss": 0.8723,
+      "step": 18519
+    },
+    {
+      "epoch": 0.6735280212386806,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0883220135775225e-06,
+      "loss": 0.8955,
+      "step": 18520
+    },
+    {
+      "epoch": 0.6735643888424192,
+      "grad_norm": 0.0,
+      "learning_rate": 5.087296053600675e-06,
+      "loss": 0.9583,
+      "step": 18521
+    },
+    {
+      "epoch": 0.6736007564461578,
+      "grad_norm": 0.0,
+      "learning_rate": 5.086270161780622e-06,
+      "loss": 0.9981,
+      "step": 18522
+    },
+    {
+      "epoch": 0.6736371240498964,
+      "grad_norm": 0.0,
+      "learning_rate": 5.085244338131605e-06,
+      "loss": 0.9879,
+      "step": 18523
+    },
+    {
+      "epoch": 0.673673491653635,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0842185826678425e-06,
+      "loss": 1.0185,
+      "step": 18524
+    },
+    {
+      "epoch": 0.6737098592573735,
+      "grad_norm": 0.0,
+      "learning_rate": 5.083192895403577e-06,
+      "loss": 0.9225,
+      "step": 18525
+    },
+    {
+      "epoch": 0.6737462268611121,
+      "grad_norm": 0.0,
+      "learning_rate": 5.082167276353033e-06,
+      "loss": 0.9604,
+      "step": 18526
+    },
+    {
+      "epoch": 0.6737825944648507,
+      "grad_norm": 0.0,
+      "learning_rate": 5.081141725530441e-06,
+      "loss": 0.9191,
+      "step": 18527
+    },
+    {
+      "epoch": 0.6738189620685893,
+      "grad_norm": 0.0,
+      "learning_rate": 5.080116242950025e-06,
+      "loss": 1.0778,
+      "step": 18528
+    },
+    {
+      "epoch": 0.6738553296723279,
+      "grad_norm": 0.0,
+      "learning_rate": 5.079090828626019e-06,
+      "loss": 0.9043,
+      "step": 18529
+    },
+    {
+      "epoch": 0.6738916972760665,
+      "grad_norm": 0.0,
+      "learning_rate": 5.078065482572651e-06,
+      "loss": 0.8684,
+      "step": 18530
+    },
+    {
+      "epoch": 0.673928064879805,
+      "grad_norm": 0.0,
+      "learning_rate": 5.077040204804132e-06,
+      "loss": 1.0085,
+      "step": 18531
+    },
+    {
+      "epoch": 0.6739644324835437,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0760149953347e-06,
+      "loss": 0.8144,
+      "step": 18532
+    },
+    {
+      "epoch": 0.6740008000872822,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0749898541785735e-06,
+      "loss": 0.8766,
+      "step": 18533
+    },
+    {
+      "epoch": 0.6740371676910208,
+      "grad_norm": 0.0,
+      "learning_rate": 5.073964781349976e-06,
+      "loss": 0.9412,
+      "step": 18534
+    },
+    {
+      "epoch": 0.6740735352947594,
+      "grad_norm": 0.0,
+      "learning_rate": 5.072939776863123e-06,
+      "loss": 0.881,
+      "step": 18535
+    },
+    {
+      "epoch": 0.674109902898498,
+      "grad_norm": 0.0,
+      "learning_rate": 5.071914840732248e-06,
+      "loss": 1.0115,
+      "step": 18536
+    },
+    {
+      "epoch": 0.6741462705022366,
+      "grad_norm": 0.0,
+      "learning_rate": 5.070889972971554e-06,
+      "loss": 1.0305,
+      "step": 18537
+    },
+    {
+      "epoch": 0.6741826381059752,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0698651735952746e-06,
+      "loss": 0.8515,
+      "step": 18538
+    },
+    {
+      "epoch": 0.6742190057097138,
+      "grad_norm": 0.0,
+      "learning_rate": 5.068840442617617e-06,
+      "loss": 1.0197,
+      "step": 18539
+    },
+    {
+      "epoch": 0.6742553733134524,
+      "grad_norm": 0.0,
+      "learning_rate": 5.067815780052805e-06,
+      "loss": 0.9117,
+      "step": 18540
+    },
+    {
+      "epoch": 0.674291740917191,
+      "grad_norm": 0.0,
+      "learning_rate": 5.066791185915046e-06,
+      "loss": 0.9562,
+      "step": 18541
+    },
+    {
+      "epoch": 0.6743281085209295,
+      "grad_norm": 0.0,
+      "learning_rate": 5.06576666021857e-06,
+      "loss": 0.9589,
+      "step": 18542
+    },
+    {
+      "epoch": 0.6743644761246681,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0647422029775704e-06,
+      "loss": 0.8417,
+      "step": 18543
+    },
+    {
+      "epoch": 0.6744008437284067,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0637178142062745e-06,
+      "loss": 0.9857,
+      "step": 18544
+    },
+    {
+      "epoch": 0.6744372113321453,
+      "grad_norm": 0.0,
+      "learning_rate": 5.062693493918891e-06,
+      "loss": 1.0214,
+      "step": 18545
+    },
+    {
+      "epoch": 0.6744735789358839,
+      "grad_norm": 0.0,
+      "learning_rate": 5.061669242129631e-06,
+      "loss": 0.827,
+      "step": 18546
+    },
+    {
+      "epoch": 0.6745099465396225,
+      "grad_norm": 0.0,
+      "learning_rate": 5.060645058852698e-06,
+      "loss": 1.0021,
+      "step": 18547
+    },
+    {
+      "epoch": 0.6745463141433611,
+      "grad_norm": 0.0,
+      "learning_rate": 5.059620944102313e-06,
+      "loss": 0.9241,
+      "step": 18548
+    },
+    {
+      "epoch": 0.6745826817470997,
+      "grad_norm": 0.0,
+      "learning_rate": 5.058596897892681e-06,
+      "loss": 0.8466,
+      "step": 18549
+    },
+    {
+      "epoch": 0.6746190493508383,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0575729202379965e-06,
+      "loss": 0.9352,
+      "step": 18550
+    },
+    {
+      "epoch": 0.6746554169545769,
+      "grad_norm": 0.0,
+      "learning_rate": 5.056549011152481e-06,
+      "loss": 0.9671,
+      "step": 18551
+    },
+    {
+      "epoch": 0.6746917845583155,
+      "grad_norm": 0.0,
+      "learning_rate": 5.055525170650333e-06,
+      "loss": 1.036,
+      "step": 18552
+    },
+    {
+      "epoch": 0.6747281521620541,
+      "grad_norm": 0.0,
+      "learning_rate": 5.054501398745759e-06,
+      "loss": 0.9998,
+      "step": 18553
+    },
+    {
+      "epoch": 0.6747645197657927,
+      "grad_norm": 0.0,
+      "learning_rate": 5.053477695452958e-06,
+      "loss": 0.9468,
+      "step": 18554
+    },
+    {
+      "epoch": 0.6748008873695313,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0524540607861416e-06,
+      "loss": 0.9608,
+      "step": 18555
+    },
+    {
+      "epoch": 0.6748372549732699,
+      "grad_norm": 0.0,
+      "learning_rate": 5.051430494759498e-06,
+      "loss": 0.9829,
+      "step": 18556
+    },
+    {
+      "epoch": 0.6748736225770084,
+      "grad_norm": 0.0,
+      "learning_rate": 5.05040699738724e-06,
+      "loss": 0.9955,
+      "step": 18557
+    },
+    {
+      "epoch": 0.674909990180747,
+      "grad_norm": 0.0,
+      "learning_rate": 5.049383568683562e-06,
+      "loss": 0.9332,
+      "step": 18558
+    },
+    {
+      "epoch": 0.6749463577844855,
+      "grad_norm": 0.0,
+      "learning_rate": 5.048360208662664e-06,
+      "loss": 0.8419,
+      "step": 18559
+    },
+    {
+      "epoch": 0.6749827253882241,
+      "grad_norm": 0.0,
+      "learning_rate": 5.047336917338738e-06,
+      "loss": 1.0316,
+      "step": 18560
+    },
+    {
+      "epoch": 0.6750190929919627,
+      "grad_norm": 0.0,
+      "learning_rate": 5.046313694725988e-06,
+      "loss": 0.9224,
+      "step": 18561
+    },
+    {
+      "epoch": 0.6750554605957013,
+      "grad_norm": 0.0,
+      "learning_rate": 5.045290540838609e-06,
+      "loss": 0.9566,
+      "step": 18562
+    },
+    {
+      "epoch": 0.6750918281994399,
+      "grad_norm": 0.0,
+      "learning_rate": 5.044267455690792e-06,
+      "loss": 0.7957,
+      "step": 18563
+    },
+    {
+      "epoch": 0.6751281958031785,
+      "grad_norm": 0.0,
+      "learning_rate": 5.043244439296735e-06,
+      "loss": 0.9507,
+      "step": 18564
+    },
+    {
+      "epoch": 0.6751645634069171,
+      "grad_norm": 0.0,
+      "learning_rate": 5.042221491670627e-06,
+      "loss": 0.9784,
+      "step": 18565
+    },
+    {
+      "epoch": 0.6752009310106557,
+      "grad_norm": 0.0,
+      "learning_rate": 5.041198612826659e-06,
+      "loss": 0.9782,
+      "step": 18566
+    },
+    {
+      "epoch": 0.6752372986143943,
+      "grad_norm": 0.0,
+      "learning_rate": 5.040175802779028e-06,
+      "loss": 1.0283,
+      "step": 18567
+    },
+    {
+      "epoch": 0.6752736662181329,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0391530615419205e-06,
+      "loss": 1.0229,
+      "step": 18568
+    },
+    {
+      "epoch": 0.6753100338218715,
+      "grad_norm": 0.0,
+      "learning_rate": 5.038130389129525e-06,
+      "loss": 1.0415,
+      "step": 18569
+    },
+    {
+      "epoch": 0.6753464014256101,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0371077855560324e-06,
+      "loss": 0.986,
+      "step": 18570
+    },
+    {
+      "epoch": 0.6753827690293487,
+      "grad_norm": 0.0,
+      "learning_rate": 5.036085250835626e-06,
+      "loss": 0.9825,
+      "step": 18571
+    },
+    {
+      "epoch": 0.6754191366330873,
+      "grad_norm": 0.0,
+      "learning_rate": 5.035062784982494e-06,
+      "loss": 0.9197,
+      "step": 18572
+    },
+    {
+      "epoch": 0.6754555042368259,
+      "grad_norm": 0.0,
+      "learning_rate": 5.034040388010819e-06,
+      "loss": 0.929,
+      "step": 18573
+    },
+    {
+      "epoch": 0.6754918718405645,
+      "grad_norm": 0.0,
+      "learning_rate": 5.033018059934791e-06,
+      "loss": 0.9146,
+      "step": 18574
+    },
+    {
+      "epoch": 0.6755282394443031,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0319958007685895e-06,
+      "loss": 0.8803,
+      "step": 18575
+    },
+    {
+      "epoch": 0.6755646070480416,
+      "grad_norm": 0.0,
+      "learning_rate": 5.030973610526398e-06,
+      "loss": 0.9587,
+      "step": 18576
+    },
+    {
+      "epoch": 0.6756009746517802,
+      "grad_norm": 0.0,
+      "learning_rate": 5.029951489222398e-06,
+      "loss": 1.097,
+      "step": 18577
+    },
+    {
+      "epoch": 0.6756373422555187,
+      "grad_norm": 0.0,
+      "learning_rate": 5.028929436870768e-06,
+      "loss": 0.8969,
+      "step": 18578
+    },
+    {
+      "epoch": 0.6756737098592573,
+      "grad_norm": 0.0,
+      "learning_rate": 5.027907453485687e-06,
+      "loss": 1.0273,
+      "step": 18579
+    },
+    {
+      "epoch": 0.6757100774629959,
+      "grad_norm": 0.0,
+      "learning_rate": 5.02688553908134e-06,
+      "loss": 1.015,
+      "step": 18580
+    },
+    {
+      "epoch": 0.6757464450667345,
+      "grad_norm": 0.0,
+      "learning_rate": 5.025863693671899e-06,
+      "loss": 0.9447,
+      "step": 18581
+    },
+    {
+      "epoch": 0.6757828126704731,
+      "grad_norm": 0.0,
+      "learning_rate": 5.024841917271541e-06,
+      "loss": 0.9246,
+      "step": 18582
+    },
+    {
+      "epoch": 0.6758191802742117,
+      "grad_norm": 0.0,
+      "learning_rate": 5.023820209894441e-06,
+      "loss": 0.9307,
+      "step": 18583
+    },
+    {
+      "epoch": 0.6758555478779503,
+      "grad_norm": 0.0,
+      "learning_rate": 5.022798571554781e-06,
+      "loss": 0.9401,
+      "step": 18584
+    },
+    {
+      "epoch": 0.6758919154816889,
+      "grad_norm": 0.0,
+      "learning_rate": 5.021777002266722e-06,
+      "loss": 0.9237,
+      "step": 18585
+    },
+    {
+      "epoch": 0.6759282830854275,
+      "grad_norm": 0.0,
+      "learning_rate": 5.020755502044447e-06,
+      "loss": 0.9763,
+      "step": 18586
+    },
+    {
+      "epoch": 0.6759646506891661,
+      "grad_norm": 0.0,
+      "learning_rate": 5.019734070902127e-06,
+      "loss": 0.9853,
+      "step": 18587
+    },
+    {
+      "epoch": 0.6760010182929047,
+      "grad_norm": 0.0,
+      "learning_rate": 5.018712708853929e-06,
+      "loss": 0.9087,
+      "step": 18588
+    },
+    {
+      "epoch": 0.6760373858966433,
+      "grad_norm": 0.0,
+      "learning_rate": 5.017691415914026e-06,
+      "loss": 0.9589,
+      "step": 18589
+    },
+    {
+      "epoch": 0.6760737535003819,
+      "grad_norm": 0.0,
+      "learning_rate": 5.016670192096585e-06,
+      "loss": 0.9168,
+      "step": 18590
+    },
+    {
+      "epoch": 0.6761101211041205,
+      "grad_norm": 0.0,
+      "learning_rate": 5.015649037415773e-06,
+      "loss": 1.0253,
+      "step": 18591
+    },
+    {
+      "epoch": 0.6761464887078591,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0146279518857556e-06,
+      "loss": 0.9689,
+      "step": 18592
+    },
+    {
+      "epoch": 0.6761828563115976,
+      "grad_norm": 0.0,
+      "learning_rate": 5.013606935520708e-06,
+      "loss": 0.7814,
+      "step": 18593
+    },
+    {
+      "epoch": 0.6762192239153362,
+      "grad_norm": 0.0,
+      "learning_rate": 5.012585988334786e-06,
+      "loss": 0.9543,
+      "step": 18594
+    },
+    {
+      "epoch": 0.6762555915190748,
+      "grad_norm": 0.0,
+      "learning_rate": 5.01156511034216e-06,
+      "loss": 1.0431,
+      "step": 18595
+    },
+    {
+      "epoch": 0.6762919591228134,
+      "grad_norm": 0.0,
+      "learning_rate": 5.010544301556985e-06,
+      "loss": 0.8902,
+      "step": 18596
+    },
+    {
+      "epoch": 0.676328326726552,
+      "grad_norm": 0.0,
+      "learning_rate": 5.009523561993437e-06,
+      "loss": 0.9276,
+      "step": 18597
+    },
+    {
+      "epoch": 0.6763646943302906,
+      "grad_norm": 0.0,
+      "learning_rate": 5.008502891665662e-06,
+      "loss": 0.9933,
+      "step": 18598
+    },
+    {
+      "epoch": 0.6764010619340292,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0074822905878305e-06,
+      "loss": 0.9697,
+      "step": 18599
+    },
+    {
+      "epoch": 0.6764374295377678,
+      "grad_norm": 0.0,
+      "learning_rate": 5.006461758774099e-06,
+      "loss": 0.9484,
+      "step": 18600
+    },
+    {
+      "epoch": 0.6764737971415063,
+      "grad_norm": 0.0,
+      "learning_rate": 5.005441296238625e-06,
+      "loss": 1.044,
+      "step": 18601
+    },
+    {
+      "epoch": 0.676510164745245,
+      "grad_norm": 0.0,
+      "learning_rate": 5.004420902995564e-06,
+      "loss": 0.7763,
+      "step": 18602
+    },
+    {
+      "epoch": 0.6765465323489835,
+      "grad_norm": 0.0,
+      "learning_rate": 5.003400579059081e-06,
+      "loss": 0.9705,
+      "step": 18603
+    },
+    {
+      "epoch": 0.6765828999527221,
+      "grad_norm": 0.0,
+      "learning_rate": 5.00238032444332e-06,
+      "loss": 0.9057,
+      "step": 18604
+    },
+    {
+      "epoch": 0.6766192675564607,
+      "grad_norm": 0.0,
+      "learning_rate": 5.001360139162444e-06,
+      "loss": 0.9494,
+      "step": 18605
+    },
+    {
+      "epoch": 0.6766556351601993,
+      "grad_norm": 0.0,
+      "learning_rate": 5.000340023230606e-06,
+      "loss": 0.9023,
+      "step": 18606
+    },
+    {
+      "epoch": 0.6766920027639379,
+      "grad_norm": 0.0,
+      "learning_rate": 4.999319976661954e-06,
+      "loss": 0.8468,
+      "step": 18607
+    },
+    {
+      "epoch": 0.6767283703676765,
+      "grad_norm": 0.0,
+      "learning_rate": 4.998299999470643e-06,
+      "loss": 1.1923,
+      "step": 18608
+    },
+    {
+      "epoch": 0.676764737971415,
+      "grad_norm": 0.0,
+      "learning_rate": 4.997280091670823e-06,
+      "loss": 0.8757,
+      "step": 18609
+    },
+    {
+      "epoch": 0.6768011055751536,
+      "grad_norm": 0.0,
+      "learning_rate": 4.996260253276643e-06,
+      "loss": 1.0313,
+      "step": 18610
+    },
+    {
+      "epoch": 0.6768374731788922,
+      "grad_norm": 0.0,
+      "learning_rate": 4.995240484302251e-06,
+      "loss": 1.0161,
+      "step": 18611
+    },
+    {
+      "epoch": 0.6768738407826308,
+      "grad_norm": 0.0,
+      "learning_rate": 4.994220784761798e-06,
+      "loss": 1.025,
+      "step": 18612
+    },
+    {
+      "epoch": 0.6769102083863694,
+      "grad_norm": 0.0,
+      "learning_rate": 4.993201154669432e-06,
+      "loss": 1.0492,
+      "step": 18613
+    },
+    {
+      "epoch": 0.676946575990108,
+      "grad_norm": 0.0,
+      "learning_rate": 4.992181594039294e-06,
+      "loss": 0.9492,
+      "step": 18614
+    },
+    {
+      "epoch": 0.6769829435938466,
+      "grad_norm": 0.0,
+      "learning_rate": 4.991162102885529e-06,
+      "loss": 0.9546,
+      "step": 18615
+    },
+    {
+      "epoch": 0.6770193111975852,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9901426812222886e-06,
+      "loss": 0.8752,
+      "step": 18616
+    },
+    {
+      "epoch": 0.6770556788013238,
+      "grad_norm": 0.0,
+      "learning_rate": 4.989123329063704e-06,
+      "loss": 0.8794,
+      "step": 18617
+    },
+    {
+      "epoch": 0.6770920464050624,
+      "grad_norm": 0.0,
+      "learning_rate": 4.988104046423928e-06,
+      "loss": 0.9311,
+      "step": 18618
+    },
+    {
+      "epoch": 0.677128414008801,
+      "grad_norm": 0.0,
+      "learning_rate": 4.987084833317097e-06,
+      "loss": 0.9925,
+      "step": 18619
+    },
+    {
+      "epoch": 0.6771647816125396,
+      "grad_norm": 0.0,
+      "learning_rate": 4.986065689757352e-06,
+      "loss": 0.8897,
+      "step": 18620
+    },
+    {
+      "epoch": 0.6772011492162782,
+      "grad_norm": 0.0,
+      "learning_rate": 4.985046615758827e-06,
+      "loss": 0.9588,
+      "step": 18621
+    },
+    {
+      "epoch": 0.6772375168200168,
+      "grad_norm": 0.0,
+      "learning_rate": 4.984027611335674e-06,
+      "loss": 0.8469,
+      "step": 18622
+    },
+    {
+      "epoch": 0.6772738844237554,
+      "grad_norm": 0.0,
+      "learning_rate": 4.983008676502011e-06,
+      "loss": 0.9772,
+      "step": 18623
+    },
+    {
+      "epoch": 0.677310252027494,
+      "grad_norm": 0.0,
+      "learning_rate": 4.98198981127199e-06,
+      "loss": 0.9046,
+      "step": 18624
+    },
+    {
+      "epoch": 0.6773466196312325,
+      "grad_norm": 0.0,
+      "learning_rate": 4.980971015659741e-06,
+      "loss": 0.8684,
+      "step": 18625
+    },
+    {
+      "epoch": 0.677382987234971,
+      "grad_norm": 0.0,
+      "learning_rate": 4.979952289679397e-06,
+      "loss": 0.9375,
+      "step": 18626
+    },
+    {
+      "epoch": 0.6774193548387096,
+      "grad_norm": 0.0,
+      "learning_rate": 4.978933633345093e-06,
+      "loss": 0.9982,
+      "step": 18627
+    },
+    {
+      "epoch": 0.6774557224424482,
+      "grad_norm": 0.0,
+      "learning_rate": 4.977915046670958e-06,
+      "loss": 0.9641,
+      "step": 18628
+    },
+    {
+      "epoch": 0.6774920900461868,
+      "grad_norm": 0.0,
+      "learning_rate": 4.976896529671133e-06,
+      "loss": 1.079,
+      "step": 18629
+    },
+    {
+      "epoch": 0.6775284576499254,
+      "grad_norm": 0.0,
+      "learning_rate": 4.975878082359734e-06,
+      "loss": 0.9504,
+      "step": 18630
+    },
+    {
+      "epoch": 0.677564825253664,
+      "grad_norm": 0.0,
+      "learning_rate": 4.974859704750903e-06,
+      "loss": 0.7716,
+      "step": 18631
+    },
+    {
+      "epoch": 0.6776011928574026,
+      "grad_norm": 0.0,
+      "learning_rate": 4.973841396858763e-06,
+      "loss": 0.7793,
+      "step": 18632
+    },
+    {
+      "epoch": 0.6776375604611412,
+      "grad_norm": 0.0,
+      "learning_rate": 4.972823158697443e-06,
+      "loss": 0.9967,
+      "step": 18633
+    },
+    {
+      "epoch": 0.6776739280648798,
+      "grad_norm": 0.0,
+      "learning_rate": 4.971804990281065e-06,
+      "loss": 0.9829,
+      "step": 18634
+    },
+    {
+      "epoch": 0.6777102956686184,
+      "grad_norm": 0.0,
+      "learning_rate": 4.970786891623766e-06,
+      "loss": 0.9178,
+      "step": 18635
+    },
+    {
+      "epoch": 0.677746663272357,
+      "grad_norm": 0.0,
+      "learning_rate": 4.969768862739656e-06,
+      "loss": 0.984,
+      "step": 18636
+    },
+    {
+      "epoch": 0.6777830308760956,
+      "grad_norm": 0.0,
+      "learning_rate": 4.968750903642869e-06,
+      "loss": 1.0997,
+      "step": 18637
+    },
+    {
+      "epoch": 0.6778193984798342,
+      "grad_norm": 0.0,
+      "learning_rate": 4.967733014347527e-06,
+      "loss": 0.9879,
+      "step": 18638
+    },
+    {
+      "epoch": 0.6778557660835728,
+      "grad_norm": 0.0,
+      "learning_rate": 4.966715194867747e-06,
+      "loss": 0.9721,
+      "step": 18639
+    },
+    {
+      "epoch": 0.6778921336873114,
+      "grad_norm": 0.0,
+      "learning_rate": 4.96569744521765e-06,
+      "loss": 0.9904,
+      "step": 18640
+    },
+    {
+      "epoch": 0.67792850129105,
+      "grad_norm": 0.0,
+      "learning_rate": 4.964679765411362e-06,
+      "loss": 1.0213,
+      "step": 18641
+    },
+    {
+      "epoch": 0.6779648688947886,
+      "grad_norm": 0.0,
+      "learning_rate": 4.963662155462998e-06,
+      "loss": 0.8381,
+      "step": 18642
+    },
+    {
+      "epoch": 0.6780012364985271,
+      "grad_norm": 0.0,
+      "learning_rate": 4.962644615386676e-06,
+      "loss": 0.9261,
+      "step": 18643
+    },
+    {
+      "epoch": 0.6780376041022657,
+      "grad_norm": 0.0,
+      "learning_rate": 4.961627145196512e-06,
+      "loss": 1.0273,
+      "step": 18644
+    },
+    {
+      "epoch": 0.6780739717060043,
+      "grad_norm": 0.0,
+      "learning_rate": 4.960609744906624e-06,
+      "loss": 0.9039,
+      "step": 18645
+    },
+    {
+      "epoch": 0.6781103393097428,
+      "grad_norm": 0.0,
+      "learning_rate": 4.959592414531126e-06,
+      "loss": 0.9852,
+      "step": 18646
+    },
+    {
+      "epoch": 0.6781467069134814,
+      "grad_norm": 0.0,
+      "learning_rate": 4.958575154084126e-06,
+      "loss": 0.9229,
+      "step": 18647
+    },
+    {
+      "epoch": 0.67818307451722,
+      "grad_norm": 0.0,
+      "learning_rate": 4.957557963579753e-06,
+      "loss": 0.8923,
+      "step": 18648
+    },
+    {
+      "epoch": 0.6782194421209586,
+      "grad_norm": 0.0,
+      "learning_rate": 4.956540843032101e-06,
+      "loss": 0.9174,
+      "step": 18649
+    },
+    {
+      "epoch": 0.6782558097246972,
+      "grad_norm": 0.0,
+      "learning_rate": 4.955523792455291e-06,
+      "loss": 0.9659,
+      "step": 18650
+    },
+    {
+      "epoch": 0.6782921773284358,
+      "grad_norm": 0.0,
+      "learning_rate": 4.954506811863433e-06,
+      "loss": 0.823,
+      "step": 18651
+    },
+    {
+      "epoch": 0.6783285449321744,
+      "grad_norm": 0.0,
+      "learning_rate": 4.953489901270634e-06,
+      "loss": 1.0042,
+      "step": 18652
+    },
+    {
+      "epoch": 0.678364912535913,
+      "grad_norm": 0.0,
+      "learning_rate": 4.952473060690997e-06,
+      "loss": 0.9498,
+      "step": 18653
+    },
+    {
+      "epoch": 0.6784012801396516,
+      "grad_norm": 0.0,
+      "learning_rate": 4.951456290138645e-06,
+      "loss": 0.9112,
+      "step": 18654
+    },
+    {
+      "epoch": 0.6784376477433902,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9504395896276645e-06,
+      "loss": 0.9946,
+      "step": 18655
+    },
+    {
+      "epoch": 0.6784740153471288,
+      "grad_norm": 0.0,
+      "learning_rate": 4.949422959172175e-06,
+      "loss": 0.8607,
+      "step": 18656
+    },
+    {
+      "epoch": 0.6785103829508674,
+      "grad_norm": 0.0,
+      "learning_rate": 4.948406398786277e-06,
+      "loss": 0.9271,
+      "step": 18657
+    },
+    {
+      "epoch": 0.678546750554606,
+      "grad_norm": 0.0,
+      "learning_rate": 4.947389908484071e-06,
+      "loss": 0.9829,
+      "step": 18658
+    },
+    {
+      "epoch": 0.6785831181583446,
+      "grad_norm": 0.0,
+      "learning_rate": 4.946373488279659e-06,
+      "loss": 1.0359,
+      "step": 18659
+    },
+    {
+      "epoch": 0.6786194857620831,
+      "grad_norm": 0.0,
+      "learning_rate": 4.945357138187148e-06,
+      "loss": 1.0115,
+      "step": 18660
+    },
+    {
+      "epoch": 0.6786558533658217,
+      "grad_norm": 0.0,
+      "learning_rate": 4.944340858220635e-06,
+      "loss": 0.9602,
+      "step": 18661
+    },
+    {
+      "epoch": 0.6786922209695603,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9433246483942195e-06,
+      "loss": 0.951,
+      "step": 18662
+    },
+    {
+      "epoch": 0.6787285885732989,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9423085087219965e-06,
+      "loss": 0.7488,
+      "step": 18663
+    },
+    {
+      "epoch": 0.6787649561770375,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9412924392180735e-06,
+      "loss": 0.8847,
+      "step": 18664
+    },
+    {
+      "epoch": 0.6788013237807761,
+      "grad_norm": 0.0,
+      "learning_rate": 4.940276439896538e-06,
+      "loss": 0.9412,
+      "step": 18665
+    },
+    {
+      "epoch": 0.6788376913845147,
+      "grad_norm": 0.0,
+      "learning_rate": 4.939260510771484e-06,
+      "loss": 1.1023,
+      "step": 18666
+    },
+    {
+      "epoch": 0.6788740589882533,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9382446518570185e-06,
+      "loss": 1.0001,
+      "step": 18667
+    },
+    {
+      "epoch": 0.6789104265919919,
+      "grad_norm": 0.0,
+      "learning_rate": 4.937228863167218e-06,
+      "loss": 0.9171,
+      "step": 18668
+    },
+    {
+      "epoch": 0.6789467941957305,
+      "grad_norm": 0.0,
+      "learning_rate": 4.936213144716189e-06,
+      "loss": 1.0699,
+      "step": 18669
+    },
+    {
+      "epoch": 0.678983161799469,
+      "grad_norm": 0.0,
+      "learning_rate": 4.935197496518016e-06,
+      "loss": 0.875,
+      "step": 18670
+    },
+    {
+      "epoch": 0.6790195294032076,
+      "grad_norm": 0.0,
+      "learning_rate": 4.934181918586793e-06,
+      "loss": 0.958,
+      "step": 18671
+    },
+    {
+      "epoch": 0.6790558970069462,
+      "grad_norm": 0.0,
+      "learning_rate": 4.933166410936605e-06,
+      "loss": 0.9322,
+      "step": 18672
+    },
+    {
+      "epoch": 0.6790922646106848,
+      "grad_norm": 0.0,
+      "learning_rate": 4.932150973581547e-06,
+      "loss": 0.8702,
+      "step": 18673
+    },
+    {
+      "epoch": 0.6791286322144234,
+      "grad_norm": 0.0,
+      "learning_rate": 4.931135606535704e-06,
+      "loss": 0.9518,
+      "step": 18674
+    },
+    {
+      "epoch": 0.679164999818162,
+      "grad_norm": 0.0,
+      "learning_rate": 4.930120309813163e-06,
+      "loss": 1.0566,
+      "step": 18675
+    },
+    {
+      "epoch": 0.6792013674219006,
+      "grad_norm": 0.0,
+      "learning_rate": 4.92910508342801e-06,
+      "loss": 0.8961,
+      "step": 18676
+    },
+    {
+      "epoch": 0.6792377350256391,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9280899273943284e-06,
+      "loss": 0.9711,
+      "step": 18677
+    },
+    {
+      "epoch": 0.6792741026293777,
+      "grad_norm": 0.0,
+      "learning_rate": 4.927074841726201e-06,
+      "loss": 1.059,
+      "step": 18678
+    },
+    {
+      "epoch": 0.6793104702331163,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9260598264377135e-06,
+      "loss": 1.0019,
+      "step": 18679
+    },
+    {
+      "epoch": 0.6793468378368549,
+      "grad_norm": 0.0,
+      "learning_rate": 4.925044881542949e-06,
+      "loss": 0.8809,
+      "step": 18680
+    },
+    {
+      "epoch": 0.6793832054405935,
+      "grad_norm": 0.0,
+      "learning_rate": 4.924030007055985e-06,
+      "loss": 0.9431,
+      "step": 18681
+    },
+    {
+      "epoch": 0.6794195730443321,
+      "grad_norm": 0.0,
+      "learning_rate": 4.923015202990901e-06,
+      "loss": 0.8835,
+      "step": 18682
+    },
+    {
+      "epoch": 0.6794559406480707,
+      "grad_norm": 0.0,
+      "learning_rate": 4.922000469361784e-06,
+      "loss": 0.9817,
+      "step": 18683
+    },
+    {
+      "epoch": 0.6794923082518093,
+      "grad_norm": 0.0,
+      "learning_rate": 4.920985806182702e-06,
+      "loss": 0.9796,
+      "step": 18684
+    },
+    {
+      "epoch": 0.6795286758555479,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9199712134677316e-06,
+      "loss": 0.8193,
+      "step": 18685
+    },
+    {
+      "epoch": 0.6795650434592865,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9189566912309575e-06,
+      "loss": 0.9425,
+      "step": 18686
+    },
+    {
+      "epoch": 0.6796014110630251,
+      "grad_norm": 0.0,
+      "learning_rate": 4.91794223948645e-06,
+      "loss": 0.9364,
+      "step": 18687
+    },
+    {
+      "epoch": 0.6796377786667637,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9169278582482835e-06,
+      "loss": 0.8742,
+      "step": 18688
+    },
+    {
+      "epoch": 0.6796741462705023,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9159135475305305e-06,
+      "loss": 1.0401,
+      "step": 18689
+    },
+    {
+      "epoch": 0.6797105138742409,
+      "grad_norm": 0.0,
+      "learning_rate": 4.914899307347264e-06,
+      "loss": 0.8396,
+      "step": 18690
+    },
+    {
+      "epoch": 0.6797468814779795,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9138851377125515e-06,
+      "loss": 0.9136,
+      "step": 18691
+    },
+    {
+      "epoch": 0.679783249081718,
+      "grad_norm": 0.0,
+      "learning_rate": 4.91287103864047e-06,
+      "loss": 0.8197,
+      "step": 18692
+    },
+    {
+      "epoch": 0.6798196166854565,
+      "grad_norm": 0.0,
+      "learning_rate": 4.911857010145085e-06,
+      "loss": 1.1158,
+      "step": 18693
+    },
+    {
+      "epoch": 0.6798559842891951,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9108430522404646e-06,
+      "loss": 0.7723,
+      "step": 18694
+    },
+    {
+      "epoch": 0.6798923518929337,
+      "grad_norm": 0.0,
+      "learning_rate": 4.909829164940673e-06,
+      "loss": 0.8507,
+      "step": 18695
+    },
+    {
+      "epoch": 0.6799287194966723,
+      "grad_norm": 0.0,
+      "learning_rate": 4.908815348259786e-06,
+      "loss": 1.0339,
+      "step": 18696
+    },
+    {
+      "epoch": 0.6799650871004109,
+      "grad_norm": 0.0,
+      "learning_rate": 4.907801602211857e-06,
+      "loss": 1.0602,
+      "step": 18697
+    },
+    {
+      "epoch": 0.6800014547041495,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9067879268109585e-06,
+      "loss": 0.8789,
+      "step": 18698
+    },
+    {
+      "epoch": 0.6800378223078881,
+      "grad_norm": 0.0,
+      "learning_rate": 4.905774322071152e-06,
+      "loss": 0.9697,
+      "step": 18699
+    },
+    {
+      "epoch": 0.6800741899116267,
+      "grad_norm": 0.0,
+      "learning_rate": 4.904760788006498e-06,
+      "loss": 0.8957,
+      "step": 18700
+    },
+    {
+      "epoch": 0.6801105575153653,
+      "grad_norm": 0.0,
+      "learning_rate": 4.903747324631055e-06,
+      "loss": 0.9688,
+      "step": 18701
+    },
+    {
+      "epoch": 0.6801469251191039,
+      "grad_norm": 0.0,
+      "learning_rate": 4.902733931958896e-06,
+      "loss": 0.9713,
+      "step": 18702
+    },
+    {
+      "epoch": 0.6801832927228425,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9017206100040675e-06,
+      "loss": 0.9609,
+      "step": 18703
+    },
+    {
+      "epoch": 0.6802196603265811,
+      "grad_norm": 0.0,
+      "learning_rate": 4.900707358780627e-06,
+      "loss": 0.9508,
+      "step": 18704
+    },
+    {
+      "epoch": 0.6802560279303197,
+      "grad_norm": 0.0,
+      "learning_rate": 4.899694178302642e-06,
+      "loss": 1.012,
+      "step": 18705
+    },
+    {
+      "epoch": 0.6802923955340583,
+      "grad_norm": 0.0,
+      "learning_rate": 4.898681068584164e-06,
+      "loss": 0.9641,
+      "step": 18706
+    },
+    {
+      "epoch": 0.6803287631377969,
+      "grad_norm": 0.0,
+      "learning_rate": 4.897668029639248e-06,
+      "loss": 1.0051,
+      "step": 18707
+    },
+    {
+      "epoch": 0.6803651307415355,
+      "grad_norm": 0.0,
+      "learning_rate": 4.896655061481949e-06,
+      "loss": 0.9024,
+      "step": 18708
+    },
+    {
+      "epoch": 0.6804014983452741,
+      "grad_norm": 0.0,
+      "learning_rate": 4.89564216412632e-06,
+      "loss": 1.0187,
+      "step": 18709
+    },
+    {
+      "epoch": 0.6804378659490126,
+      "grad_norm": 0.0,
+      "learning_rate": 4.89462933758641e-06,
+      "loss": 0.9823,
+      "step": 18710
+    },
+    {
+      "epoch": 0.6804742335527512,
+      "grad_norm": 0.0,
+      "learning_rate": 4.893616581876278e-06,
+      "loss": 0.9796,
+      "step": 18711
+    },
+    {
+      "epoch": 0.6805106011564898,
+      "grad_norm": 0.0,
+      "learning_rate": 4.892603897009971e-06,
+      "loss": 0.8795,
+      "step": 18712
+    },
+    {
+      "epoch": 0.6805469687602284,
+      "grad_norm": 0.0,
+      "learning_rate": 4.891591283001539e-06,
+      "loss": 0.9174,
+      "step": 18713
+    },
+    {
+      "epoch": 0.680583336363967,
+      "grad_norm": 0.0,
+      "learning_rate": 4.890578739865025e-06,
+      "loss": 0.983,
+      "step": 18714
+    },
+    {
+      "epoch": 0.6806197039677055,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8895662676144905e-06,
+      "loss": 0.9809,
+      "step": 18715
+    },
+    {
+      "epoch": 0.6806560715714441,
+      "grad_norm": 0.0,
+      "learning_rate": 4.888553866263965e-06,
+      "loss": 0.9562,
+      "step": 18716
+    },
+    {
+      "epoch": 0.6806924391751827,
+      "grad_norm": 0.0,
+      "learning_rate": 4.887541535827507e-06,
+      "loss": 0.7704,
+      "step": 18717
+    },
+    {
+      "epoch": 0.6807288067789213,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8865292763191565e-06,
+      "loss": 0.9768,
+      "step": 18718
+    },
+    {
+      "epoch": 0.6807651743826599,
+      "grad_norm": 0.0,
+      "learning_rate": 4.885517087752957e-06,
+      "loss": 0.9031,
+      "step": 18719
+    },
+    {
+      "epoch": 0.6808015419863985,
+      "grad_norm": 0.0,
+      "learning_rate": 4.884504970142947e-06,
+      "loss": 0.9226,
+      "step": 18720
+    },
+    {
+      "epoch": 0.6808379095901371,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8834929235031805e-06,
+      "loss": 0.9902,
+      "step": 18721
+    },
+    {
+      "epoch": 0.6808742771938757,
+      "grad_norm": 0.0,
+      "learning_rate": 4.882480947847687e-06,
+      "loss": 0.9368,
+      "step": 18722
+    },
+    {
+      "epoch": 0.6809106447976143,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8814690431905056e-06,
+      "loss": 0.9279,
+      "step": 18723
+    },
+    {
+      "epoch": 0.6809470124013529,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8804572095456826e-06,
+      "loss": 0.8513,
+      "step": 18724
+    },
+    {
+      "epoch": 0.6809833800050915,
+      "grad_norm": 0.0,
+      "learning_rate": 4.879445446927251e-06,
+      "loss": 1.103,
+      "step": 18725
+    },
+    {
+      "epoch": 0.6810197476088301,
+      "grad_norm": 0.0,
+      "learning_rate": 4.87843375534925e-06,
+      "loss": 0.8252,
+      "step": 18726
+    },
+    {
+      "epoch": 0.6810561152125686,
+      "grad_norm": 0.0,
+      "learning_rate": 4.87742213482571e-06,
+      "loss": 0.9272,
+      "step": 18727
+    },
+    {
+      "epoch": 0.6810924828163072,
+      "grad_norm": 0.0,
+      "learning_rate": 4.876410585370678e-06,
+      "loss": 0.9392,
+      "step": 18728
+    },
+    {
+      "epoch": 0.6811288504200458,
+      "grad_norm": 0.0,
+      "learning_rate": 4.875399106998172e-06,
+      "loss": 0.9808,
+      "step": 18729
+    },
+    {
+      "epoch": 0.6811652180237844,
+      "grad_norm": 0.0,
+      "learning_rate": 4.874387699722236e-06,
+      "loss": 0.9915,
+      "step": 18730
+    },
+    {
+      "epoch": 0.681201585627523,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8733763635568985e-06,
+      "loss": 0.9256,
+      "step": 18731
+    },
+    {
+      "epoch": 0.6812379532312616,
+      "grad_norm": 0.0,
+      "learning_rate": 4.87236509851619e-06,
+      "loss": 0.9164,
+      "step": 18732
+    },
+    {
+      "epoch": 0.6812743208350002,
+      "grad_norm": 0.0,
+      "learning_rate": 4.871353904614136e-06,
+      "loss": 0.9347,
+      "step": 18733
+    },
+    {
+      "epoch": 0.6813106884387388,
+      "grad_norm": 0.0,
+      "learning_rate": 4.87034278186478e-06,
+      "loss": 1.0627,
+      "step": 18734
+    },
+    {
+      "epoch": 0.6813470560424774,
+      "grad_norm": 0.0,
+      "learning_rate": 4.86933173028213e-06,
+      "loss": 1.0615,
+      "step": 18735
+    },
+    {
+      "epoch": 0.681383423646216,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8683207498802276e-06,
+      "loss": 1.0391,
+      "step": 18736
+    },
+    {
+      "epoch": 0.6814197912499546,
+      "grad_norm": 0.0,
+      "learning_rate": 4.867309840673093e-06,
+      "loss": 0.9487,
+      "step": 18737
+    },
+    {
+      "epoch": 0.6814561588536932,
+      "grad_norm": 0.0,
+      "learning_rate": 4.866299002674754e-06,
+      "loss": 1.0108,
+      "step": 18738
+    },
+    {
+      "epoch": 0.6814925264574317,
+      "grad_norm": 0.0,
+      "learning_rate": 4.865288235899227e-06,
+      "loss": 1.0748,
+      "step": 18739
+    },
+    {
+      "epoch": 0.6815288940611703,
+      "grad_norm": 0.0,
+      "learning_rate": 4.864277540360545e-06,
+      "loss": 0.9741,
+      "step": 18740
+    },
+    {
+      "epoch": 0.6815652616649089,
+      "grad_norm": 0.0,
+      "learning_rate": 4.863266916072728e-06,
+      "loss": 0.8259,
+      "step": 18741
+    },
+    {
+      "epoch": 0.6816016292686475,
+      "grad_norm": 0.0,
+      "learning_rate": 4.862256363049789e-06,
+      "loss": 0.941,
+      "step": 18742
+    },
+    {
+      "epoch": 0.6816379968723861,
+      "grad_norm": 0.0,
+      "learning_rate": 4.861245881305756e-06,
+      "loss": 0.8371,
+      "step": 18743
+    },
+    {
+      "epoch": 0.6816743644761246,
+      "grad_norm": 0.0,
+      "learning_rate": 4.860235470854645e-06,
+      "loss": 0.8762,
+      "step": 18744
+    },
+    {
+      "epoch": 0.6817107320798632,
+      "grad_norm": 0.0,
+      "learning_rate": 4.859225131710476e-06,
+      "loss": 0.8478,
+      "step": 18745
+    },
+    {
+      "epoch": 0.6817470996836018,
+      "grad_norm": 0.0,
+      "learning_rate": 4.858214863887259e-06,
+      "loss": 0.8734,
+      "step": 18746
+    },
+    {
+      "epoch": 0.6817834672873404,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8572046673990245e-06,
+      "loss": 0.9217,
+      "step": 18747
+    },
+    {
+      "epoch": 0.681819834891079,
+      "grad_norm": 0.0,
+      "learning_rate": 4.85619454225977e-06,
+      "loss": 0.937,
+      "step": 18748
+    },
+    {
+      "epoch": 0.6818562024948176,
+      "grad_norm": 0.0,
+      "learning_rate": 4.855184488483523e-06,
+      "loss": 0.9089,
+      "step": 18749
+    },
+    {
+      "epoch": 0.6818925700985562,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8541745060842905e-06,
+      "loss": 0.929,
+      "step": 18750
+    },
+    {
+      "epoch": 0.6819289377022948,
+      "grad_norm": 0.0,
+      "learning_rate": 4.853164595076086e-06,
+      "loss": 1.0016,
+      "step": 18751
+    },
+    {
+      "epoch": 0.6819653053060334,
+      "grad_norm": 0.0,
+      "learning_rate": 4.852154755472917e-06,
+      "loss": 1.0219,
+      "step": 18752
+    },
+    {
+      "epoch": 0.682001672909772,
+      "grad_norm": 0.0,
+      "learning_rate": 4.851144987288805e-06,
+      "loss": 0.9713,
+      "step": 18753
+    },
+    {
+      "epoch": 0.6820380405135106,
+      "grad_norm": 0.0,
+      "learning_rate": 4.850135290537742e-06,
+      "loss": 1.0671,
+      "step": 18754
+    },
+    {
+      "epoch": 0.6820744081172492,
+      "grad_norm": 0.0,
+      "learning_rate": 4.849125665233751e-06,
+      "loss": 0.9463,
+      "step": 18755
+    },
+    {
+      "epoch": 0.6821107757209878,
+      "grad_norm": 0.0,
+      "learning_rate": 4.848116111390832e-06,
+      "loss": 0.9272,
+      "step": 18756
+    },
+    {
+      "epoch": 0.6821471433247264,
+      "grad_norm": 0.0,
+      "learning_rate": 4.847106629022992e-06,
+      "loss": 0.9503,
+      "step": 18757
+    },
+    {
+      "epoch": 0.682183510928465,
+      "grad_norm": 0.0,
+      "learning_rate": 4.846097218144233e-06,
+      "loss": 0.9627,
+      "step": 18758
+    },
+    {
+      "epoch": 0.6822198785322036,
+      "grad_norm": 0.0,
+      "learning_rate": 4.845087878768566e-06,
+      "loss": 0.9514,
+      "step": 18759
+    },
+    {
+      "epoch": 0.6822562461359422,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8440786109099955e-06,
+      "loss": 0.9537,
+      "step": 18760
+    },
+    {
+      "epoch": 0.6822926137396806,
+      "grad_norm": 0.0,
+      "learning_rate": 4.843069414582511e-06,
+      "loss": 0.9462,
+      "step": 18761
+    },
+    {
+      "epoch": 0.6823289813434192,
+      "grad_norm": 0.0,
+      "learning_rate": 4.842060289800126e-06,
+      "loss": 0.8401,
+      "step": 18762
+    },
+    {
+      "epoch": 0.6823653489471578,
+      "grad_norm": 0.0,
+      "learning_rate": 4.841051236576834e-06,
+      "loss": 0.9181,
+      "step": 18763
+    },
+    {
+      "epoch": 0.6824017165508964,
+      "grad_norm": 0.0,
+      "learning_rate": 4.840042254926639e-06,
+      "loss": 1.0169,
+      "step": 18764
+    },
+    {
+      "epoch": 0.682438084154635,
+      "grad_norm": 0.0,
+      "learning_rate": 4.839033344863533e-06,
+      "loss": 1.103,
+      "step": 18765
+    },
+    {
+      "epoch": 0.6824744517583736,
+      "grad_norm": 0.0,
+      "learning_rate": 4.838024506401523e-06,
+      "loss": 0.8895,
+      "step": 18766
+    },
+    {
+      "epoch": 0.6825108193621122,
+      "grad_norm": 0.0,
+      "learning_rate": 4.837015739554592e-06,
+      "loss": 0.9261,
+      "step": 18767
+    },
+    {
+      "epoch": 0.6825471869658508,
+      "grad_norm": 0.0,
+      "learning_rate": 4.836007044336747e-06,
+      "loss": 0.9635,
+      "step": 18768
+    },
+    {
+      "epoch": 0.6825835545695894,
+      "grad_norm": 0.0,
+      "learning_rate": 4.834998420761976e-06,
+      "loss": 0.9914,
+      "step": 18769
+    },
+    {
+      "epoch": 0.682619922173328,
+      "grad_norm": 0.0,
+      "learning_rate": 4.833989868844276e-06,
+      "loss": 1.0419,
+      "step": 18770
+    },
+    {
+      "epoch": 0.6826562897770666,
+      "grad_norm": 0.0,
+      "learning_rate": 4.832981388597631e-06,
+      "loss": 0.905,
+      "step": 18771
+    },
+    {
+      "epoch": 0.6826926573808052,
+      "grad_norm": 0.0,
+      "learning_rate": 4.831972980036042e-06,
+      "loss": 0.8594,
+      "step": 18772
+    },
+    {
+      "epoch": 0.6827290249845438,
+      "grad_norm": 0.0,
+      "learning_rate": 4.830964643173497e-06,
+      "loss": 0.9077,
+      "step": 18773
+    },
+    {
+      "epoch": 0.6827653925882824,
+      "grad_norm": 0.0,
+      "learning_rate": 4.829956378023982e-06,
+      "loss": 1.0235,
+      "step": 18774
+    },
+    {
+      "epoch": 0.682801760192021,
+      "grad_norm": 0.0,
+      "learning_rate": 4.828948184601488e-06,
+      "loss": 0.929,
+      "step": 18775
+    },
+    {
+      "epoch": 0.6828381277957596,
+      "grad_norm": 0.0,
+      "learning_rate": 4.827940062920001e-06,
+      "loss": 1.0663,
+      "step": 18776
+    },
+    {
+      "epoch": 0.6828744953994982,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8269320129935036e-06,
+      "loss": 0.9642,
+      "step": 18777
+    },
+    {
+      "epoch": 0.6829108630032367,
+      "grad_norm": 0.0,
+      "learning_rate": 4.825924034835988e-06,
+      "loss": 0.934,
+      "step": 18778
+    },
+    {
+      "epoch": 0.6829472306069753,
+      "grad_norm": 0.0,
+      "learning_rate": 4.82491612846144e-06,
+      "loss": 0.8816,
+      "step": 18779
+    },
+    {
+      "epoch": 0.6829835982107139,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8239082938838314e-06,
+      "loss": 0.9421,
+      "step": 18780
+    },
+    {
+      "epoch": 0.6830199658144525,
+      "grad_norm": 0.0,
+      "learning_rate": 4.822900531117153e-06,
+      "loss": 1.0018,
+      "step": 18781
+    },
+    {
+      "epoch": 0.683056333418191,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8218928401753864e-06,
+      "loss": 0.8484,
+      "step": 18782
+    },
+    {
+      "epoch": 0.6830927010219296,
+      "grad_norm": 0.0,
+      "learning_rate": 4.820885221072509e-06,
+      "loss": 0.9373,
+      "step": 18783
+    },
+    {
+      "epoch": 0.6831290686256682,
+      "grad_norm": 0.0,
+      "learning_rate": 4.819877673822496e-06,
+      "loss": 0.9378,
+      "step": 18784
+    },
+    {
+      "epoch": 0.6831654362294068,
+      "grad_norm": 0.0,
+      "learning_rate": 4.818870198439336e-06,
+      "loss": 1.0765,
+      "step": 18785
+    },
+    {
+      "epoch": 0.6832018038331454,
+      "grad_norm": 0.0,
+      "learning_rate": 4.817862794937e-06,
+      "loss": 0.9904,
+      "step": 18786
+    },
+    {
+      "epoch": 0.683238171436884,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8168554633294665e-06,
+      "loss": 1.0297,
+      "step": 18787
+    },
+    {
+      "epoch": 0.6832745390406226,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8158482036307085e-06,
+      "loss": 0.9681,
+      "step": 18788
+    },
+    {
+      "epoch": 0.6833109066443612,
+      "grad_norm": 0.0,
+      "learning_rate": 4.814841015854702e-06,
+      "loss": 0.9953,
+      "step": 18789
+    },
+    {
+      "epoch": 0.6833472742480998,
+      "grad_norm": 0.0,
+      "learning_rate": 4.813833900015415e-06,
+      "loss": 1.0016,
+      "step": 18790
+    },
+    {
+      "epoch": 0.6833836418518384,
+      "grad_norm": 0.0,
+      "learning_rate": 4.812826856126829e-06,
+      "loss": 0.9893,
+      "step": 18791
+    },
+    {
+      "epoch": 0.683420009455577,
+      "grad_norm": 0.0,
+      "learning_rate": 4.811819884202909e-06,
+      "loss": 0.9306,
+      "step": 18792
+    },
+    {
+      "epoch": 0.6834563770593156,
+      "grad_norm": 0.0,
+      "learning_rate": 4.810812984257629e-06,
+      "loss": 0.9587,
+      "step": 18793
+    },
+    {
+      "epoch": 0.6834927446630541,
+      "grad_norm": 0.0,
+      "learning_rate": 4.809806156304952e-06,
+      "loss": 0.9629,
+      "step": 18794
+    },
+    {
+      "epoch": 0.6835291122667927,
+      "grad_norm": 0.0,
+      "learning_rate": 4.808799400358859e-06,
+      "loss": 0.9752,
+      "step": 18795
+    },
+    {
+      "epoch": 0.6835654798705313,
+      "grad_norm": 0.0,
+      "learning_rate": 4.807792716433301e-06,
+      "loss": 0.8969,
+      "step": 18796
+    },
+    {
+      "epoch": 0.6836018474742699,
+      "grad_norm": 0.0,
+      "learning_rate": 4.806786104542255e-06,
+      "loss": 0.9244,
+      "step": 18797
+    },
+    {
+      "epoch": 0.6836382150780085,
+      "grad_norm": 0.0,
+      "learning_rate": 4.805779564699687e-06,
+      "loss": 0.9789,
+      "step": 18798
+    },
+    {
+      "epoch": 0.6836745826817471,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8047730969195525e-06,
+      "loss": 0.9841,
+      "step": 18799
+    },
+    {
+      "epoch": 0.6837109502854857,
+      "grad_norm": 0.0,
+      "learning_rate": 4.803766701215823e-06,
+      "loss": 0.8596,
+      "step": 18800
+    },
+    {
+      "epoch": 0.6837473178892243,
+      "grad_norm": 0.0,
+      "learning_rate": 4.802760377602458e-06,
+      "loss": 0.8587,
+      "step": 18801
+    },
+    {
+      "epoch": 0.6837836854929629,
+      "grad_norm": 0.0,
+      "learning_rate": 4.801754126093419e-06,
+      "loss": 0.9168,
+      "step": 18802
+    },
+    {
+      "epoch": 0.6838200530967015,
+      "grad_norm": 0.0,
+      "learning_rate": 4.800747946702662e-06,
+      "loss": 0.9445,
+      "step": 18803
+    },
+    {
+      "epoch": 0.6838564207004401,
+      "grad_norm": 0.0,
+      "learning_rate": 4.799741839444154e-06,
+      "loss": 0.8124,
+      "step": 18804
+    },
+    {
+      "epoch": 0.6838927883041787,
+      "grad_norm": 0.0,
+      "learning_rate": 4.79873580433185e-06,
+      "loss": 0.862,
+      "step": 18805
+    },
+    {
+      "epoch": 0.6839291559079173,
+      "grad_norm": 0.0,
+      "learning_rate": 4.797729841379707e-06,
+      "loss": 1.0119,
+      "step": 18806
+    },
+    {
+      "epoch": 0.6839655235116558,
+      "grad_norm": 0.0,
+      "learning_rate": 4.796723950601678e-06,
+      "loss": 0.9215,
+      "step": 18807
+    },
+    {
+      "epoch": 0.6840018911153944,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7957181320117295e-06,
+      "loss": 0.8896,
+      "step": 18808
+    },
+    {
+      "epoch": 0.684038258719133,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7947123856237995e-06,
+      "loss": 0.9719,
+      "step": 18809
+    },
+    {
+      "epoch": 0.6840746263228716,
+      "grad_norm": 0.0,
+      "learning_rate": 4.793706711451855e-06,
+      "loss": 0.9049,
+      "step": 18810
+    },
+    {
+      "epoch": 0.6841109939266101,
+      "grad_norm": 0.0,
+      "learning_rate": 4.792701109509842e-06,
+      "loss": 0.8594,
+      "step": 18811
+    },
+    {
+      "epoch": 0.6841473615303487,
+      "grad_norm": 0.0,
+      "learning_rate": 4.791695579811713e-06,
+      "loss": 1.0306,
+      "step": 18812
+    },
+    {
+      "epoch": 0.6841837291340873,
+      "grad_norm": 0.0,
+      "learning_rate": 4.790690122371414e-06,
+      "loss": 0.9604,
+      "step": 18813
+    },
+    {
+      "epoch": 0.6842200967378259,
+      "grad_norm": 0.0,
+      "learning_rate": 4.789684737202908e-06,
+      "loss": 1.1068,
+      "step": 18814
+    },
+    {
+      "epoch": 0.6842564643415645,
+      "grad_norm": 0.0,
+      "learning_rate": 4.788679424320124e-06,
+      "loss": 0.9221,
+      "step": 18815
+    },
+    {
+      "epoch": 0.6842928319453031,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7876741837370245e-06,
+      "loss": 0.8621,
+      "step": 18816
+    },
+    {
+      "epoch": 0.6843291995490417,
+      "grad_norm": 0.0,
+      "learning_rate": 4.786669015467549e-06,
+      "loss": 0.9634,
+      "step": 18817
+    },
+    {
+      "epoch": 0.6843655671527803,
+      "grad_norm": 0.0,
+      "learning_rate": 4.785663919525645e-06,
+      "loss": 0.9074,
+      "step": 18818
+    },
+    {
+      "epoch": 0.6844019347565189,
+      "grad_norm": 0.0,
+      "learning_rate": 4.784658895925256e-06,
+      "loss": 0.9055,
+      "step": 18819
+    },
+    {
+      "epoch": 0.6844383023602575,
+      "grad_norm": 0.0,
+      "learning_rate": 4.783653944680324e-06,
+      "loss": 0.9391,
+      "step": 18820
+    },
+    {
+      "epoch": 0.6844746699639961,
+      "grad_norm": 0.0,
+      "learning_rate": 4.782649065804794e-06,
+      "loss": 1.0194,
+      "step": 18821
+    },
+    {
+      "epoch": 0.6845110375677347,
+      "grad_norm": 0.0,
+      "learning_rate": 4.781644259312602e-06,
+      "loss": 0.9891,
+      "step": 18822
+    },
+    {
+      "epoch": 0.6845474051714733,
+      "grad_norm": 0.0,
+      "learning_rate": 4.780639525217695e-06,
+      "loss": 0.8978,
+      "step": 18823
+    },
+    {
+      "epoch": 0.6845837727752119,
+      "grad_norm": 0.0,
+      "learning_rate": 4.779634863534009e-06,
+      "loss": 0.9648,
+      "step": 18824
+    },
+    {
+      "epoch": 0.6846201403789505,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7786302742754845e-06,
+      "loss": 0.94,
+      "step": 18825
+    },
+    {
+      "epoch": 0.6846565079826891,
+      "grad_norm": 0.0,
+      "learning_rate": 4.77762575745605e-06,
+      "loss": 1.1192,
+      "step": 18826
+    },
+    {
+      "epoch": 0.6846928755864277,
+      "grad_norm": 0.0,
+      "learning_rate": 4.776621313089659e-06,
+      "loss": 0.8984,
+      "step": 18827
+    },
+    {
+      "epoch": 0.6847292431901661,
+      "grad_norm": 0.0,
+      "learning_rate": 4.775616941190227e-06,
+      "loss": 1.0001,
+      "step": 18828
+    },
+    {
+      "epoch": 0.6847656107939047,
+      "grad_norm": 0.0,
+      "learning_rate": 4.774612641771702e-06,
+      "loss": 0.8637,
+      "step": 18829
+    },
+    {
+      "epoch": 0.6848019783976433,
+      "grad_norm": 0.0,
+      "learning_rate": 4.773608414848012e-06,
+      "loss": 0.9075,
+      "step": 18830
+    },
+    {
+      "epoch": 0.6848383460013819,
+      "grad_norm": 0.0,
+      "learning_rate": 4.77260426043309e-06,
+      "loss": 0.8389,
+      "step": 18831
+    },
+    {
+      "epoch": 0.6848747136051205,
+      "grad_norm": 0.0,
+      "learning_rate": 4.771600178540864e-06,
+      "loss": 1.0799,
+      "step": 18832
+    },
+    {
+      "epoch": 0.6849110812088591,
+      "grad_norm": 0.0,
+      "learning_rate": 4.770596169185275e-06,
+      "loss": 0.8747,
+      "step": 18833
+    },
+    {
+      "epoch": 0.6849474488125977,
+      "grad_norm": 0.0,
+      "learning_rate": 4.769592232380237e-06,
+      "loss": 0.96,
+      "step": 18834
+    },
+    {
+      "epoch": 0.6849838164163363,
+      "grad_norm": 0.0,
+      "learning_rate": 4.768588368139689e-06,
+      "loss": 0.8412,
+      "step": 18835
+    },
+    {
+      "epoch": 0.6850201840200749,
+      "grad_norm": 0.0,
+      "learning_rate": 4.767584576477555e-06,
+      "loss": 0.9768,
+      "step": 18836
+    },
+    {
+      "epoch": 0.6850565516238135,
+      "grad_norm": 0.0,
+      "learning_rate": 4.766580857407762e-06,
+      "loss": 0.9823,
+      "step": 18837
+    },
+    {
+      "epoch": 0.6850929192275521,
+      "grad_norm": 0.0,
+      "learning_rate": 4.765577210944234e-06,
+      "loss": 0.9101,
+      "step": 18838
+    },
+    {
+      "epoch": 0.6851292868312907,
+      "grad_norm": 0.0,
+      "learning_rate": 4.764573637100892e-06,
+      "loss": 0.8923,
+      "step": 18839
+    },
+    {
+      "epoch": 0.6851656544350293,
+      "grad_norm": 0.0,
+      "learning_rate": 4.76357013589167e-06,
+      "loss": 0.843,
+      "step": 18840
+    },
+    {
+      "epoch": 0.6852020220387679,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7625667073304745e-06,
+      "loss": 1.0471,
+      "step": 18841
+    },
+    {
+      "epoch": 0.6852383896425065,
+      "grad_norm": 0.0,
+      "learning_rate": 4.76156335143124e-06,
+      "loss": 0.955,
+      "step": 18842
+    },
+    {
+      "epoch": 0.6852747572462451,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7605600682078816e-06,
+      "loss": 0.9279,
+      "step": 18843
+    },
+    {
+      "epoch": 0.6853111248499837,
+      "grad_norm": 0.0,
+      "learning_rate": 4.759556857674317e-06,
+      "loss": 0.9425,
+      "step": 18844
+    },
+    {
+      "epoch": 0.6853474924537222,
+      "grad_norm": 0.0,
+      "learning_rate": 4.758553719844463e-06,
+      "loss": 0.9249,
+      "step": 18845
+    },
+    {
+      "epoch": 0.6853838600574608,
+      "grad_norm": 0.0,
+      "learning_rate": 4.757550654732247e-06,
+      "loss": 0.974,
+      "step": 18846
+    },
+    {
+      "epoch": 0.6854202276611994,
+      "grad_norm": 0.0,
+      "learning_rate": 4.756547662351571e-06,
+      "loss": 0.9979,
+      "step": 18847
+    },
+    {
+      "epoch": 0.685456595264938,
+      "grad_norm": 0.0,
+      "learning_rate": 4.755544742716358e-06,
+      "loss": 0.8906,
+      "step": 18848
+    },
+    {
+      "epoch": 0.6854929628686766,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7545418958405235e-06,
+      "loss": 1.0299,
+      "step": 18849
+    },
+    {
+      "epoch": 0.6855293304724152,
+      "grad_norm": 0.0,
+      "learning_rate": 4.753539121737976e-06,
+      "loss": 0.9552,
+      "step": 18850
+    },
+    {
+      "epoch": 0.6855656980761538,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7525364204226265e-06,
+      "loss": 0.9417,
+      "step": 18851
+    },
+    {
+      "epoch": 0.6856020656798923,
+      "grad_norm": 0.0,
+      "learning_rate": 4.751533791908396e-06,
+      "loss": 0.9844,
+      "step": 18852
+    },
+    {
+      "epoch": 0.6856384332836309,
+      "grad_norm": 0.0,
+      "learning_rate": 4.750531236209181e-06,
+      "loss": 1.0951,
+      "step": 18853
+    },
+    {
+      "epoch": 0.6856748008873695,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7495287533389e-06,
+      "loss": 0.9664,
+      "step": 18854
+    },
+    {
+      "epoch": 0.6857111684911081,
+      "grad_norm": 0.0,
+      "learning_rate": 4.748526343311458e-06,
+      "loss": 0.9951,
+      "step": 18855
+    },
+    {
+      "epoch": 0.6857475360948467,
+      "grad_norm": 0.0,
+      "learning_rate": 4.747524006140762e-06,
+      "loss": 0.8639,
+      "step": 18856
+    },
+    {
+      "epoch": 0.6857839036985853,
+      "grad_norm": 0.0,
+      "learning_rate": 4.746521741840719e-06,
+      "loss": 0.959,
+      "step": 18857
+    },
+    {
+      "epoch": 0.6858202713023239,
+      "grad_norm": 0.0,
+      "learning_rate": 4.74551955042523e-06,
+      "loss": 1.002,
+      "step": 18858
+    },
+    {
+      "epoch": 0.6858566389060625,
+      "grad_norm": 0.0,
+      "learning_rate": 4.744517431908208e-06,
+      "loss": 0.996,
+      "step": 18859
+    },
+    {
+      "epoch": 0.6858930065098011,
+      "grad_norm": 0.0,
+      "learning_rate": 4.743515386303542e-06,
+      "loss": 1.0282,
+      "step": 18860
+    },
+    {
+      "epoch": 0.6859293741135397,
+      "grad_norm": 0.0,
+      "learning_rate": 4.742513413625147e-06,
+      "loss": 0.9272,
+      "step": 18861
+    },
+    {
+      "epoch": 0.6859657417172782,
+      "grad_norm": 0.0,
+      "learning_rate": 4.74151151388692e-06,
+      "loss": 0.9555,
+      "step": 18862
+    },
+    {
+      "epoch": 0.6860021093210168,
+      "grad_norm": 0.0,
+      "learning_rate": 4.740509687102758e-06,
+      "loss": 1.0286,
+      "step": 18863
+    },
+    {
+      "epoch": 0.6860384769247554,
+      "grad_norm": 0.0,
+      "learning_rate": 4.739507933286559e-06,
+      "loss": 0.8935,
+      "step": 18864
+    },
+    {
+      "epoch": 0.686074844528494,
+      "grad_norm": 0.0,
+      "learning_rate": 4.738506252452231e-06,
+      "loss": 0.9405,
+      "step": 18865
+    },
+    {
+      "epoch": 0.6861112121322326,
+      "grad_norm": 0.0,
+      "learning_rate": 4.737504644613656e-06,
+      "loss": 0.9828,
+      "step": 18866
+    },
+    {
+      "epoch": 0.6861475797359712,
+      "grad_norm": 0.0,
+      "learning_rate": 4.736503109784743e-06,
+      "loss": 0.8604,
+      "step": 18867
+    },
+    {
+      "epoch": 0.6861839473397098,
+      "grad_norm": 0.0,
+      "learning_rate": 4.735501647979379e-06,
+      "loss": 0.9125,
+      "step": 18868
+    },
+    {
+      "epoch": 0.6862203149434484,
+      "grad_norm": 0.0,
+      "learning_rate": 4.734500259211462e-06,
+      "loss": 0.998,
+      "step": 18869
+    },
+    {
+      "epoch": 0.686256682547187,
+      "grad_norm": 0.0,
+      "learning_rate": 4.733498943494879e-06,
+      "loss": 1.032,
+      "step": 18870
+    },
+    {
+      "epoch": 0.6862930501509256,
+      "grad_norm": 0.0,
+      "learning_rate": 4.732497700843529e-06,
+      "loss": 1.0994,
+      "step": 18871
+    },
+    {
+      "epoch": 0.6863294177546642,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7314965312713e-06,
+      "loss": 0.9709,
+      "step": 18872
+    },
+    {
+      "epoch": 0.6863657853584028,
+      "grad_norm": 0.0,
+      "learning_rate": 4.730495434792082e-06,
+      "loss": 1.0135,
+      "step": 18873
+    },
+    {
+      "epoch": 0.6864021529621414,
+      "grad_norm": 0.0,
+      "learning_rate": 4.729494411419763e-06,
+      "loss": 1.038,
+      "step": 18874
+    },
+    {
+      "epoch": 0.68643852056588,
+      "grad_norm": 0.0,
+      "learning_rate": 4.728493461168232e-06,
+      "loss": 0.9858,
+      "step": 18875
+    },
+    {
+      "epoch": 0.6864748881696185,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7274925840513745e-06,
+      "loss": 0.9422,
+      "step": 18876
+    },
+    {
+      "epoch": 0.6865112557733571,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7264917800830726e-06,
+      "loss": 0.9969,
+      "step": 18877
+    },
+    {
+      "epoch": 0.6865476233770956,
+      "grad_norm": 0.0,
+      "learning_rate": 4.725491049277222e-06,
+      "loss": 1.0303,
+      "step": 18878
+    },
+    {
+      "epoch": 0.6865839909808342,
+      "grad_norm": 0.0,
+      "learning_rate": 4.724490391647694e-06,
+      "loss": 0.8011,
+      "step": 18879
+    },
+    {
+      "epoch": 0.6866203585845728,
+      "grad_norm": 0.0,
+      "learning_rate": 4.723489807208379e-06,
+      "loss": 1.0148,
+      "step": 18880
+    },
+    {
+      "epoch": 0.6866567261883114,
+      "grad_norm": 0.0,
+      "learning_rate": 4.722489295973156e-06,
+      "loss": 0.9991,
+      "step": 18881
+    },
+    {
+      "epoch": 0.68669309379205,
+      "grad_norm": 0.0,
+      "learning_rate": 4.721488857955908e-06,
+      "loss": 0.9136,
+      "step": 18882
+    },
+    {
+      "epoch": 0.6867294613957886,
+      "grad_norm": 0.0,
+      "learning_rate": 4.720488493170509e-06,
+      "loss": 0.9756,
+      "step": 18883
+    },
+    {
+      "epoch": 0.6867658289995272,
+      "grad_norm": 0.0,
+      "learning_rate": 4.719488201630844e-06,
+      "loss": 0.9534,
+      "step": 18884
+    },
+    {
+      "epoch": 0.6868021966032658,
+      "grad_norm": 0.0,
+      "learning_rate": 4.718487983350788e-06,
+      "loss": 1.0129,
+      "step": 18885
+    },
+    {
+      "epoch": 0.6868385642070044,
+      "grad_norm": 0.0,
+      "learning_rate": 4.717487838344219e-06,
+      "loss": 1.0732,
+      "step": 18886
+    },
+    {
+      "epoch": 0.686874931810743,
+      "grad_norm": 0.0,
+      "learning_rate": 4.716487766625011e-06,
+      "loss": 0.9749,
+      "step": 18887
+    },
+    {
+      "epoch": 0.6869112994144816,
+      "grad_norm": 0.0,
+      "learning_rate": 4.715487768207039e-06,
+      "loss": 1.0089,
+      "step": 18888
+    },
+    {
+      "epoch": 0.6869476670182202,
+      "grad_norm": 0.0,
+      "learning_rate": 4.714487843104172e-06,
+      "loss": 0.9704,
+      "step": 18889
+    },
+    {
+      "epoch": 0.6869840346219588,
+      "grad_norm": 0.0,
+      "learning_rate": 4.713487991330291e-06,
+      "loss": 0.9237,
+      "step": 18890
+    },
+    {
+      "epoch": 0.6870204022256974,
+      "grad_norm": 0.0,
+      "learning_rate": 4.712488212899263e-06,
+      "loss": 0.965,
+      "step": 18891
+    },
+    {
+      "epoch": 0.687056769829436,
+      "grad_norm": 0.0,
+      "learning_rate": 4.711488507824961e-06,
+      "loss": 0.903,
+      "step": 18892
+    },
+    {
+      "epoch": 0.6870931374331746,
+      "grad_norm": 0.0,
+      "learning_rate": 4.710488876121248e-06,
+      "loss": 0.9636,
+      "step": 18893
+    },
+    {
+      "epoch": 0.6871295050369132,
+      "grad_norm": 0.0,
+      "learning_rate": 4.709489317802003e-06,
+      "loss": 0.9786,
+      "step": 18894
+    },
+    {
+      "epoch": 0.6871658726406517,
+      "grad_norm": 0.0,
+      "learning_rate": 4.708489832881085e-06,
+      "loss": 0.8233,
+      "step": 18895
+    },
+    {
+      "epoch": 0.6872022402443903,
+      "grad_norm": 0.0,
+      "learning_rate": 4.707490421372359e-06,
+      "loss": 1.0048,
+      "step": 18896
+    },
+    {
+      "epoch": 0.6872386078481288,
+      "grad_norm": 0.0,
+      "learning_rate": 4.706491083289702e-06,
+      "loss": 0.9039,
+      "step": 18897
+    },
+    {
+      "epoch": 0.6872749754518674,
+      "grad_norm": 0.0,
+      "learning_rate": 4.705491818646962e-06,
+      "loss": 0.962,
+      "step": 18898
+    },
+    {
+      "epoch": 0.687311343055606,
+      "grad_norm": 0.0,
+      "learning_rate": 4.704492627458016e-06,
+      "loss": 1.0305,
+      "step": 18899
+    },
+    {
+      "epoch": 0.6873477106593446,
+      "grad_norm": 0.0,
+      "learning_rate": 4.703493509736722e-06,
+      "loss": 0.9232,
+      "step": 18900
+    },
+    {
+      "epoch": 0.6873840782630832,
+      "grad_norm": 0.0,
+      "learning_rate": 4.70249446549694e-06,
+      "loss": 0.8877,
+      "step": 18901
+    },
+    {
+      "epoch": 0.6874204458668218,
+      "grad_norm": 0.0,
+      "learning_rate": 4.701495494752527e-06,
+      "loss": 1.0007,
+      "step": 18902
+    },
+    {
+      "epoch": 0.6874568134705604,
+      "grad_norm": 0.0,
+      "learning_rate": 4.700496597517351e-06,
+      "loss": 0.9717,
+      "step": 18903
+    },
+    {
+      "epoch": 0.687493181074299,
+      "grad_norm": 0.0,
+      "learning_rate": 4.699497773805265e-06,
+      "loss": 0.9712,
+      "step": 18904
+    },
+    {
+      "epoch": 0.6875295486780376,
+      "grad_norm": 0.0,
+      "learning_rate": 4.698499023630128e-06,
+      "loss": 0.8464,
+      "step": 18905
+    },
+    {
+      "epoch": 0.6875659162817762,
+      "grad_norm": 0.0,
+      "learning_rate": 4.69750034700579e-06,
+      "loss": 0.948,
+      "step": 18906
+    },
+    {
+      "epoch": 0.6876022838855148,
+      "grad_norm": 0.0,
+      "learning_rate": 4.696501743946119e-06,
+      "loss": 0.9034,
+      "step": 18907
+    },
+    {
+      "epoch": 0.6876386514892534,
+      "grad_norm": 0.0,
+      "learning_rate": 4.695503214464955e-06,
+      "loss": 0.8853,
+      "step": 18908
+    },
+    {
+      "epoch": 0.687675019092992,
+      "grad_norm": 0.0,
+      "learning_rate": 4.694504758576161e-06,
+      "loss": 0.9235,
+      "step": 18909
+    },
+    {
+      "epoch": 0.6877113866967306,
+      "grad_norm": 0.0,
+      "learning_rate": 4.693506376293587e-06,
+      "loss": 0.9941,
+      "step": 18910
+    },
+    {
+      "epoch": 0.6877477543004692,
+      "grad_norm": 0.0,
+      "learning_rate": 4.692508067631082e-06,
+      "loss": 0.9544,
+      "step": 18911
+    },
+    {
+      "epoch": 0.6877841219042077,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6915098326024945e-06,
+      "loss": 1.003,
+      "step": 18912
+    },
+    {
+      "epoch": 0.6878204895079463,
+      "grad_norm": 0.0,
+      "learning_rate": 4.690511671221683e-06,
+      "loss": 0.9485,
+      "step": 18913
+    },
+    {
+      "epoch": 0.6878568571116849,
+      "grad_norm": 0.0,
+      "learning_rate": 4.689513583502485e-06,
+      "loss": 0.9961,
+      "step": 18914
+    },
+    {
+      "epoch": 0.6878932247154235,
+      "grad_norm": 0.0,
+      "learning_rate": 4.688515569458748e-06,
+      "loss": 1.018,
+      "step": 18915
+    },
+    {
+      "epoch": 0.6879295923191621,
+      "grad_norm": 0.0,
+      "learning_rate": 4.687517629104325e-06,
+      "loss": 0.9027,
+      "step": 18916
+    },
+    {
+      "epoch": 0.6879659599229007,
+      "grad_norm": 0.0,
+      "learning_rate": 4.686519762453058e-06,
+      "loss": 0.8708,
+      "step": 18917
+    },
+    {
+      "epoch": 0.6880023275266393,
+      "grad_norm": 0.0,
+      "learning_rate": 4.685521969518789e-06,
+      "loss": 0.8771,
+      "step": 18918
+    },
+    {
+      "epoch": 0.6880386951303779,
+      "grad_norm": 0.0,
+      "learning_rate": 4.684524250315364e-06,
+      "loss": 0.938,
+      "step": 18919
+    },
+    {
+      "epoch": 0.6880750627341164,
+      "grad_norm": 0.0,
+      "learning_rate": 4.683526604856623e-06,
+      "loss": 0.8387,
+      "step": 18920
+    },
+    {
+      "epoch": 0.688111430337855,
+      "grad_norm": 0.0,
+      "learning_rate": 4.682529033156403e-06,
+      "loss": 0.8967,
+      "step": 18921
+    },
+    {
+      "epoch": 0.6881477979415936,
+      "grad_norm": 0.0,
+      "learning_rate": 4.681531535228553e-06,
+      "loss": 0.9705,
+      "step": 18922
+    },
+    {
+      "epoch": 0.6881841655453322,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6805341110869065e-06,
+      "loss": 0.9625,
+      "step": 18923
+    },
+    {
+      "epoch": 0.6882205331490708,
+      "grad_norm": 0.0,
+      "learning_rate": 4.679536760745301e-06,
+      "loss": 0.8748,
+      "step": 18924
+    },
+    {
+      "epoch": 0.6882569007528094,
+      "grad_norm": 0.0,
+      "learning_rate": 4.678539484217571e-06,
+      "loss": 1.0387,
+      "step": 18925
+    },
+    {
+      "epoch": 0.688293268356548,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6775422815175635e-06,
+      "loss": 0.9856,
+      "step": 18926
+    },
+    {
+      "epoch": 0.6883296359602866,
+      "grad_norm": 0.0,
+      "learning_rate": 4.676545152659098e-06,
+      "loss": 0.9069,
+      "step": 18927
+    },
+    {
+      "epoch": 0.6883660035640252,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6755480976560175e-06,
+      "loss": 0.9086,
+      "step": 18928
+    },
+    {
+      "epoch": 0.6884023711677637,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6745511165221545e-06,
+      "loss": 0.9382,
+      "step": 18929
+    },
+    {
+      "epoch": 0.6884387387715023,
+      "grad_norm": 0.0,
+      "learning_rate": 4.673554209271338e-06,
+      "loss": 0.9085,
+      "step": 18930
+    },
+    {
+      "epoch": 0.6884751063752409,
+      "grad_norm": 0.0,
+      "learning_rate": 4.672557375917396e-06,
+      "loss": 0.979,
+      "step": 18931
+    },
+    {
+      "epoch": 0.6885114739789795,
+      "grad_norm": 0.0,
+      "learning_rate": 4.67156061647417e-06,
+      "loss": 0.9299,
+      "step": 18932
+    },
+    {
+      "epoch": 0.6885478415827181,
+      "grad_norm": 0.0,
+      "learning_rate": 4.670563930955475e-06,
+      "loss": 0.8581,
+      "step": 18933
+    },
+    {
+      "epoch": 0.6885842091864567,
+      "grad_norm": 0.0,
+      "learning_rate": 4.669567319375142e-06,
+      "loss": 1.0074,
+      "step": 18934
+    },
+    {
+      "epoch": 0.6886205767901953,
+      "grad_norm": 0.0,
+      "learning_rate": 4.668570781747003e-06,
+      "loss": 0.8794,
+      "step": 18935
+    },
+    {
+      "epoch": 0.6886569443939339,
+      "grad_norm": 0.0,
+      "learning_rate": 4.667574318084879e-06,
+      "loss": 0.9988,
+      "step": 18936
+    },
+    {
+      "epoch": 0.6886933119976725,
+      "grad_norm": 0.0,
+      "learning_rate": 4.666577928402597e-06,
+      "loss": 0.9319,
+      "step": 18937
+    },
+    {
+      "epoch": 0.6887296796014111,
+      "grad_norm": 0.0,
+      "learning_rate": 4.665581612713974e-06,
+      "loss": 0.7525,
+      "step": 18938
+    },
+    {
+      "epoch": 0.6887660472051497,
+      "grad_norm": 0.0,
+      "learning_rate": 4.664585371032846e-06,
+      "loss": 0.8951,
+      "step": 18939
+    },
+    {
+      "epoch": 0.6888024148088883,
+      "grad_norm": 0.0,
+      "learning_rate": 4.663589203373018e-06,
+      "loss": 0.8894,
+      "step": 18940
+    },
+    {
+      "epoch": 0.6888387824126269,
+      "grad_norm": 0.0,
+      "learning_rate": 4.662593109748323e-06,
+      "loss": 0.9126,
+      "step": 18941
+    },
+    {
+      "epoch": 0.6888751500163655,
+      "grad_norm": 0.0,
+      "learning_rate": 4.661597090172575e-06,
+      "loss": 0.7275,
+      "step": 18942
+    },
+    {
+      "epoch": 0.688911517620104,
+      "grad_norm": 0.0,
+      "learning_rate": 4.660601144659594e-06,
+      "loss": 0.93,
+      "step": 18943
+    },
+    {
+      "epoch": 0.6889478852238426,
+      "grad_norm": 0.0,
+      "learning_rate": 4.659605273223191e-06,
+      "loss": 0.9591,
+      "step": 18944
+    },
+    {
+      "epoch": 0.6889842528275812,
+      "grad_norm": 0.0,
+      "learning_rate": 4.658609475877197e-06,
+      "loss": 0.9487,
+      "step": 18945
+    },
+    {
+      "epoch": 0.6890206204313197,
+      "grad_norm": 0.0,
+      "learning_rate": 4.657613752635409e-06,
+      "loss": 0.9544,
+      "step": 18946
+    },
+    {
+      "epoch": 0.6890569880350583,
+      "grad_norm": 0.0,
+      "learning_rate": 4.656618103511655e-06,
+      "loss": 1.1137,
+      "step": 18947
+    },
+    {
+      "epoch": 0.6890933556387969,
+      "grad_norm": 0.0,
+      "learning_rate": 4.655622528519744e-06,
+      "loss": 1.0467,
+      "step": 18948
+    },
+    {
+      "epoch": 0.6891297232425355,
+      "grad_norm": 0.0,
+      "learning_rate": 4.654627027673489e-06,
+      "loss": 0.9447,
+      "step": 18949
+    },
+    {
+      "epoch": 0.6891660908462741,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6536316009866935e-06,
+      "loss": 0.9073,
+      "step": 18950
+    },
+    {
+      "epoch": 0.6892024584500127,
+      "grad_norm": 0.0,
+      "learning_rate": 4.652636248473179e-06,
+      "loss": 0.9814,
+      "step": 18951
+    },
+    {
+      "epoch": 0.6892388260537513,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6516409701467525e-06,
+      "loss": 0.9857,
+      "step": 18952
+    },
+    {
+      "epoch": 0.6892751936574899,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6506457660212124e-06,
+      "loss": 0.8991,
+      "step": 18953
+    },
+    {
+      "epoch": 0.6893115612612285,
+      "grad_norm": 0.0,
+      "learning_rate": 4.649650636110377e-06,
+      "loss": 0.9872,
+      "step": 18954
+    },
+    {
+      "epoch": 0.6893479288649671,
+      "grad_norm": 0.0,
+      "learning_rate": 4.648655580428047e-06,
+      "loss": 0.8931,
+      "step": 18955
+    },
+    {
+      "epoch": 0.6893842964687057,
+      "grad_norm": 0.0,
+      "learning_rate": 4.647660598988028e-06,
+      "loss": 1.0165,
+      "step": 18956
+    },
+    {
+      "epoch": 0.6894206640724443,
+      "grad_norm": 0.0,
+      "learning_rate": 4.646665691804121e-06,
+      "loss": 0.9688,
+      "step": 18957
+    },
+    {
+      "epoch": 0.6894570316761829,
+      "grad_norm": 0.0,
+      "learning_rate": 4.64567085889014e-06,
+      "loss": 0.994,
+      "step": 18958
+    },
+    {
+      "epoch": 0.6894933992799215,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6446761002598704e-06,
+      "loss": 1.059,
+      "step": 18959
+    },
+    {
+      "epoch": 0.6895297668836601,
+      "grad_norm": 0.0,
+      "learning_rate": 4.643681415927127e-06,
+      "loss": 1.0216,
+      "step": 18960
+    },
+    {
+      "epoch": 0.6895661344873987,
+      "grad_norm": 0.0,
+      "learning_rate": 4.642686805905703e-06,
+      "loss": 0.9147,
+      "step": 18961
+    },
+    {
+      "epoch": 0.6896025020911372,
+      "grad_norm": 0.0,
+      "learning_rate": 4.641692270209399e-06,
+      "loss": 0.9401,
+      "step": 18962
+    },
+    {
+      "epoch": 0.6896388696948758,
+      "grad_norm": 0.0,
+      "learning_rate": 4.640697808852008e-06,
+      "loss": 0.845,
+      "step": 18963
+    },
+    {
+      "epoch": 0.6896752372986144,
+      "grad_norm": 0.0,
+      "learning_rate": 4.63970342184734e-06,
+      "loss": 0.9501,
+      "step": 18964
+    },
+    {
+      "epoch": 0.689711604902353,
+      "grad_norm": 0.0,
+      "learning_rate": 4.638709109209173e-06,
+      "loss": 0.9467,
+      "step": 18965
+    },
+    {
+      "epoch": 0.6897479725060915,
+      "grad_norm": 0.0,
+      "learning_rate": 4.637714870951314e-06,
+      "loss": 0.9646,
+      "step": 18966
+    },
+    {
+      "epoch": 0.6897843401098301,
+      "grad_norm": 0.0,
+      "learning_rate": 4.636720707087554e-06,
+      "loss": 1.0275,
+      "step": 18967
+    },
+    {
+      "epoch": 0.6898207077135687,
+      "grad_norm": 0.0,
+      "learning_rate": 4.635726617631685e-06,
+      "loss": 0.869,
+      "step": 18968
+    },
+    {
+      "epoch": 0.6898570753173073,
+      "grad_norm": 0.0,
+      "learning_rate": 4.634732602597494e-06,
+      "loss": 0.8716,
+      "step": 18969
+    },
+    {
+      "epoch": 0.6898934429210459,
+      "grad_norm": 0.0,
+      "learning_rate": 4.633738661998779e-06,
+      "loss": 0.9647,
+      "step": 18970
+    },
+    {
+      "epoch": 0.6899298105247845,
+      "grad_norm": 0.0,
+      "learning_rate": 4.63274479584933e-06,
+      "loss": 0.9294,
+      "step": 18971
+    },
+    {
+      "epoch": 0.6899661781285231,
+      "grad_norm": 0.0,
+      "learning_rate": 4.631751004162925e-06,
+      "loss": 0.9232,
+      "step": 18972
+    },
+    {
+      "epoch": 0.6900025457322617,
+      "grad_norm": 0.0,
+      "learning_rate": 4.630757286953363e-06,
+      "loss": 1.0441,
+      "step": 18973
+    },
+    {
+      "epoch": 0.6900389133360003,
+      "grad_norm": 0.0,
+      "learning_rate": 4.629763644234424e-06,
+      "loss": 0.9312,
+      "step": 18974
+    },
+    {
+      "epoch": 0.6900752809397389,
+      "grad_norm": 0.0,
+      "learning_rate": 4.628770076019895e-06,
+      "loss": 1.0049,
+      "step": 18975
+    },
+    {
+      "epoch": 0.6901116485434775,
+      "grad_norm": 0.0,
+      "learning_rate": 4.627776582323558e-06,
+      "loss": 1.0267,
+      "step": 18976
+    },
+    {
+      "epoch": 0.6901480161472161,
+      "grad_norm": 0.0,
+      "learning_rate": 4.626783163159205e-06,
+      "loss": 0.9742,
+      "step": 18977
+    },
+    {
+      "epoch": 0.6901843837509547,
+      "grad_norm": 0.0,
+      "learning_rate": 4.625789818540604e-06,
+      "loss": 0.9073,
+      "step": 18978
+    },
+    {
+      "epoch": 0.6902207513546932,
+      "grad_norm": 0.0,
+      "learning_rate": 4.624796548481549e-06,
+      "loss": 0.8782,
+      "step": 18979
+    },
+    {
+      "epoch": 0.6902571189584318,
+      "grad_norm": 0.0,
+      "learning_rate": 4.623803352995816e-06,
+      "loss": 1.0052,
+      "step": 18980
+    },
+    {
+      "epoch": 0.6902934865621704,
+      "grad_norm": 0.0,
+      "learning_rate": 4.622810232097183e-06,
+      "loss": 0.8945,
+      "step": 18981
+    },
+    {
+      "epoch": 0.690329854165909,
+      "grad_norm": 0.0,
+      "learning_rate": 4.621817185799426e-06,
+      "loss": 0.9391,
+      "step": 18982
+    },
+    {
+      "epoch": 0.6903662217696476,
+      "grad_norm": 0.0,
+      "learning_rate": 4.620824214116327e-06,
+      "loss": 0.9638,
+      "step": 18983
+    },
+    {
+      "epoch": 0.6904025893733862,
+      "grad_norm": 0.0,
+      "learning_rate": 4.619831317061662e-06,
+      "loss": 1.0133,
+      "step": 18984
+    },
+    {
+      "epoch": 0.6904389569771248,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6188384946492014e-06,
+      "loss": 0.986,
+      "step": 18985
+    },
+    {
+      "epoch": 0.6904753245808634,
+      "grad_norm": 0.0,
+      "learning_rate": 4.617845746892724e-06,
+      "loss": 1.0305,
+      "step": 18986
+    },
+    {
+      "epoch": 0.690511692184602,
+      "grad_norm": 0.0,
+      "learning_rate": 4.616853073805999e-06,
+      "loss": 0.9154,
+      "step": 18987
+    },
+    {
+      "epoch": 0.6905480597883406,
+      "grad_norm": 0.0,
+      "learning_rate": 4.615860475402798e-06,
+      "loss": 0.9346,
+      "step": 18988
+    },
+    {
+      "epoch": 0.6905844273920791,
+      "grad_norm": 0.0,
+      "learning_rate": 4.614867951696896e-06,
+      "loss": 1.1032,
+      "step": 18989
+    },
+    {
+      "epoch": 0.6906207949958177,
+      "grad_norm": 0.0,
+      "learning_rate": 4.613875502702065e-06,
+      "loss": 0.9263,
+      "step": 18990
+    },
+    {
+      "epoch": 0.6906571625995563,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6128831284320615e-06,
+      "loss": 0.9655,
+      "step": 18991
+    },
+    {
+      "epoch": 0.6906935302032949,
+      "grad_norm": 0.0,
+      "learning_rate": 4.611890828900666e-06,
+      "loss": 0.9779,
+      "step": 18992
+    },
+    {
+      "epoch": 0.6907298978070335,
+      "grad_norm": 0.0,
+      "learning_rate": 4.61089860412164e-06,
+      "loss": 0.9465,
+      "step": 18993
+    },
+    {
+      "epoch": 0.6907662654107721,
+      "grad_norm": 0.0,
+      "learning_rate": 4.60990645410875e-06,
+      "loss": 0.9797,
+      "step": 18994
+    },
+    {
+      "epoch": 0.6908026330145107,
+      "grad_norm": 0.0,
+      "learning_rate": 4.608914378875756e-06,
+      "loss": 0.9831,
+      "step": 18995
+    },
+    {
+      "epoch": 0.6908390006182492,
+      "grad_norm": 0.0,
+      "learning_rate": 4.60792237843643e-06,
+      "loss": 0.9849,
+      "step": 18996
+    },
+    {
+      "epoch": 0.6908753682219878,
+      "grad_norm": 0.0,
+      "learning_rate": 4.606930452804531e-06,
+      "loss": 0.8902,
+      "step": 18997
+    },
+    {
+      "epoch": 0.6909117358257264,
+      "grad_norm": 0.0,
+      "learning_rate": 4.60593860199382e-06,
+      "loss": 0.9684,
+      "step": 18998
+    },
+    {
+      "epoch": 0.690948103429465,
+      "grad_norm": 0.0,
+      "learning_rate": 4.604946826018057e-06,
+      "loss": 0.8691,
+      "step": 18999
+    },
+    {
+      "epoch": 0.6909844710332036,
+      "grad_norm": 0.0,
+      "learning_rate": 4.603955124891004e-06,
+      "loss": 0.9294,
+      "step": 19000
+    },
+    {
+      "epoch": 0.6910208386369422,
+      "grad_norm": 0.0,
+      "learning_rate": 4.602963498626412e-06,
+      "loss": 0.8936,
+      "step": 19001
+    },
+    {
+      "epoch": 0.6910572062406808,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6019719472380494e-06,
+      "loss": 0.9993,
+      "step": 19002
+    },
+    {
+      "epoch": 0.6910935738444194,
+      "grad_norm": 0.0,
+      "learning_rate": 4.600980470739667e-06,
+      "loss": 0.9578,
+      "step": 19003
+    },
+    {
+      "epoch": 0.691129941448158,
+      "grad_norm": 0.0,
+      "learning_rate": 4.599989069145019e-06,
+      "loss": 0.9743,
+      "step": 19004
+    },
+    {
+      "epoch": 0.6911663090518966,
+      "grad_norm": 0.0,
+      "learning_rate": 4.598997742467859e-06,
+      "loss": 0.8444,
+      "step": 19005
+    },
+    {
+      "epoch": 0.6912026766556352,
+      "grad_norm": 0.0,
+      "learning_rate": 4.598006490721948e-06,
+      "loss": 0.9488,
+      "step": 19006
+    },
+    {
+      "epoch": 0.6912390442593738,
+      "grad_norm": 0.0,
+      "learning_rate": 4.597015313921026e-06,
+      "loss": 1.028,
+      "step": 19007
+    },
+    {
+      "epoch": 0.6912754118631124,
+      "grad_norm": 0.0,
+      "learning_rate": 4.596024212078854e-06,
+      "loss": 1.0058,
+      "step": 19008
+    },
+    {
+      "epoch": 0.691311779466851,
+      "grad_norm": 0.0,
+      "learning_rate": 4.595033185209184e-06,
+      "loss": 0.844,
+      "step": 19009
+    },
+    {
+      "epoch": 0.6913481470705896,
+      "grad_norm": 0.0,
+      "learning_rate": 4.594042233325752e-06,
+      "loss": 0.9715,
+      "step": 19010
+    },
+    {
+      "epoch": 0.6913845146743282,
+      "grad_norm": 0.0,
+      "learning_rate": 4.593051356442317e-06,
+      "loss": 0.955,
+      "step": 19011
+    },
+    {
+      "epoch": 0.6914208822780668,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5920605545726225e-06,
+      "loss": 0.9473,
+      "step": 19012
+    },
+    {
+      "epoch": 0.6914572498818052,
+      "grad_norm": 0.0,
+      "learning_rate": 4.591069827730416e-06,
+      "loss": 0.9597,
+      "step": 19013
+    },
+    {
+      "epoch": 0.6914936174855438,
+      "grad_norm": 0.0,
+      "learning_rate": 4.590079175929437e-06,
+      "loss": 0.8702,
+      "step": 19014
+    },
+    {
+      "epoch": 0.6915299850892824,
+      "grad_norm": 0.0,
+      "learning_rate": 4.589088599183437e-06,
+      "loss": 0.9512,
+      "step": 19015
+    },
+    {
+      "epoch": 0.691566352693021,
+      "grad_norm": 0.0,
+      "learning_rate": 4.588098097506156e-06,
+      "loss": 0.9694,
+      "step": 19016
+    },
+    {
+      "epoch": 0.6916027202967596,
+      "grad_norm": 0.0,
+      "learning_rate": 4.587107670911336e-06,
+      "loss": 0.9181,
+      "step": 19017
+    },
+    {
+      "epoch": 0.6916390879004982,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5861173194127175e-06,
+      "loss": 0.9682,
+      "step": 19018
+    },
+    {
+      "epoch": 0.6916754555042368,
+      "grad_norm": 0.0,
+      "learning_rate": 4.585127043024039e-06,
+      "loss": 0.9296,
+      "step": 19019
+    },
+    {
+      "epoch": 0.6917118231079754,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5841368417590355e-06,
+      "loss": 0.8719,
+      "step": 19020
+    },
+    {
+      "epoch": 0.691748190711714,
+      "grad_norm": 0.0,
+      "learning_rate": 4.583146715631455e-06,
+      "loss": 0.8446,
+      "step": 19021
+    },
+    {
+      "epoch": 0.6917845583154526,
+      "grad_norm": 0.0,
+      "learning_rate": 4.582156664655028e-06,
+      "loss": 0.9633,
+      "step": 19022
+    },
+    {
+      "epoch": 0.6918209259191912,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5811666888434895e-06,
+      "loss": 1.0132,
+      "step": 19023
+    },
+    {
+      "epoch": 0.6918572935229298,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5801767882105706e-06,
+      "loss": 0.9801,
+      "step": 19024
+    },
+    {
+      "epoch": 0.6918936611266684,
+      "grad_norm": 0.0,
+      "learning_rate": 4.579186962770018e-06,
+      "loss": 0.8363,
+      "step": 19025
+    },
+    {
+      "epoch": 0.691930028730407,
+      "grad_norm": 0.0,
+      "learning_rate": 4.578197212535547e-06,
+      "loss": 0.9175,
+      "step": 19026
+    },
+    {
+      "epoch": 0.6919663963341456,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5772075375209e-06,
+      "loss": 0.9489,
+      "step": 19027
+    },
+    {
+      "epoch": 0.6920027639378842,
+      "grad_norm": 0.0,
+      "learning_rate": 4.576217937739805e-06,
+      "loss": 0.9285,
+      "step": 19028
+    },
+    {
+      "epoch": 0.6920391315416228,
+      "grad_norm": 0.0,
+      "learning_rate": 4.575228413205992e-06,
+      "loss": 0.979,
+      "step": 19029
+    },
+    {
+      "epoch": 0.6920754991453613,
+      "grad_norm": 0.0,
+      "learning_rate": 4.574238963933187e-06,
+      "loss": 1.0404,
+      "step": 19030
+    },
+    {
+      "epoch": 0.6921118667490999,
+      "grad_norm": 0.0,
+      "learning_rate": 4.573249589935118e-06,
+      "loss": 0.879,
+      "step": 19031
+    },
+    {
+      "epoch": 0.6921482343528385,
+      "grad_norm": 0.0,
+      "learning_rate": 4.572260291225512e-06,
+      "loss": 1.0219,
+      "step": 19032
+    },
+    {
+      "epoch": 0.692184601956577,
+      "grad_norm": 0.0,
+      "learning_rate": 4.57127106781809e-06,
+      "loss": 0.9857,
+      "step": 19033
+    },
+    {
+      "epoch": 0.6922209695603156,
+      "grad_norm": 0.0,
+      "learning_rate": 4.570281919726583e-06,
+      "loss": 0.8691,
+      "step": 19034
+    },
+    {
+      "epoch": 0.6922573371640542,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5692928469647105e-06,
+      "loss": 0.9821,
+      "step": 19035
+    },
+    {
+      "epoch": 0.6922937047677928,
+      "grad_norm": 0.0,
+      "learning_rate": 4.568303849546194e-06,
+      "loss": 0.9944,
+      "step": 19036
+    },
+    {
+      "epoch": 0.6923300723715314,
+      "grad_norm": 0.0,
+      "learning_rate": 4.567314927484752e-06,
+      "loss": 0.8314,
+      "step": 19037
+    },
+    {
+      "epoch": 0.69236643997527,
+      "grad_norm": 0.0,
+      "learning_rate": 4.566326080794115e-06,
+      "loss": 0.9848,
+      "step": 19038
+    },
+    {
+      "epoch": 0.6924028075790086,
+      "grad_norm": 0.0,
+      "learning_rate": 4.565337309487986e-06,
+      "loss": 0.9334,
+      "step": 19039
+    },
+    {
+      "epoch": 0.6924391751827472,
+      "grad_norm": 0.0,
+      "learning_rate": 4.564348613580095e-06,
+      "loss": 0.9619,
+      "step": 19040
+    },
+    {
+      "epoch": 0.6924755427864858,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5633599930841555e-06,
+      "loss": 0.9518,
+      "step": 19041
+    },
+    {
+      "epoch": 0.6925119103902244,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5623714480138815e-06,
+      "loss": 0.9113,
+      "step": 19042
+    },
+    {
+      "epoch": 0.692548277993963,
+      "grad_norm": 0.0,
+      "learning_rate": 4.561382978382986e-06,
+      "loss": 0.8631,
+      "step": 19043
+    },
+    {
+      "epoch": 0.6925846455977016,
+      "grad_norm": 0.0,
+      "learning_rate": 4.560394584205192e-06,
+      "loss": 0.9148,
+      "step": 19044
+    },
+    {
+      "epoch": 0.6926210132014402,
+      "grad_norm": 0.0,
+      "learning_rate": 4.559406265494198e-06,
+      "loss": 0.9396,
+      "step": 19045
+    },
+    {
+      "epoch": 0.6926573808051788,
+      "grad_norm": 0.0,
+      "learning_rate": 4.558418022263726e-06,
+      "loss": 0.842,
+      "step": 19046
+    },
+    {
+      "epoch": 0.6926937484089173,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5574298545274845e-06,
+      "loss": 1.0097,
+      "step": 19047
+    },
+    {
+      "epoch": 0.6927301160126559,
+      "grad_norm": 0.0,
+      "learning_rate": 4.556441762299182e-06,
+      "loss": 1.0247,
+      "step": 19048
+    },
+    {
+      "epoch": 0.6927664836163945,
+      "grad_norm": 0.0,
+      "learning_rate": 4.555453745592525e-06,
+      "loss": 1.0043,
+      "step": 19049
+    },
+    {
+      "epoch": 0.6928028512201331,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5544658044212195e-06,
+      "loss": 1.0129,
+      "step": 19050
+    },
+    {
+      "epoch": 0.6928392188238717,
+      "grad_norm": 0.0,
+      "learning_rate": 4.553477938798983e-06,
+      "loss": 0.991,
+      "step": 19051
+    },
+    {
+      "epoch": 0.6928755864276103,
+      "grad_norm": 0.0,
+      "learning_rate": 4.552490148739505e-06,
+      "loss": 0.8525,
+      "step": 19052
+    },
+    {
+      "epoch": 0.6929119540313489,
+      "grad_norm": 0.0,
+      "learning_rate": 4.551502434256501e-06,
+      "loss": 0.9169,
+      "step": 19053
+    },
+    {
+      "epoch": 0.6929483216350875,
+      "grad_norm": 0.0,
+      "learning_rate": 4.550514795363669e-06,
+      "loss": 0.9598,
+      "step": 19054
+    },
+    {
+      "epoch": 0.6929846892388261,
+      "grad_norm": 0.0,
+      "learning_rate": 4.549527232074713e-06,
+      "loss": 0.9923,
+      "step": 19055
+    },
+    {
+      "epoch": 0.6930210568425647,
+      "grad_norm": 0.0,
+      "learning_rate": 4.54853974440333e-06,
+      "loss": 1.0113,
+      "step": 19056
+    },
+    {
+      "epoch": 0.6930574244463032,
+      "grad_norm": 0.0,
+      "learning_rate": 4.547552332363231e-06,
+      "loss": 0.9366,
+      "step": 19057
+    },
+    {
+      "epoch": 0.6930937920500418,
+      "grad_norm": 0.0,
+      "learning_rate": 4.546564995968099e-06,
+      "loss": 0.9672,
+      "step": 19058
+    },
+    {
+      "epoch": 0.6931301596537804,
+      "grad_norm": 0.0,
+      "learning_rate": 4.545577735231645e-06,
+      "loss": 0.9883,
+      "step": 19059
+    },
+    {
+      "epoch": 0.693166527257519,
+      "grad_norm": 0.0,
+      "learning_rate": 4.544590550167561e-06,
+      "loss": 0.9156,
+      "step": 19060
+    },
+    {
+      "epoch": 0.6932028948612576,
+      "grad_norm": 0.0,
+      "learning_rate": 4.543603440789542e-06,
+      "loss": 0.8482,
+      "step": 19061
+    },
+    {
+      "epoch": 0.6932392624649962,
+      "grad_norm": 0.0,
+      "learning_rate": 4.542616407111281e-06,
+      "loss": 0.8704,
+      "step": 19062
+    },
+    {
+      "epoch": 0.6932756300687347,
+      "grad_norm": 0.0,
+      "learning_rate": 4.541629449146482e-06,
+      "loss": 0.7636,
+      "step": 19063
+    },
+    {
+      "epoch": 0.6933119976724733,
+      "grad_norm": 0.0,
+      "learning_rate": 4.540642566908821e-06,
+      "loss": 1.0171,
+      "step": 19064
+    },
+    {
+      "epoch": 0.6933483652762119,
+      "grad_norm": 0.0,
+      "learning_rate": 4.539655760412004e-06,
+      "loss": 1.0185,
+      "step": 19065
+    },
+    {
+      "epoch": 0.6933847328799505,
+      "grad_norm": 0.0,
+      "learning_rate": 4.538669029669715e-06,
+      "loss": 0.8908,
+      "step": 19066
+    },
+    {
+      "epoch": 0.6934211004836891,
+      "grad_norm": 0.0,
+      "learning_rate": 4.537682374695644e-06,
+      "loss": 0.968,
+      "step": 19067
+    },
+    {
+      "epoch": 0.6934574680874277,
+      "grad_norm": 0.0,
+      "learning_rate": 4.536695795503482e-06,
+      "loss": 0.9675,
+      "step": 19068
+    },
+    {
+      "epoch": 0.6934938356911663,
+      "grad_norm": 0.0,
+      "learning_rate": 4.535709292106909e-06,
+      "loss": 0.9473,
+      "step": 19069
+    },
+    {
+      "epoch": 0.6935302032949049,
+      "grad_norm": 0.0,
+      "learning_rate": 4.534722864519625e-06,
+      "loss": 1.0115,
+      "step": 19070
+    },
+    {
+      "epoch": 0.6935665708986435,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5337365127553e-06,
+      "loss": 0.9072,
+      "step": 19071
+    },
+    {
+      "epoch": 0.6936029385023821,
+      "grad_norm": 0.0,
+      "learning_rate": 4.532750236827628e-06,
+      "loss": 0.9338,
+      "step": 19072
+    },
+    {
+      "epoch": 0.6936393061061207,
+      "grad_norm": 0.0,
+      "learning_rate": 4.53176403675029e-06,
+      "loss": 1.0314,
+      "step": 19073
+    },
+    {
+      "epoch": 0.6936756737098593,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5307779125369675e-06,
+      "loss": 0.9515,
+      "step": 19074
+    },
+    {
+      "epoch": 0.6937120413135979,
+      "grad_norm": 0.0,
+      "learning_rate": 4.529791864201338e-06,
+      "loss": 0.975,
+      "step": 19075
+    },
+    {
+      "epoch": 0.6937484089173365,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5288058917570925e-06,
+      "loss": 0.9929,
+      "step": 19076
+    },
+    {
+      "epoch": 0.6937847765210751,
+      "grad_norm": 0.0,
+      "learning_rate": 4.527819995217895e-06,
+      "loss": 1.0409,
+      "step": 19077
+    },
+    {
+      "epoch": 0.6938211441248137,
+      "grad_norm": 0.0,
+      "learning_rate": 4.526834174597435e-06,
+      "loss": 0.9942,
+      "step": 19078
+    },
+    {
+      "epoch": 0.6938575117285523,
+      "grad_norm": 0.0,
+      "learning_rate": 4.525848429909385e-06,
+      "loss": 1.0386,
+      "step": 19079
+    },
+    {
+      "epoch": 0.6938938793322907,
+      "grad_norm": 0.0,
+      "learning_rate": 4.524862761167422e-06,
+      "loss": 0.9315,
+      "step": 19080
+    },
+    {
+      "epoch": 0.6939302469360293,
+      "grad_norm": 0.0,
+      "learning_rate": 4.523877168385216e-06,
+      "loss": 1.0369,
+      "step": 19081
+    },
+    {
+      "epoch": 0.6939666145397679,
+      "grad_norm": 0.0,
+      "learning_rate": 4.522891651576449e-06,
+      "loss": 1.0059,
+      "step": 19082
+    },
+    {
+      "epoch": 0.6940029821435065,
+      "grad_norm": 0.0,
+      "learning_rate": 4.521906210754789e-06,
+      "loss": 1.0976,
+      "step": 19083
+    },
+    {
+      "epoch": 0.6940393497472451,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5209208459339075e-06,
+      "loss": 0.935,
+      "step": 19084
+    },
+    {
+      "epoch": 0.6940757173509837,
+      "grad_norm": 0.0,
+      "learning_rate": 4.519935557127475e-06,
+      "loss": 0.7985,
+      "step": 19085
+    },
+    {
+      "epoch": 0.6941120849547223,
+      "grad_norm": 0.0,
+      "learning_rate": 4.518950344349162e-06,
+      "loss": 1.0199,
+      "step": 19086
+    },
+    {
+      "epoch": 0.6941484525584609,
+      "grad_norm": 0.0,
+      "learning_rate": 4.517965207612637e-06,
+      "loss": 0.9447,
+      "step": 19087
+    },
+    {
+      "epoch": 0.6941848201621995,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5169801469315625e-06,
+      "loss": 0.9686,
+      "step": 19088
+    },
+    {
+      "epoch": 0.6942211877659381,
+      "grad_norm": 0.0,
+      "learning_rate": 4.515995162319615e-06,
+      "loss": 1.0072,
+      "step": 19089
+    },
+    {
+      "epoch": 0.6942575553696767,
+      "grad_norm": 0.0,
+      "learning_rate": 4.515010253790448e-06,
+      "loss": 0.8565,
+      "step": 19090
+    },
+    {
+      "epoch": 0.6942939229734153,
+      "grad_norm": 0.0,
+      "learning_rate": 4.514025421357735e-06,
+      "loss": 0.9728,
+      "step": 19091
+    },
+    {
+      "epoch": 0.6943302905771539,
+      "grad_norm": 0.0,
+      "learning_rate": 4.513040665035137e-06,
+      "loss": 0.8463,
+      "step": 19092
+    },
+    {
+      "epoch": 0.6943666581808925,
+      "grad_norm": 0.0,
+      "learning_rate": 4.512055984836313e-06,
+      "loss": 0.907,
+      "step": 19093
+    },
+    {
+      "epoch": 0.6944030257846311,
+      "grad_norm": 0.0,
+      "learning_rate": 4.511071380774923e-06,
+      "loss": 0.9686,
+      "step": 19094
+    },
+    {
+      "epoch": 0.6944393933883697,
+      "grad_norm": 0.0,
+      "learning_rate": 4.510086852864632e-06,
+      "loss": 0.7356,
+      "step": 19095
+    },
+    {
+      "epoch": 0.6944757609921083,
+      "grad_norm": 0.0,
+      "learning_rate": 4.509102401119099e-06,
+      "loss": 0.8534,
+      "step": 19096
+    },
+    {
+      "epoch": 0.6945121285958468,
+      "grad_norm": 0.0,
+      "learning_rate": 4.508118025551979e-06,
+      "loss": 0.9535,
+      "step": 19097
+    },
+    {
+      "epoch": 0.6945484961995854,
+      "grad_norm": 0.0,
+      "learning_rate": 4.507133726176928e-06,
+      "loss": 0.9311,
+      "step": 19098
+    },
+    {
+      "epoch": 0.694584863803324,
+      "grad_norm": 0.0,
+      "learning_rate": 4.506149503007605e-06,
+      "loss": 0.8979,
+      "step": 19099
+    },
+    {
+      "epoch": 0.6946212314070626,
+      "grad_norm": 0.0,
+      "learning_rate": 4.505165356057658e-06,
+      "loss": 0.9401,
+      "step": 19100
+    },
+    {
+      "epoch": 0.6946575990108012,
+      "grad_norm": 0.0,
+      "learning_rate": 4.504181285340749e-06,
+      "loss": 1.001,
+      "step": 19101
+    },
+    {
+      "epoch": 0.6946939666145397,
+      "grad_norm": 0.0,
+      "learning_rate": 4.503197290870529e-06,
+      "loss": 0.9854,
+      "step": 19102
+    },
+    {
+      "epoch": 0.6947303342182783,
+      "grad_norm": 0.0,
+      "learning_rate": 4.502213372660646e-06,
+      "loss": 0.8094,
+      "step": 19103
+    },
+    {
+      "epoch": 0.6947667018220169,
+      "grad_norm": 0.0,
+      "learning_rate": 4.501229530724749e-06,
+      "loss": 1.0231,
+      "step": 19104
+    },
+    {
+      "epoch": 0.6948030694257555,
+      "grad_norm": 0.0,
+      "learning_rate": 4.500245765076498e-06,
+      "loss": 0.9895,
+      "step": 19105
+    },
+    {
+      "epoch": 0.6948394370294941,
+      "grad_norm": 0.0,
+      "learning_rate": 4.499262075729529e-06,
+      "loss": 0.92,
+      "step": 19106
+    },
+    {
+      "epoch": 0.6948758046332327,
+      "grad_norm": 0.0,
+      "learning_rate": 4.498278462697492e-06,
+      "loss": 0.8673,
+      "step": 19107
+    },
+    {
+      "epoch": 0.6949121722369713,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4972949259940425e-06,
+      "loss": 0.87,
+      "step": 19108
+    },
+    {
+      "epoch": 0.6949485398407099,
+      "grad_norm": 0.0,
+      "learning_rate": 4.496311465632811e-06,
+      "loss": 0.9322,
+      "step": 19109
+    },
+    {
+      "epoch": 0.6949849074444485,
+      "grad_norm": 0.0,
+      "learning_rate": 4.495328081627453e-06,
+      "loss": 0.9055,
+      "step": 19110
+    },
+    {
+      "epoch": 0.6950212750481871,
+      "grad_norm": 0.0,
+      "learning_rate": 4.494344773991607e-06,
+      "loss": 0.8579,
+      "step": 19111
+    },
+    {
+      "epoch": 0.6950576426519257,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4933615427389165e-06,
+      "loss": 0.962,
+      "step": 19112
+    },
+    {
+      "epoch": 0.6950940102556643,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4923783878830175e-06,
+      "loss": 1.0075,
+      "step": 19113
+    },
+    {
+      "epoch": 0.6951303778594028,
+      "grad_norm": 0.0,
+      "learning_rate": 4.491395309437558e-06,
+      "loss": 0.9772,
+      "step": 19114
+    },
+    {
+      "epoch": 0.6951667454631414,
+      "grad_norm": 0.0,
+      "learning_rate": 4.490412307416173e-06,
+      "loss": 1.0683,
+      "step": 19115
+    },
+    {
+      "epoch": 0.69520311306688,
+      "grad_norm": 0.0,
+      "learning_rate": 4.489429381832501e-06,
+      "loss": 0.8156,
+      "step": 19116
+    },
+    {
+      "epoch": 0.6952394806706186,
+      "grad_norm": 0.0,
+      "learning_rate": 4.488446532700173e-06,
+      "loss": 0.9398,
+      "step": 19117
+    },
+    {
+      "epoch": 0.6952758482743572,
+      "grad_norm": 0.0,
+      "learning_rate": 4.487463760032837e-06,
+      "loss": 0.884,
+      "step": 19118
+    },
+    {
+      "epoch": 0.6953122158780958,
+      "grad_norm": 0.0,
+      "learning_rate": 4.486481063844112e-06,
+      "loss": 1.0265,
+      "step": 19119
+    },
+    {
+      "epoch": 0.6953485834818344,
+      "grad_norm": 0.0,
+      "learning_rate": 4.485498444147645e-06,
+      "loss": 0.9734,
+      "step": 19120
+    },
+    {
+      "epoch": 0.695384951085573,
+      "grad_norm": 0.0,
+      "learning_rate": 4.484515900957063e-06,
+      "loss": 0.9717,
+      "step": 19121
+    },
+    {
+      "epoch": 0.6954213186893116,
+      "grad_norm": 0.0,
+      "learning_rate": 4.483533434285996e-06,
+      "loss": 0.9442,
+      "step": 19122
+    },
+    {
+      "epoch": 0.6954576862930502,
+      "grad_norm": 0.0,
+      "learning_rate": 4.482551044148072e-06,
+      "loss": 0.89,
+      "step": 19123
+    },
+    {
+      "epoch": 0.6954940538967888,
+      "grad_norm": 0.0,
+      "learning_rate": 4.481568730556932e-06,
+      "loss": 0.9812,
+      "step": 19124
+    },
+    {
+      "epoch": 0.6955304215005274,
+      "grad_norm": 0.0,
+      "learning_rate": 4.480586493526194e-06,
+      "loss": 0.8378,
+      "step": 19125
+    },
+    {
+      "epoch": 0.695566789104266,
+      "grad_norm": 0.0,
+      "learning_rate": 4.479604333069482e-06,
+      "loss": 0.9081,
+      "step": 19126
+    },
+    {
+      "epoch": 0.6956031567080045,
+      "grad_norm": 0.0,
+      "learning_rate": 4.47862224920043e-06,
+      "loss": 1.0027,
+      "step": 19127
+    },
+    {
+      "epoch": 0.6956395243117431,
+      "grad_norm": 0.0,
+      "learning_rate": 4.477640241932663e-06,
+      "loss": 0.9213,
+      "step": 19128
+    },
+    {
+      "epoch": 0.6956758919154817,
+      "grad_norm": 0.0,
+      "learning_rate": 4.476658311279801e-06,
+      "loss": 0.8818,
+      "step": 19129
+    },
+    {
+      "epoch": 0.6957122595192203,
+      "grad_norm": 0.0,
+      "learning_rate": 4.475676457255469e-06,
+      "loss": 0.9365,
+      "step": 19130
+    },
+    {
+      "epoch": 0.6957486271229588,
+      "grad_norm": 0.0,
+      "learning_rate": 4.474694679873287e-06,
+      "loss": 0.9359,
+      "step": 19131
+    },
+    {
+      "epoch": 0.6957849947266974,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4737129791468745e-06,
+      "loss": 1.0223,
+      "step": 19132
+    },
+    {
+      "epoch": 0.695821362330436,
+      "grad_norm": 0.0,
+      "learning_rate": 4.472731355089857e-06,
+      "loss": 0.8936,
+      "step": 19133
+    },
+    {
+      "epoch": 0.6958577299341746,
+      "grad_norm": 0.0,
+      "learning_rate": 4.471749807715849e-06,
+      "loss": 0.7656,
+      "step": 19134
+    },
+    {
+      "epoch": 0.6958940975379132,
+      "grad_norm": 0.0,
+      "learning_rate": 4.47076833703847e-06,
+      "loss": 1.1398,
+      "step": 19135
+    },
+    {
+      "epoch": 0.6959304651416518,
+      "grad_norm": 0.0,
+      "learning_rate": 4.469786943071332e-06,
+      "loss": 0.9998,
+      "step": 19136
+    },
+    {
+      "epoch": 0.6959668327453904,
+      "grad_norm": 0.0,
+      "learning_rate": 4.46880562582806e-06,
+      "loss": 1.0254,
+      "step": 19137
+    },
+    {
+      "epoch": 0.696003200349129,
+      "grad_norm": 0.0,
+      "learning_rate": 4.467824385322256e-06,
+      "loss": 0.9224,
+      "step": 19138
+    },
+    {
+      "epoch": 0.6960395679528676,
+      "grad_norm": 0.0,
+      "learning_rate": 4.466843221567542e-06,
+      "loss": 0.9089,
+      "step": 19139
+    },
+    {
+      "epoch": 0.6960759355566062,
+      "grad_norm": 0.0,
+      "learning_rate": 4.465862134577527e-06,
+      "loss": 0.9705,
+      "step": 19140
+    },
+    {
+      "epoch": 0.6961123031603448,
+      "grad_norm": 0.0,
+      "learning_rate": 4.464881124365825e-06,
+      "loss": 1.0129,
+      "step": 19141
+    },
+    {
+      "epoch": 0.6961486707640834,
+      "grad_norm": 0.0,
+      "learning_rate": 4.46390019094604e-06,
+      "loss": 0.9307,
+      "step": 19142
+    },
+    {
+      "epoch": 0.696185038367822,
+      "grad_norm": 0.0,
+      "learning_rate": 4.462919334331792e-06,
+      "loss": 0.9156,
+      "step": 19143
+    },
+    {
+      "epoch": 0.6962214059715606,
+      "grad_norm": 0.0,
+      "learning_rate": 4.461938554536679e-06,
+      "loss": 0.9239,
+      "step": 19144
+    },
+    {
+      "epoch": 0.6962577735752992,
+      "grad_norm": 0.0,
+      "learning_rate": 4.460957851574307e-06,
+      "loss": 1.0035,
+      "step": 19145
+    },
+    {
+      "epoch": 0.6962941411790378,
+      "grad_norm": 0.0,
+      "learning_rate": 4.459977225458289e-06,
+      "loss": 1.0132,
+      "step": 19146
+    },
+    {
+      "epoch": 0.6963305087827762,
+      "grad_norm": 0.0,
+      "learning_rate": 4.458996676202228e-06,
+      "loss": 0.9906,
+      "step": 19147
+    },
+    {
+      "epoch": 0.6963668763865148,
+      "grad_norm": 0.0,
+      "learning_rate": 4.458016203819726e-06,
+      "loss": 1.0146,
+      "step": 19148
+    },
+    {
+      "epoch": 0.6964032439902534,
+      "grad_norm": 0.0,
+      "learning_rate": 4.457035808324382e-06,
+      "loss": 0.8786,
+      "step": 19149
+    },
+    {
+      "epoch": 0.696439611593992,
+      "grad_norm": 0.0,
+      "learning_rate": 4.456055489729809e-06,
+      "loss": 0.9114,
+      "step": 19150
+    },
+    {
+      "epoch": 0.6964759791977306,
+      "grad_norm": 0.0,
+      "learning_rate": 4.455075248049594e-06,
+      "loss": 0.9992,
+      "step": 19151
+    },
+    {
+      "epoch": 0.6965123468014692,
+      "grad_norm": 0.0,
+      "learning_rate": 4.454095083297345e-06,
+      "loss": 0.8973,
+      "step": 19152
+    },
+    {
+      "epoch": 0.6965487144052078,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4531149954866594e-06,
+      "loss": 0.8468,
+      "step": 19153
+    },
+    {
+      "epoch": 0.6965850820089464,
+      "grad_norm": 0.0,
+      "learning_rate": 4.452134984631133e-06,
+      "loss": 1.0013,
+      "step": 19154
+    },
+    {
+      "epoch": 0.696621449612685,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4511550507443565e-06,
+      "loss": 0.8426,
+      "step": 19155
+    },
+    {
+      "epoch": 0.6966578172164236,
+      "grad_norm": 0.0,
+      "learning_rate": 4.450175193839941e-06,
+      "loss": 1.019,
+      "step": 19156
+    },
+    {
+      "epoch": 0.6966941848201622,
+      "grad_norm": 0.0,
+      "learning_rate": 4.449195413931462e-06,
+      "loss": 0.9245,
+      "step": 19157
+    },
+    {
+      "epoch": 0.6967305524239008,
+      "grad_norm": 0.0,
+      "learning_rate": 4.448215711032525e-06,
+      "loss": 0.9708,
+      "step": 19158
+    },
+    {
+      "epoch": 0.6967669200276394,
+      "grad_norm": 0.0,
+      "learning_rate": 4.447236085156718e-06,
+      "loss": 0.8478,
+      "step": 19159
+    },
+    {
+      "epoch": 0.696803287631378,
+      "grad_norm": 0.0,
+      "learning_rate": 4.446256536317631e-06,
+      "loss": 0.9349,
+      "step": 19160
+    },
+    {
+      "epoch": 0.6968396552351166,
+      "grad_norm": 0.0,
+      "learning_rate": 4.445277064528852e-06,
+      "loss": 0.9897,
+      "step": 19161
+    },
+    {
+      "epoch": 0.6968760228388552,
+      "grad_norm": 0.0,
+      "learning_rate": 4.444297669803981e-06,
+      "loss": 0.7461,
+      "step": 19162
+    },
+    {
+      "epoch": 0.6969123904425938,
+      "grad_norm": 0.0,
+      "learning_rate": 4.443318352156593e-06,
+      "loss": 0.8754,
+      "step": 19163
+    },
+    {
+      "epoch": 0.6969487580463323,
+      "grad_norm": 0.0,
+      "learning_rate": 4.442339111600274e-06,
+      "loss": 0.9044,
+      "step": 19164
+    },
+    {
+      "epoch": 0.6969851256500709,
+      "grad_norm": 0.0,
+      "learning_rate": 4.441359948148619e-06,
+      "loss": 0.9006,
+      "step": 19165
+    },
+    {
+      "epoch": 0.6970214932538095,
+      "grad_norm": 0.0,
+      "learning_rate": 4.440380861815209e-06,
+      "loss": 0.9507,
+      "step": 19166
+    },
+    {
+      "epoch": 0.6970578608575481,
+      "grad_norm": 0.0,
+      "learning_rate": 4.439401852613625e-06,
+      "loss": 1.0288,
+      "step": 19167
+    },
+    {
+      "epoch": 0.6970942284612867,
+      "grad_norm": 0.0,
+      "learning_rate": 4.438422920557448e-06,
+      "loss": 0.9227,
+      "step": 19168
+    },
+    {
+      "epoch": 0.6971305960650253,
+      "grad_norm": 0.0,
+      "learning_rate": 4.437444065660269e-06,
+      "loss": 1.0013,
+      "step": 19169
+    },
+    {
+      "epoch": 0.6971669636687639,
+      "grad_norm": 0.0,
+      "learning_rate": 4.436465287935655e-06,
+      "loss": 0.8841,
+      "step": 19170
+    },
+    {
+      "epoch": 0.6972033312725024,
+      "grad_norm": 0.0,
+      "learning_rate": 4.435486587397194e-06,
+      "loss": 1.0059,
+      "step": 19171
+    },
+    {
+      "epoch": 0.697239698876241,
+      "grad_norm": 0.0,
+      "learning_rate": 4.434507964058462e-06,
+      "loss": 0.9114,
+      "step": 19172
+    },
+    {
+      "epoch": 0.6972760664799796,
+      "grad_norm": 0.0,
+      "learning_rate": 4.433529417933036e-06,
+      "loss": 0.9761,
+      "step": 19173
+    },
+    {
+      "epoch": 0.6973124340837182,
+      "grad_norm": 0.0,
+      "learning_rate": 4.432550949034487e-06,
+      "loss": 1.0443,
+      "step": 19174
+    },
+    {
+      "epoch": 0.6973488016874568,
+      "grad_norm": 0.0,
+      "learning_rate": 4.431572557376402e-06,
+      "loss": 0.8581,
+      "step": 19175
+    },
+    {
+      "epoch": 0.6973851692911954,
+      "grad_norm": 0.0,
+      "learning_rate": 4.430594242972342e-06,
+      "loss": 0.8646,
+      "step": 19176
+    },
+    {
+      "epoch": 0.697421536894934,
+      "grad_norm": 0.0,
+      "learning_rate": 4.429616005835886e-06,
+      "loss": 1.009,
+      "step": 19177
+    },
+    {
+      "epoch": 0.6974579044986726,
+      "grad_norm": 0.0,
+      "learning_rate": 4.428637845980604e-06,
+      "loss": 0.9722,
+      "step": 19178
+    },
+    {
+      "epoch": 0.6974942721024112,
+      "grad_norm": 0.0,
+      "learning_rate": 4.427659763420069e-06,
+      "loss": 0.7505,
+      "step": 19179
+    },
+    {
+      "epoch": 0.6975306397061498,
+      "grad_norm": 0.0,
+      "learning_rate": 4.426681758167844e-06,
+      "loss": 1.0602,
+      "step": 19180
+    },
+    {
+      "epoch": 0.6975670073098883,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4257038302375076e-06,
+      "loss": 1.016,
+      "step": 19181
+    },
+    {
+      "epoch": 0.6976033749136269,
+      "grad_norm": 0.0,
+      "learning_rate": 4.424725979642623e-06,
+      "loss": 0.9827,
+      "step": 19182
+    },
+    {
+      "epoch": 0.6976397425173655,
+      "grad_norm": 0.0,
+      "learning_rate": 4.42374820639675e-06,
+      "loss": 1.0453,
+      "step": 19183
+    },
+    {
+      "epoch": 0.6976761101211041,
+      "grad_norm": 0.0,
+      "learning_rate": 4.422770510513461e-06,
+      "loss": 0.8817,
+      "step": 19184
+    },
+    {
+      "epoch": 0.6977124777248427,
+      "grad_norm": 0.0,
+      "learning_rate": 4.421792892006318e-06,
+      "loss": 1.0309,
+      "step": 19185
+    },
+    {
+      "epoch": 0.6977488453285813,
+      "grad_norm": 0.0,
+      "learning_rate": 4.420815350888885e-06,
+      "loss": 0.9741,
+      "step": 19186
+    },
+    {
+      "epoch": 0.6977852129323199,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4198378871747195e-06,
+      "loss": 0.9252,
+      "step": 19187
+    },
+    {
+      "epoch": 0.6978215805360585,
+      "grad_norm": 0.0,
+      "learning_rate": 4.418860500877394e-06,
+      "loss": 0.9282,
+      "step": 19188
+    },
+    {
+      "epoch": 0.6978579481397971,
+      "grad_norm": 0.0,
+      "learning_rate": 4.417883192010451e-06,
+      "loss": 0.8653,
+      "step": 19189
+    },
+    {
+      "epoch": 0.6978943157435357,
+      "grad_norm": 0.0,
+      "learning_rate": 4.416905960587464e-06,
+      "loss": 0.9323,
+      "step": 19190
+    },
+    {
+      "epoch": 0.6979306833472743,
+      "grad_norm": 0.0,
+      "learning_rate": 4.415928806621985e-06,
+      "loss": 0.9941,
+      "step": 19191
+    },
+    {
+      "epoch": 0.6979670509510129,
+      "grad_norm": 0.0,
+      "learning_rate": 4.414951730127571e-06,
+      "loss": 1.0008,
+      "step": 19192
+    },
+    {
+      "epoch": 0.6980034185547515,
+      "grad_norm": 0.0,
+      "learning_rate": 4.413974731117775e-06,
+      "loss": 0.8454,
+      "step": 19193
+    },
+    {
+      "epoch": 0.69803978615849,
+      "grad_norm": 0.0,
+      "learning_rate": 4.412997809606156e-06,
+      "loss": 0.957,
+      "step": 19194
+    },
+    {
+      "epoch": 0.6980761537622286,
+      "grad_norm": 0.0,
+      "learning_rate": 4.412020965606266e-06,
+      "loss": 0.8986,
+      "step": 19195
+    },
+    {
+      "epoch": 0.6981125213659672,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4110441991316574e-06,
+      "loss": 0.9721,
+      "step": 19196
+    },
+    {
+      "epoch": 0.6981488889697058,
+      "grad_norm": 0.0,
+      "learning_rate": 4.410067510195881e-06,
+      "loss": 1.0663,
+      "step": 19197
+    },
+    {
+      "epoch": 0.6981852565734443,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4090908988124855e-06,
+      "loss": 0.9048,
+      "step": 19198
+    },
+    {
+      "epoch": 0.6982216241771829,
+      "grad_norm": 0.0,
+      "learning_rate": 4.408114364995018e-06,
+      "loss": 0.8725,
+      "step": 19199
+    },
+    {
+      "epoch": 0.6982579917809215,
+      "grad_norm": 0.0,
+      "learning_rate": 4.407137908757035e-06,
+      "loss": 0.8306,
+      "step": 19200
+    },
+    {
+      "epoch": 0.6982943593846601,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4061615301120805e-06,
+      "loss": 0.8814,
+      "step": 19201
+    },
+    {
+      "epoch": 0.6983307269883987,
+      "grad_norm": 0.0,
+      "learning_rate": 4.405185229073692e-06,
+      "loss": 0.8528,
+      "step": 19202
+    },
+    {
+      "epoch": 0.6983670945921373,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4042090056554236e-06,
+      "loss": 0.9103,
+      "step": 19203
+    },
+    {
+      "epoch": 0.6984034621958759,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4032328598708154e-06,
+      "loss": 0.9123,
+      "step": 19204
+    },
+    {
+      "epoch": 0.6984398297996145,
+      "grad_norm": 0.0,
+      "learning_rate": 4.40225679173341e-06,
+      "loss": 0.8996,
+      "step": 19205
+    },
+    {
+      "epoch": 0.6984761974033531,
+      "grad_norm": 0.0,
+      "learning_rate": 4.401280801256747e-06,
+      "loss": 0.973,
+      "step": 19206
+    },
+    {
+      "epoch": 0.6985125650070917,
+      "grad_norm": 0.0,
+      "learning_rate": 4.400304888454376e-06,
+      "loss": 0.8767,
+      "step": 19207
+    },
+    {
+      "epoch": 0.6985489326108303,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3993290533398235e-06,
+      "loss": 1.0758,
+      "step": 19208
+    },
+    {
+      "epoch": 0.6985853002145689,
+      "grad_norm": 0.0,
+      "learning_rate": 4.398353295926636e-06,
+      "loss": 0.8817,
+      "step": 19209
+    },
+    {
+      "epoch": 0.6986216678183075,
+      "grad_norm": 0.0,
+      "learning_rate": 4.39737761622835e-06,
+      "loss": 0.922,
+      "step": 19210
+    },
+    {
+      "epoch": 0.6986580354220461,
+      "grad_norm": 0.0,
+      "learning_rate": 4.396402014258501e-06,
+      "loss": 1.0037,
+      "step": 19211
+    },
+    {
+      "epoch": 0.6986944030257847,
+      "grad_norm": 0.0,
+      "learning_rate": 4.395426490030619e-06,
+      "loss": 0.8343,
+      "step": 19212
+    },
+    {
+      "epoch": 0.6987307706295233,
+      "grad_norm": 0.0,
+      "learning_rate": 4.394451043558245e-06,
+      "loss": 0.9027,
+      "step": 19213
+    },
+    {
+      "epoch": 0.6987671382332619,
+      "grad_norm": 0.0,
+      "learning_rate": 4.393475674854911e-06,
+      "loss": 0.9778,
+      "step": 19214
+    },
+    {
+      "epoch": 0.6988035058370003,
+      "grad_norm": 0.0,
+      "learning_rate": 4.392500383934149e-06,
+      "loss": 0.8983,
+      "step": 19215
+    },
+    {
+      "epoch": 0.698839873440739,
+      "grad_norm": 0.0,
+      "learning_rate": 4.391525170809482e-06,
+      "loss": 0.9687,
+      "step": 19216
+    },
+    {
+      "epoch": 0.6988762410444775,
+      "grad_norm": 0.0,
+      "learning_rate": 4.390550035494454e-06,
+      "loss": 0.9922,
+      "step": 19217
+    },
+    {
+      "epoch": 0.6989126086482161,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3895749780025775e-06,
+      "loss": 0.9607,
+      "step": 19218
+    },
+    {
+      "epoch": 0.6989489762519547,
+      "grad_norm": 0.0,
+      "learning_rate": 4.388599998347393e-06,
+      "loss": 1.0494,
+      "step": 19219
+    },
+    {
+      "epoch": 0.6989853438556933,
+      "grad_norm": 0.0,
+      "learning_rate": 4.387625096542425e-06,
+      "loss": 1.0233,
+      "step": 19220
+    },
+    {
+      "epoch": 0.6990217114594319,
+      "grad_norm": 0.0,
+      "learning_rate": 4.386650272601188e-06,
+      "loss": 0.9134,
+      "step": 19221
+    },
+    {
+      "epoch": 0.6990580790631705,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3856755265372185e-06,
+      "loss": 1.0243,
+      "step": 19222
+    },
+    {
+      "epoch": 0.6990944466669091,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3847008583640345e-06,
+      "loss": 0.9613,
+      "step": 19223
+    },
+    {
+      "epoch": 0.6991308142706477,
+      "grad_norm": 0.0,
+      "learning_rate": 4.383726268095161e-06,
+      "loss": 0.9197,
+      "step": 19224
+    },
+    {
+      "epoch": 0.6991671818743863,
+      "grad_norm": 0.0,
+      "learning_rate": 4.382751755744113e-06,
+      "loss": 1.0999,
+      "step": 19225
+    },
+    {
+      "epoch": 0.6992035494781249,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3817773213244165e-06,
+      "loss": 0.9683,
+      "step": 19226
+    },
+    {
+      "epoch": 0.6992399170818635,
+      "grad_norm": 0.0,
+      "learning_rate": 4.380802964849589e-06,
+      "loss": 0.985,
+      "step": 19227
+    },
+    {
+      "epoch": 0.6992762846856021,
+      "grad_norm": 0.0,
+      "learning_rate": 4.379828686333148e-06,
+      "loss": 0.9108,
+      "step": 19228
+    },
+    {
+      "epoch": 0.6993126522893407,
+      "grad_norm": 0.0,
+      "learning_rate": 4.37885448578861e-06,
+      "loss": 0.8483,
+      "step": 19229
+    },
+    {
+      "epoch": 0.6993490198930793,
+      "grad_norm": 0.0,
+      "learning_rate": 4.377880363229491e-06,
+      "loss": 0.9734,
+      "step": 19230
+    },
+    {
+      "epoch": 0.6993853874968179,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3769063186693015e-06,
+      "loss": 0.9171,
+      "step": 19231
+    },
+    {
+      "epoch": 0.6994217551005564,
+      "grad_norm": 0.0,
+      "learning_rate": 4.375932352121562e-06,
+      "loss": 0.9195,
+      "step": 19232
+    },
+    {
+      "epoch": 0.699458122704295,
+      "grad_norm": 0.0,
+      "learning_rate": 4.374958463599783e-06,
+      "loss": 1.0241,
+      "step": 19233
+    },
+    {
+      "epoch": 0.6994944903080336,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3739846531174726e-06,
+      "loss": 0.8773,
+      "step": 19234
+    },
+    {
+      "epoch": 0.6995308579117722,
+      "grad_norm": 0.0,
+      "learning_rate": 4.37301092068814e-06,
+      "loss": 0.9352,
+      "step": 19235
+    },
+    {
+      "epoch": 0.6995672255155108,
+      "grad_norm": 0.0,
+      "learning_rate": 4.372037266325304e-06,
+      "loss": 0.9949,
+      "step": 19236
+    },
+    {
+      "epoch": 0.6996035931192494,
+      "grad_norm": 0.0,
+      "learning_rate": 4.371063690042459e-06,
+      "loss": 0.8236,
+      "step": 19237
+    },
+    {
+      "epoch": 0.699639960722988,
+      "grad_norm": 0.0,
+      "learning_rate": 4.370090191853123e-06,
+      "loss": 0.889,
+      "step": 19238
+    },
+    {
+      "epoch": 0.6996763283267265,
+      "grad_norm": 0.0,
+      "learning_rate": 4.369116771770796e-06,
+      "loss": 0.9487,
+      "step": 19239
+    },
+    {
+      "epoch": 0.6997126959304651,
+      "grad_norm": 0.0,
+      "learning_rate": 4.368143429808986e-06,
+      "loss": 0.9938,
+      "step": 19240
+    },
+    {
+      "epoch": 0.6997490635342037,
+      "grad_norm": 0.0,
+      "learning_rate": 4.367170165981196e-06,
+      "loss": 1.0193,
+      "step": 19241
+    },
+    {
+      "epoch": 0.6997854311379423,
+      "grad_norm": 0.0,
+      "learning_rate": 4.366196980300927e-06,
+      "loss": 0.9344,
+      "step": 19242
+    },
+    {
+      "epoch": 0.6998217987416809,
+      "grad_norm": 0.0,
+      "learning_rate": 4.365223872781682e-06,
+      "loss": 0.9487,
+      "step": 19243
+    },
+    {
+      "epoch": 0.6998581663454195,
+      "grad_norm": 0.0,
+      "learning_rate": 4.364250843436957e-06,
+      "loss": 0.9023,
+      "step": 19244
+    },
+    {
+      "epoch": 0.6998945339491581,
+      "grad_norm": 0.0,
+      "learning_rate": 4.363277892280259e-06,
+      "loss": 0.8934,
+      "step": 19245
+    },
+    {
+      "epoch": 0.6999309015528967,
+      "grad_norm": 0.0,
+      "learning_rate": 4.362305019325084e-06,
+      "loss": 0.9361,
+      "step": 19246
+    },
+    {
+      "epoch": 0.6999672691566353,
+      "grad_norm": 0.0,
+      "learning_rate": 4.361332224584926e-06,
+      "loss": 1.0453,
+      "step": 19247
+    },
+    {
+      "epoch": 0.7000036367603738,
+      "grad_norm": 0.0,
+      "learning_rate": 4.360359508073281e-06,
+      "loss": 0.9438,
+      "step": 19248
+    },
+    {
+      "epoch": 0.7000400043641124,
+      "grad_norm": 0.0,
+      "learning_rate": 4.359386869803654e-06,
+      "loss": 1.0435,
+      "step": 19249
+    },
+    {
+      "epoch": 0.700076371967851,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3584143097895226e-06,
+      "loss": 0.9664,
+      "step": 19250
+    },
+    {
+      "epoch": 0.7001127395715896,
+      "grad_norm": 0.0,
+      "learning_rate": 4.357441828044392e-06,
+      "loss": 0.9384,
+      "step": 19251
+    },
+    {
+      "epoch": 0.7001491071753282,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3564694245817505e-06,
+      "loss": 0.9853,
+      "step": 19252
+    },
+    {
+      "epoch": 0.7001854747790668,
+      "grad_norm": 0.0,
+      "learning_rate": 4.355497099415089e-06,
+      "loss": 0.9292,
+      "step": 19253
+    },
+    {
+      "epoch": 0.7002218423828054,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3545248525578935e-06,
+      "loss": 0.9436,
+      "step": 19254
+    },
+    {
+      "epoch": 0.700258209986544,
+      "grad_norm": 0.0,
+      "learning_rate": 4.353552684023662e-06,
+      "loss": 0.9767,
+      "step": 19255
+    },
+    {
+      "epoch": 0.7002945775902826,
+      "grad_norm": 0.0,
+      "learning_rate": 4.352580593825869e-06,
+      "loss": 0.9846,
+      "step": 19256
+    },
+    {
+      "epoch": 0.7003309451940212,
+      "grad_norm": 0.0,
+      "learning_rate": 4.351608581978012e-06,
+      "loss": 0.9478,
+      "step": 19257
+    },
+    {
+      "epoch": 0.7003673127977598,
+      "grad_norm": 0.0,
+      "learning_rate": 4.350636648493571e-06,
+      "loss": 0.9802,
+      "step": 19258
+    },
+    {
+      "epoch": 0.7004036804014984,
+      "grad_norm": 0.0,
+      "learning_rate": 4.349664793386032e-06,
+      "loss": 0.9604,
+      "step": 19259
+    },
+    {
+      "epoch": 0.700440048005237,
+      "grad_norm": 0.0,
+      "learning_rate": 4.348693016668875e-06,
+      "loss": 0.9641,
+      "step": 19260
+    },
+    {
+      "epoch": 0.7004764156089756,
+      "grad_norm": 0.0,
+      "learning_rate": 4.347721318355582e-06,
+      "loss": 0.8409,
+      "step": 19261
+    },
+    {
+      "epoch": 0.7005127832127142,
+      "grad_norm": 0.0,
+      "learning_rate": 4.346749698459643e-06,
+      "loss": 0.9484,
+      "step": 19262
+    },
+    {
+      "epoch": 0.7005491508164527,
+      "grad_norm": 0.0,
+      "learning_rate": 4.345778156994526e-06,
+      "loss": 0.9072,
+      "step": 19263
+    },
+    {
+      "epoch": 0.7005855184201913,
+      "grad_norm": 0.0,
+      "learning_rate": 4.344806693973715e-06,
+      "loss": 0.9744,
+      "step": 19264
+    },
+    {
+      "epoch": 0.7006218860239298,
+      "grad_norm": 0.0,
+      "learning_rate": 4.34383530941069e-06,
+      "loss": 0.9604,
+      "step": 19265
+    },
+    {
+      "epoch": 0.7006582536276684,
+      "grad_norm": 0.0,
+      "learning_rate": 4.342864003318924e-06,
+      "loss": 0.8359,
+      "step": 19266
+    },
+    {
+      "epoch": 0.700694621231407,
+      "grad_norm": 0.0,
+      "learning_rate": 4.34189277571189e-06,
+      "loss": 0.919,
+      "step": 19267
+    },
+    {
+      "epoch": 0.7007309888351456,
+      "grad_norm": 0.0,
+      "learning_rate": 4.340921626603073e-06,
+      "loss": 0.8595,
+      "step": 19268
+    },
+    {
+      "epoch": 0.7007673564388842,
+      "grad_norm": 0.0,
+      "learning_rate": 4.339950556005933e-06,
+      "loss": 0.7295,
+      "step": 19269
+    },
+    {
+      "epoch": 0.7008037240426228,
+      "grad_norm": 0.0,
+      "learning_rate": 4.33897956393395e-06,
+      "loss": 0.9418,
+      "step": 19270
+    },
+    {
+      "epoch": 0.7008400916463614,
+      "grad_norm": 0.0,
+      "learning_rate": 4.338008650400596e-06,
+      "loss": 1.0385,
+      "step": 19271
+    },
+    {
+      "epoch": 0.7008764592501,
+      "grad_norm": 0.0,
+      "learning_rate": 4.337037815419337e-06,
+      "loss": 1.0865,
+      "step": 19272
+    },
+    {
+      "epoch": 0.7009128268538386,
+      "grad_norm": 0.0,
+      "learning_rate": 4.336067059003641e-06,
+      "loss": 1.0183,
+      "step": 19273
+    },
+    {
+      "epoch": 0.7009491944575772,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3350963811669855e-06,
+      "loss": 0.9135,
+      "step": 19274
+    },
+    {
+      "epoch": 0.7009855620613158,
+      "grad_norm": 0.0,
+      "learning_rate": 4.334125781922822e-06,
+      "loss": 0.905,
+      "step": 19275
+    },
+    {
+      "epoch": 0.7010219296650544,
+      "grad_norm": 0.0,
+      "learning_rate": 4.33315526128463e-06,
+      "loss": 0.9598,
+      "step": 19276
+    },
+    {
+      "epoch": 0.701058297268793,
+      "grad_norm": 0.0,
+      "learning_rate": 4.332184819265867e-06,
+      "loss": 0.9404,
+      "step": 19277
+    },
+    {
+      "epoch": 0.7010946648725316,
+      "grad_norm": 0.0,
+      "learning_rate": 4.331214455879998e-06,
+      "loss": 0.8177,
+      "step": 19278
+    },
+    {
+      "epoch": 0.7011310324762702,
+      "grad_norm": 0.0,
+      "learning_rate": 4.330244171140486e-06,
+      "loss": 0.8701,
+      "step": 19279
+    },
+    {
+      "epoch": 0.7011674000800088,
+      "grad_norm": 0.0,
+      "learning_rate": 4.329273965060789e-06,
+      "loss": 0.8927,
+      "step": 19280
+    },
+    {
+      "epoch": 0.7012037676837474,
+      "grad_norm": 0.0,
+      "learning_rate": 4.328303837654376e-06,
+      "loss": 0.8948,
+      "step": 19281
+    },
+    {
+      "epoch": 0.7012401352874859,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3273337889346935e-06,
+      "loss": 0.8366,
+      "step": 19282
+    },
+    {
+      "epoch": 0.7012765028912245,
+      "grad_norm": 0.0,
+      "learning_rate": 4.32636381891521e-06,
+      "loss": 1.0464,
+      "step": 19283
+    },
+    {
+      "epoch": 0.701312870494963,
+      "grad_norm": 0.0,
+      "learning_rate": 4.325393927609378e-06,
+      "loss": 0.9437,
+      "step": 19284
+    },
+    {
+      "epoch": 0.7013492380987016,
+      "grad_norm": 0.0,
+      "learning_rate": 4.324424115030654e-06,
+      "loss": 0.9544,
+      "step": 19285
+    },
+    {
+      "epoch": 0.7013856057024402,
+      "grad_norm": 0.0,
+      "learning_rate": 4.32345438119249e-06,
+      "loss": 0.9063,
+      "step": 19286
+    },
+    {
+      "epoch": 0.7014219733061788,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3224847261083495e-06,
+      "loss": 0.9273,
+      "step": 19287
+    },
+    {
+      "epoch": 0.7014583409099174,
+      "grad_norm": 0.0,
+      "learning_rate": 4.321515149791671e-06,
+      "loss": 0.9027,
+      "step": 19288
+    },
+    {
+      "epoch": 0.701494708513656,
+      "grad_norm": 0.0,
+      "learning_rate": 4.320545652255916e-06,
+      "loss": 0.9472,
+      "step": 19289
+    },
+    {
+      "epoch": 0.7015310761173946,
+      "grad_norm": 0.0,
+      "learning_rate": 4.319576233514533e-06,
+      "loss": 0.9482,
+      "step": 19290
+    },
+    {
+      "epoch": 0.7015674437211332,
+      "grad_norm": 0.0,
+      "learning_rate": 4.318606893580969e-06,
+      "loss": 0.9602,
+      "step": 19291
+    },
+    {
+      "epoch": 0.7016038113248718,
+      "grad_norm": 0.0,
+      "learning_rate": 4.317637632468671e-06,
+      "loss": 0.7617,
+      "step": 19292
+    },
+    {
+      "epoch": 0.7016401789286104,
+      "grad_norm": 0.0,
+      "learning_rate": 4.316668450191092e-06,
+      "loss": 0.843,
+      "step": 19293
+    },
+    {
+      "epoch": 0.701676546532349,
+      "grad_norm": 0.0,
+      "learning_rate": 4.315699346761674e-06,
+      "loss": 0.8753,
+      "step": 19294
+    },
+    {
+      "epoch": 0.7017129141360876,
+      "grad_norm": 0.0,
+      "learning_rate": 4.314730322193863e-06,
+      "loss": 1.0852,
+      "step": 19295
+    },
+    {
+      "epoch": 0.7017492817398262,
+      "grad_norm": 0.0,
+      "learning_rate": 4.313761376501102e-06,
+      "loss": 0.9043,
+      "step": 19296
+    },
+    {
+      "epoch": 0.7017856493435648,
+      "grad_norm": 0.0,
+      "learning_rate": 4.312792509696835e-06,
+      "loss": 0.9647,
+      "step": 19297
+    },
+    {
+      "epoch": 0.7018220169473034,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3118237217945024e-06,
+      "loss": 1.0908,
+      "step": 19298
+    },
+    {
+      "epoch": 0.7018583845510419,
+      "grad_norm": 0.0,
+      "learning_rate": 4.310855012807542e-06,
+      "loss": 0.9731,
+      "step": 19299
+    },
+    {
+      "epoch": 0.7018947521547805,
+      "grad_norm": 0.0,
+      "learning_rate": 4.309886382749403e-06,
+      "loss": 0.859,
+      "step": 19300
+    },
+    {
+      "epoch": 0.7019311197585191,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3089178316335115e-06,
+      "loss": 0.9073,
+      "step": 19301
+    },
+    {
+      "epoch": 0.7019674873622577,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3079493594733126e-06,
+      "loss": 0.8868,
+      "step": 19302
+    },
+    {
+      "epoch": 0.7020038549659963,
+      "grad_norm": 0.0,
+      "learning_rate": 4.30698096628224e-06,
+      "loss": 0.9372,
+      "step": 19303
+    },
+    {
+      "epoch": 0.7020402225697349,
+      "grad_norm": 0.0,
+      "learning_rate": 4.306012652073731e-06,
+      "loss": 0.9918,
+      "step": 19304
+    },
+    {
+      "epoch": 0.7020765901734735,
+      "grad_norm": 0.0,
+      "learning_rate": 4.305044416861213e-06,
+      "loss": 0.9122,
+      "step": 19305
+    },
+    {
+      "epoch": 0.702112957777212,
+      "grad_norm": 0.0,
+      "learning_rate": 4.304076260658127e-06,
+      "loss": 0.9711,
+      "step": 19306
+    },
+    {
+      "epoch": 0.7021493253809507,
+      "grad_norm": 0.0,
+      "learning_rate": 4.303108183477902e-06,
+      "loss": 0.847,
+      "step": 19307
+    },
+    {
+      "epoch": 0.7021856929846892,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3021401853339676e-06,
+      "loss": 0.825,
+      "step": 19308
+    },
+    {
+      "epoch": 0.7022220605884278,
+      "grad_norm": 0.0,
+      "learning_rate": 4.301172266239754e-06,
+      "loss": 1.0858,
+      "step": 19309
+    },
+    {
+      "epoch": 0.7022584281921664,
+      "grad_norm": 0.0,
+      "learning_rate": 4.300204426208689e-06,
+      "loss": 1.0594,
+      "step": 19310
+    },
+    {
+      "epoch": 0.702294795795905,
+      "grad_norm": 0.0,
+      "learning_rate": 4.299236665254199e-06,
+      "loss": 0.9622,
+      "step": 19311
+    },
+    {
+      "epoch": 0.7023311633996436,
+      "grad_norm": 0.0,
+      "learning_rate": 4.298268983389714e-06,
+      "loss": 0.9118,
+      "step": 19312
+    },
+    {
+      "epoch": 0.7023675310033822,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2973013806286575e-06,
+      "loss": 0.9373,
+      "step": 19313
+    },
+    {
+      "epoch": 0.7024038986071208,
+      "grad_norm": 0.0,
+      "learning_rate": 4.296333856984454e-06,
+      "loss": 1.0166,
+      "step": 19314
+    },
+    {
+      "epoch": 0.7024402662108594,
+      "grad_norm": 0.0,
+      "learning_rate": 4.295366412470521e-06,
+      "loss": 0.9953,
+      "step": 19315
+    },
+    {
+      "epoch": 0.7024766338145979,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2943990471002925e-06,
+      "loss": 0.9697,
+      "step": 19316
+    },
+    {
+      "epoch": 0.7025130014183365,
+      "grad_norm": 0.0,
+      "learning_rate": 4.293431760887179e-06,
+      "loss": 0.8814,
+      "step": 19317
+    },
+    {
+      "epoch": 0.7025493690220751,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2924645538446e-06,
+      "loss": 0.9286,
+      "step": 19318
+    },
+    {
+      "epoch": 0.7025857366258137,
+      "grad_norm": 0.0,
+      "learning_rate": 4.291497425985983e-06,
+      "loss": 0.8756,
+      "step": 19319
+    },
+    {
+      "epoch": 0.7026221042295523,
+      "grad_norm": 0.0,
+      "learning_rate": 4.290530377324734e-06,
+      "loss": 1.0487,
+      "step": 19320
+    },
+    {
+      "epoch": 0.7026584718332909,
+      "grad_norm": 0.0,
+      "learning_rate": 4.289563407874279e-06,
+      "loss": 0.9423,
+      "step": 19321
+    },
+    {
+      "epoch": 0.7026948394370295,
+      "grad_norm": 0.0,
+      "learning_rate": 4.288596517648029e-06,
+      "loss": 0.9739,
+      "step": 19322
+    },
+    {
+      "epoch": 0.7027312070407681,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2876297066594e-06,
+      "loss": 0.8831,
+      "step": 19323
+    },
+    {
+      "epoch": 0.7027675746445067,
+      "grad_norm": 0.0,
+      "learning_rate": 4.286662974921799e-06,
+      "loss": 1.0008,
+      "step": 19324
+    },
+    {
+      "epoch": 0.7028039422482453,
+      "grad_norm": 0.0,
+      "learning_rate": 4.285696322448648e-06,
+      "loss": 0.9375,
+      "step": 19325
+    },
+    {
+      "epoch": 0.7028403098519839,
+      "grad_norm": 0.0,
+      "learning_rate": 4.284729749253353e-06,
+      "loss": 0.9761,
+      "step": 19326
+    },
+    {
+      "epoch": 0.7028766774557225,
+      "grad_norm": 0.0,
+      "learning_rate": 4.283763255349322e-06,
+      "loss": 0.9364,
+      "step": 19327
+    },
+    {
+      "epoch": 0.7029130450594611,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2827968407499675e-06,
+      "loss": 0.8954,
+      "step": 19328
+    },
+    {
+      "epoch": 0.7029494126631997,
+      "grad_norm": 0.0,
+      "learning_rate": 4.281830505468694e-06,
+      "loss": 0.9801,
+      "step": 19329
+    },
+    {
+      "epoch": 0.7029857802669383,
+      "grad_norm": 0.0,
+      "learning_rate": 4.280864249518905e-06,
+      "loss": 0.8727,
+      "step": 19330
+    },
+    {
+      "epoch": 0.7030221478706769,
+      "grad_norm": 0.0,
+      "learning_rate": 4.279898072914016e-06,
+      "loss": 0.8736,
+      "step": 19331
+    },
+    {
+      "epoch": 0.7030585154744153,
+      "grad_norm": 0.0,
+      "learning_rate": 4.278931975667424e-06,
+      "loss": 0.8571,
+      "step": 19332
+    },
+    {
+      "epoch": 0.7030948830781539,
+      "grad_norm": 0.0,
+      "learning_rate": 4.277965957792534e-06,
+      "loss": 1.055,
+      "step": 19333
+    },
+    {
+      "epoch": 0.7031312506818925,
+      "grad_norm": 0.0,
+      "learning_rate": 4.277000019302744e-06,
+      "loss": 0.8007,
+      "step": 19334
+    },
+    {
+      "epoch": 0.7031676182856311,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2760341602114665e-06,
+      "loss": 0.8996,
+      "step": 19335
+    },
+    {
+      "epoch": 0.7032039858893697,
+      "grad_norm": 0.0,
+      "learning_rate": 4.27506838053209e-06,
+      "loss": 1.0205,
+      "step": 19336
+    },
+    {
+      "epoch": 0.7032403534931083,
+      "grad_norm": 0.0,
+      "learning_rate": 4.274102680278013e-06,
+      "loss": 0.793,
+      "step": 19337
+    },
+    {
+      "epoch": 0.7032767210968469,
+      "grad_norm": 0.0,
+      "learning_rate": 4.273137059462642e-06,
+      "loss": 0.9419,
+      "step": 19338
+    },
+    {
+      "epoch": 0.7033130887005855,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2721715180993686e-06,
+      "loss": 0.851,
+      "step": 19339
+    },
+    {
+      "epoch": 0.7033494563043241,
+      "grad_norm": 0.0,
+      "learning_rate": 4.271206056201588e-06,
+      "loss": 0.9327,
+      "step": 19340
+    },
+    {
+      "epoch": 0.7033858239080627,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2702406737826956e-06,
+      "loss": 0.8961,
+      "step": 19341
+    },
+    {
+      "epoch": 0.7034221915118013,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2692753708560845e-06,
+      "loss": 0.8316,
+      "step": 19342
+    },
+    {
+      "epoch": 0.7034585591155399,
+      "grad_norm": 0.0,
+      "learning_rate": 4.268310147435143e-06,
+      "loss": 0.8691,
+      "step": 19343
+    },
+    {
+      "epoch": 0.7034949267192785,
+      "grad_norm": 0.0,
+      "learning_rate": 4.26734500353327e-06,
+      "loss": 0.9373,
+      "step": 19344
+    },
+    {
+      "epoch": 0.7035312943230171,
+      "grad_norm": 0.0,
+      "learning_rate": 4.266379939163851e-06,
+      "loss": 0.9174,
+      "step": 19345
+    },
+    {
+      "epoch": 0.7035676619267557,
+      "grad_norm": 0.0,
+      "learning_rate": 4.265414954340276e-06,
+      "loss": 0.9899,
+      "step": 19346
+    },
+    {
+      "epoch": 0.7036040295304943,
+      "grad_norm": 0.0,
+      "learning_rate": 4.264450049075928e-06,
+      "loss": 0.9423,
+      "step": 19347
+    },
+    {
+      "epoch": 0.7036403971342329,
+      "grad_norm": 0.0,
+      "learning_rate": 4.263485223384205e-06,
+      "loss": 0.8898,
+      "step": 19348
+    },
+    {
+      "epoch": 0.7036767647379714,
+      "grad_norm": 0.0,
+      "learning_rate": 4.262520477278478e-06,
+      "loss": 0.9667,
+      "step": 19349
+    },
+    {
+      "epoch": 0.70371313234171,
+      "grad_norm": 0.0,
+      "learning_rate": 4.261555810772143e-06,
+      "loss": 1.001,
+      "step": 19350
+    },
+    {
+      "epoch": 0.7037494999454486,
+      "grad_norm": 0.0,
+      "learning_rate": 4.260591223878579e-06,
+      "loss": 0.8451,
+      "step": 19351
+    },
+    {
+      "epoch": 0.7037858675491871,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2596267166111705e-06,
+      "loss": 0.8622,
+      "step": 19352
+    },
+    {
+      "epoch": 0.7038222351529257,
+      "grad_norm": 0.0,
+      "learning_rate": 4.258662288983292e-06,
+      "loss": 0.7784,
+      "step": 19353
+    },
+    {
+      "epoch": 0.7038586027566643,
+      "grad_norm": 0.0,
+      "learning_rate": 4.257697941008334e-06,
+      "loss": 0.9771,
+      "step": 19354
+    },
+    {
+      "epoch": 0.7038949703604029,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2567336726996674e-06,
+      "loss": 0.9247,
+      "step": 19355
+    },
+    {
+      "epoch": 0.7039313379641415,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2557694840706695e-06,
+      "loss": 0.9498,
+      "step": 19356
+    },
+    {
+      "epoch": 0.7039677055678801,
+      "grad_norm": 0.0,
+      "learning_rate": 4.254805375134723e-06,
+      "loss": 0.9162,
+      "step": 19357
+    },
+    {
+      "epoch": 0.7040040731716187,
+      "grad_norm": 0.0,
+      "learning_rate": 4.253841345905201e-06,
+      "loss": 0.8824,
+      "step": 19358
+    },
+    {
+      "epoch": 0.7040404407753573,
+      "grad_norm": 0.0,
+      "learning_rate": 4.252877396395477e-06,
+      "loss": 1.0415,
+      "step": 19359
+    },
+    {
+      "epoch": 0.7040768083790959,
+      "grad_norm": 0.0,
+      "learning_rate": 4.251913526618922e-06,
+      "loss": 0.8857,
+      "step": 19360
+    },
+    {
+      "epoch": 0.7041131759828345,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2509497365889185e-06,
+      "loss": 0.9372,
+      "step": 19361
+    },
+    {
+      "epoch": 0.7041495435865731,
+      "grad_norm": 0.0,
+      "learning_rate": 4.249986026318823e-06,
+      "loss": 0.8804,
+      "step": 19362
+    },
+    {
+      "epoch": 0.7041859111903117,
+      "grad_norm": 0.0,
+      "learning_rate": 4.249022395822017e-06,
+      "loss": 1.0396,
+      "step": 19363
+    },
+    {
+      "epoch": 0.7042222787940503,
+      "grad_norm": 0.0,
+      "learning_rate": 4.248058845111866e-06,
+      "loss": 0.8725,
+      "step": 19364
+    },
+    {
+      "epoch": 0.7042586463977889,
+      "grad_norm": 0.0,
+      "learning_rate": 4.247095374201736e-06,
+      "loss": 1.0579,
+      "step": 19365
+    },
+    {
+      "epoch": 0.7042950140015274,
+      "grad_norm": 0.0,
+      "learning_rate": 4.246131983104993e-06,
+      "loss": 0.875,
+      "step": 19366
+    },
+    {
+      "epoch": 0.704331381605266,
+      "grad_norm": 0.0,
+      "learning_rate": 4.245168671835013e-06,
+      "loss": 0.9433,
+      "step": 19367
+    },
+    {
+      "epoch": 0.7043677492090046,
+      "grad_norm": 0.0,
+      "learning_rate": 4.244205440405144e-06,
+      "loss": 0.8795,
+      "step": 19368
+    },
+    {
+      "epoch": 0.7044041168127432,
+      "grad_norm": 0.0,
+      "learning_rate": 4.243242288828763e-06,
+      "loss": 0.8414,
+      "step": 19369
+    },
+    {
+      "epoch": 0.7044404844164818,
+      "grad_norm": 0.0,
+      "learning_rate": 4.242279217119226e-06,
+      "loss": 0.943,
+      "step": 19370
+    },
+    {
+      "epoch": 0.7044768520202204,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2413162252898964e-06,
+      "loss": 0.7867,
+      "step": 19371
+    },
+    {
+      "epoch": 0.704513219623959,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2403533133541305e-06,
+      "loss": 0.913,
+      "step": 19372
+    },
+    {
+      "epoch": 0.7045495872276976,
+      "grad_norm": 0.0,
+      "learning_rate": 4.239390481325297e-06,
+      "loss": 0.9899,
+      "step": 19373
+    },
+    {
+      "epoch": 0.7045859548314362,
+      "grad_norm": 0.0,
+      "learning_rate": 4.238427729216744e-06,
+      "loss": 0.9231,
+      "step": 19374
+    },
+    {
+      "epoch": 0.7046223224351748,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2374650570418285e-06,
+      "loss": 0.8656,
+      "step": 19375
+    },
+    {
+      "epoch": 0.7046586900389133,
+      "grad_norm": 0.0,
+      "learning_rate": 4.236502464813913e-06,
+      "loss": 0.82,
+      "step": 19376
+    },
+    {
+      "epoch": 0.704695057642652,
+      "grad_norm": 0.0,
+      "learning_rate": 4.235539952546348e-06,
+      "loss": 0.9445,
+      "step": 19377
+    },
+    {
+      "epoch": 0.7047314252463905,
+      "grad_norm": 0.0,
+      "learning_rate": 4.234577520252488e-06,
+      "loss": 0.8972,
+      "step": 19378
+    },
+    {
+      "epoch": 0.7047677928501291,
+      "grad_norm": 0.0,
+      "learning_rate": 4.23361516794568e-06,
+      "loss": 1.0003,
+      "step": 19379
+    },
+    {
+      "epoch": 0.7048041604538677,
+      "grad_norm": 0.0,
+      "learning_rate": 4.232652895639289e-06,
+      "loss": 0.9285,
+      "step": 19380
+    },
+    {
+      "epoch": 0.7048405280576063,
+      "grad_norm": 0.0,
+      "learning_rate": 4.231690703346648e-06,
+      "loss": 1.0096,
+      "step": 19381
+    },
+    {
+      "epoch": 0.7048768956613449,
+      "grad_norm": 0.0,
+      "learning_rate": 4.230728591081118e-06,
+      "loss": 0.9168,
+      "step": 19382
+    },
+    {
+      "epoch": 0.7049132632650834,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2297665588560434e-06,
+      "loss": 0.9877,
+      "step": 19383
+    },
+    {
+      "epoch": 0.704949630868822,
+      "grad_norm": 0.0,
+      "learning_rate": 4.228804606684771e-06,
+      "loss": 1.0684,
+      "step": 19384
+    },
+    {
+      "epoch": 0.7049859984725606,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2278427345806426e-06,
+      "loss": 1.0012,
+      "step": 19385
+    },
+    {
+      "epoch": 0.7050223660762992,
+      "grad_norm": 0.0,
+      "learning_rate": 4.226880942557015e-06,
+      "loss": 1.0477,
+      "step": 19386
+    },
+    {
+      "epoch": 0.7050587336800378,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2259192306272145e-06,
+      "loss": 0.9458,
+      "step": 19387
+    },
+    {
+      "epoch": 0.7050951012837764,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2249575988045965e-06,
+      "loss": 0.8999,
+      "step": 19388
+    },
+    {
+      "epoch": 0.705131468887515,
+      "grad_norm": 0.0,
+      "learning_rate": 4.223996047102498e-06,
+      "loss": 0.868,
+      "step": 19389
+    },
+    {
+      "epoch": 0.7051678364912536,
+      "grad_norm": 0.0,
+      "learning_rate": 4.223034575534259e-06,
+      "loss": 0.9689,
+      "step": 19390
+    },
+    {
+      "epoch": 0.7052042040949922,
+      "grad_norm": 0.0,
+      "learning_rate": 4.222073184113216e-06,
+      "loss": 0.8717,
+      "step": 19391
+    },
+    {
+      "epoch": 0.7052405716987308,
+      "grad_norm": 0.0,
+      "learning_rate": 4.221111872852713e-06,
+      "loss": 1.0239,
+      "step": 19392
+    },
+    {
+      "epoch": 0.7052769393024694,
+      "grad_norm": 0.0,
+      "learning_rate": 4.220150641766088e-06,
+      "loss": 0.9938,
+      "step": 19393
+    },
+    {
+      "epoch": 0.705313306906208,
+      "grad_norm": 0.0,
+      "learning_rate": 4.219189490866665e-06,
+      "loss": 0.9993,
+      "step": 19394
+    },
+    {
+      "epoch": 0.7053496745099466,
+      "grad_norm": 0.0,
+      "learning_rate": 4.218228420167788e-06,
+      "loss": 0.9243,
+      "step": 19395
+    },
+    {
+      "epoch": 0.7053860421136852,
+      "grad_norm": 0.0,
+      "learning_rate": 4.217267429682789e-06,
+      "loss": 0.8112,
+      "step": 19396
+    },
+    {
+      "epoch": 0.7054224097174238,
+      "grad_norm": 0.0,
+      "learning_rate": 4.216306519425e-06,
+      "loss": 0.9703,
+      "step": 19397
+    },
+    {
+      "epoch": 0.7054587773211624,
+      "grad_norm": 0.0,
+      "learning_rate": 4.215345689407748e-06,
+      "loss": 0.9355,
+      "step": 19398
+    },
+    {
+      "epoch": 0.705495144924901,
+      "grad_norm": 0.0,
+      "learning_rate": 4.214384939644375e-06,
+      "loss": 0.8675,
+      "step": 19399
+    },
+    {
+      "epoch": 0.7055315125286394,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2134242701481955e-06,
+      "loss": 0.9411,
+      "step": 19400
+    },
+    {
+      "epoch": 0.705567880132378,
+      "grad_norm": 0.0,
+      "learning_rate": 4.212463680932547e-06,
+      "loss": 0.9716,
+      "step": 19401
+    },
+    {
+      "epoch": 0.7056042477361166,
+      "grad_norm": 0.0,
+      "learning_rate": 4.211503172010753e-06,
+      "loss": 0.8062,
+      "step": 19402
+    },
+    {
+      "epoch": 0.7056406153398552,
+      "grad_norm": 0.0,
+      "learning_rate": 4.210542743396141e-06,
+      "loss": 1.0023,
+      "step": 19403
+    },
+    {
+      "epoch": 0.7056769829435938,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2095823951020294e-06,
+      "loss": 1.0795,
+      "step": 19404
+    },
+    {
+      "epoch": 0.7057133505473324,
+      "grad_norm": 0.0,
+      "learning_rate": 4.208622127141752e-06,
+      "loss": 0.9095,
+      "step": 19405
+    },
+    {
+      "epoch": 0.705749718151071,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2076619395286244e-06,
+      "loss": 0.9718,
+      "step": 19406
+    },
+    {
+      "epoch": 0.7057860857548096,
+      "grad_norm": 0.0,
+      "learning_rate": 4.206701832275969e-06,
+      "loss": 0.8922,
+      "step": 19407
+    },
+    {
+      "epoch": 0.7058224533585482,
+      "grad_norm": 0.0,
+      "learning_rate": 4.205741805397107e-06,
+      "loss": 0.9834,
+      "step": 19408
+    },
+    {
+      "epoch": 0.7058588209622868,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2047818589053546e-06,
+      "loss": 0.9834,
+      "step": 19409
+    },
+    {
+      "epoch": 0.7058951885660254,
+      "grad_norm": 0.0,
+      "learning_rate": 4.20382199281403e-06,
+      "loss": 0.9016,
+      "step": 19410
+    },
+    {
+      "epoch": 0.705931556169764,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2028622071364535e-06,
+      "loss": 1.0266,
+      "step": 19411
+    },
+    {
+      "epoch": 0.7059679237735026,
+      "grad_norm": 0.0,
+      "learning_rate": 4.201902501885944e-06,
+      "loss": 1.0312,
+      "step": 19412
+    },
+    {
+      "epoch": 0.7060042913772412,
+      "grad_norm": 0.0,
+      "learning_rate": 4.200942877075802e-06,
+      "loss": 0.944,
+      "step": 19413
+    },
+    {
+      "epoch": 0.7060406589809798,
+      "grad_norm": 0.0,
+      "learning_rate": 4.199983332719355e-06,
+      "loss": 0.9157,
+      "step": 19414
+    },
+    {
+      "epoch": 0.7060770265847184,
+      "grad_norm": 0.0,
+      "learning_rate": 4.199023868829909e-06,
+      "loss": 0.8984,
+      "step": 19415
+    },
+    {
+      "epoch": 0.706113394188457,
+      "grad_norm": 0.0,
+      "learning_rate": 4.198064485420777e-06,
+      "loss": 1.0761,
+      "step": 19416
+    },
+    {
+      "epoch": 0.7061497617921955,
+      "grad_norm": 0.0,
+      "learning_rate": 4.197105182505265e-06,
+      "loss": 0.9138,
+      "step": 19417
+    },
+    {
+      "epoch": 0.7061861293959341,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1961459600966916e-06,
+      "loss": 0.9562,
+      "step": 19418
+    },
+    {
+      "epoch": 0.7062224969996727,
+      "grad_norm": 0.0,
+      "learning_rate": 4.195186818208351e-06,
+      "loss": 0.926,
+      "step": 19419
+    },
+    {
+      "epoch": 0.7062588646034113,
+      "grad_norm": 0.0,
+      "learning_rate": 4.194227756853562e-06,
+      "loss": 0.9395,
+      "step": 19420
+    },
+    {
+      "epoch": 0.7062952322071498,
+      "grad_norm": 0.0,
+      "learning_rate": 4.193268776045625e-06,
+      "loss": 0.985,
+      "step": 19421
+    },
+    {
+      "epoch": 0.7063315998108884,
+      "grad_norm": 0.0,
+      "learning_rate": 4.192309875797845e-06,
+      "loss": 0.9521,
+      "step": 19422
+    },
+    {
+      "epoch": 0.706367967414627,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1913510561235225e-06,
+      "loss": 0.9593,
+      "step": 19423
+    },
+    {
+      "epoch": 0.7064043350183656,
+      "grad_norm": 0.0,
+      "learning_rate": 4.190392317035966e-06,
+      "loss": 1.0017,
+      "step": 19424
+    },
+    {
+      "epoch": 0.7064407026221042,
+      "grad_norm": 0.0,
+      "learning_rate": 4.189433658548473e-06,
+      "loss": 0.8714,
+      "step": 19425
+    },
+    {
+      "epoch": 0.7064770702258428,
+      "grad_norm": 0.0,
+      "learning_rate": 4.188475080674345e-06,
+      "loss": 0.9197,
+      "step": 19426
+    },
+    {
+      "epoch": 0.7065134378295814,
+      "grad_norm": 0.0,
+      "learning_rate": 4.187516583426874e-06,
+      "loss": 0.8835,
+      "step": 19427
+    },
+    {
+      "epoch": 0.70654980543332,
+      "grad_norm": 0.0,
+      "learning_rate": 4.186558166819372e-06,
+      "loss": 0.9561,
+      "step": 19428
+    },
+    {
+      "epoch": 0.7065861730370586,
+      "grad_norm": 0.0,
+      "learning_rate": 4.18559983086512e-06,
+      "loss": 0.8983,
+      "step": 19429
+    },
+    {
+      "epoch": 0.7066225406407972,
+      "grad_norm": 0.0,
+      "learning_rate": 4.184641575577425e-06,
+      "loss": 0.8464,
+      "step": 19430
+    },
+    {
+      "epoch": 0.7066589082445358,
+      "grad_norm": 0.0,
+      "learning_rate": 4.18368340096958e-06,
+      "loss": 0.9515,
+      "step": 19431
+    },
+    {
+      "epoch": 0.7066952758482744,
+      "grad_norm": 0.0,
+      "learning_rate": 4.182725307054869e-06,
+      "loss": 0.9707,
+      "step": 19432
+    },
+    {
+      "epoch": 0.7067316434520129,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1817672938465935e-06,
+      "loss": 1.0201,
+      "step": 19433
+    },
+    {
+      "epoch": 0.7067680110557515,
+      "grad_norm": 0.0,
+      "learning_rate": 4.180809361358043e-06,
+      "loss": 0.9454,
+      "step": 19434
+    },
+    {
+      "epoch": 0.7068043786594901,
+      "grad_norm": 0.0,
+      "learning_rate": 4.179851509602505e-06,
+      "loss": 0.8299,
+      "step": 19435
+    },
+    {
+      "epoch": 0.7068407462632287,
+      "grad_norm": 0.0,
+      "learning_rate": 4.178893738593267e-06,
+      "loss": 1.1086,
+      "step": 19436
+    },
+    {
+      "epoch": 0.7068771138669673,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1779360483436215e-06,
+      "loss": 1.0379,
+      "step": 19437
+    },
+    {
+      "epoch": 0.7069134814707059,
+      "grad_norm": 0.0,
+      "learning_rate": 4.176978438866854e-06,
+      "loss": 0.8739,
+      "step": 19438
+    },
+    {
+      "epoch": 0.7069498490744445,
+      "grad_norm": 0.0,
+      "learning_rate": 4.176020910176249e-06,
+      "loss": 0.9823,
+      "step": 19439
+    },
+    {
+      "epoch": 0.7069862166781831,
+      "grad_norm": 0.0,
+      "learning_rate": 4.175063462285091e-06,
+      "loss": 0.8636,
+      "step": 19440
+    },
+    {
+      "epoch": 0.7070225842819217,
+      "grad_norm": 0.0,
+      "learning_rate": 4.174106095206662e-06,
+      "loss": 0.9549,
+      "step": 19441
+    },
+    {
+      "epoch": 0.7070589518856603,
+      "grad_norm": 0.0,
+      "learning_rate": 4.173148808954242e-06,
+      "loss": 0.8693,
+      "step": 19442
+    },
+    {
+      "epoch": 0.7070953194893989,
+      "grad_norm": 0.0,
+      "learning_rate": 4.172191603541118e-06,
+      "loss": 1.0707,
+      "step": 19443
+    },
+    {
+      "epoch": 0.7071316870931375,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1712344789805684e-06,
+      "loss": 0.9688,
+      "step": 19444
+    },
+    {
+      "epoch": 0.707168054696876,
+      "grad_norm": 0.0,
+      "learning_rate": 4.170277435285869e-06,
+      "loss": 0.9531,
+      "step": 19445
+    },
+    {
+      "epoch": 0.7072044223006146,
+      "grad_norm": 0.0,
+      "learning_rate": 4.169320472470295e-06,
+      "loss": 0.8735,
+      "step": 19446
+    },
+    {
+      "epoch": 0.7072407899043532,
+      "grad_norm": 0.0,
+      "learning_rate": 4.168363590547135e-06,
+      "loss": 0.9194,
+      "step": 19447
+    },
+    {
+      "epoch": 0.7072771575080918,
+      "grad_norm": 0.0,
+      "learning_rate": 4.167406789529649e-06,
+      "loss": 0.9893,
+      "step": 19448
+    },
+    {
+      "epoch": 0.7073135251118304,
+      "grad_norm": 0.0,
+      "learning_rate": 4.166450069431122e-06,
+      "loss": 0.8007,
+      "step": 19449
+    },
+    {
+      "epoch": 0.7073498927155689,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1654934302648234e-06,
+      "loss": 0.8569,
+      "step": 19450
+    },
+    {
+      "epoch": 0.7073862603193075,
+      "grad_norm": 0.0,
+      "learning_rate": 4.164536872044025e-06,
+      "loss": 0.9149,
+      "step": 19451
+    },
+    {
+      "epoch": 0.7074226279230461,
+      "grad_norm": 0.0,
+      "learning_rate": 4.163580394781999e-06,
+      "loss": 0.9501,
+      "step": 19452
+    },
+    {
+      "epoch": 0.7074589955267847,
+      "grad_norm": 0.0,
+      "learning_rate": 4.162623998492012e-06,
+      "loss": 0.8985,
+      "step": 19453
+    },
+    {
+      "epoch": 0.7074953631305233,
+      "grad_norm": 0.0,
+      "learning_rate": 4.161667683187337e-06,
+      "loss": 0.9959,
+      "step": 19454
+    },
+    {
+      "epoch": 0.7075317307342619,
+      "grad_norm": 0.0,
+      "learning_rate": 4.160711448881236e-06,
+      "loss": 0.9952,
+      "step": 19455
+    },
+    {
+      "epoch": 0.7075680983380005,
+      "grad_norm": 0.0,
+      "learning_rate": 4.159755295586982e-06,
+      "loss": 0.8787,
+      "step": 19456
+    },
+    {
+      "epoch": 0.7076044659417391,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1587992233178366e-06,
+      "loss": 1.0075,
+      "step": 19457
+    },
+    {
+      "epoch": 0.7076408335454777,
+      "grad_norm": 0.0,
+      "learning_rate": 4.157843232087064e-06,
+      "loss": 0.9855,
+      "step": 19458
+    },
+    {
+      "epoch": 0.7076772011492163,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1568873219079245e-06,
+      "loss": 0.8956,
+      "step": 19459
+    },
+    {
+      "epoch": 0.7077135687529549,
+      "grad_norm": 0.0,
+      "learning_rate": 4.15593149279369e-06,
+      "loss": 1.0026,
+      "step": 19460
+    },
+    {
+      "epoch": 0.7077499363566935,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1549757447576076e-06,
+      "loss": 0.8819,
+      "step": 19461
+    },
+    {
+      "epoch": 0.7077863039604321,
+      "grad_norm": 0.0,
+      "learning_rate": 4.154020077812947e-06,
+      "loss": 1.0597,
+      "step": 19462
+    },
+    {
+      "epoch": 0.7078226715641707,
+      "grad_norm": 0.0,
+      "learning_rate": 4.153064491972963e-06,
+      "loss": 0.9455,
+      "step": 19463
+    },
+    {
+      "epoch": 0.7078590391679093,
+      "grad_norm": 0.0,
+      "learning_rate": 4.152108987250915e-06,
+      "loss": 0.8489,
+      "step": 19464
+    },
+    {
+      "epoch": 0.7078954067716479,
+      "grad_norm": 0.0,
+      "learning_rate": 4.151153563660053e-06,
+      "loss": 0.9882,
+      "step": 19465
+    },
+    {
+      "epoch": 0.7079317743753865,
+      "grad_norm": 0.0,
+      "learning_rate": 4.150198221213645e-06,
+      "loss": 0.9943,
+      "step": 19466
+    },
+    {
+      "epoch": 0.7079681419791249,
+      "grad_norm": 0.0,
+      "learning_rate": 4.149242959924928e-06,
+      "loss": 0.9473,
+      "step": 19467
+    },
+    {
+      "epoch": 0.7080045095828635,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1482877798071694e-06,
+      "loss": 0.7682,
+      "step": 19468
+    },
+    {
+      "epoch": 0.7080408771866021,
+      "grad_norm": 0.0,
+      "learning_rate": 4.147332680873615e-06,
+      "loss": 0.8955,
+      "step": 19469
+    },
+    {
+      "epoch": 0.7080772447903407,
+      "grad_norm": 0.0,
+      "learning_rate": 4.146377663137514e-06,
+      "loss": 0.9604,
+      "step": 19470
+    },
+    {
+      "epoch": 0.7081136123940793,
+      "grad_norm": 0.0,
+      "learning_rate": 4.145422726612119e-06,
+      "loss": 0.9929,
+      "step": 19471
+    },
+    {
+      "epoch": 0.7081499799978179,
+      "grad_norm": 0.0,
+      "learning_rate": 4.144467871310674e-06,
+      "loss": 1.005,
+      "step": 19472
+    },
+    {
+      "epoch": 0.7081863476015565,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1435130972464355e-06,
+      "loss": 0.993,
+      "step": 19473
+    },
+    {
+      "epoch": 0.7082227152052951,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1425584044326376e-06,
+      "loss": 0.9087,
+      "step": 19474
+    },
+    {
+      "epoch": 0.7082590828090337,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1416037928825334e-06,
+      "loss": 0.894,
+      "step": 19475
+    },
+    {
+      "epoch": 0.7082954504127723,
+      "grad_norm": 0.0,
+      "learning_rate": 4.140649262609366e-06,
+      "loss": 1.0149,
+      "step": 19476
+    },
+    {
+      "epoch": 0.7083318180165109,
+      "grad_norm": 0.0,
+      "learning_rate": 4.139694813626376e-06,
+      "loss": 0.9449,
+      "step": 19477
+    },
+    {
+      "epoch": 0.7083681856202495,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1387404459468015e-06,
+      "loss": 0.9006,
+      "step": 19478
+    },
+    {
+      "epoch": 0.7084045532239881,
+      "grad_norm": 0.0,
+      "learning_rate": 4.137786159583895e-06,
+      "loss": 0.9769,
+      "step": 19479
+    },
+    {
+      "epoch": 0.7084409208277267,
+      "grad_norm": 0.0,
+      "learning_rate": 4.136831954550882e-06,
+      "loss": 0.8299,
+      "step": 19480
+    },
+    {
+      "epoch": 0.7084772884314653,
+      "grad_norm": 0.0,
+      "learning_rate": 4.135877830861009e-06,
+      "loss": 0.8404,
+      "step": 19481
+    },
+    {
+      "epoch": 0.7085136560352039,
+      "grad_norm": 0.0,
+      "learning_rate": 4.134923788527512e-06,
+      "loss": 0.9928,
+      "step": 19482
+    },
+    {
+      "epoch": 0.7085500236389425,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1339698275636255e-06,
+      "loss": 1.0165,
+      "step": 19483
+    },
+    {
+      "epoch": 0.708586391242681,
+      "grad_norm": 0.0,
+      "learning_rate": 4.133015947982581e-06,
+      "loss": 0.8324,
+      "step": 19484
+    },
+    {
+      "epoch": 0.7086227588464196,
+      "grad_norm": 0.0,
+      "learning_rate": 4.132062149797623e-06,
+      "loss": 1.0511,
+      "step": 19485
+    },
+    {
+      "epoch": 0.7086591264501582,
+      "grad_norm": 0.0,
+      "learning_rate": 4.131108433021971e-06,
+      "loss": 0.8532,
+      "step": 19486
+    },
+    {
+      "epoch": 0.7086954940538968,
+      "grad_norm": 0.0,
+      "learning_rate": 4.130154797668865e-06,
+      "loss": 0.9546,
+      "step": 19487
+    },
+    {
+      "epoch": 0.7087318616576354,
+      "grad_norm": 0.0,
+      "learning_rate": 4.129201243751534e-06,
+      "loss": 1.0105,
+      "step": 19488
+    },
+    {
+      "epoch": 0.708768229261374,
+      "grad_norm": 0.0,
+      "learning_rate": 4.128247771283206e-06,
+      "loss": 0.9389,
+      "step": 19489
+    },
+    {
+      "epoch": 0.7088045968651125,
+      "grad_norm": 0.0,
+      "learning_rate": 4.127294380277109e-06,
+      "loss": 1.0198,
+      "step": 19490
+    },
+    {
+      "epoch": 0.7088409644688511,
+      "grad_norm": 0.0,
+      "learning_rate": 4.126341070746467e-06,
+      "loss": 0.9506,
+      "step": 19491
+    },
+    {
+      "epoch": 0.7088773320725897,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1253878427045155e-06,
+      "loss": 1.009,
+      "step": 19492
+    },
+    {
+      "epoch": 0.7089136996763283,
+      "grad_norm": 0.0,
+      "learning_rate": 4.124434696164465e-06,
+      "loss": 0.8508,
+      "step": 19493
+    },
+    {
+      "epoch": 0.7089500672800669,
+      "grad_norm": 0.0,
+      "learning_rate": 4.123481631139552e-06,
+      "loss": 0.9623,
+      "step": 19494
+    },
+    {
+      "epoch": 0.7089864348838055,
+      "grad_norm": 0.0,
+      "learning_rate": 4.122528647642993e-06,
+      "loss": 0.8275,
+      "step": 19495
+    },
+    {
+      "epoch": 0.7090228024875441,
+      "grad_norm": 0.0,
+      "learning_rate": 4.12157574568801e-06,
+      "loss": 1.0628,
+      "step": 19496
+    },
+    {
+      "epoch": 0.7090591700912827,
+      "grad_norm": 0.0,
+      "learning_rate": 4.120622925287819e-06,
+      "loss": 0.9157,
+      "step": 19497
+    },
+    {
+      "epoch": 0.7090955376950213,
+      "grad_norm": 0.0,
+      "learning_rate": 4.119670186455651e-06,
+      "loss": 0.9858,
+      "step": 19498
+    },
+    {
+      "epoch": 0.7091319052987599,
+      "grad_norm": 0.0,
+      "learning_rate": 4.118717529204709e-06,
+      "loss": 0.9019,
+      "step": 19499
+    },
+    {
+      "epoch": 0.7091682729024985,
+      "grad_norm": 0.0,
+      "learning_rate": 4.11776495354822e-06,
+      "loss": 0.9555,
+      "step": 19500
+    },
+    {
+      "epoch": 0.709204640506237,
+      "grad_norm": 0.0,
+      "learning_rate": 4.116812459499397e-06,
+      "loss": 1.03,
+      "step": 19501
+    },
+    {
+      "epoch": 0.7092410081099756,
+      "grad_norm": 0.0,
+      "learning_rate": 4.115860047071453e-06,
+      "loss": 0.9821,
+      "step": 19502
+    },
+    {
+      "epoch": 0.7092773757137142,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1149077162776e-06,
+      "loss": 1.0162,
+      "step": 19503
+    },
+    {
+      "epoch": 0.7093137433174528,
+      "grad_norm": 0.0,
+      "learning_rate": 4.113955467131054e-06,
+      "loss": 0.9718,
+      "step": 19504
+    },
+    {
+      "epoch": 0.7093501109211914,
+      "grad_norm": 0.0,
+      "learning_rate": 4.113003299645028e-06,
+      "loss": 1.0073,
+      "step": 19505
+    },
+    {
+      "epoch": 0.70938647852493,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1120512138327265e-06,
+      "loss": 0.9493,
+      "step": 19506
+    },
+    {
+      "epoch": 0.7094228461286686,
+      "grad_norm": 0.0,
+      "learning_rate": 4.11109920970736e-06,
+      "loss": 1.0163,
+      "step": 19507
+    },
+    {
+      "epoch": 0.7094592137324072,
+      "grad_norm": 0.0,
+      "learning_rate": 4.110147287282138e-06,
+      "loss": 0.8798,
+      "step": 19508
+    },
+    {
+      "epoch": 0.7094955813361458,
+      "grad_norm": 0.0,
+      "learning_rate": 4.109195446570265e-06,
+      "loss": 0.7788,
+      "step": 19509
+    },
+    {
+      "epoch": 0.7095319489398844,
+      "grad_norm": 0.0,
+      "learning_rate": 4.108243687584944e-06,
+      "loss": 1.0249,
+      "step": 19510
+    },
+    {
+      "epoch": 0.709568316543623,
+      "grad_norm": 0.0,
+      "learning_rate": 4.10729201033939e-06,
+      "loss": 0.9026,
+      "step": 19511
+    },
+    {
+      "epoch": 0.7096046841473616,
+      "grad_norm": 0.0,
+      "learning_rate": 4.106340414846791e-06,
+      "loss": 0.8302,
+      "step": 19512
+    },
+    {
+      "epoch": 0.7096410517511001,
+      "grad_norm": 0.0,
+      "learning_rate": 4.105388901120361e-06,
+      "loss": 0.8628,
+      "step": 19513
+    },
+    {
+      "epoch": 0.7096774193548387,
+      "grad_norm": 0.0,
+      "learning_rate": 4.104437469173296e-06,
+      "loss": 0.8706,
+      "step": 19514
+    },
+    {
+      "epoch": 0.7097137869585773,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1034861190187956e-06,
+      "loss": 0.9885,
+      "step": 19515
+    },
+    {
+      "epoch": 0.7097501545623159,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1025348506700565e-06,
+      "loss": 0.9185,
+      "step": 19516
+    },
+    {
+      "epoch": 0.7097865221660544,
+      "grad_norm": 0.0,
+      "learning_rate": 4.101583664140286e-06,
+      "loss": 0.8973,
+      "step": 19517
+    },
+    {
+      "epoch": 0.709822889769793,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1006325594426664e-06,
+      "loss": 0.8836,
+      "step": 19518
+    },
+    {
+      "epoch": 0.7098592573735316,
+      "grad_norm": 0.0,
+      "learning_rate": 4.099681536590404e-06,
+      "loss": 0.9666,
+      "step": 19519
+    },
+    {
+      "epoch": 0.7098956249772702,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0987305955966885e-06,
+      "loss": 0.9427,
+      "step": 19520
+    },
+    {
+      "epoch": 0.7099319925810088,
+      "grad_norm": 0.0,
+      "learning_rate": 4.097779736474713e-06,
+      "loss": 0.9583,
+      "step": 19521
+    },
+    {
+      "epoch": 0.7099683601847474,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0968289592376664e-06,
+      "loss": 0.7793,
+      "step": 19522
+    },
+    {
+      "epoch": 0.710004727788486,
+      "grad_norm": 0.0,
+      "learning_rate": 4.095878263898745e-06,
+      "loss": 0.9831,
+      "step": 19523
+    },
+    {
+      "epoch": 0.7100410953922246,
+      "grad_norm": 0.0,
+      "learning_rate": 4.094927650471136e-06,
+      "loss": 1.0304,
+      "step": 19524
+    },
+    {
+      "epoch": 0.7100774629959632,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0939771189680285e-06,
+      "loss": 0.8834,
+      "step": 19525
+    },
+    {
+      "epoch": 0.7101138305997018,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0930266694026055e-06,
+      "loss": 0.9347,
+      "step": 19526
+    },
+    {
+      "epoch": 0.7101501982034404,
+      "grad_norm": 0.0,
+      "learning_rate": 4.092076301788063e-06,
+      "loss": 0.9036,
+      "step": 19527
+    },
+    {
+      "epoch": 0.710186565807179,
+      "grad_norm": 0.0,
+      "learning_rate": 4.091126016137577e-06,
+      "loss": 0.8918,
+      "step": 19528
+    },
+    {
+      "epoch": 0.7102229334109176,
+      "grad_norm": 0.0,
+      "learning_rate": 4.090175812464328e-06,
+      "loss": 0.8996,
+      "step": 19529
+    },
+    {
+      "epoch": 0.7102593010146562,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0892256907815145e-06,
+      "loss": 1.0278,
+      "step": 19530
+    },
+    {
+      "epoch": 0.7102956686183948,
+      "grad_norm": 0.0,
+      "learning_rate": 4.088275651102299e-06,
+      "loss": 1.1038,
+      "step": 19531
+    },
+    {
+      "epoch": 0.7103320362221334,
+      "grad_norm": 0.0,
+      "learning_rate": 4.087325693439877e-06,
+      "loss": 1.0309,
+      "step": 19532
+    },
+    {
+      "epoch": 0.710368403825872,
+      "grad_norm": 0.0,
+      "learning_rate": 4.08637581780742e-06,
+      "loss": 0.886,
+      "step": 19533
+    },
+    {
+      "epoch": 0.7104047714296104,
+      "grad_norm": 0.0,
+      "learning_rate": 4.085426024218109e-06,
+      "loss": 1.0242,
+      "step": 19534
+    },
+    {
+      "epoch": 0.710441139033349,
+      "grad_norm": 0.0,
+      "learning_rate": 4.084476312685116e-06,
+      "loss": 1.0389,
+      "step": 19535
+    },
+    {
+      "epoch": 0.7104775066370876,
+      "grad_norm": 0.0,
+      "learning_rate": 4.083526683221626e-06,
+      "loss": 0.9683,
+      "step": 19536
+    },
+    {
+      "epoch": 0.7105138742408262,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0825771358408075e-06,
+      "loss": 1.1034,
+      "step": 19537
+    },
+    {
+      "epoch": 0.7105502418445648,
+      "grad_norm": 0.0,
+      "learning_rate": 4.081627670555836e-06,
+      "loss": 0.9046,
+      "step": 19538
+    },
+    {
+      "epoch": 0.7105866094483034,
+      "grad_norm": 0.0,
+      "learning_rate": 4.080678287379884e-06,
+      "loss": 0.9272,
+      "step": 19539
+    },
+    {
+      "epoch": 0.710622977052042,
+      "grad_norm": 0.0,
+      "learning_rate": 4.079728986326122e-06,
+      "loss": 0.9678,
+      "step": 19540
+    },
+    {
+      "epoch": 0.7106593446557806,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0787797674077166e-06,
+      "loss": 0.9314,
+      "step": 19541
+    },
+    {
+      "epoch": 0.7106957122595192,
+      "grad_norm": 0.0,
+      "learning_rate": 4.077830630637846e-06,
+      "loss": 0.994,
+      "step": 19542
+    },
+    {
+      "epoch": 0.7107320798632578,
+      "grad_norm": 0.0,
+      "learning_rate": 4.076881576029671e-06,
+      "loss": 0.8964,
+      "step": 19543
+    },
+    {
+      "epoch": 0.7107684474669964,
+      "grad_norm": 0.0,
+      "learning_rate": 4.075932603596361e-06,
+      "loss": 1.0626,
+      "step": 19544
+    },
+    {
+      "epoch": 0.710804815070735,
+      "grad_norm": 0.0,
+      "learning_rate": 4.074983713351077e-06,
+      "loss": 0.9243,
+      "step": 19545
+    },
+    {
+      "epoch": 0.7108411826744736,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0740349053069946e-06,
+      "loss": 1.0201,
+      "step": 19546
+    },
+    {
+      "epoch": 0.7108775502782122,
+      "grad_norm": 0.0,
+      "learning_rate": 4.073086179477268e-06,
+      "loss": 0.9429,
+      "step": 19547
+    },
+    {
+      "epoch": 0.7109139178819508,
+      "grad_norm": 0.0,
+      "learning_rate": 4.072137535875057e-06,
+      "loss": 0.9272,
+      "step": 19548
+    },
+    {
+      "epoch": 0.7109502854856894,
+      "grad_norm": 0.0,
+      "learning_rate": 4.07118897451353e-06,
+      "loss": 0.9857,
+      "step": 19549
+    },
+    {
+      "epoch": 0.710986653089428,
+      "grad_norm": 0.0,
+      "learning_rate": 4.070240495405845e-06,
+      "loss": 0.8589,
+      "step": 19550
+    },
+    {
+      "epoch": 0.7110230206931665,
+      "grad_norm": 0.0,
+      "learning_rate": 4.06929209856516e-06,
+      "loss": 0.9377,
+      "step": 19551
+    },
+    {
+      "epoch": 0.7110593882969051,
+      "grad_norm": 0.0,
+      "learning_rate": 4.068343784004632e-06,
+      "loss": 1.0684,
+      "step": 19552
+    },
+    {
+      "epoch": 0.7110957559006437,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0673955517374174e-06,
+      "loss": 0.8977,
+      "step": 19553
+    },
+    {
+      "epoch": 0.7111321235043823,
+      "grad_norm": 0.0,
+      "learning_rate": 4.06644740177667e-06,
+      "loss": 1.0113,
+      "step": 19554
+    },
+    {
+      "epoch": 0.7111684911081209,
+      "grad_norm": 0.0,
+      "learning_rate": 4.065499334135549e-06,
+      "loss": 0.9831,
+      "step": 19555
+    },
+    {
+      "epoch": 0.7112048587118595,
+      "grad_norm": 0.0,
+      "learning_rate": 4.064551348827205e-06,
+      "loss": 0.9802,
+      "step": 19556
+    },
+    {
+      "epoch": 0.711241226315598,
+      "grad_norm": 0.0,
+      "learning_rate": 4.06360344586479e-06,
+      "loss": 0.8607,
+      "step": 19557
+    },
+    {
+      "epoch": 0.7112775939193366,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0626556252614515e-06,
+      "loss": 0.8936,
+      "step": 19558
+    },
+    {
+      "epoch": 0.7113139615230752,
+      "grad_norm": 0.0,
+      "learning_rate": 4.061707887030348e-06,
+      "loss": 0.9137,
+      "step": 19559
+    },
+    {
+      "epoch": 0.7113503291268138,
+      "grad_norm": 0.0,
+      "learning_rate": 4.060760231184616e-06,
+      "loss": 0.961,
+      "step": 19560
+    },
+    {
+      "epoch": 0.7113866967305524,
+      "grad_norm": 0.0,
+      "learning_rate": 4.059812657737412e-06,
+      "loss": 0.988,
+      "step": 19561
+    },
+    {
+      "epoch": 0.711423064334291,
+      "grad_norm": 0.0,
+      "learning_rate": 4.05886516670188e-06,
+      "loss": 1.0719,
+      "step": 19562
+    },
+    {
+      "epoch": 0.7114594319380296,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0579177580911644e-06,
+      "loss": 0.9728,
+      "step": 19563
+    },
+    {
+      "epoch": 0.7114957995417682,
+      "grad_norm": 0.0,
+      "learning_rate": 4.056970431918406e-06,
+      "loss": 0.8356,
+      "step": 19564
+    },
+    {
+      "epoch": 0.7115321671455068,
+      "grad_norm": 0.0,
+      "learning_rate": 4.056023188196757e-06,
+      "loss": 0.9638,
+      "step": 19565
+    },
+    {
+      "epoch": 0.7115685347492454,
+      "grad_norm": 0.0,
+      "learning_rate": 4.05507602693935e-06,
+      "loss": 0.9898,
+      "step": 19566
+    },
+    {
+      "epoch": 0.711604902352984,
+      "grad_norm": 0.0,
+      "learning_rate": 4.054128948159324e-06,
+      "loss": 0.8643,
+      "step": 19567
+    },
+    {
+      "epoch": 0.7116412699567225,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0531819518698265e-06,
+      "loss": 0.9158,
+      "step": 19568
+    },
+    {
+      "epoch": 0.7116776375604611,
+      "grad_norm": 0.0,
+      "learning_rate": 4.052235038083993e-06,
+      "loss": 1.0174,
+      "step": 19569
+    },
+    {
+      "epoch": 0.7117140051641997,
+      "grad_norm": 0.0,
+      "learning_rate": 4.051288206814959e-06,
+      "loss": 1.0265,
+      "step": 19570
+    },
+    {
+      "epoch": 0.7117503727679383,
+      "grad_norm": 0.0,
+      "learning_rate": 4.050341458075857e-06,
+      "loss": 0.9555,
+      "step": 19571
+    },
+    {
+      "epoch": 0.7117867403716769,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0493947918798334e-06,
+      "loss": 0.8686,
+      "step": 19572
+    },
+    {
+      "epoch": 0.7118231079754155,
+      "grad_norm": 0.0,
+      "learning_rate": 4.048448208240006e-06,
+      "loss": 0.9442,
+      "step": 19573
+    },
+    {
+      "epoch": 0.7118594755791541,
+      "grad_norm": 0.0,
+      "learning_rate": 4.047501707169521e-06,
+      "loss": 1.0844,
+      "step": 19574
+    },
+    {
+      "epoch": 0.7118958431828927,
+      "grad_norm": 0.0,
+      "learning_rate": 4.046555288681504e-06,
+      "loss": 0.9302,
+      "step": 19575
+    },
+    {
+      "epoch": 0.7119322107866313,
+      "grad_norm": 0.0,
+      "learning_rate": 4.045608952789085e-06,
+      "loss": 0.9135,
+      "step": 19576
+    },
+    {
+      "epoch": 0.7119685783903699,
+      "grad_norm": 0.0,
+      "learning_rate": 4.04466269950539e-06,
+      "loss": 0.8743,
+      "step": 19577
+    },
+    {
+      "epoch": 0.7120049459941085,
+      "grad_norm": 0.0,
+      "learning_rate": 4.043716528843557e-06,
+      "loss": 0.9221,
+      "step": 19578
+    },
+    {
+      "epoch": 0.7120413135978471,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0427704408167e-06,
+      "loss": 0.7426,
+      "step": 19579
+    },
+    {
+      "epoch": 0.7120776812015857,
+      "grad_norm": 0.0,
+      "learning_rate": 4.041824435437954e-06,
+      "loss": 1.0129,
+      "step": 19580
+    },
+    {
+      "epoch": 0.7121140488053243,
+      "grad_norm": 0.0,
+      "learning_rate": 4.040878512720442e-06,
+      "loss": 0.9692,
+      "step": 19581
+    },
+    {
+      "epoch": 0.7121504164090628,
+      "grad_norm": 0.0,
+      "learning_rate": 4.039932672677283e-06,
+      "loss": 1.1064,
+      "step": 19582
+    },
+    {
+      "epoch": 0.7121867840128014,
+      "grad_norm": 0.0,
+      "learning_rate": 4.038986915321601e-06,
+      "loss": 0.9118,
+      "step": 19583
+    },
+    {
+      "epoch": 0.71222315161654,
+      "grad_norm": 0.0,
+      "learning_rate": 4.038041240666524e-06,
+      "loss": 0.9341,
+      "step": 19584
+    },
+    {
+      "epoch": 0.7122595192202785,
+      "grad_norm": 0.0,
+      "learning_rate": 4.037095648725162e-06,
+      "loss": 0.931,
+      "step": 19585
+    },
+    {
+      "epoch": 0.7122958868240171,
+      "grad_norm": 0.0,
+      "learning_rate": 4.036150139510635e-06,
+      "loss": 0.9948,
+      "step": 19586
+    },
+    {
+      "epoch": 0.7123322544277557,
+      "grad_norm": 0.0,
+      "learning_rate": 4.035204713036066e-06,
+      "loss": 0.7988,
+      "step": 19587
+    },
+    {
+      "epoch": 0.7123686220314943,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0342593693145685e-06,
+      "loss": 0.9778,
+      "step": 19588
+    },
+    {
+      "epoch": 0.7124049896352329,
+      "grad_norm": 0.0,
+      "learning_rate": 4.033314108359259e-06,
+      "loss": 0.9189,
+      "step": 19589
+    },
+    {
+      "epoch": 0.7124413572389715,
+      "grad_norm": 0.0,
+      "learning_rate": 4.032368930183246e-06,
+      "loss": 0.9633,
+      "step": 19590
+    },
+    {
+      "epoch": 0.7124777248427101,
+      "grad_norm": 0.0,
+      "learning_rate": 4.031423834799655e-06,
+      "loss": 0.9652,
+      "step": 19591
+    },
+    {
+      "epoch": 0.7125140924464487,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0304788222215815e-06,
+      "loss": 0.971,
+      "step": 19592
+    },
+    {
+      "epoch": 0.7125504600501873,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0295338924621495e-06,
+      "loss": 0.795,
+      "step": 19593
+    },
+    {
+      "epoch": 0.7125868276539259,
+      "grad_norm": 0.0,
+      "learning_rate": 4.028589045534463e-06,
+      "loss": 0.8584,
+      "step": 19594
+    },
+    {
+      "epoch": 0.7126231952576645,
+      "grad_norm": 0.0,
+      "learning_rate": 4.027644281451632e-06,
+      "loss": 0.9549,
+      "step": 19595
+    },
+    {
+      "epoch": 0.7126595628614031,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0266996002267585e-06,
+      "loss": 1.0063,
+      "step": 19596
+    },
+    {
+      "epoch": 0.7126959304651417,
+      "grad_norm": 0.0,
+      "learning_rate": 4.025755001872962e-06,
+      "loss": 1.0368,
+      "step": 19597
+    },
+    {
+      "epoch": 0.7127322980688803,
+      "grad_norm": 0.0,
+      "learning_rate": 4.02481048640333e-06,
+      "loss": 0.8783,
+      "step": 19598
+    },
+    {
+      "epoch": 0.7127686656726189,
+      "grad_norm": 0.0,
+      "learning_rate": 4.023866053830978e-06,
+      "loss": 0.9007,
+      "step": 19599
+    },
+    {
+      "epoch": 0.7128050332763575,
+      "grad_norm": 0.0,
+      "learning_rate": 4.022921704169006e-06,
+      "loss": 0.9833,
+      "step": 19600
+    },
+    {
+      "epoch": 0.7128414008800961,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0219774374305155e-06,
+      "loss": 0.8324,
+      "step": 19601
+    },
+    {
+      "epoch": 0.7128777684838346,
+      "grad_norm": 0.0,
+      "learning_rate": 4.021033253628601e-06,
+      "loss": 0.9792,
+      "step": 19602
+    },
+    {
+      "epoch": 0.7129141360875731,
+      "grad_norm": 0.0,
+      "learning_rate": 4.020089152776373e-06,
+      "loss": 0.8954,
+      "step": 19603
+    },
+    {
+      "epoch": 0.7129505036913117,
+      "grad_norm": 0.0,
+      "learning_rate": 4.019145134886926e-06,
+      "loss": 0.9559,
+      "step": 19604
+    },
+    {
+      "epoch": 0.7129868712950503,
+      "grad_norm": 0.0,
+      "learning_rate": 4.018201199973347e-06,
+      "loss": 1.0125,
+      "step": 19605
+    },
+    {
+      "epoch": 0.7130232388987889,
+      "grad_norm": 0.0,
+      "learning_rate": 4.017257348048743e-06,
+      "loss": 0.974,
+      "step": 19606
+    },
+    {
+      "epoch": 0.7130596065025275,
+      "grad_norm": 0.0,
+      "learning_rate": 4.016313579126206e-06,
+      "loss": 0.9913,
+      "step": 19607
+    },
+    {
+      "epoch": 0.7130959741062661,
+      "grad_norm": 0.0,
+      "learning_rate": 4.015369893218826e-06,
+      "loss": 0.8726,
+      "step": 19608
+    },
+    {
+      "epoch": 0.7131323417100047,
+      "grad_norm": 0.0,
+      "learning_rate": 4.014426290339697e-06,
+      "loss": 0.9063,
+      "step": 19609
+    },
+    {
+      "epoch": 0.7131687093137433,
+      "grad_norm": 0.0,
+      "learning_rate": 4.013482770501917e-06,
+      "loss": 0.9059,
+      "step": 19610
+    },
+    {
+      "epoch": 0.7132050769174819,
+      "grad_norm": 0.0,
+      "learning_rate": 4.012539333718561e-06,
+      "loss": 0.9094,
+      "step": 19611
+    },
+    {
+      "epoch": 0.7132414445212205,
+      "grad_norm": 0.0,
+      "learning_rate": 4.011595980002732e-06,
+      "loss": 0.9461,
+      "step": 19612
+    },
+    {
+      "epoch": 0.7132778121249591,
+      "grad_norm": 0.0,
+      "learning_rate": 4.010652709367512e-06,
+      "loss": 0.9684,
+      "step": 19613
+    },
+    {
+      "epoch": 0.7133141797286977,
+      "grad_norm": 0.0,
+      "learning_rate": 4.009709521825988e-06,
+      "loss": 0.8982,
+      "step": 19614
+    },
+    {
+      "epoch": 0.7133505473324363,
+      "grad_norm": 0.0,
+      "learning_rate": 4.008766417391241e-06,
+      "loss": 0.9345,
+      "step": 19615
+    },
+    {
+      "epoch": 0.7133869149361749,
+      "grad_norm": 0.0,
+      "learning_rate": 4.007823396076363e-06,
+      "loss": 0.9283,
+      "step": 19616
+    },
+    {
+      "epoch": 0.7134232825399135,
+      "grad_norm": 0.0,
+      "learning_rate": 4.006880457894434e-06,
+      "loss": 0.8111,
+      "step": 19617
+    },
+    {
+      "epoch": 0.713459650143652,
+      "grad_norm": 0.0,
+      "learning_rate": 4.005937602858536e-06,
+      "loss": 0.9603,
+      "step": 19618
+    },
+    {
+      "epoch": 0.7134960177473906,
+      "grad_norm": 0.0,
+      "learning_rate": 4.004994830981749e-06,
+      "loss": 0.9178,
+      "step": 19619
+    },
+    {
+      "epoch": 0.7135323853511292,
+      "grad_norm": 0.0,
+      "learning_rate": 4.004052142277153e-06,
+      "loss": 1.0065,
+      "step": 19620
+    },
+    {
+      "epoch": 0.7135687529548678,
+      "grad_norm": 0.0,
+      "learning_rate": 4.003109536757822e-06,
+      "loss": 0.8848,
+      "step": 19621
+    },
+    {
+      "epoch": 0.7136051205586064,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0021670144368415e-06,
+      "loss": 1.0515,
+      "step": 19622
+    },
+    {
+      "epoch": 0.713641488162345,
+      "grad_norm": 0.0,
+      "learning_rate": 4.001224575327288e-06,
+      "loss": 0.804,
+      "step": 19623
+    },
+    {
+      "epoch": 0.7136778557660836,
+      "grad_norm": 0.0,
+      "learning_rate": 4.000282219442225e-06,
+      "loss": 0.8781,
+      "step": 19624
+    },
+    {
+      "epoch": 0.7137142233698222,
+      "grad_norm": 0.0,
+      "learning_rate": 3.999339946794737e-06,
+      "loss": 0.8642,
+      "step": 19625
+    },
+    {
+      "epoch": 0.7137505909735608,
+      "grad_norm": 0.0,
+      "learning_rate": 3.998397757397894e-06,
+      "loss": 0.8236,
+      "step": 19626
+    },
+    {
+      "epoch": 0.7137869585772993,
+      "grad_norm": 0.0,
+      "learning_rate": 3.997455651264767e-06,
+      "loss": 0.9206,
+      "step": 19627
+    },
+    {
+      "epoch": 0.7138233261810379,
+      "grad_norm": 0.0,
+      "learning_rate": 3.996513628408423e-06,
+      "loss": 0.8948,
+      "step": 19628
+    },
+    {
+      "epoch": 0.7138596937847765,
+      "grad_norm": 0.0,
+      "learning_rate": 3.995571688841941e-06,
+      "loss": 0.9977,
+      "step": 19629
+    },
+    {
+      "epoch": 0.7138960613885151,
+      "grad_norm": 0.0,
+      "learning_rate": 3.994629832578375e-06,
+      "loss": 0.9492,
+      "step": 19630
+    },
+    {
+      "epoch": 0.7139324289922537,
+      "grad_norm": 0.0,
+      "learning_rate": 3.993688059630804e-06,
+      "loss": 0.8276,
+      "step": 19631
+    },
+    {
+      "epoch": 0.7139687965959923,
+      "grad_norm": 0.0,
+      "learning_rate": 3.992746370012289e-06,
+      "loss": 0.9391,
+      "step": 19632
+    },
+    {
+      "epoch": 0.7140051641997309,
+      "grad_norm": 0.0,
+      "learning_rate": 3.991804763735895e-06,
+      "loss": 0.9753,
+      "step": 19633
+    },
+    {
+      "epoch": 0.7140415318034695,
+      "grad_norm": 0.0,
+      "learning_rate": 3.990863240814682e-06,
+      "loss": 1.0649,
+      "step": 19634
+    },
+    {
+      "epoch": 0.714077899407208,
+      "grad_norm": 0.0,
+      "learning_rate": 3.989921801261719e-06,
+      "loss": 1.0247,
+      "step": 19635
+    },
+    {
+      "epoch": 0.7141142670109466,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9889804450900645e-06,
+      "loss": 0.9693,
+      "step": 19636
+    },
+    {
+      "epoch": 0.7141506346146852,
+      "grad_norm": 0.0,
+      "learning_rate": 3.988039172312778e-06,
+      "loss": 0.9912,
+      "step": 19637
+    },
+    {
+      "epoch": 0.7141870022184238,
+      "grad_norm": 0.0,
+      "learning_rate": 3.987097982942917e-06,
+      "loss": 0.911,
+      "step": 19638
+    },
+    {
+      "epoch": 0.7142233698221624,
+      "grad_norm": 0.0,
+      "learning_rate": 3.98615687699354e-06,
+      "loss": 1.0354,
+      "step": 19639
+    },
+    {
+      "epoch": 0.714259737425901,
+      "grad_norm": 0.0,
+      "learning_rate": 3.985215854477701e-06,
+      "loss": 0.8353,
+      "step": 19640
+    },
+    {
+      "epoch": 0.7142961050296396,
+      "grad_norm": 0.0,
+      "learning_rate": 3.984274915408461e-06,
+      "loss": 1.0431,
+      "step": 19641
+    },
+    {
+      "epoch": 0.7143324726333782,
+      "grad_norm": 0.0,
+      "learning_rate": 3.983334059798877e-06,
+      "loss": 0.8816,
+      "step": 19642
+    },
+    {
+      "epoch": 0.7143688402371168,
+      "grad_norm": 0.0,
+      "learning_rate": 3.982393287661988e-06,
+      "loss": 1.0922,
+      "step": 19643
+    },
+    {
+      "epoch": 0.7144052078408554,
+      "grad_norm": 0.0,
+      "learning_rate": 3.981452599010857e-06,
+      "loss": 0.8674,
+      "step": 19644
+    },
+    {
+      "epoch": 0.714441575444594,
+      "grad_norm": 0.0,
+      "learning_rate": 3.980511993858532e-06,
+      "loss": 0.9849,
+      "step": 19645
+    },
+    {
+      "epoch": 0.7144779430483326,
+      "grad_norm": 0.0,
+      "learning_rate": 3.979571472218063e-06,
+      "loss": 0.9234,
+      "step": 19646
+    },
+    {
+      "epoch": 0.7145143106520712,
+      "grad_norm": 0.0,
+      "learning_rate": 3.978631034102496e-06,
+      "loss": 0.9106,
+      "step": 19647
+    },
+    {
+      "epoch": 0.7145506782558098,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9776906795248816e-06,
+      "loss": 0.9229,
+      "step": 19648
+    },
+    {
+      "epoch": 0.7145870458595484,
+      "grad_norm": 0.0,
+      "learning_rate": 3.976750408498266e-06,
+      "loss": 0.9162,
+      "step": 19649
+    },
+    {
+      "epoch": 0.714623413463287,
+      "grad_norm": 0.0,
+      "learning_rate": 3.97581022103569e-06,
+      "loss": 0.8671,
+      "step": 19650
+    },
+    {
+      "epoch": 0.7146597810670255,
+      "grad_norm": 0.0,
+      "learning_rate": 3.974870117150203e-06,
+      "loss": 0.9354,
+      "step": 19651
+    },
+    {
+      "epoch": 0.714696148670764,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9739300968548435e-06,
+      "loss": 0.9177,
+      "step": 19652
+    },
+    {
+      "epoch": 0.7147325162745026,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9729901601626504e-06,
+      "loss": 0.9105,
+      "step": 19653
+    },
+    {
+      "epoch": 0.7147688838782412,
+      "grad_norm": 0.0,
+      "learning_rate": 3.972050307086671e-06,
+      "loss": 0.876,
+      "step": 19654
+    },
+    {
+      "epoch": 0.7148052514819798,
+      "grad_norm": 0.0,
+      "learning_rate": 3.971110537639941e-06,
+      "loss": 1.002,
+      "step": 19655
+    },
+    {
+      "epoch": 0.7148416190857184,
+      "grad_norm": 0.0,
+      "learning_rate": 3.970170851835499e-06,
+      "loss": 0.7429,
+      "step": 19656
+    },
+    {
+      "epoch": 0.714877986689457,
+      "grad_norm": 0.0,
+      "learning_rate": 3.969231249686377e-06,
+      "loss": 1.0115,
+      "step": 19657
+    },
+    {
+      "epoch": 0.7149143542931956,
+      "grad_norm": 0.0,
+      "learning_rate": 3.968291731205623e-06,
+      "loss": 0.9242,
+      "step": 19658
+    },
+    {
+      "epoch": 0.7149507218969342,
+      "grad_norm": 0.0,
+      "learning_rate": 3.967352296406255e-06,
+      "loss": 0.9338,
+      "step": 19659
+    },
+    {
+      "epoch": 0.7149870895006728,
+      "grad_norm": 0.0,
+      "learning_rate": 3.966412945301319e-06,
+      "loss": 0.913,
+      "step": 19660
+    },
+    {
+      "epoch": 0.7150234571044114,
+      "grad_norm": 0.0,
+      "learning_rate": 3.965473677903846e-06,
+      "loss": 0.8617,
+      "step": 19661
+    },
+    {
+      "epoch": 0.71505982470815,
+      "grad_norm": 0.0,
+      "learning_rate": 3.964534494226857e-06,
+      "loss": 0.8411,
+      "step": 19662
+    },
+    {
+      "epoch": 0.7150961923118886,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9635953942833915e-06,
+      "loss": 0.9464,
+      "step": 19663
+    },
+    {
+      "epoch": 0.7151325599156272,
+      "grad_norm": 0.0,
+      "learning_rate": 3.962656378086477e-06,
+      "loss": 0.9132,
+      "step": 19664
+    },
+    {
+      "epoch": 0.7151689275193658,
+      "grad_norm": 0.0,
+      "learning_rate": 3.961717445649138e-06,
+      "loss": 0.958,
+      "step": 19665
+    },
+    {
+      "epoch": 0.7152052951231044,
+      "grad_norm": 0.0,
+      "learning_rate": 3.960778596984399e-06,
+      "loss": 0.8939,
+      "step": 19666
+    },
+    {
+      "epoch": 0.715241662726843,
+      "grad_norm": 0.0,
+      "learning_rate": 3.959839832105292e-06,
+      "loss": 0.9252,
+      "step": 19667
+    },
+    {
+      "epoch": 0.7152780303305816,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9589011510248375e-06,
+      "loss": 0.9296,
+      "step": 19668
+    },
+    {
+      "epoch": 0.7153143979343201,
+      "grad_norm": 0.0,
+      "learning_rate": 3.957962553756058e-06,
+      "loss": 0.8318,
+      "step": 19669
+    },
+    {
+      "epoch": 0.7153507655380587,
+      "grad_norm": 0.0,
+      "learning_rate": 3.957024040311973e-06,
+      "loss": 0.964,
+      "step": 19670
+    },
+    {
+      "epoch": 0.7153871331417972,
+      "grad_norm": 0.0,
+      "learning_rate": 3.956085610705611e-06,
+      "loss": 0.9106,
+      "step": 19671
+    },
+    {
+      "epoch": 0.7154235007455358,
+      "grad_norm": 0.0,
+      "learning_rate": 3.95514726494998e-06,
+      "loss": 0.9395,
+      "step": 19672
+    },
+    {
+      "epoch": 0.7154598683492744,
+      "grad_norm": 0.0,
+      "learning_rate": 3.954209003058108e-06,
+      "loss": 1.0707,
+      "step": 19673
+    },
+    {
+      "epoch": 0.715496235953013,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9532708250430075e-06,
+      "loss": 0.8603,
+      "step": 19674
+    },
+    {
+      "epoch": 0.7155326035567516,
+      "grad_norm": 0.0,
+      "learning_rate": 3.952332730917696e-06,
+      "loss": 0.9192,
+      "step": 19675
+    },
+    {
+      "epoch": 0.7155689711604902,
+      "grad_norm": 0.0,
+      "learning_rate": 3.951394720695183e-06,
+      "loss": 0.9832,
+      "step": 19676
+    },
+    {
+      "epoch": 0.7156053387642288,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9504567943884945e-06,
+      "loss": 0.8461,
+      "step": 19677
+    },
+    {
+      "epoch": 0.7156417063679674,
+      "grad_norm": 0.0,
+      "learning_rate": 3.949518952010626e-06,
+      "loss": 0.8917,
+      "step": 19678
+    },
+    {
+      "epoch": 0.715678073971706,
+      "grad_norm": 0.0,
+      "learning_rate": 3.948581193574602e-06,
+      "loss": 0.8909,
+      "step": 19679
+    },
+    {
+      "epoch": 0.7157144415754446,
+      "grad_norm": 0.0,
+      "learning_rate": 3.947643519093428e-06,
+      "loss": 1.0484,
+      "step": 19680
+    },
+    {
+      "epoch": 0.7157508091791832,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9467059285801135e-06,
+      "loss": 0.9989,
+      "step": 19681
+    },
+    {
+      "epoch": 0.7157871767829218,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9457684220476655e-06,
+      "loss": 0.9986,
+      "step": 19682
+    },
+    {
+      "epoch": 0.7158235443866604,
+      "grad_norm": 0.0,
+      "learning_rate": 3.944830999509091e-06,
+      "loss": 1.0049,
+      "step": 19683
+    },
+    {
+      "epoch": 0.715859911990399,
+      "grad_norm": 0.0,
+      "learning_rate": 3.943893660977395e-06,
+      "loss": 0.9373,
+      "step": 19684
+    },
+    {
+      "epoch": 0.7158962795941376,
+      "grad_norm": 0.0,
+      "learning_rate": 3.942956406465578e-06,
+      "loss": 0.9252,
+      "step": 19685
+    },
+    {
+      "epoch": 0.7159326471978761,
+      "grad_norm": 0.0,
+      "learning_rate": 3.94201923598665e-06,
+      "loss": 0.8619,
+      "step": 19686
+    },
+    {
+      "epoch": 0.7159690148016147,
+      "grad_norm": 0.0,
+      "learning_rate": 3.941082149553611e-06,
+      "loss": 0.8909,
+      "step": 19687
+    },
+    {
+      "epoch": 0.7160053824053533,
+      "grad_norm": 0.0,
+      "learning_rate": 3.94014514717946e-06,
+      "loss": 1.0226,
+      "step": 19688
+    },
+    {
+      "epoch": 0.7160417500090919,
+      "grad_norm": 0.0,
+      "learning_rate": 3.939208228877195e-06,
+      "loss": 0.9359,
+      "step": 19689
+    },
+    {
+      "epoch": 0.7160781176128305,
+      "grad_norm": 0.0,
+      "learning_rate": 3.938271394659823e-06,
+      "loss": 0.8777,
+      "step": 19690
+    },
+    {
+      "epoch": 0.7161144852165691,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9373346445403285e-06,
+      "loss": 0.9283,
+      "step": 19691
+    },
+    {
+      "epoch": 0.7161508528203077,
+      "grad_norm": 0.0,
+      "learning_rate": 3.936397978531717e-06,
+      "loss": 0.8223,
+      "step": 19692
+    },
+    {
+      "epoch": 0.7161872204240463,
+      "grad_norm": 0.0,
+      "learning_rate": 3.93546139664698e-06,
+      "loss": 0.9845,
+      "step": 19693
+    },
+    {
+      "epoch": 0.7162235880277849,
+      "grad_norm": 0.0,
+      "learning_rate": 3.934524898899112e-06,
+      "loss": 0.8441,
+      "step": 19694
+    },
+    {
+      "epoch": 0.7162599556315234,
+      "grad_norm": 0.0,
+      "learning_rate": 3.933588485301102e-06,
+      "loss": 0.9138,
+      "step": 19695
+    },
+    {
+      "epoch": 0.716296323235262,
+      "grad_norm": 0.0,
+      "learning_rate": 3.932652155865951e-06,
+      "loss": 0.9951,
+      "step": 19696
+    },
+    {
+      "epoch": 0.7163326908390006,
+      "grad_norm": 0.0,
+      "learning_rate": 3.931715910606636e-06,
+      "loss": 1.0333,
+      "step": 19697
+    },
+    {
+      "epoch": 0.7163690584427392,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9307797495361564e-06,
+      "loss": 0.9562,
+      "step": 19698
+    },
+    {
+      "epoch": 0.7164054260464778,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9298436726674985e-06,
+      "loss": 0.9224,
+      "step": 19699
+    },
+    {
+      "epoch": 0.7164417936502164,
+      "grad_norm": 0.0,
+      "learning_rate": 3.928907680013645e-06,
+      "loss": 0.8701,
+      "step": 19700
+    },
+    {
+      "epoch": 0.716478161253955,
+      "grad_norm": 0.0,
+      "learning_rate": 3.927971771587584e-06,
+      "loss": 0.9505,
+      "step": 19701
+    },
+    {
+      "epoch": 0.7165145288576935,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9270359474022965e-06,
+      "loss": 1.064,
+      "step": 19702
+    },
+    {
+      "epoch": 0.7165508964614321,
+      "grad_norm": 0.0,
+      "learning_rate": 3.926100207470777e-06,
+      "loss": 0.9363,
+      "step": 19703
+    },
+    {
+      "epoch": 0.7165872640651707,
+      "grad_norm": 0.0,
+      "learning_rate": 3.925164551805991e-06,
+      "loss": 0.8924,
+      "step": 19704
+    },
+    {
+      "epoch": 0.7166236316689093,
+      "grad_norm": 0.0,
+      "learning_rate": 3.924228980420931e-06,
+      "loss": 0.9837,
+      "step": 19705
+    },
+    {
+      "epoch": 0.7166599992726479,
+      "grad_norm": 0.0,
+      "learning_rate": 3.923293493328574e-06,
+      "loss": 0.9404,
+      "step": 19706
+    },
+    {
+      "epoch": 0.7166963668763865,
+      "grad_norm": 0.0,
+      "learning_rate": 3.922358090541898e-06,
+      "loss": 0.9915,
+      "step": 19707
+    },
+    {
+      "epoch": 0.7167327344801251,
+      "grad_norm": 0.0,
+      "learning_rate": 3.921422772073878e-06,
+      "loss": 0.9804,
+      "step": 19708
+    },
+    {
+      "epoch": 0.7167691020838637,
+      "grad_norm": 0.0,
+      "learning_rate": 3.920487537937499e-06,
+      "loss": 1.024,
+      "step": 19709
+    },
+    {
+      "epoch": 0.7168054696876023,
+      "grad_norm": 0.0,
+      "learning_rate": 3.919552388145721e-06,
+      "loss": 1.035,
+      "step": 19710
+    },
+    {
+      "epoch": 0.7168418372913409,
+      "grad_norm": 0.0,
+      "learning_rate": 3.918617322711532e-06,
+      "loss": 0.9104,
+      "step": 19711
+    },
+    {
+      "epoch": 0.7168782048950795,
+      "grad_norm": 0.0,
+      "learning_rate": 3.917682341647898e-06,
+      "loss": 0.9146,
+      "step": 19712
+    },
+    {
+      "epoch": 0.7169145724988181,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9167474449677934e-06,
+      "loss": 1.0169,
+      "step": 19713
+    },
+    {
+      "epoch": 0.7169509401025567,
+      "grad_norm": 0.0,
+      "learning_rate": 3.915812632684183e-06,
+      "loss": 0.9746,
+      "step": 19714
+    },
+    {
+      "epoch": 0.7169873077062953,
+      "grad_norm": 0.0,
+      "learning_rate": 3.914877904810043e-06,
+      "loss": 0.9401,
+      "step": 19715
+    },
+    {
+      "epoch": 0.7170236753100339,
+      "grad_norm": 0.0,
+      "learning_rate": 3.913943261358338e-06,
+      "loss": 0.9158,
+      "step": 19716
+    },
+    {
+      "epoch": 0.7170600429137725,
+      "grad_norm": 0.0,
+      "learning_rate": 3.913008702342036e-06,
+      "loss": 0.8432,
+      "step": 19717
+    },
+    {
+      "epoch": 0.717096410517511,
+      "grad_norm": 0.0,
+      "learning_rate": 3.912074227774102e-06,
+      "loss": 0.9268,
+      "step": 19718
+    },
+    {
+      "epoch": 0.7171327781212495,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9111398376675e-06,
+      "loss": 0.8292,
+      "step": 19719
+    },
+    {
+      "epoch": 0.7171691457249881,
+      "grad_norm": 0.0,
+      "learning_rate": 3.910205532035195e-06,
+      "loss": 0.9861,
+      "step": 19720
+    },
+    {
+      "epoch": 0.7172055133287267,
+      "grad_norm": 0.0,
+      "learning_rate": 3.909271310890142e-06,
+      "loss": 0.8862,
+      "step": 19721
+    },
+    {
+      "epoch": 0.7172418809324653,
+      "grad_norm": 0.0,
+      "learning_rate": 3.908337174245318e-06,
+      "loss": 0.9559,
+      "step": 19722
+    },
+    {
+      "epoch": 0.7172782485362039,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9074031221136635e-06,
+      "loss": 0.9324,
+      "step": 19723
+    },
+    {
+      "epoch": 0.7173146161399425,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9064691545081515e-06,
+      "loss": 0.9946,
+      "step": 19724
+    },
+    {
+      "epoch": 0.7173509837436811,
+      "grad_norm": 0.0,
+      "learning_rate": 3.905535271441733e-06,
+      "loss": 0.9273,
+      "step": 19725
+    },
+    {
+      "epoch": 0.7173873513474197,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9046014729273665e-06,
+      "loss": 1.083,
+      "step": 19726
+    },
+    {
+      "epoch": 0.7174237189511583,
+      "grad_norm": 0.0,
+      "learning_rate": 3.903667758978003e-06,
+      "loss": 0.9729,
+      "step": 19727
+    },
+    {
+      "epoch": 0.7174600865548969,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9027341296066056e-06,
+      "loss": 0.8114,
+      "step": 19728
+    },
+    {
+      "epoch": 0.7174964541586355,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9018005848261155e-06,
+      "loss": 0.8706,
+      "step": 19729
+    },
+    {
+      "epoch": 0.7175328217623741,
+      "grad_norm": 0.0,
+      "learning_rate": 3.900867124649492e-06,
+      "loss": 1.0325,
+      "step": 19730
+    },
+    {
+      "epoch": 0.7175691893661127,
+      "grad_norm": 0.0,
+      "learning_rate": 3.899933749089685e-06,
+      "loss": 0.9809,
+      "step": 19731
+    },
+    {
+      "epoch": 0.7176055569698513,
+      "grad_norm": 0.0,
+      "learning_rate": 3.899000458159642e-06,
+      "loss": 0.899,
+      "step": 19732
+    },
+    {
+      "epoch": 0.7176419245735899,
+      "grad_norm": 0.0,
+      "learning_rate": 3.898067251872308e-06,
+      "loss": 0.9255,
+      "step": 19733
+    },
+    {
+      "epoch": 0.7176782921773285,
+      "grad_norm": 0.0,
+      "learning_rate": 3.897134130240637e-06,
+      "loss": 0.8471,
+      "step": 19734
+    },
+    {
+      "epoch": 0.7177146597810671,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8962010932775715e-06,
+      "loss": 0.9907,
+      "step": 19735
+    },
+    {
+      "epoch": 0.7177510273848056,
+      "grad_norm": 0.0,
+      "learning_rate": 3.895268140996055e-06,
+      "loss": 0.8661,
+      "step": 19736
+    },
+    {
+      "epoch": 0.7177873949885442,
+      "grad_norm": 0.0,
+      "learning_rate": 3.894335273409031e-06,
+      "loss": 0.8998,
+      "step": 19737
+    },
+    {
+      "epoch": 0.7178237625922828,
+      "grad_norm": 0.0,
+      "learning_rate": 3.893402490529443e-06,
+      "loss": 0.8861,
+      "step": 19738
+    },
+    {
+      "epoch": 0.7178601301960214,
+      "grad_norm": 0.0,
+      "learning_rate": 3.892469792370232e-06,
+      "loss": 0.9478,
+      "step": 19739
+    },
+    {
+      "epoch": 0.71789649779976,
+      "grad_norm": 0.0,
+      "learning_rate": 3.891537178944333e-06,
+      "loss": 0.9481,
+      "step": 19740
+    },
+    {
+      "epoch": 0.7179328654034985,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8906046502646955e-06,
+      "loss": 0.9537,
+      "step": 19741
+    },
+    {
+      "epoch": 0.7179692330072371,
+      "grad_norm": 0.0,
+      "learning_rate": 3.889672206344244e-06,
+      "loss": 0.9676,
+      "step": 19742
+    },
+    {
+      "epoch": 0.7180056006109757,
+      "grad_norm": 0.0,
+      "learning_rate": 3.888739847195925e-06,
+      "loss": 0.8783,
+      "step": 19743
+    },
+    {
+      "epoch": 0.7180419682147143,
+      "grad_norm": 0.0,
+      "learning_rate": 3.887807572832669e-06,
+      "loss": 0.9971,
+      "step": 19744
+    },
+    {
+      "epoch": 0.7180783358184529,
+      "grad_norm": 0.0,
+      "learning_rate": 3.886875383267411e-06,
+      "loss": 0.9234,
+      "step": 19745
+    },
+    {
+      "epoch": 0.7181147034221915,
+      "grad_norm": 0.0,
+      "learning_rate": 3.88594327851308e-06,
+      "loss": 1.0716,
+      "step": 19746
+    },
+    {
+      "epoch": 0.7181510710259301,
+      "grad_norm": 0.0,
+      "learning_rate": 3.885011258582616e-06,
+      "loss": 0.8691,
+      "step": 19747
+    },
+    {
+      "epoch": 0.7181874386296687,
+      "grad_norm": 0.0,
+      "learning_rate": 3.884079323488944e-06,
+      "loss": 0.9288,
+      "step": 19748
+    },
+    {
+      "epoch": 0.7182238062334073,
+      "grad_norm": 0.0,
+      "learning_rate": 3.883147473244994e-06,
+      "loss": 0.9405,
+      "step": 19749
+    },
+    {
+      "epoch": 0.7182601738371459,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8822157078636934e-06,
+      "loss": 1.0351,
+      "step": 19750
+    },
+    {
+      "epoch": 0.7182965414408845,
+      "grad_norm": 0.0,
+      "learning_rate": 3.881284027357971e-06,
+      "loss": 0.9722,
+      "step": 19751
+    },
+    {
+      "epoch": 0.7183329090446231,
+      "grad_norm": 0.0,
+      "learning_rate": 3.880352431740747e-06,
+      "loss": 0.9138,
+      "step": 19752
+    },
+    {
+      "epoch": 0.7183692766483616,
+      "grad_norm": 0.0,
+      "learning_rate": 3.879420921024955e-06,
+      "loss": 0.8253,
+      "step": 19753
+    },
+    {
+      "epoch": 0.7184056442521002,
+      "grad_norm": 0.0,
+      "learning_rate": 3.878489495223513e-06,
+      "loss": 1.0055,
+      "step": 19754
+    },
+    {
+      "epoch": 0.7184420118558388,
+      "grad_norm": 0.0,
+      "learning_rate": 3.877558154349343e-06,
+      "loss": 0.8255,
+      "step": 19755
+    },
+    {
+      "epoch": 0.7184783794595774,
+      "grad_norm": 0.0,
+      "learning_rate": 3.876626898415369e-06,
+      "loss": 0.8883,
+      "step": 19756
+    },
+    {
+      "epoch": 0.718514747063316,
+      "grad_norm": 0.0,
+      "learning_rate": 3.875695727434508e-06,
+      "loss": 0.9559,
+      "step": 19757
+    },
+    {
+      "epoch": 0.7185511146670546,
+      "grad_norm": 0.0,
+      "learning_rate": 3.874764641419679e-06,
+      "loss": 0.9541,
+      "step": 19758
+    },
+    {
+      "epoch": 0.7185874822707932,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8738336403837975e-06,
+      "loss": 0.9786,
+      "step": 19759
+    },
+    {
+      "epoch": 0.7186238498745318,
+      "grad_norm": 0.0,
+      "learning_rate": 3.872902724339785e-06,
+      "loss": 1.0325,
+      "step": 19760
+    },
+    {
+      "epoch": 0.7186602174782704,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8719718933005535e-06,
+      "loss": 1.0277,
+      "step": 19761
+    },
+    {
+      "epoch": 0.718696585082009,
+      "grad_norm": 0.0,
+      "learning_rate": 3.871041147279019e-06,
+      "loss": 0.9127,
+      "step": 19762
+    },
+    {
+      "epoch": 0.7187329526857476,
+      "grad_norm": 0.0,
+      "learning_rate": 3.870110486288091e-06,
+      "loss": 0.9595,
+      "step": 19763
+    },
+    {
+      "epoch": 0.7187693202894861,
+      "grad_norm": 0.0,
+      "learning_rate": 3.869179910340685e-06,
+      "loss": 0.9968,
+      "step": 19764
+    },
+    {
+      "epoch": 0.7188056878932247,
+      "grad_norm": 0.0,
+      "learning_rate": 3.868249419449704e-06,
+      "loss": 1.0495,
+      "step": 19765
+    },
+    {
+      "epoch": 0.7188420554969633,
+      "grad_norm": 0.0,
+      "learning_rate": 3.867319013628066e-06,
+      "loss": 0.873,
+      "step": 19766
+    },
+    {
+      "epoch": 0.7188784231007019,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8663886928886765e-06,
+      "loss": 0.9471,
+      "step": 19767
+    },
+    {
+      "epoch": 0.7189147907044405,
+      "grad_norm": 0.0,
+      "learning_rate": 3.865458457244441e-06,
+      "loss": 0.8892,
+      "step": 19768
+    },
+    {
+      "epoch": 0.7189511583081791,
+      "grad_norm": 0.0,
+      "learning_rate": 3.864528306708262e-06,
+      "loss": 1.0472,
+      "step": 19769
+    },
+    {
+      "epoch": 0.7189875259119176,
+      "grad_norm": 0.0,
+      "learning_rate": 3.863598241293055e-06,
+      "loss": 0.9981,
+      "step": 19770
+    },
+    {
+      "epoch": 0.7190238935156562,
+      "grad_norm": 0.0,
+      "learning_rate": 3.862668261011708e-06,
+      "loss": 0.9248,
+      "step": 19771
+    },
+    {
+      "epoch": 0.7190602611193948,
+      "grad_norm": 0.0,
+      "learning_rate": 3.861738365877136e-06,
+      "loss": 0.8937,
+      "step": 19772
+    },
+    {
+      "epoch": 0.7190966287231334,
+      "grad_norm": 0.0,
+      "learning_rate": 3.860808555902234e-06,
+      "loss": 1.0285,
+      "step": 19773
+    },
+    {
+      "epoch": 0.719132996326872,
+      "grad_norm": 0.0,
+      "learning_rate": 3.859878831099903e-06,
+      "loss": 0.9028,
+      "step": 19774
+    },
+    {
+      "epoch": 0.7191693639306106,
+      "grad_norm": 0.0,
+      "learning_rate": 3.858949191483039e-06,
+      "loss": 0.9722,
+      "step": 19775
+    },
+    {
+      "epoch": 0.7192057315343492,
+      "grad_norm": 0.0,
+      "learning_rate": 3.858019637064548e-06,
+      "loss": 0.9627,
+      "step": 19776
+    },
+    {
+      "epoch": 0.7192420991380878,
+      "grad_norm": 0.0,
+      "learning_rate": 3.857090167857318e-06,
+      "loss": 1.007,
+      "step": 19777
+    },
+    {
+      "epoch": 0.7192784667418264,
+      "grad_norm": 0.0,
+      "learning_rate": 3.856160783874241e-06,
+      "loss": 0.9007,
+      "step": 19778
+    },
+    {
+      "epoch": 0.719314834345565,
+      "grad_norm": 0.0,
+      "learning_rate": 3.855231485128221e-06,
+      "loss": 0.9326,
+      "step": 19779
+    },
+    {
+      "epoch": 0.7193512019493036,
+      "grad_norm": 0.0,
+      "learning_rate": 3.854302271632146e-06,
+      "loss": 0.9278,
+      "step": 19780
+    },
+    {
+      "epoch": 0.7193875695530422,
+      "grad_norm": 0.0,
+      "learning_rate": 3.853373143398907e-06,
+      "loss": 0.8792,
+      "step": 19781
+    },
+    {
+      "epoch": 0.7194239371567808,
+      "grad_norm": 0.0,
+      "learning_rate": 3.852444100441391e-06,
+      "loss": 0.8886,
+      "step": 19782
+    },
+    {
+      "epoch": 0.7194603047605194,
+      "grad_norm": 0.0,
+      "learning_rate": 3.851515142772497e-06,
+      "loss": 0.9537,
+      "step": 19783
+    },
+    {
+      "epoch": 0.719496672364258,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8505862704051015e-06,
+      "loss": 0.8975,
+      "step": 19784
+    },
+    {
+      "epoch": 0.7195330399679966,
+      "grad_norm": 0.0,
+      "learning_rate": 3.849657483352099e-06,
+      "loss": 0.9323,
+      "step": 19785
+    },
+    {
+      "epoch": 0.7195694075717352,
+      "grad_norm": 0.0,
+      "learning_rate": 3.848728781626373e-06,
+      "loss": 0.9249,
+      "step": 19786
+    },
+    {
+      "epoch": 0.7196057751754736,
+      "grad_norm": 0.0,
+      "learning_rate": 3.847800165240807e-06,
+      "loss": 0.8254,
+      "step": 19787
+    },
+    {
+      "epoch": 0.7196421427792122,
+      "grad_norm": 0.0,
+      "learning_rate": 3.846871634208283e-06,
+      "loss": 0.9148,
+      "step": 19788
+    },
+    {
+      "epoch": 0.7196785103829508,
+      "grad_norm": 0.0,
+      "learning_rate": 3.84594318854169e-06,
+      "loss": 0.9201,
+      "step": 19789
+    },
+    {
+      "epoch": 0.7197148779866894,
+      "grad_norm": 0.0,
+      "learning_rate": 3.845014828253898e-06,
+      "loss": 1.0034,
+      "step": 19790
+    },
+    {
+      "epoch": 0.719751245590428,
+      "grad_norm": 0.0,
+      "learning_rate": 3.844086553357795e-06,
+      "loss": 0.8603,
+      "step": 19791
+    },
+    {
+      "epoch": 0.7197876131941666,
+      "grad_norm": 0.0,
+      "learning_rate": 3.843158363866259e-06,
+      "loss": 0.9765,
+      "step": 19792
+    },
+    {
+      "epoch": 0.7198239807979052,
+      "grad_norm": 0.0,
+      "learning_rate": 3.842230259792164e-06,
+      "loss": 0.9813,
+      "step": 19793
+    },
+    {
+      "epoch": 0.7198603484016438,
+      "grad_norm": 0.0,
+      "learning_rate": 3.841302241148383e-06,
+      "loss": 0.913,
+      "step": 19794
+    },
+    {
+      "epoch": 0.7198967160053824,
+      "grad_norm": 0.0,
+      "learning_rate": 3.840374307947804e-06,
+      "loss": 0.8772,
+      "step": 19795
+    },
+    {
+      "epoch": 0.719933083609121,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8394464602032875e-06,
+      "loss": 0.9805,
+      "step": 19796
+    },
+    {
+      "epoch": 0.7199694512128596,
+      "grad_norm": 0.0,
+      "learning_rate": 3.838518697927708e-06,
+      "loss": 0.9353,
+      "step": 19797
+    },
+    {
+      "epoch": 0.7200058188165982,
+      "grad_norm": 0.0,
+      "learning_rate": 3.837591021133942e-06,
+      "loss": 0.8794,
+      "step": 19798
+    },
+    {
+      "epoch": 0.7200421864203368,
+      "grad_norm": 0.0,
+      "learning_rate": 3.836663429834858e-06,
+      "loss": 0.9205,
+      "step": 19799
+    },
+    {
+      "epoch": 0.7200785540240754,
+      "grad_norm": 0.0,
+      "learning_rate": 3.835735924043325e-06,
+      "loss": 0.9579,
+      "step": 19800
+    },
+    {
+      "epoch": 0.720114921627814,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8348085037722045e-06,
+      "loss": 0.975,
+      "step": 19801
+    },
+    {
+      "epoch": 0.7201512892315526,
+      "grad_norm": 0.0,
+      "learning_rate": 3.833881169034378e-06,
+      "loss": 0.8826,
+      "step": 19802
+    },
+    {
+      "epoch": 0.7201876568352911,
+      "grad_norm": 0.0,
+      "learning_rate": 3.832953919842693e-06,
+      "loss": 0.984,
+      "step": 19803
+    },
+    {
+      "epoch": 0.7202240244390297,
+      "grad_norm": 0.0,
+      "learning_rate": 3.832026756210027e-06,
+      "loss": 0.9698,
+      "step": 19804
+    },
+    {
+      "epoch": 0.7202603920427683,
+      "grad_norm": 0.0,
+      "learning_rate": 3.831099678149238e-06,
+      "loss": 0.9307,
+      "step": 19805
+    },
+    {
+      "epoch": 0.7202967596465069,
+      "grad_norm": 0.0,
+      "learning_rate": 3.830172685673189e-06,
+      "loss": 0.9057,
+      "step": 19806
+    },
+    {
+      "epoch": 0.7203331272502455,
+      "grad_norm": 0.0,
+      "learning_rate": 3.829245778794737e-06,
+      "loss": 0.9553,
+      "step": 19807
+    },
+    {
+      "epoch": 0.720369494853984,
+      "grad_norm": 0.0,
+      "learning_rate": 3.82831895752675e-06,
+      "loss": 1.0498,
+      "step": 19808
+    },
+    {
+      "epoch": 0.7204058624577226,
+      "grad_norm": 0.0,
+      "learning_rate": 3.827392221882076e-06,
+      "loss": 1.0483,
+      "step": 19809
+    },
+    {
+      "epoch": 0.7204422300614612,
+      "grad_norm": 0.0,
+      "learning_rate": 3.826465571873578e-06,
+      "loss": 0.9927,
+      "step": 19810
+    },
+    {
+      "epoch": 0.7204785976651998,
+      "grad_norm": 0.0,
+      "learning_rate": 3.825539007514114e-06,
+      "loss": 0.9394,
+      "step": 19811
+    },
+    {
+      "epoch": 0.7205149652689384,
+      "grad_norm": 0.0,
+      "learning_rate": 3.824612528816535e-06,
+      "loss": 0.9178,
+      "step": 19812
+    },
+    {
+      "epoch": 0.720551332872677,
+      "grad_norm": 0.0,
+      "learning_rate": 3.823686135793691e-06,
+      "loss": 0.9987,
+      "step": 19813
+    },
+    {
+      "epoch": 0.7205877004764156,
+      "grad_norm": 0.0,
+      "learning_rate": 3.822759828458442e-06,
+      "loss": 1.0225,
+      "step": 19814
+    },
+    {
+      "epoch": 0.7206240680801542,
+      "grad_norm": 0.0,
+      "learning_rate": 3.821833606823641e-06,
+      "loss": 0.8452,
+      "step": 19815
+    },
+    {
+      "epoch": 0.7206604356838928,
+      "grad_norm": 0.0,
+      "learning_rate": 3.820907470902125e-06,
+      "loss": 0.878,
+      "step": 19816
+    },
+    {
+      "epoch": 0.7206968032876314,
+      "grad_norm": 0.0,
+      "learning_rate": 3.819981420706756e-06,
+      "loss": 1.0127,
+      "step": 19817
+    },
+    {
+      "epoch": 0.72073317089137,
+      "grad_norm": 0.0,
+      "learning_rate": 3.819055456250374e-06,
+      "loss": 1.0062,
+      "step": 19818
+    },
+    {
+      "epoch": 0.7207695384951086,
+      "grad_norm": 0.0,
+      "learning_rate": 3.81812957754583e-06,
+      "loss": 0.936,
+      "step": 19819
+    },
+    {
+      "epoch": 0.7208059060988471,
+      "grad_norm": 0.0,
+      "learning_rate": 3.817203784605963e-06,
+      "loss": 0.8535,
+      "step": 19820
+    },
+    {
+      "epoch": 0.7208422737025857,
+      "grad_norm": 0.0,
+      "learning_rate": 3.816278077443627e-06,
+      "loss": 0.9565,
+      "step": 19821
+    },
+    {
+      "epoch": 0.7208786413063243,
+      "grad_norm": 0.0,
+      "learning_rate": 3.815352456071653e-06,
+      "loss": 0.7804,
+      "step": 19822
+    },
+    {
+      "epoch": 0.7209150089100629,
+      "grad_norm": 0.0,
+      "learning_rate": 3.814426920502893e-06,
+      "loss": 0.9022,
+      "step": 19823
+    },
+    {
+      "epoch": 0.7209513765138015,
+      "grad_norm": 0.0,
+      "learning_rate": 3.813501470750183e-06,
+      "loss": 0.9379,
+      "step": 19824
+    },
+    {
+      "epoch": 0.7209877441175401,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8125761068263625e-06,
+      "loss": 0.9742,
+      "step": 19825
+    },
+    {
+      "epoch": 0.7210241117212787,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8116508287442665e-06,
+      "loss": 0.9193,
+      "step": 19826
+    },
+    {
+      "epoch": 0.7210604793250173,
+      "grad_norm": 0.0,
+      "learning_rate": 3.810725636516741e-06,
+      "loss": 0.993,
+      "step": 19827
+    },
+    {
+      "epoch": 0.7210968469287559,
+      "grad_norm": 0.0,
+      "learning_rate": 3.80980053015661e-06,
+      "loss": 0.926,
+      "step": 19828
+    },
+    {
+      "epoch": 0.7211332145324945,
+      "grad_norm": 0.0,
+      "learning_rate": 3.808875509676717e-06,
+      "loss": 0.9206,
+      "step": 19829
+    },
+    {
+      "epoch": 0.7211695821362331,
+      "grad_norm": 0.0,
+      "learning_rate": 3.807950575089893e-06,
+      "loss": 0.9411,
+      "step": 19830
+    },
+    {
+      "epoch": 0.7212059497399717,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8070257264089683e-06,
+      "loss": 0.9203,
+      "step": 19831
+    },
+    {
+      "epoch": 0.7212423173437102,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8061009636467716e-06,
+      "loss": 0.912,
+      "step": 19832
+    },
+    {
+      "epoch": 0.7212786849474488,
+      "grad_norm": 0.0,
+      "learning_rate": 3.80517628681614e-06,
+      "loss": 0.9559,
+      "step": 19833
+    },
+    {
+      "epoch": 0.7213150525511874,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8042516959299015e-06,
+      "loss": 0.9981,
+      "step": 19834
+    },
+    {
+      "epoch": 0.721351420154926,
+      "grad_norm": 0.0,
+      "learning_rate": 3.803327191000873e-06,
+      "loss": 0.921,
+      "step": 19835
+    },
+    {
+      "epoch": 0.7213877877586646,
+      "grad_norm": 0.0,
+      "learning_rate": 3.802402772041891e-06,
+      "loss": 0.8398,
+      "step": 19836
+    },
+    {
+      "epoch": 0.7214241553624031,
+      "grad_norm": 0.0,
+      "learning_rate": 3.801478439065778e-06,
+      "loss": 0.9662,
+      "step": 19837
+    },
+    {
+      "epoch": 0.7214605229661417,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8005541920853563e-06,
+      "loss": 0.9732,
+      "step": 19838
+    },
+    {
+      "epoch": 0.7214968905698803,
+      "grad_norm": 0.0,
+      "learning_rate": 3.799630031113446e-06,
+      "loss": 0.9965,
+      "step": 19839
+    },
+    {
+      "epoch": 0.7215332581736189,
+      "grad_norm": 0.0,
+      "learning_rate": 3.798705956162879e-06,
+      "loss": 1.087,
+      "step": 19840
+    },
+    {
+      "epoch": 0.7215696257773575,
+      "grad_norm": 0.0,
+      "learning_rate": 3.797781967246461e-06,
+      "loss": 0.9045,
+      "step": 19841
+    },
+    {
+      "epoch": 0.7216059933810961,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7968580643770224e-06,
+      "loss": 0.9853,
+      "step": 19842
+    },
+    {
+      "epoch": 0.7216423609848347,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7959342475673776e-06,
+      "loss": 1.0198,
+      "step": 19843
+    },
+    {
+      "epoch": 0.7216787285885733,
+      "grad_norm": 0.0,
+      "learning_rate": 3.795010516830342e-06,
+      "loss": 0.8392,
+      "step": 19844
+    },
+    {
+      "epoch": 0.7217150961923119,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7940868721787282e-06,
+      "loss": 0.9801,
+      "step": 19845
+    },
+    {
+      "epoch": 0.7217514637960505,
+      "grad_norm": 0.0,
+      "learning_rate": 3.793163313625358e-06,
+      "loss": 0.9692,
+      "step": 19846
+    },
+    {
+      "epoch": 0.7217878313997891,
+      "grad_norm": 0.0,
+      "learning_rate": 3.79223984118304e-06,
+      "loss": 0.8066,
+      "step": 19847
+    },
+    {
+      "epoch": 0.7218241990035277,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7913164548645866e-06,
+      "loss": 1.0576,
+      "step": 19848
+    },
+    {
+      "epoch": 0.7218605666072663,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7903931546828086e-06,
+      "loss": 0.9262,
+      "step": 19849
+    },
+    {
+      "epoch": 0.7218969342110049,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7894699406505155e-06,
+      "loss": 0.918,
+      "step": 19850
+    },
+    {
+      "epoch": 0.7219333018147435,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7885468127805114e-06,
+      "loss": 0.8477,
+      "step": 19851
+    },
+    {
+      "epoch": 0.7219696694184821,
+      "grad_norm": 0.0,
+      "learning_rate": 3.787623771085611e-06,
+      "loss": 0.8046,
+      "step": 19852
+    },
+    {
+      "epoch": 0.7220060370222207,
+      "grad_norm": 0.0,
+      "learning_rate": 3.78670081557862e-06,
+      "loss": 1.0175,
+      "step": 19853
+    },
+    {
+      "epoch": 0.7220424046259591,
+      "grad_norm": 0.0,
+      "learning_rate": 3.785777946272332e-06,
+      "loss": 0.9265,
+      "step": 19854
+    },
+    {
+      "epoch": 0.7220787722296977,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7848551631795626e-06,
+      "loss": 0.9905,
+      "step": 19855
+    },
+    {
+      "epoch": 0.7221151398334363,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7839324663131093e-06,
+      "loss": 0.9104,
+      "step": 19856
+    },
+    {
+      "epoch": 0.7221515074371749,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7830098556857743e-06,
+      "loss": 1.0537,
+      "step": 19857
+    },
+    {
+      "epoch": 0.7221878750409135,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7820873313103534e-06,
+      "loss": 0.9919,
+      "step": 19858
+    },
+    {
+      "epoch": 0.7222242426446521,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7811648931996515e-06,
+      "loss": 0.9726,
+      "step": 19859
+    },
+    {
+      "epoch": 0.7222606102483907,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7802425413664636e-06,
+      "loss": 1.0542,
+      "step": 19860
+    },
+    {
+      "epoch": 0.7222969778521293,
+      "grad_norm": 0.0,
+      "learning_rate": 3.779320275823585e-06,
+      "loss": 0.8118,
+      "step": 19861
+    },
+    {
+      "epoch": 0.7223333454558679,
+      "grad_norm": 0.0,
+      "learning_rate": 3.778398096583812e-06,
+      "loss": 0.8991,
+      "step": 19862
+    },
+    {
+      "epoch": 0.7223697130596065,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7774760036599387e-06,
+      "loss": 0.8504,
+      "step": 19863
+    },
+    {
+      "epoch": 0.7224060806633451,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7765539970647534e-06,
+      "loss": 1.0178,
+      "step": 19864
+    },
+    {
+      "epoch": 0.7224424482670837,
+      "grad_norm": 0.0,
+      "learning_rate": 3.775632076811054e-06,
+      "loss": 0.8716,
+      "step": 19865
+    },
+    {
+      "epoch": 0.7224788158708223,
+      "grad_norm": 0.0,
+      "learning_rate": 3.77471024291163e-06,
+      "loss": 0.9572,
+      "step": 19866
+    },
+    {
+      "epoch": 0.7225151834745609,
+      "grad_norm": 0.0,
+      "learning_rate": 3.773788495379268e-06,
+      "loss": 0.9361,
+      "step": 19867
+    },
+    {
+      "epoch": 0.7225515510782995,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7728668342267538e-06,
+      "loss": 1.0262,
+      "step": 19868
+    },
+    {
+      "epoch": 0.7225879186820381,
+      "grad_norm": 0.0,
+      "learning_rate": 3.771945259466884e-06,
+      "loss": 0.8007,
+      "step": 19869
+    },
+    {
+      "epoch": 0.7226242862857767,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7710237711124287e-06,
+      "loss": 1.0035,
+      "step": 19870
+    },
+    {
+      "epoch": 0.7226606538895152,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7701023691761864e-06,
+      "loss": 0.9081,
+      "step": 19871
+    },
+    {
+      "epoch": 0.7226970214932538,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7691810536709374e-06,
+      "loss": 0.9971,
+      "step": 19872
+    },
+    {
+      "epoch": 0.7227333890969924,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7682598246094537e-06,
+      "loss": 0.9561,
+      "step": 19873
+    },
+    {
+      "epoch": 0.722769756700731,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7673386820045264e-06,
+      "loss": 1.0951,
+      "step": 19874
+    },
+    {
+      "epoch": 0.7228061243044696,
+      "grad_norm": 0.0,
+      "learning_rate": 3.766417625868932e-06,
+      "loss": 0.9528,
+      "step": 19875
+    },
+    {
+      "epoch": 0.7228424919082082,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7654966562154495e-06,
+      "loss": 1.0937,
+      "step": 19876
+    },
+    {
+      "epoch": 0.7228788595119467,
+      "grad_norm": 0.0,
+      "learning_rate": 3.764575773056852e-06,
+      "loss": 0.8796,
+      "step": 19877
+    },
+    {
+      "epoch": 0.7229152271156853,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7636549764059217e-06,
+      "loss": 0.9399,
+      "step": 19878
+    },
+    {
+      "epoch": 0.7229515947194239,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7627342662754297e-06,
+      "loss": 0.9717,
+      "step": 19879
+    },
+    {
+      "epoch": 0.7229879623231625,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7618136426781514e-06,
+      "loss": 0.8881,
+      "step": 19880
+    },
+    {
+      "epoch": 0.7230243299269011,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7608931056268537e-06,
+      "loss": 0.9645,
+      "step": 19881
+    },
+    {
+      "epoch": 0.7230606975306397,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7599726551343195e-06,
+      "loss": 0.9362,
+      "step": 19882
+    },
+    {
+      "epoch": 0.7230970651343783,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7590522912133045e-06,
+      "loss": 0.871,
+      "step": 19883
+    },
+    {
+      "epoch": 0.7231334327381169,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7581320138765864e-06,
+      "loss": 0.9081,
+      "step": 19884
+    },
+    {
+      "epoch": 0.7231698003418555,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7572118231369314e-06,
+      "loss": 0.9861,
+      "step": 19885
+    },
+    {
+      "epoch": 0.7232061679455941,
+      "grad_norm": 0.0,
+      "learning_rate": 3.756291719007106e-06,
+      "loss": 0.8418,
+      "step": 19886
+    },
+    {
+      "epoch": 0.7232425355493326,
+      "grad_norm": 0.0,
+      "learning_rate": 3.755371701499869e-06,
+      "loss": 0.9236,
+      "step": 19887
+    },
+    {
+      "epoch": 0.7232789031530712,
+      "grad_norm": 0.0,
+      "learning_rate": 3.754451770627997e-06,
+      "loss": 1.0383,
+      "step": 19888
+    },
+    {
+      "epoch": 0.7233152707568098,
+      "grad_norm": 0.0,
+      "learning_rate": 3.753531926404239e-06,
+      "loss": 0.9119,
+      "step": 19889
+    },
+    {
+      "epoch": 0.7233516383605484,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7526121688413663e-06,
+      "loss": 0.8127,
+      "step": 19890
+    },
+    {
+      "epoch": 0.723388005964287,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7516924979521364e-06,
+      "loss": 0.9551,
+      "step": 19891
+    },
+    {
+      "epoch": 0.7234243735680256,
+      "grad_norm": 0.0,
+      "learning_rate": 3.750772913749308e-06,
+      "loss": 0.7394,
+      "step": 19892
+    },
+    {
+      "epoch": 0.7234607411717642,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7498534162456377e-06,
+      "loss": 0.8148,
+      "step": 19893
+    },
+    {
+      "epoch": 0.7234971087755028,
+      "grad_norm": 0.0,
+      "learning_rate": 3.748934005453885e-06,
+      "loss": 0.9724,
+      "step": 19894
+    },
+    {
+      "epoch": 0.7235334763792414,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7480146813868024e-06,
+      "loss": 0.9493,
+      "step": 19895
+    },
+    {
+      "epoch": 0.72356984398298,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7470954440571426e-06,
+      "loss": 0.8869,
+      "step": 19896
+    },
+    {
+      "epoch": 0.7236062115867186,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7461762934776657e-06,
+      "loss": 1.1165,
+      "step": 19897
+    },
+    {
+      "epoch": 0.7236425791904572,
+      "grad_norm": 0.0,
+      "learning_rate": 3.745257229661119e-06,
+      "loss": 0.9077,
+      "step": 19898
+    },
+    {
+      "epoch": 0.7236789467941958,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7443382526202553e-06,
+      "loss": 1.002,
+      "step": 19899
+    },
+    {
+      "epoch": 0.7237153143979344,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7434193623678183e-06,
+      "loss": 1.0403,
+      "step": 19900
+    },
+    {
+      "epoch": 0.723751682001673,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7425005589165674e-06,
+      "loss": 0.9622,
+      "step": 19901
+    },
+    {
+      "epoch": 0.7237880496054115,
+      "grad_norm": 0.0,
+      "learning_rate": 3.741581842279236e-06,
+      "loss": 0.895,
+      "step": 19902
+    },
+    {
+      "epoch": 0.7238244172091501,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7406632124685816e-06,
+      "loss": 0.9935,
+      "step": 19903
+    },
+    {
+      "epoch": 0.7238607848128886,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7397446694973427e-06,
+      "loss": 0.982,
+      "step": 19904
+    },
+    {
+      "epoch": 0.7238971524166272,
+      "grad_norm": 0.0,
+      "learning_rate": 3.738826213378266e-06,
+      "loss": 0.8584,
+      "step": 19905
+    },
+    {
+      "epoch": 0.7239335200203658,
+      "grad_norm": 0.0,
+      "learning_rate": 3.737907844124088e-06,
+      "loss": 0.85,
+      "step": 19906
+    },
+    {
+      "epoch": 0.7239698876241044,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7369895617475605e-06,
+      "loss": 0.6927,
+      "step": 19907
+    },
+    {
+      "epoch": 0.724006255227843,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7360713662614103e-06,
+      "loss": 1.0014,
+      "step": 19908
+    },
+    {
+      "epoch": 0.7240426228315816,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7351532576783857e-06,
+      "loss": 0.8503,
+      "step": 19909
+    },
+    {
+      "epoch": 0.7240789904353202,
+      "grad_norm": 0.0,
+      "learning_rate": 3.734235236011221e-06,
+      "loss": 0.952,
+      "step": 19910
+    },
+    {
+      "epoch": 0.7241153580390588,
+      "grad_norm": 0.0,
+      "learning_rate": 3.733317301272653e-06,
+      "loss": 0.9239,
+      "step": 19911
+    },
+    {
+      "epoch": 0.7241517256427974,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7323994534754147e-06,
+      "loss": 0.9068,
+      "step": 19912
+    },
+    {
+      "epoch": 0.724188093246536,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7314816926322396e-06,
+      "loss": 0.9108,
+      "step": 19913
+    },
+    {
+      "epoch": 0.7242244608502746,
+      "grad_norm": 0.0,
+      "learning_rate": 3.730564018755869e-06,
+      "loss": 1.0424,
+      "step": 19914
+    },
+    {
+      "epoch": 0.7242608284540132,
+      "grad_norm": 0.0,
+      "learning_rate": 3.72964643185902e-06,
+      "loss": 1.015,
+      "step": 19915
+    },
+    {
+      "epoch": 0.7242971960577518,
+      "grad_norm": 0.0,
+      "learning_rate": 3.728728931954433e-06,
+      "loss": 0.9015,
+      "step": 19916
+    },
+    {
+      "epoch": 0.7243335636614904,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7278115190548357e-06,
+      "loss": 0.8691,
+      "step": 19917
+    },
+    {
+      "epoch": 0.724369931265229,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7268941931729542e-06,
+      "loss": 1.092,
+      "step": 19918
+    },
+    {
+      "epoch": 0.7244062988689676,
+      "grad_norm": 0.0,
+      "learning_rate": 3.725976954321511e-06,
+      "loss": 0.9621,
+      "step": 19919
+    },
+    {
+      "epoch": 0.7244426664727062,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7250598025132433e-06,
+      "loss": 1.0192,
+      "step": 19920
+    },
+    {
+      "epoch": 0.7244790340764447,
+      "grad_norm": 0.0,
+      "learning_rate": 3.724142737760861e-06,
+      "loss": 0.8661,
+      "step": 19921
+    },
+    {
+      "epoch": 0.7245154016801832,
+      "grad_norm": 0.0,
+      "learning_rate": 3.723225760077097e-06,
+      "loss": 0.9924,
+      "step": 19922
+    },
+    {
+      "epoch": 0.7245517692839218,
+      "grad_norm": 0.0,
+      "learning_rate": 3.722308869474671e-06,
+      "loss": 0.9811,
+      "step": 19923
+    },
+    {
+      "epoch": 0.7245881368876604,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7213920659663015e-06,
+      "loss": 0.9634,
+      "step": 19924
+    },
+    {
+      "epoch": 0.724624504491399,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7204753495647063e-06,
+      "loss": 0.944,
+      "step": 19925
+    },
+    {
+      "epoch": 0.7246608720951376,
+      "grad_norm": 0.0,
+      "learning_rate": 3.71955872028261e-06,
+      "loss": 1.0397,
+      "step": 19926
+    },
+    {
+      "epoch": 0.7246972396988762,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7186421781327243e-06,
+      "loss": 0.8459,
+      "step": 19927
+    },
+    {
+      "epoch": 0.7247336073026148,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7177257231277675e-06,
+      "loss": 0.9577,
+      "step": 19928
+    },
+    {
+      "epoch": 0.7247699749063534,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7168093552804516e-06,
+      "loss": 0.9162,
+      "step": 19929
+    },
+    {
+      "epoch": 0.724806342510092,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7158930746034927e-06,
+      "loss": 0.9209,
+      "step": 19930
+    },
+    {
+      "epoch": 0.7248427101138306,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7149768811096e-06,
+      "loss": 0.9551,
+      "step": 19931
+    },
+    {
+      "epoch": 0.7248790777175692,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7140607748114833e-06,
+      "loss": 0.9083,
+      "step": 19932
+    },
+    {
+      "epoch": 0.7249154453213078,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7131447557218613e-06,
+      "loss": 1.0125,
+      "step": 19933
+    },
+    {
+      "epoch": 0.7249518129250464,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7122288238534287e-06,
+      "loss": 0.9784,
+      "step": 19934
+    },
+    {
+      "epoch": 0.724988180528785,
+      "grad_norm": 0.0,
+      "learning_rate": 3.711312979218904e-06,
+      "loss": 0.8595,
+      "step": 19935
+    },
+    {
+      "epoch": 0.7250245481325236,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7103972218309894e-06,
+      "loss": 0.811,
+      "step": 19936
+    },
+    {
+      "epoch": 0.7250609157362622,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7094815517023897e-06,
+      "loss": 0.8461,
+      "step": 19937
+    },
+    {
+      "epoch": 0.7250972833400007,
+      "grad_norm": 0.0,
+      "learning_rate": 3.708565968845804e-06,
+      "loss": 0.9548,
+      "step": 19938
+    },
+    {
+      "epoch": 0.7251336509437393,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7076504732739473e-06,
+      "loss": 1.0409,
+      "step": 19939
+    },
+    {
+      "epoch": 0.7251700185474779,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7067350649995037e-06,
+      "loss": 0.9296,
+      "step": 19940
+    },
+    {
+      "epoch": 0.7252063861512165,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7058197440351872e-06,
+      "loss": 0.8561,
+      "step": 19941
+    },
+    {
+      "epoch": 0.7252427537549551,
+      "grad_norm": 0.0,
+      "learning_rate": 3.704904510393691e-06,
+      "loss": 0.8971,
+      "step": 19942
+    },
+    {
+      "epoch": 0.7252791213586937,
+      "grad_norm": 0.0,
+      "learning_rate": 3.703989364087713e-06,
+      "loss": 0.8442,
+      "step": 19943
+    },
+    {
+      "epoch": 0.7253154889624323,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7030743051299467e-06,
+      "loss": 0.9898,
+      "step": 19944
+    },
+    {
+      "epoch": 0.7253518565661708,
+      "grad_norm": 0.0,
+      "learning_rate": 3.702159333533093e-06,
+      "loss": 0.9407,
+      "step": 19945
+    },
+    {
+      "epoch": 0.7253882241699094,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7012444493098442e-06,
+      "loss": 0.9125,
+      "step": 19946
+    },
+    {
+      "epoch": 0.725424591773648,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7003296524728917e-06,
+      "loss": 0.8903,
+      "step": 19947
+    },
+    {
+      "epoch": 0.7254609593773866,
+      "grad_norm": 0.0,
+      "learning_rate": 3.699414943034926e-06,
+      "loss": 0.9915,
+      "step": 19948
+    },
+    {
+      "epoch": 0.7254973269811252,
+      "grad_norm": 0.0,
+      "learning_rate": 3.698500321008639e-06,
+      "loss": 1.0042,
+      "step": 19949
+    },
+    {
+      "epoch": 0.7255336945848638,
+      "grad_norm": 0.0,
+      "learning_rate": 3.69758578640672e-06,
+      "loss": 0.894,
+      "step": 19950
+    },
+    {
+      "epoch": 0.7255700621886024,
+      "grad_norm": 0.0,
+      "learning_rate": 3.696671339241853e-06,
+      "loss": 0.8757,
+      "step": 19951
+    },
+    {
+      "epoch": 0.725606429792341,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6957569795267344e-06,
+      "loss": 0.9148,
+      "step": 19952
+    },
+    {
+      "epoch": 0.7256427973960796,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6948427072740357e-06,
+      "loss": 0.9012,
+      "step": 19953
+    },
+    {
+      "epoch": 0.7256791649998182,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6939285224964526e-06,
+      "loss": 0.9024,
+      "step": 19954
+    },
+    {
+      "epoch": 0.7257155326035567,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6930144252066635e-06,
+      "loss": 0.8713,
+      "step": 19955
+    },
+    {
+      "epoch": 0.7257519002072953,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6921004154173523e-06,
+      "loss": 1.0403,
+      "step": 19956
+    },
+    {
+      "epoch": 0.7257882678110339,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6911864931411936e-06,
+      "loss": 0.9985,
+      "step": 19957
+    },
+    {
+      "epoch": 0.7258246354147725,
+      "grad_norm": 0.0,
+      "learning_rate": 3.690272658390874e-06,
+      "loss": 0.8564,
+      "step": 19958
+    },
+    {
+      "epoch": 0.7258610030185111,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6893589111790707e-06,
+      "loss": 0.8391,
+      "step": 19959
+    },
+    {
+      "epoch": 0.7258973706222497,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6884452515184575e-06,
+      "loss": 0.9426,
+      "step": 19960
+    },
+    {
+      "epoch": 0.7259337382259883,
+      "grad_norm": 0.0,
+      "learning_rate": 3.687531679421712e-06,
+      "loss": 0.9048,
+      "step": 19961
+    },
+    {
+      "epoch": 0.7259701058297269,
+      "grad_norm": 0.0,
+      "learning_rate": 3.686618194901508e-06,
+      "loss": 1.005,
+      "step": 19962
+    },
+    {
+      "epoch": 0.7260064734334655,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6857047979705164e-06,
+      "loss": 0.9589,
+      "step": 19963
+    },
+    {
+      "epoch": 0.7260428410372041,
+      "grad_norm": 0.0,
+      "learning_rate": 3.684791488641414e-06,
+      "loss": 0.8444,
+      "step": 19964
+    },
+    {
+      "epoch": 0.7260792086409427,
+      "grad_norm": 0.0,
+      "learning_rate": 3.68387826692687e-06,
+      "loss": 0.9754,
+      "step": 19965
+    },
+    {
+      "epoch": 0.7261155762446813,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6829651328395544e-06,
+      "loss": 0.9583,
+      "step": 19966
+    },
+    {
+      "epoch": 0.7261519438484199,
+      "grad_norm": 0.0,
+      "learning_rate": 3.682052086392134e-06,
+      "loss": 0.9213,
+      "step": 19967
+    },
+    {
+      "epoch": 0.7261883114521585,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6811391275972776e-06,
+      "loss": 0.9484,
+      "step": 19968
+    },
+    {
+      "epoch": 0.726224679055897,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6802262564676506e-06,
+      "loss": 0.84,
+      "step": 19969
+    },
+    {
+      "epoch": 0.7262610466596356,
+      "grad_norm": 0.0,
+      "learning_rate": 3.679313473015914e-06,
+      "loss": 0.8925,
+      "step": 19970
+    },
+    {
+      "epoch": 0.7262974142633742,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6784007772547413e-06,
+      "loss": 0.8148,
+      "step": 19971
+    },
+    {
+      "epoch": 0.7263337818671127,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6774881691967824e-06,
+      "loss": 0.9478,
+      "step": 19972
+    },
+    {
+      "epoch": 0.7263701494708513,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6765756488547078e-06,
+      "loss": 0.8844,
+      "step": 19973
+    },
+    {
+      "epoch": 0.7264065170745899,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6756632162411755e-06,
+      "loss": 0.9891,
+      "step": 19974
+    },
+    {
+      "epoch": 0.7264428846783285,
+      "grad_norm": 0.0,
+      "learning_rate": 3.674750871368842e-06,
+      "loss": 0.9636,
+      "step": 19975
+    },
+    {
+      "epoch": 0.7264792522820671,
+      "grad_norm": 0.0,
+      "learning_rate": 3.673838614250362e-06,
+      "loss": 1.0448,
+      "step": 19976
+    },
+    {
+      "epoch": 0.7265156198858057,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6729264448983994e-06,
+      "loss": 0.9082,
+      "step": 19977
+    },
+    {
+      "epoch": 0.7265519874895443,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6720143633256057e-06,
+      "loss": 0.8347,
+      "step": 19978
+    },
+    {
+      "epoch": 0.7265883550932829,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6711023695446336e-06,
+      "loss": 0.7829,
+      "step": 19979
+    },
+    {
+      "epoch": 0.7266247226970215,
+      "grad_norm": 0.0,
+      "learning_rate": 3.670190463568134e-06,
+      "loss": 0.9081,
+      "step": 19980
+    },
+    {
+      "epoch": 0.7266610903007601,
+      "grad_norm": 0.0,
+      "learning_rate": 3.669278645408768e-06,
+      "loss": 1.0299,
+      "step": 19981
+    },
+    {
+      "epoch": 0.7266974579044987,
+      "grad_norm": 0.0,
+      "learning_rate": 3.668366915079171e-06,
+      "loss": 0.8911,
+      "step": 19982
+    },
+    {
+      "epoch": 0.7267338255082373,
+      "grad_norm": 0.0,
+      "learning_rate": 3.667455272592003e-06,
+      "loss": 0.8967,
+      "step": 19983
+    },
+    {
+      "epoch": 0.7267701931119759,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6665437179599094e-06,
+      "loss": 0.9143,
+      "step": 19984
+    },
+    {
+      "epoch": 0.7268065607157145,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6656322511955358e-06,
+      "loss": 0.9007,
+      "step": 19985
+    },
+    {
+      "epoch": 0.7268429283194531,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6647208723115257e-06,
+      "loss": 0.9315,
+      "step": 19986
+    },
+    {
+      "epoch": 0.7268792959231917,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6638095813205265e-06,
+      "loss": 1.0021,
+      "step": 19987
+    },
+    {
+      "epoch": 0.7269156635269302,
+      "grad_norm": 0.0,
+      "learning_rate": 3.662898378235179e-06,
+      "loss": 0.8444,
+      "step": 19988
+    },
+    {
+      "epoch": 0.7269520311306688,
+      "grad_norm": 0.0,
+      "learning_rate": 3.661987263068122e-06,
+      "loss": 0.9121,
+      "step": 19989
+    },
+    {
+      "epoch": 0.7269883987344073,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6610762358320018e-06,
+      "loss": 0.8562,
+      "step": 19990
+    },
+    {
+      "epoch": 0.727024766338146,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6601652965394562e-06,
+      "loss": 1.0424,
+      "step": 19991
+    },
+    {
+      "epoch": 0.7270611339418845,
+      "grad_norm": 0.0,
+      "learning_rate": 3.659254445203122e-06,
+      "loss": 0.9632,
+      "step": 19992
+    },
+    {
+      "epoch": 0.7270975015456231,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6583436818356354e-06,
+      "loss": 0.9657,
+      "step": 19993
+    },
+    {
+      "epoch": 0.7271338691493617,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6574330064496335e-06,
+      "loss": 0.8165,
+      "step": 19994
+    },
+    {
+      "epoch": 0.7271702367531003,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6565224190577454e-06,
+      "loss": 0.8831,
+      "step": 19995
+    },
+    {
+      "epoch": 0.7272066043568389,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6556119196726127e-06,
+      "loss": 0.8455,
+      "step": 19996
+    },
+    {
+      "epoch": 0.7272429719605775,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6547015083068617e-06,
+      "loss": 1.0101,
+      "step": 19997
+    },
+    {
+      "epoch": 0.7272793395643161,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6537911849731255e-06,
+      "loss": 1.0183,
+      "step": 19998
+    },
+    {
+      "epoch": 0.7273157071680547,
+      "grad_norm": 0.0,
+      "learning_rate": 3.652880949684029e-06,
+      "loss": 1.0,
+      "step": 19999
+    },
+    {
+      "epoch": 0.7273520747717933,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6519708024522107e-06,
+      "loss": 0.9744,
+      "step": 20000
+    },
+    {
+      "epoch": 0.7273884423755319,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6510607432902846e-06,
+      "loss": 0.9996,
+      "step": 20001
+    },
+    {
+      "epoch": 0.7274248099792705,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6501507722108853e-06,
+      "loss": 0.9131,
+      "step": 20002
+    },
+    {
+      "epoch": 0.7274611775830091,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6492408892266352e-06,
+      "loss": 0.9355,
+      "step": 20003
+    },
+    {
+      "epoch": 0.7274975451867477,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6483310943501573e-06,
+      "loss": 0.9313,
+      "step": 20004
+    },
+    {
+      "epoch": 0.7275339127904862,
+      "grad_norm": 0.0,
+      "learning_rate": 3.64742138759407e-06,
+      "loss": 0.9402,
+      "step": 20005
+    },
+    {
+      "epoch": 0.7275702803942248,
+      "grad_norm": 0.0,
+      "learning_rate": 3.646511768971004e-06,
+      "loss": 0.9205,
+      "step": 20006
+    },
+    {
+      "epoch": 0.7276066479979634,
+      "grad_norm": 0.0,
+      "learning_rate": 3.645602238493571e-06,
+      "loss": 0.9331,
+      "step": 20007
+    },
+    {
+      "epoch": 0.727643015601702,
+      "grad_norm": 0.0,
+      "learning_rate": 3.644692796174386e-06,
+      "loss": 0.9784,
+      "step": 20008
+    },
+    {
+      "epoch": 0.7276793832054406,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6437834420260764e-06,
+      "loss": 1.0656,
+      "step": 20009
+    },
+    {
+      "epoch": 0.7277157508091792,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6428741760612517e-06,
+      "loss": 0.8469,
+      "step": 20010
+    },
+    {
+      "epoch": 0.7277521184129178,
+      "grad_norm": 0.0,
+      "learning_rate": 3.64196499829253e-06,
+      "loss": 0.9455,
+      "step": 20011
+    },
+    {
+      "epoch": 0.7277884860166564,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6410559087325182e-06,
+      "loss": 0.8575,
+      "step": 20012
+    },
+    {
+      "epoch": 0.727824853620395,
+      "grad_norm": 0.0,
+      "learning_rate": 3.640146907393841e-06,
+      "loss": 0.9268,
+      "step": 20013
+    },
+    {
+      "epoch": 0.7278612212241335,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6392379942890955e-06,
+      "loss": 0.9594,
+      "step": 20014
+    },
+    {
+      "epoch": 0.7278975888278721,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6383291694309007e-06,
+      "loss": 0.986,
+      "step": 20015
+    },
+    {
+      "epoch": 0.7279339564316107,
+      "grad_norm": 0.0,
+      "learning_rate": 3.637420432831863e-06,
+      "loss": 0.8942,
+      "step": 20016
+    },
+    {
+      "epoch": 0.7279703240353493,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6365117845045896e-06,
+      "loss": 0.8863,
+      "step": 20017
+    },
+    {
+      "epoch": 0.7280066916390879,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6356032244616836e-06,
+      "loss": 0.9676,
+      "step": 20018
+    },
+    {
+      "epoch": 0.7280430592428265,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6346947527157582e-06,
+      "loss": 0.8126,
+      "step": 20019
+    },
+    {
+      "epoch": 0.7280794268465651,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6337863692794064e-06,
+      "loss": 0.8628,
+      "step": 20020
+    },
+    {
+      "epoch": 0.7281157944503037,
+      "grad_norm": 0.0,
+      "learning_rate": 3.632878074165239e-06,
+      "loss": 1.0024,
+      "step": 20021
+    },
+    {
+      "epoch": 0.7281521620540422,
+      "grad_norm": 0.0,
+      "learning_rate": 3.631969867385855e-06,
+      "loss": 0.9537,
+      "step": 20022
+    },
+    {
+      "epoch": 0.7281885296577808,
+      "grad_norm": 0.0,
+      "learning_rate": 3.631061748953855e-06,
+      "loss": 0.9091,
+      "step": 20023
+    },
+    {
+      "epoch": 0.7282248972615194,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6301537188818316e-06,
+      "loss": 0.8763,
+      "step": 20024
+    },
+    {
+      "epoch": 0.728261264865258,
+      "grad_norm": 0.0,
+      "learning_rate": 3.629245777182392e-06,
+      "loss": 0.9128,
+      "step": 20025
+    },
+    {
+      "epoch": 0.7282976324689966,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6283379238681326e-06,
+      "loss": 0.8331,
+      "step": 20026
+    },
+    {
+      "epoch": 0.7283340000727352,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6274301589516357e-06,
+      "loss": 0.8682,
+      "step": 20027
+    },
+    {
+      "epoch": 0.7283703676764738,
+      "grad_norm": 0.0,
+      "learning_rate": 3.626522482445508e-06,
+      "loss": 0.8745,
+      "step": 20028
+    },
+    {
+      "epoch": 0.7284067352802124,
+      "grad_norm": 0.0,
+      "learning_rate": 3.625614894362339e-06,
+      "loss": 0.8881,
+      "step": 20029
+    },
+    {
+      "epoch": 0.728443102883951,
+      "grad_norm": 0.0,
+      "learning_rate": 3.624707394714718e-06,
+      "loss": 0.8784,
+      "step": 20030
+    },
+    {
+      "epoch": 0.7284794704876896,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6237999835152336e-06,
+      "loss": 0.9074,
+      "step": 20031
+    },
+    {
+      "epoch": 0.7285158380914282,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6228926607764846e-06,
+      "loss": 1.0919,
+      "step": 20032
+    },
+    {
+      "epoch": 0.7285522056951668,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6219854265110456e-06,
+      "loss": 0.9932,
+      "step": 20033
+    },
+    {
+      "epoch": 0.7285885732989054,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6210782807315124e-06,
+      "loss": 0.9599,
+      "step": 20034
+    },
+    {
+      "epoch": 0.728624940902644,
+      "grad_norm": 0.0,
+      "learning_rate": 3.620171223450468e-06,
+      "loss": 0.9705,
+      "step": 20035
+    },
+    {
+      "epoch": 0.7286613085063826,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6192642546804956e-06,
+      "loss": 0.935,
+      "step": 20036
+    },
+    {
+      "epoch": 0.7286976761101212,
+      "grad_norm": 0.0,
+      "learning_rate": 3.618357374434176e-06,
+      "loss": 0.9142,
+      "step": 20037
+    },
+    {
+      "epoch": 0.7287340437138597,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6174505827241003e-06,
+      "loss": 0.9853,
+      "step": 20038
+    },
+    {
+      "epoch": 0.7287704113175982,
+      "grad_norm": 0.0,
+      "learning_rate": 3.616543879562834e-06,
+      "loss": 0.9412,
+      "step": 20039
+    },
+    {
+      "epoch": 0.7288067789213368,
+      "grad_norm": 0.0,
+      "learning_rate": 3.615637264962969e-06,
+      "loss": 0.9559,
+      "step": 20040
+    },
+    {
+      "epoch": 0.7288431465250754,
+      "grad_norm": 0.0,
+      "learning_rate": 3.614730738937078e-06,
+      "loss": 1.0156,
+      "step": 20041
+    },
+    {
+      "epoch": 0.728879514128814,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6138243014977394e-06,
+      "loss": 0.9435,
+      "step": 20042
+    },
+    {
+      "epoch": 0.7289158817325526,
+      "grad_norm": 0.0,
+      "learning_rate": 3.612917952657523e-06,
+      "loss": 0.9061,
+      "step": 20043
+    },
+    {
+      "epoch": 0.7289522493362912,
+      "grad_norm": 0.0,
+      "learning_rate": 3.612011692429013e-06,
+      "loss": 0.9772,
+      "step": 20044
+    },
+    {
+      "epoch": 0.7289886169400298,
+      "grad_norm": 0.0,
+      "learning_rate": 3.611105520824779e-06,
+      "loss": 0.8883,
+      "step": 20045
+    },
+    {
+      "epoch": 0.7290249845437684,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6101994378573855e-06,
+      "loss": 0.9317,
+      "step": 20046
+    },
+    {
+      "epoch": 0.729061352147507,
+      "grad_norm": 0.0,
+      "learning_rate": 3.609293443539412e-06,
+      "loss": 0.9054,
+      "step": 20047
+    },
+    {
+      "epoch": 0.7290977197512456,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6083875378834253e-06,
+      "loss": 0.9502,
+      "step": 20048
+    },
+    {
+      "epoch": 0.7291340873549842,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6074817209019918e-06,
+      "loss": 1.0272,
+      "step": 20049
+    },
+    {
+      "epoch": 0.7291704549587228,
+      "grad_norm": 0.0,
+      "learning_rate": 3.606575992607677e-06,
+      "loss": 0.9445,
+      "step": 20050
+    },
+    {
+      "epoch": 0.7292068225624614,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6056703530130566e-06,
+      "loss": 0.8939,
+      "step": 20051
+    },
+    {
+      "epoch": 0.7292431901662,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6047648021306805e-06,
+      "loss": 1.0699,
+      "step": 20052
+    },
+    {
+      "epoch": 0.7292795577699386,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6038593399731213e-06,
+      "loss": 0.9431,
+      "step": 20053
+    },
+    {
+      "epoch": 0.7293159253736772,
+      "grad_norm": 0.0,
+      "learning_rate": 3.60295396655294e-06,
+      "loss": 1.0642,
+      "step": 20054
+    },
+    {
+      "epoch": 0.7293522929774158,
+      "grad_norm": 0.0,
+      "learning_rate": 3.602048681882696e-06,
+      "loss": 1.0016,
+      "step": 20055
+    },
+    {
+      "epoch": 0.7293886605811543,
+      "grad_norm": 0.0,
+      "learning_rate": 3.601143485974946e-06,
+      "loss": 1.0059,
+      "step": 20056
+    },
+    {
+      "epoch": 0.7294250281848929,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6002383788422548e-06,
+      "loss": 0.8754,
+      "step": 20057
+    },
+    {
+      "epoch": 0.7294613957886315,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5993333604971748e-06,
+      "loss": 0.9035,
+      "step": 20058
+    },
+    {
+      "epoch": 0.72949776339237,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5984284309522642e-06,
+      "loss": 0.9617,
+      "step": 20059
+    },
+    {
+      "epoch": 0.7295341309961086,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5975235902200766e-06,
+      "loss": 0.7985,
+      "step": 20060
+    },
+    {
+      "epoch": 0.7295704985998472,
+      "grad_norm": 0.0,
+      "learning_rate": 3.596618838313165e-06,
+      "loss": 0.9352,
+      "step": 20061
+    },
+    {
+      "epoch": 0.7296068662035858,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5957141752440785e-06,
+      "loss": 0.9828,
+      "step": 20062
+    },
+    {
+      "epoch": 0.7296432338073244,
+      "grad_norm": 0.0,
+      "learning_rate": 3.594809601025374e-06,
+      "loss": 0.937,
+      "step": 20063
+    },
+    {
+      "epoch": 0.729679601411063,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5939051156696025e-06,
+      "loss": 0.7836,
+      "step": 20064
+    },
+    {
+      "epoch": 0.7297159690148016,
+      "grad_norm": 0.0,
+      "learning_rate": 3.593000719189301e-06,
+      "loss": 0.9228,
+      "step": 20065
+    },
+    {
+      "epoch": 0.7297523366185402,
+      "grad_norm": 0.0,
+      "learning_rate": 3.592096411597028e-06,
+      "loss": 0.9426,
+      "step": 20066
+    },
+    {
+      "epoch": 0.7297887042222788,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5911921929053263e-06,
+      "loss": 0.9316,
+      "step": 20067
+    },
+    {
+      "epoch": 0.7298250718260174,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5902880631267387e-06,
+      "loss": 1.0654,
+      "step": 20068
+    },
+    {
+      "epoch": 0.729861439429756,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5893840222738074e-06,
+      "loss": 0.9354,
+      "step": 20069
+    },
+    {
+      "epoch": 0.7298978070334946,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5884800703590818e-06,
+      "loss": 0.9508,
+      "step": 20070
+    },
+    {
+      "epoch": 0.7299341746372332,
+      "grad_norm": 0.0,
+      "learning_rate": 3.587576207395097e-06,
+      "loss": 1.0056,
+      "step": 20071
+    },
+    {
+      "epoch": 0.7299705422409717,
+      "grad_norm": 0.0,
+      "learning_rate": 3.586672433394395e-06,
+      "loss": 0.8156,
+      "step": 20072
+    },
+    {
+      "epoch": 0.7300069098447103,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5857687483695124e-06,
+      "loss": 0.9182,
+      "step": 20073
+    },
+    {
+      "epoch": 0.7300432774484489,
+      "grad_norm": 0.0,
+      "learning_rate": 3.584865152332989e-06,
+      "loss": 0.9117,
+      "step": 20074
+    },
+    {
+      "epoch": 0.7300796450521875,
+      "grad_norm": 0.0,
+      "learning_rate": 3.583961645297356e-06,
+      "loss": 0.9315,
+      "step": 20075
+    },
+    {
+      "epoch": 0.7301160126559261,
+      "grad_norm": 0.0,
+      "learning_rate": 3.583058227275156e-06,
+      "loss": 0.9588,
+      "step": 20076
+    },
+    {
+      "epoch": 0.7301523802596647,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5821548982789176e-06,
+      "loss": 0.9066,
+      "step": 20077
+    },
+    {
+      "epoch": 0.7301887478634033,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5812516583211755e-06,
+      "loss": 0.9687,
+      "step": 20078
+    },
+    {
+      "epoch": 0.7302251154671419,
+      "grad_norm": 0.0,
+      "learning_rate": 3.580348507414455e-06,
+      "loss": 0.9154,
+      "step": 20079
+    },
+    {
+      "epoch": 0.7302614830708805,
+      "grad_norm": 0.0,
+      "learning_rate": 3.579445445571298e-06,
+      "loss": 0.8513,
+      "step": 20080
+    },
+    {
+      "epoch": 0.730297850674619,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5785424728042197e-06,
+      "loss": 0.9366,
+      "step": 20081
+    },
+    {
+      "epoch": 0.7303342182783576,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5776395891257554e-06,
+      "loss": 0.9716,
+      "step": 20082
+    },
+    {
+      "epoch": 0.7303705858820962,
+      "grad_norm": 0.0,
+      "learning_rate": 3.576736794548434e-06,
+      "loss": 0.9852,
+      "step": 20083
+    },
+    {
+      "epoch": 0.7304069534858348,
+      "grad_norm": 0.0,
+      "learning_rate": 3.57583408908477e-06,
+      "loss": 0.8808,
+      "step": 20084
+    },
+    {
+      "epoch": 0.7304433210895734,
+      "grad_norm": 0.0,
+      "learning_rate": 3.574931472747296e-06,
+      "loss": 1.1274,
+      "step": 20085
+    },
+    {
+      "epoch": 0.730479688693312,
+      "grad_norm": 0.0,
+      "learning_rate": 3.574028945548532e-06,
+      "loss": 0.9654,
+      "step": 20086
+    },
+    {
+      "epoch": 0.7305160562970506,
+      "grad_norm": 0.0,
+      "learning_rate": 3.573126507500999e-06,
+      "loss": 0.986,
+      "step": 20087
+    },
+    {
+      "epoch": 0.7305524239007892,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5722241586172147e-06,
+      "loss": 0.8691,
+      "step": 20088
+    },
+    {
+      "epoch": 0.7305887915045277,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5713218989097033e-06,
+      "loss": 1.0139,
+      "step": 20089
+    },
+    {
+      "epoch": 0.7306251591082663,
+      "grad_norm": 0.0,
+      "learning_rate": 3.570419728390978e-06,
+      "loss": 0.9171,
+      "step": 20090
+    },
+    {
+      "epoch": 0.7306615267120049,
+      "grad_norm": 0.0,
+      "learning_rate": 3.569517647073558e-06,
+      "loss": 0.9259,
+      "step": 20091
+    },
+    {
+      "epoch": 0.7306978943157435,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5686156549699534e-06,
+      "loss": 0.8856,
+      "step": 20092
+    },
+    {
+      "epoch": 0.7307342619194821,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5677137520926883e-06,
+      "loss": 0.886,
+      "step": 20093
+    },
+    {
+      "epoch": 0.7307706295232207,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5668119384542623e-06,
+      "loss": 0.8817,
+      "step": 20094
+    },
+    {
+      "epoch": 0.7308069971269593,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5659102140671964e-06,
+      "loss": 0.8839,
+      "step": 20095
+    },
+    {
+      "epoch": 0.7308433647306979,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5650085789439968e-06,
+      "loss": 0.9668,
+      "step": 20096
+    },
+    {
+      "epoch": 0.7308797323344365,
+      "grad_norm": 0.0,
+      "learning_rate": 3.564107033097174e-06,
+      "loss": 0.8378,
+      "step": 20097
+    },
+    {
+      "epoch": 0.7309160999381751,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5632055765392304e-06,
+      "loss": 0.9563,
+      "step": 20098
+    },
+    {
+      "epoch": 0.7309524675419137,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5623042092826842e-06,
+      "loss": 1.0041,
+      "step": 20099
+    },
+    {
+      "epoch": 0.7309888351456523,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5614029313400266e-06,
+      "loss": 1.0229,
+      "step": 20100
+    },
+    {
+      "epoch": 0.7310252027493909,
+      "grad_norm": 0.0,
+      "learning_rate": 3.560501742723772e-06,
+      "loss": 0.9773,
+      "step": 20101
+    },
+    {
+      "epoch": 0.7310615703531295,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5596006434464182e-06,
+      "loss": 0.9337,
+      "step": 20102
+    },
+    {
+      "epoch": 0.7310979379568681,
+      "grad_norm": 0.0,
+      "learning_rate": 3.558699633520468e-06,
+      "loss": 0.9663,
+      "step": 20103
+    },
+    {
+      "epoch": 0.7311343055606067,
+      "grad_norm": 0.0,
+      "learning_rate": 3.557798712958421e-06,
+      "loss": 0.9779,
+      "step": 20104
+    },
+    {
+      "epoch": 0.7311706731643453,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5568978817727783e-06,
+      "loss": 1.0203,
+      "step": 20105
+    },
+    {
+      "epoch": 0.7312070407680837,
+      "grad_norm": 0.0,
+      "learning_rate": 3.555997139976034e-06,
+      "loss": 0.9504,
+      "step": 20106
+    },
+    {
+      "epoch": 0.7312434083718223,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5550964875806836e-06,
+      "loss": 1.0359,
+      "step": 20107
+    },
+    {
+      "epoch": 0.7312797759755609,
+      "grad_norm": 0.0,
+      "learning_rate": 3.55419592459923e-06,
+      "loss": 0.9295,
+      "step": 20108
+    },
+    {
+      "epoch": 0.7313161435792995,
+      "grad_norm": 0.0,
+      "learning_rate": 3.553295451044161e-06,
+      "loss": 1.0365,
+      "step": 20109
+    },
+    {
+      "epoch": 0.7313525111830381,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5523950669279718e-06,
+      "loss": 1.0037,
+      "step": 20110
+    },
+    {
+      "epoch": 0.7313888787867767,
+      "grad_norm": 0.0,
+      "learning_rate": 3.551494772263149e-06,
+      "loss": 1.0656,
+      "step": 20111
+    },
+    {
+      "epoch": 0.7314252463905153,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5505945670621945e-06,
+      "loss": 0.9428,
+      "step": 20112
+    },
+    {
+      "epoch": 0.7314616139942539,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5496944513375833e-06,
+      "loss": 0.9814,
+      "step": 20113
+    },
+    {
+      "epoch": 0.7314979815979925,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5487944251018115e-06,
+      "loss": 0.9651,
+      "step": 20114
+    },
+    {
+      "epoch": 0.7315343492017311,
+      "grad_norm": 0.0,
+      "learning_rate": 3.547894488367365e-06,
+      "loss": 1.0645,
+      "step": 20115
+    },
+    {
+      "epoch": 0.7315707168054697,
+      "grad_norm": 0.0,
+      "learning_rate": 3.546994641146728e-06,
+      "loss": 0.8924,
+      "step": 20116
+    },
+    {
+      "epoch": 0.7316070844092083,
+      "grad_norm": 0.0,
+      "learning_rate": 3.546094883452382e-06,
+      "loss": 1.0279,
+      "step": 20117
+    },
+    {
+      "epoch": 0.7316434520129469,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5451952152968195e-06,
+      "loss": 0.9337,
+      "step": 20118
+    },
+    {
+      "epoch": 0.7316798196166855,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5442956366925074e-06,
+      "loss": 0.9155,
+      "step": 20119
+    },
+    {
+      "epoch": 0.7317161872204241,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5433961476519383e-06,
+      "loss": 0.9373,
+      "step": 20120
+    },
+    {
+      "epoch": 0.7317525548241627,
+      "grad_norm": 0.0,
+      "learning_rate": 3.542496748187586e-06,
+      "loss": 0.9806,
+      "step": 20121
+    },
+    {
+      "epoch": 0.7317889224279013,
+      "grad_norm": 0.0,
+      "learning_rate": 3.54159743831193e-06,
+      "loss": 1.0125,
+      "step": 20122
+    },
+    {
+      "epoch": 0.7318252900316398,
+      "grad_norm": 0.0,
+      "learning_rate": 3.540698218037446e-06,
+      "loss": 0.8862,
+      "step": 20123
+    },
+    {
+      "epoch": 0.7318616576353784,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5397990873766063e-06,
+      "loss": 0.8843,
+      "step": 20124
+    },
+    {
+      "epoch": 0.731898025239117,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5389000463418956e-06,
+      "loss": 0.8849,
+      "step": 20125
+    },
+    {
+      "epoch": 0.7319343928428556,
+      "grad_norm": 0.0,
+      "learning_rate": 3.538001094945772e-06,
+      "loss": 0.9623,
+      "step": 20126
+    },
+    {
+      "epoch": 0.7319707604465941,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5371022332007188e-06,
+      "loss": 0.8884,
+      "step": 20127
+    },
+    {
+      "epoch": 0.7320071280503327,
+      "grad_norm": 0.0,
+      "learning_rate": 3.536203461119202e-06,
+      "loss": 1.0224,
+      "step": 20128
+    },
+    {
+      "epoch": 0.7320434956540713,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5353047787136918e-06,
+      "loss": 1.0092,
+      "step": 20129
+    },
+    {
+      "epoch": 0.7320798632578099,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5344061859966527e-06,
+      "loss": 0.8718,
+      "step": 20130
+    },
+    {
+      "epoch": 0.7321162308615485,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5335076829805603e-06,
+      "loss": 0.8715,
+      "step": 20131
+    },
+    {
+      "epoch": 0.7321525984652871,
+      "grad_norm": 0.0,
+      "learning_rate": 3.532609269677867e-06,
+      "loss": 0.9599,
+      "step": 20132
+    },
+    {
+      "epoch": 0.7321889660690257,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5317109461010477e-06,
+      "loss": 0.9918,
+      "step": 20133
+    },
+    {
+      "epoch": 0.7322253336727643,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5308127122625625e-06,
+      "loss": 0.8778,
+      "step": 20134
+    },
+    {
+      "epoch": 0.7322617012765029,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5299145681748714e-06,
+      "loss": 0.8143,
+      "step": 20135
+    },
+    {
+      "epoch": 0.7322980688802415,
+      "grad_norm": 0.0,
+      "learning_rate": 3.529016513850433e-06,
+      "loss": 0.9589,
+      "step": 20136
+    },
+    {
+      "epoch": 0.7323344364839801,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5281185493017157e-06,
+      "loss": 0.8752,
+      "step": 20137
+    },
+    {
+      "epoch": 0.7323708040877187,
+      "grad_norm": 0.0,
+      "learning_rate": 3.527220674541164e-06,
+      "loss": 0.9805,
+      "step": 20138
+    },
+    {
+      "epoch": 0.7324071716914573,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5263228895812462e-06,
+      "loss": 1.0255,
+      "step": 20139
+    },
+    {
+      "epoch": 0.7324435392951958,
+      "grad_norm": 0.0,
+      "learning_rate": 3.525425194434412e-06,
+      "loss": 0.8378,
+      "step": 20140
+    },
+    {
+      "epoch": 0.7324799068989344,
+      "grad_norm": 0.0,
+      "learning_rate": 3.524527589113118e-06,
+      "loss": 0.9755,
+      "step": 20141
+    },
+    {
+      "epoch": 0.732516274502673,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5236300736298157e-06,
+      "loss": 0.9166,
+      "step": 20142
+    },
+    {
+      "epoch": 0.7325526421064116,
+      "grad_norm": 0.0,
+      "learning_rate": 3.522732647996955e-06,
+      "loss": 0.9507,
+      "step": 20143
+    },
+    {
+      "epoch": 0.7325890097101502,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5218353122269943e-06,
+      "loss": 0.9251,
+      "step": 20144
+    },
+    {
+      "epoch": 0.7326253773138888,
+      "grad_norm": 0.0,
+      "learning_rate": 3.520938066332371e-06,
+      "loss": 0.9865,
+      "step": 20145
+    },
+    {
+      "epoch": 0.7326617449176274,
+      "grad_norm": 0.0,
+      "learning_rate": 3.520040910325543e-06,
+      "loss": 0.8908,
+      "step": 20146
+    },
+    {
+      "epoch": 0.732698112521366,
+      "grad_norm": 0.0,
+      "learning_rate": 3.519143844218954e-06,
+      "loss": 0.9394,
+      "step": 20147
+    },
+    {
+      "epoch": 0.7327344801251046,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5182468680250483e-06,
+      "loss": 0.9439,
+      "step": 20148
+    },
+    {
+      "epoch": 0.7327708477288432,
+      "grad_norm": 0.0,
+      "learning_rate": 3.517349981756267e-06,
+      "loss": 0.9712,
+      "step": 20149
+    },
+    {
+      "epoch": 0.7328072153325818,
+      "grad_norm": 0.0,
+      "learning_rate": 3.516453185425065e-06,
+      "loss": 1.0008,
+      "step": 20150
+    },
+    {
+      "epoch": 0.7328435829363203,
+      "grad_norm": 0.0,
+      "learning_rate": 3.515556479043869e-06,
+      "loss": 0.8941,
+      "step": 20151
+    },
+    {
+      "epoch": 0.7328799505400589,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5146598626251292e-06,
+      "loss": 0.9546,
+      "step": 20152
+    },
+    {
+      "epoch": 0.7329163181437975,
+      "grad_norm": 0.0,
+      "learning_rate": 3.513763336181283e-06,
+      "loss": 1.0539,
+      "step": 20153
+    },
+    {
+      "epoch": 0.7329526857475361,
+      "grad_norm": 0.0,
+      "learning_rate": 3.512866899724768e-06,
+      "loss": 0.8813,
+      "step": 20154
+    },
+    {
+      "epoch": 0.7329890533512747,
+      "grad_norm": 0.0,
+      "learning_rate": 3.511970553268017e-06,
+      "loss": 0.8727,
+      "step": 20155
+    },
+    {
+      "epoch": 0.7330254209550133,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5110742968234724e-06,
+      "loss": 0.9885,
+      "step": 20156
+    },
+    {
+      "epoch": 0.7330617885587518,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5101781304035654e-06,
+      "loss": 0.9897,
+      "step": 20157
+    },
+    {
+      "epoch": 0.7330981561624904,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5092820540207294e-06,
+      "loss": 1.0163,
+      "step": 20158
+    },
+    {
+      "epoch": 0.733134523766229,
+      "grad_norm": 0.0,
+      "learning_rate": 3.508386067687396e-06,
+      "loss": 0.9097,
+      "step": 20159
+    },
+    {
+      "epoch": 0.7331708913699676,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5074901714159947e-06,
+      "loss": 0.8385,
+      "step": 20160
+    },
+    {
+      "epoch": 0.7332072589737062,
+      "grad_norm": 0.0,
+      "learning_rate": 3.506594365218956e-06,
+      "loss": 0.8552,
+      "step": 20161
+    },
+    {
+      "epoch": 0.7332436265774448,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5056986491087042e-06,
+      "loss": 0.9446,
+      "step": 20162
+    },
+    {
+      "epoch": 0.7332799941811834,
+      "grad_norm": 0.0,
+      "learning_rate": 3.504803023097676e-06,
+      "loss": 0.891,
+      "step": 20163
+    },
+    {
+      "epoch": 0.733316361784922,
+      "grad_norm": 0.0,
+      "learning_rate": 3.503907487198284e-06,
+      "loss": 0.9291,
+      "step": 20164
+    },
+    {
+      "epoch": 0.7333527293886606,
+      "grad_norm": 0.0,
+      "learning_rate": 3.503012041422962e-06,
+      "loss": 0.9035,
+      "step": 20165
+    },
+    {
+      "epoch": 0.7333890969923992,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5021166857841303e-06,
+      "loss": 0.9893,
+      "step": 20166
+    },
+    {
+      "epoch": 0.7334254645961378,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5012214202942095e-06,
+      "loss": 1.0108,
+      "step": 20167
+    },
+    {
+      "epoch": 0.7334618321998764,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5003262449656185e-06,
+      "loss": 0.9147,
+      "step": 20168
+    },
+    {
+      "epoch": 0.733498199803615,
+      "grad_norm": 0.0,
+      "learning_rate": 3.499431159810781e-06,
+      "loss": 0.9798,
+      "step": 20169
+    },
+    {
+      "epoch": 0.7335345674073536,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4985361648421146e-06,
+      "loss": 0.8895,
+      "step": 20170
+    },
+    {
+      "epoch": 0.7335709350110922,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4976412600720333e-06,
+      "loss": 0.8611,
+      "step": 20171
+    },
+    {
+      "epoch": 0.7336073026148308,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4967464455129553e-06,
+      "loss": 1.0219,
+      "step": 20172
+    },
+    {
+      "epoch": 0.7336436702185692,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4958517211772924e-06,
+      "loss": 0.7736,
+      "step": 20173
+    },
+    {
+      "epoch": 0.7336800378223078,
+      "grad_norm": 0.0,
+      "learning_rate": 3.494957087077455e-06,
+      "loss": 0.9138,
+      "step": 20174
+    },
+    {
+      "epoch": 0.7337164054260464,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4940625432258625e-06,
+      "loss": 0.8872,
+      "step": 20175
+    },
+    {
+      "epoch": 0.733752773029785,
+      "grad_norm": 0.0,
+      "learning_rate": 3.493168089634922e-06,
+      "loss": 0.7296,
+      "step": 20176
+    },
+    {
+      "epoch": 0.7337891406335236,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4922737263170412e-06,
+      "loss": 0.7427,
+      "step": 20177
+    },
+    {
+      "epoch": 0.7338255082372622,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4913794532846302e-06,
+      "loss": 0.8646,
+      "step": 20178
+    },
+    {
+      "epoch": 0.7338618758410008,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4904852705500937e-06,
+      "loss": 1.0064,
+      "step": 20179
+    },
+    {
+      "epoch": 0.7338982434447394,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4895911781258386e-06,
+      "loss": 0.9048,
+      "step": 20180
+    },
+    {
+      "epoch": 0.733934611048478,
+      "grad_norm": 0.0,
+      "learning_rate": 3.488697176024266e-06,
+      "loss": 0.9835,
+      "step": 20181
+    },
+    {
+      "epoch": 0.7339709786522166,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4878032642577884e-06,
+      "loss": 0.885,
+      "step": 20182
+    },
+    {
+      "epoch": 0.7340073462559552,
+      "grad_norm": 0.0,
+      "learning_rate": 3.486909442838794e-06,
+      "loss": 0.858,
+      "step": 20183
+    },
+    {
+      "epoch": 0.7340437138596938,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4860157117796935e-06,
+      "loss": 0.8406,
+      "step": 20184
+    },
+    {
+      "epoch": 0.7340800814634324,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4851220710928835e-06,
+      "loss": 0.971,
+      "step": 20185
+    },
+    {
+      "epoch": 0.734116449067171,
+      "grad_norm": 0.0,
+      "learning_rate": 3.48422852079076e-06,
+      "loss": 0.8631,
+      "step": 20186
+    },
+    {
+      "epoch": 0.7341528166709096,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4833350608857185e-06,
+      "loss": 1.0228,
+      "step": 20187
+    },
+    {
+      "epoch": 0.7341891842746482,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4824416913901604e-06,
+      "loss": 0.8871,
+      "step": 20188
+    },
+    {
+      "epoch": 0.7342255518783868,
+      "grad_norm": 0.0,
+      "learning_rate": 3.481548412316477e-06,
+      "loss": 0.8381,
+      "step": 20189
+    },
+    {
+      "epoch": 0.7342619194821253,
+      "grad_norm": 0.0,
+      "learning_rate": 3.480655223677061e-06,
+      "loss": 0.8671,
+      "step": 20190
+    },
+    {
+      "epoch": 0.7342982870858639,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4797621254842994e-06,
+      "loss": 0.9727,
+      "step": 20191
+    },
+    {
+      "epoch": 0.7343346546896025,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4788691177505952e-06,
+      "loss": 0.9131,
+      "step": 20192
+    },
+    {
+      "epoch": 0.7343710222933411,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4779762004883234e-06,
+      "loss": 0.9727,
+      "step": 20193
+    },
+    {
+      "epoch": 0.7344073898970797,
+      "grad_norm": 0.0,
+      "learning_rate": 3.477083373709881e-06,
+      "loss": 0.8561,
+      "step": 20194
+    },
+    {
+      "epoch": 0.7344437575008183,
+      "grad_norm": 0.0,
+      "learning_rate": 3.476190637427651e-06,
+      "loss": 0.9845,
+      "step": 20195
+    },
+    {
+      "epoch": 0.7344801251045568,
+      "grad_norm": 0.0,
+      "learning_rate": 3.475297991654021e-06,
+      "loss": 0.9502,
+      "step": 20196
+    },
+    {
+      "epoch": 0.7345164927082954,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4744054364013737e-06,
+      "loss": 1.0881,
+      "step": 20197
+    },
+    {
+      "epoch": 0.734552860312034,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4735129716820923e-06,
+      "loss": 1.0304,
+      "step": 20198
+    },
+    {
+      "epoch": 0.7345892279157726,
+      "grad_norm": 0.0,
+      "learning_rate": 3.472620597508559e-06,
+      "loss": 0.942,
+      "step": 20199
+    },
+    {
+      "epoch": 0.7346255955195112,
+      "grad_norm": 0.0,
+      "learning_rate": 3.47172831389315e-06,
+      "loss": 0.9389,
+      "step": 20200
+    },
+    {
+      "epoch": 0.7346619631232498,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4708361208482523e-06,
+      "loss": 0.9905,
+      "step": 20201
+    },
+    {
+      "epoch": 0.7346983307269884,
+      "grad_norm": 0.0,
+      "learning_rate": 3.469944018386239e-06,
+      "loss": 0.9048,
+      "step": 20202
+    },
+    {
+      "epoch": 0.734734698330727,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4690520065194876e-06,
+      "loss": 0.9431,
+      "step": 20203
+    },
+    {
+      "epoch": 0.7347710659344656,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4681600852603737e-06,
+      "loss": 1.0056,
+      "step": 20204
+    },
+    {
+      "epoch": 0.7348074335382042,
+      "grad_norm": 0.0,
+      "learning_rate": 3.46726825462127e-06,
+      "loss": 1.149,
+      "step": 20205
+    },
+    {
+      "epoch": 0.7348438011419428,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4663765146145488e-06,
+      "loss": 0.9959,
+      "step": 20206
+    },
+    {
+      "epoch": 0.7348801687456813,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4654848652525862e-06,
+      "loss": 0.8935,
+      "step": 20207
+    },
+    {
+      "epoch": 0.7349165363494199,
+      "grad_norm": 0.0,
+      "learning_rate": 3.464593306547749e-06,
+      "loss": 0.9501,
+      "step": 20208
+    },
+    {
+      "epoch": 0.7349529039531585,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4637018385124076e-06,
+      "loss": 0.8655,
+      "step": 20209
+    },
+    {
+      "epoch": 0.7349892715568971,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4628104611589265e-06,
+      "loss": 0.8942,
+      "step": 20210
+    },
+    {
+      "epoch": 0.7350256391606357,
+      "grad_norm": 0.0,
+      "learning_rate": 3.461919174499682e-06,
+      "loss": 0.856,
+      "step": 20211
+    },
+    {
+      "epoch": 0.7350620067643743,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4610279785470258e-06,
+      "loss": 0.9785,
+      "step": 20212
+    },
+    {
+      "epoch": 0.7350983743681129,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4601368733133323e-06,
+      "loss": 0.9112,
+      "step": 20213
+    },
+    {
+      "epoch": 0.7351347419718515,
+      "grad_norm": 0.0,
+      "learning_rate": 3.45924585881096e-06,
+      "loss": 0.8538,
+      "step": 20214
+    },
+    {
+      "epoch": 0.7351711095755901,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4583549350522716e-06,
+      "loss": 0.9758,
+      "step": 20215
+    },
+    {
+      "epoch": 0.7352074771793287,
+      "grad_norm": 0.0,
+      "learning_rate": 3.457464102049628e-06,
+      "loss": 0.8486,
+      "step": 20216
+    },
+    {
+      "epoch": 0.7352438447830673,
+      "grad_norm": 0.0,
+      "learning_rate": 3.456573359815386e-06,
+      "loss": 0.9696,
+      "step": 20217
+    },
+    {
+      "epoch": 0.7352802123868059,
+      "grad_norm": 0.0,
+      "learning_rate": 3.455682708361906e-06,
+      "loss": 0.9583,
+      "step": 20218
+    },
+    {
+      "epoch": 0.7353165799905444,
+      "grad_norm": 0.0,
+      "learning_rate": 3.454792147701539e-06,
+      "loss": 1.0126,
+      "step": 20219
+    },
+    {
+      "epoch": 0.735352947594283,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4539016778466485e-06,
+      "loss": 1.0044,
+      "step": 20220
+    },
+    {
+      "epoch": 0.7353893151980216,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4530112988095843e-06,
+      "loss": 0.8198,
+      "step": 20221
+    },
+    {
+      "epoch": 0.7354256828017602,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4521210106026992e-06,
+      "loss": 0.9449,
+      "step": 20222
+    },
+    {
+      "epoch": 0.7354620504054988,
+      "grad_norm": 0.0,
+      "learning_rate": 3.451230813238341e-06,
+      "loss": 0.9184,
+      "step": 20223
+    },
+    {
+      "epoch": 0.7354984180092373,
+      "grad_norm": 0.0,
+      "learning_rate": 3.450340706728871e-06,
+      "loss": 1.0275,
+      "step": 20224
+    },
+    {
+      "epoch": 0.7355347856129759,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4494506910866245e-06,
+      "loss": 0.8835,
+      "step": 20225
+    },
+    {
+      "epoch": 0.7355711532167145,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4485607663239584e-06,
+      "loss": 0.8974,
+      "step": 20226
+    },
+    {
+      "epoch": 0.7356075208204531,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4476709324532165e-06,
+      "loss": 0.9035,
+      "step": 20227
+    },
+    {
+      "epoch": 0.7356438884241917,
+      "grad_norm": 0.0,
+      "learning_rate": 3.446781189486744e-06,
+      "loss": 0.9385,
+      "step": 20228
+    },
+    {
+      "epoch": 0.7356802560279303,
+      "grad_norm": 0.0,
+      "learning_rate": 3.445891537436882e-06,
+      "loss": 0.949,
+      "step": 20229
+    },
+    {
+      "epoch": 0.7357166236316689,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4450019763159827e-06,
+      "loss": 1.0337,
+      "step": 20230
+    },
+    {
+      "epoch": 0.7357529912354075,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4441125061363723e-06,
+      "loss": 0.8953,
+      "step": 20231
+    },
+    {
+      "epoch": 0.7357893588391461,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4432231269104043e-06,
+      "loss": 0.857,
+      "step": 20232
+    },
+    {
+      "epoch": 0.7358257264428847,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4423338386504114e-06,
+      "loss": 0.8848,
+      "step": 20233
+    },
+    {
+      "epoch": 0.7358620940466233,
+      "grad_norm": 0.0,
+      "learning_rate": 3.441444641368733e-06,
+      "loss": 0.8409,
+      "step": 20234
+    },
+    {
+      "epoch": 0.7358984616503619,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4405555350777008e-06,
+      "loss": 0.9237,
+      "step": 20235
+    },
+    {
+      "epoch": 0.7359348292541005,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4396665197896584e-06,
+      "loss": 0.8327,
+      "step": 20236
+    },
+    {
+      "epoch": 0.7359711968578391,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4387775955169376e-06,
+      "loss": 1.0545,
+      "step": 20237
+    },
+    {
+      "epoch": 0.7360075644615777,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4378887622718627e-06,
+      "loss": 0.8853,
+      "step": 20238
+    },
+    {
+      "epoch": 0.7360439320653163,
+      "grad_norm": 0.0,
+      "learning_rate": 3.437000020066773e-06,
+      "loss": 0.9638,
+      "step": 20239
+    },
+    {
+      "epoch": 0.7360802996690549,
+      "grad_norm": 0.0,
+      "learning_rate": 3.436111368913997e-06,
+      "loss": 0.9891,
+      "step": 20240
+    },
+    {
+      "epoch": 0.7361166672727933,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4352228088258634e-06,
+      "loss": 0.8127,
+      "step": 20241
+    },
+    {
+      "epoch": 0.7361530348765319,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4343343398146945e-06,
+      "loss": 0.9648,
+      "step": 20242
+    },
+    {
+      "epoch": 0.7361894024802705,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4334459618928284e-06,
+      "loss": 0.9404,
+      "step": 20243
+    },
+    {
+      "epoch": 0.7362257700840091,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4325576750725766e-06,
+      "loss": 0.9595,
+      "step": 20244
+    },
+    {
+      "epoch": 0.7362621376877477,
+      "grad_norm": 0.0,
+      "learning_rate": 3.431669479366272e-06,
+      "loss": 1.0091,
+      "step": 20245
+    },
+    {
+      "epoch": 0.7362985052914863,
+      "grad_norm": 0.0,
+      "learning_rate": 3.430781374786234e-06,
+      "loss": 0.9539,
+      "step": 20246
+    },
+    {
+      "epoch": 0.7363348728952249,
+      "grad_norm": 0.0,
+      "learning_rate": 3.429893361344784e-06,
+      "loss": 1.0338,
+      "step": 20247
+    },
+    {
+      "epoch": 0.7363712404989635,
+      "grad_norm": 0.0,
+      "learning_rate": 3.429005439054238e-06,
+      "loss": 1.0328,
+      "step": 20248
+    },
+    {
+      "epoch": 0.7364076081027021,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4281176079269262e-06,
+      "loss": 0.8983,
+      "step": 20249
+    },
+    {
+      "epoch": 0.7364439757064407,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4272298679751514e-06,
+      "loss": 0.9158,
+      "step": 20250
+    },
+    {
+      "epoch": 0.7364803433101793,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4263422192112395e-06,
+      "loss": 0.8875,
+      "step": 20251
+    },
+    {
+      "epoch": 0.7365167109139179,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4254546616475027e-06,
+      "loss": 1.0415,
+      "step": 20252
+    },
+    {
+      "epoch": 0.7365530785176565,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4245671952962543e-06,
+      "loss": 0.9012,
+      "step": 20253
+    },
+    {
+      "epoch": 0.7365894461213951,
+      "grad_norm": 0.0,
+      "learning_rate": 3.423679820169804e-06,
+      "loss": 1.0083,
+      "step": 20254
+    },
+    {
+      "epoch": 0.7366258137251337,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4227925362804692e-06,
+      "loss": 0.8926,
+      "step": 20255
+    },
+    {
+      "epoch": 0.7366621813288723,
+      "grad_norm": 0.0,
+      "learning_rate": 3.421905343640558e-06,
+      "loss": 0.9503,
+      "step": 20256
+    },
+    {
+      "epoch": 0.7366985489326108,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4210182422623717e-06,
+      "loss": 0.9825,
+      "step": 20257
+    },
+    {
+      "epoch": 0.7367349165363494,
+      "grad_norm": 0.0,
+      "learning_rate": 3.420131232158226e-06,
+      "loss": 0.9548,
+      "step": 20258
+    },
+    {
+      "epoch": 0.736771284140088,
+      "grad_norm": 0.0,
+      "learning_rate": 3.419244313340424e-06,
+      "loss": 0.9488,
+      "step": 20259
+    },
+    {
+      "epoch": 0.7368076517438266,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4183574858212696e-06,
+      "loss": 1.02,
+      "step": 20260
+    },
+    {
+      "epoch": 0.7368440193475652,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4174707496130634e-06,
+      "loss": 0.9384,
+      "step": 20261
+    },
+    {
+      "epoch": 0.7368803869513038,
+      "grad_norm": 0.0,
+      "learning_rate": 3.416584104728119e-06,
+      "loss": 0.8592,
+      "step": 20262
+    },
+    {
+      "epoch": 0.7369167545550424,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4156975511787217e-06,
+      "loss": 0.8986,
+      "step": 20263
+    },
+    {
+      "epoch": 0.736953122158781,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4148110889771837e-06,
+      "loss": 0.9089,
+      "step": 20264
+    },
+    {
+      "epoch": 0.7369894897625195,
+      "grad_norm": 0.0,
+      "learning_rate": 3.413924718135797e-06,
+      "loss": 0.8737,
+      "step": 20265
+    },
+    {
+      "epoch": 0.7370258573662581,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4130384386668615e-06,
+      "loss": 0.999,
+      "step": 20266
+    },
+    {
+      "epoch": 0.7370622249699967,
+      "grad_norm": 0.0,
+      "learning_rate": 3.412152250582669e-06,
+      "loss": 0.8635,
+      "step": 20267
+    },
+    {
+      "epoch": 0.7370985925737353,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4112661538955194e-06,
+      "loss": 0.8923,
+      "step": 20268
+    },
+    {
+      "epoch": 0.7371349601774739,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4103801486177034e-06,
+      "loss": 1.0262,
+      "step": 20269
+    },
+    {
+      "epoch": 0.7371713277812125,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4094942347615146e-06,
+      "loss": 0.9948,
+      "step": 20270
+    },
+    {
+      "epoch": 0.7372076953849511,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4086084123392417e-06,
+      "loss": 0.9069,
+      "step": 20271
+    },
+    {
+      "epoch": 0.7372440629886897,
+      "grad_norm": 0.0,
+      "learning_rate": 3.407722681363176e-06,
+      "loss": 0.877,
+      "step": 20272
+    },
+    {
+      "epoch": 0.7372804305924283,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4068370418456008e-06,
+      "loss": 0.8383,
+      "step": 20273
+    },
+    {
+      "epoch": 0.7373167981961668,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4059514937988103e-06,
+      "loss": 0.9122,
+      "step": 20274
+    },
+    {
+      "epoch": 0.7373531657999054,
+      "grad_norm": 0.0,
+      "learning_rate": 3.405066037235092e-06,
+      "loss": 0.8737,
+      "step": 20275
+    },
+    {
+      "epoch": 0.737389533403644,
+      "grad_norm": 0.0,
+      "learning_rate": 3.404180672166717e-06,
+      "loss": 0.8743,
+      "step": 20276
+    },
+    {
+      "epoch": 0.7374259010073826,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4032953986059815e-06,
+      "loss": 0.8736,
+      "step": 20277
+    },
+    {
+      "epoch": 0.7374622686111212,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4024102165651616e-06,
+      "loss": 0.9465,
+      "step": 20278
+    },
+    {
+      "epoch": 0.7374986362148598,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4015251260565397e-06,
+      "loss": 0.873,
+      "step": 20279
+    },
+    {
+      "epoch": 0.7375350038185984,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4006401270923915e-06,
+      "loss": 1.0027,
+      "step": 20280
+    },
+    {
+      "epoch": 0.737571371422337,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3997552196850048e-06,
+      "loss": 0.9553,
+      "step": 20281
+    },
+    {
+      "epoch": 0.7376077390260756,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3988704038466434e-06,
+      "loss": 0.8781,
+      "step": 20282
+    },
+    {
+      "epoch": 0.7376441066298142,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3979856795895927e-06,
+      "loss": 0.8759,
+      "step": 20283
+    },
+    {
+      "epoch": 0.7376804742335528,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3971010469261245e-06,
+      "loss": 1.0017,
+      "step": 20284
+    },
+    {
+      "epoch": 0.7377168418372914,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3962165058685093e-06,
+      "loss": 0.8517,
+      "step": 20285
+    },
+    {
+      "epoch": 0.73775320944103,
+      "grad_norm": 0.0,
+      "learning_rate": 3.395332056429018e-06,
+      "loss": 1.0409,
+      "step": 20286
+    },
+    {
+      "epoch": 0.7377895770447686,
+      "grad_norm": 0.0,
+      "learning_rate": 3.394447698619927e-06,
+      "loss": 0.8628,
+      "step": 20287
+    },
+    {
+      "epoch": 0.7378259446485071,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3935634324535026e-06,
+      "loss": 0.9852,
+      "step": 20288
+    },
+    {
+      "epoch": 0.7378623122522457,
+      "grad_norm": 0.0,
+      "learning_rate": 3.392679257942012e-06,
+      "loss": 0.9494,
+      "step": 20289
+    },
+    {
+      "epoch": 0.7378986798559843,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3917951750977186e-06,
+      "loss": 0.9108,
+      "step": 20290
+    },
+    {
+      "epoch": 0.7379350474597228,
+      "grad_norm": 0.0,
+      "learning_rate": 3.390911183932898e-06,
+      "loss": 0.8867,
+      "step": 20291
+    },
+    {
+      "epoch": 0.7379714150634614,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3900272844598004e-06,
+      "loss": 0.8629,
+      "step": 20292
+    },
+    {
+      "epoch": 0.7380077826672,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3891434766906995e-06,
+      "loss": 0.797,
+      "step": 20293
+    },
+    {
+      "epoch": 0.7380441502709386,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3882597606378577e-06,
+      "loss": 1.0095,
+      "step": 20294
+    },
+    {
+      "epoch": 0.7380805178746772,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3873761363135227e-06,
+      "loss": 0.8748,
+      "step": 20295
+    },
+    {
+      "epoch": 0.7381168854784158,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3864926037299663e-06,
+      "loss": 0.8297,
+      "step": 20296
+    },
+    {
+      "epoch": 0.7381532530821544,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3856091628994412e-06,
+      "loss": 0.9545,
+      "step": 20297
+    },
+    {
+      "epoch": 0.738189620685893,
+      "grad_norm": 0.0,
+      "learning_rate": 3.384725813834204e-06,
+      "loss": 0.7924,
+      "step": 20298
+    },
+    {
+      "epoch": 0.7382259882896316,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3838425565465073e-06,
+      "loss": 0.9752,
+      "step": 20299
+    },
+    {
+      "epoch": 0.7382623558933702,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3829593910486115e-06,
+      "loss": 0.8476,
+      "step": 20300
+    },
+    {
+      "epoch": 0.7382987234971088,
+      "grad_norm": 0.0,
+      "learning_rate": 3.382076317352766e-06,
+      "loss": 1.045,
+      "step": 20301
+    },
+    {
+      "epoch": 0.7383350911008474,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3811933354712214e-06,
+      "loss": 0.9485,
+      "step": 20302
+    },
+    {
+      "epoch": 0.738371458704586,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3803104454162283e-06,
+      "loss": 0.8477,
+      "step": 20303
+    },
+    {
+      "epoch": 0.7384078263083246,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3794276472000364e-06,
+      "loss": 0.9316,
+      "step": 20304
+    },
+    {
+      "epoch": 0.7384441939120632,
+      "grad_norm": 0.0,
+      "learning_rate": 3.378544940834888e-06,
+      "loss": 0.9812,
+      "step": 20305
+    },
+    {
+      "epoch": 0.7384805615158018,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3776623263330387e-06,
+      "loss": 0.9757,
+      "step": 20306
+    },
+    {
+      "epoch": 0.7385169291195404,
+      "grad_norm": 0.0,
+      "learning_rate": 3.376779803706728e-06,
+      "loss": 1.0339,
+      "step": 20307
+    },
+    {
+      "epoch": 0.7385532967232789,
+      "grad_norm": 0.0,
+      "learning_rate": 3.375897372968201e-06,
+      "loss": 0.9323,
+      "step": 20308
+    },
+    {
+      "epoch": 0.7385896643270174,
+      "grad_norm": 0.0,
+      "learning_rate": 3.375015034129695e-06,
+      "loss": 0.893,
+      "step": 20309
+    },
+    {
+      "epoch": 0.738626031930756,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3741327872034633e-06,
+      "loss": 0.81,
+      "step": 20310
+    },
+    {
+      "epoch": 0.7386623995344946,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3732506322017324e-06,
+      "loss": 0.8078,
+      "step": 20311
+    },
+    {
+      "epoch": 0.7386987671382332,
+      "grad_norm": 0.0,
+      "learning_rate": 3.372368569136749e-06,
+      "loss": 1.0138,
+      "step": 20312
+    },
+    {
+      "epoch": 0.7387351347419718,
+      "grad_norm": 0.0,
+      "learning_rate": 3.371486598020749e-06,
+      "loss": 0.9359,
+      "step": 20313
+    },
+    {
+      "epoch": 0.7387715023457104,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3706047188659675e-06,
+      "loss": 0.8534,
+      "step": 20314
+    },
+    {
+      "epoch": 0.738807869949449,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3697229316846393e-06,
+      "loss": 0.9266,
+      "step": 20315
+    },
+    {
+      "epoch": 0.7388442375531876,
+      "grad_norm": 0.0,
+      "learning_rate": 3.368841236488999e-06,
+      "loss": 0.9536,
+      "step": 20316
+    },
+    {
+      "epoch": 0.7388806051569262,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3679596332912766e-06,
+      "loss": 0.9129,
+      "step": 20317
+    },
+    {
+      "epoch": 0.7389169727606648,
+      "grad_norm": 0.0,
+      "learning_rate": 3.367078122103703e-06,
+      "loss": 1.0004,
+      "step": 20318
+    },
+    {
+      "epoch": 0.7389533403644034,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3661967029385122e-06,
+      "loss": 0.8712,
+      "step": 20319
+    },
+    {
+      "epoch": 0.738989707968142,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3653153758079303e-06,
+      "loss": 0.9451,
+      "step": 20320
+    },
+    {
+      "epoch": 0.7390260755718806,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3644341407241845e-06,
+      "loss": 1.0125,
+      "step": 20321
+    },
+    {
+      "epoch": 0.7390624431756192,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3635529976994963e-06,
+      "loss": 0.8702,
+      "step": 20322
+    },
+    {
+      "epoch": 0.7390988107793578,
+      "grad_norm": 0.0,
+      "learning_rate": 3.362671946746101e-06,
+      "loss": 0.9202,
+      "step": 20323
+    },
+    {
+      "epoch": 0.7391351783830964,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3617909878762087e-06,
+      "loss": 1.0073,
+      "step": 20324
+    },
+    {
+      "epoch": 0.7391715459868349,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3609101211020523e-06,
+      "loss": 0.8256,
+      "step": 20325
+    },
+    {
+      "epoch": 0.7392079135905735,
+      "grad_norm": 0.0,
+      "learning_rate": 3.360029346435848e-06,
+      "loss": 0.8954,
+      "step": 20326
+    },
+    {
+      "epoch": 0.7392442811943121,
+      "grad_norm": 0.0,
+      "learning_rate": 3.359148663889815e-06,
+      "loss": 0.9674,
+      "step": 20327
+    },
+    {
+      "epoch": 0.7392806487980507,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3582680734761695e-06,
+      "loss": 0.9365,
+      "step": 20328
+    },
+    {
+      "epoch": 0.7393170164017893,
+      "grad_norm": 0.0,
+      "learning_rate": 3.357387575207137e-06,
+      "loss": 0.9243,
+      "step": 20329
+    },
+    {
+      "epoch": 0.7393533840055279,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3565071690949215e-06,
+      "loss": 0.922,
+      "step": 20330
+    },
+    {
+      "epoch": 0.7393897516092665,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3556268551517458e-06,
+      "loss": 0.9459,
+      "step": 20331
+    },
+    {
+      "epoch": 0.739426119213005,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3547466333898206e-06,
+      "loss": 0.9729,
+      "step": 20332
+    },
+    {
+      "epoch": 0.7394624868167436,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3538665038213568e-06,
+      "loss": 0.946,
+      "step": 20333
+    },
+    {
+      "epoch": 0.7394988544204822,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3529864664585664e-06,
+      "loss": 0.9744,
+      "step": 20334
+    },
+    {
+      "epoch": 0.7395352220242208,
+      "grad_norm": 0.0,
+      "learning_rate": 3.352106521313655e-06,
+      "loss": 0.854,
+      "step": 20335
+    },
+    {
+      "epoch": 0.7395715896279594,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3512266683988394e-06,
+      "loss": 0.792,
+      "step": 20336
+    },
+    {
+      "epoch": 0.739607957231698,
+      "grad_norm": 0.0,
+      "learning_rate": 3.350346907726314e-06,
+      "loss": 1.0033,
+      "step": 20337
+    },
+    {
+      "epoch": 0.7396443248354366,
+      "grad_norm": 0.0,
+      "learning_rate": 3.349467239308294e-06,
+      "loss": 0.9052,
+      "step": 20338
+    },
+    {
+      "epoch": 0.7396806924391752,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3485876631569804e-06,
+      "loss": 1.0093,
+      "step": 20339
+    },
+    {
+      "epoch": 0.7397170600429138,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3477081792845746e-06,
+      "loss": 0.9511,
+      "step": 20340
+    },
+    {
+      "epoch": 0.7397534276466524,
+      "grad_norm": 0.0,
+      "learning_rate": 3.346828787703277e-06,
+      "loss": 0.9393,
+      "step": 20341
+    },
+    {
+      "epoch": 0.7397897952503909,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3459494884252973e-06,
+      "loss": 0.9154,
+      "step": 20342
+    },
+    {
+      "epoch": 0.7398261628541295,
+      "grad_norm": 0.0,
+      "learning_rate": 3.345070281462821e-06,
+      "loss": 0.9464,
+      "step": 20343
+    },
+    {
+      "epoch": 0.7398625304578681,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3441911668280548e-06,
+      "loss": 0.9529,
+      "step": 20344
+    },
+    {
+      "epoch": 0.7398988980616067,
+      "grad_norm": 0.0,
+      "learning_rate": 3.343312144533194e-06,
+      "loss": 0.9924,
+      "step": 20345
+    },
+    {
+      "epoch": 0.7399352656653453,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3424332145904315e-06,
+      "loss": 0.9445,
+      "step": 20346
+    },
+    {
+      "epoch": 0.7399716332690839,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3415543770119596e-06,
+      "loss": 0.9515,
+      "step": 20347
+    },
+    {
+      "epoch": 0.7400080008728225,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3406756318099798e-06,
+      "loss": 1.0011,
+      "step": 20348
+    },
+    {
+      "epoch": 0.7400443684765611,
+      "grad_norm": 0.0,
+      "learning_rate": 3.339796978996671e-06,
+      "loss": 0.9592,
+      "step": 20349
+    },
+    {
+      "epoch": 0.7400807360802997,
+      "grad_norm": 0.0,
+      "learning_rate": 3.338918418584234e-06,
+      "loss": 1.0283,
+      "step": 20350
+    },
+    {
+      "epoch": 0.7401171036840383,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3380399505848516e-06,
+      "loss": 1.0443,
+      "step": 20351
+    },
+    {
+      "epoch": 0.7401534712877769,
+      "grad_norm": 0.0,
+      "learning_rate": 3.337161575010712e-06,
+      "loss": 0.8403,
+      "step": 20352
+    },
+    {
+      "epoch": 0.7401898388915155,
+      "grad_norm": 0.0,
+      "learning_rate": 3.336283291874004e-06,
+      "loss": 0.9456,
+      "step": 20353
+    },
+    {
+      "epoch": 0.7402262064952541,
+      "grad_norm": 0.0,
+      "learning_rate": 3.335405101186907e-06,
+      "loss": 0.8848,
+      "step": 20354
+    },
+    {
+      "epoch": 0.7402625740989927,
+      "grad_norm": 0.0,
+      "learning_rate": 3.334527002961615e-06,
+      "loss": 0.9363,
+      "step": 20355
+    },
+    {
+      "epoch": 0.7402989417027313,
+      "grad_norm": 0.0,
+      "learning_rate": 3.333648997210297e-06,
+      "loss": 0.9724,
+      "step": 20356
+    },
+    {
+      "epoch": 0.7403353093064698,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3327710839451434e-06,
+      "loss": 0.9048,
+      "step": 20357
+    },
+    {
+      "epoch": 0.7403716769102083,
+      "grad_norm": 0.0,
+      "learning_rate": 3.331893263178333e-06,
+      "loss": 0.9147,
+      "step": 20358
+    },
+    {
+      "epoch": 0.7404080445139469,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3310155349220417e-06,
+      "loss": 0.9473,
+      "step": 20359
+    },
+    {
+      "epoch": 0.7404444121176855,
+      "grad_norm": 0.0,
+      "learning_rate": 3.330137899188445e-06,
+      "loss": 1.0224,
+      "step": 20360
+    },
+    {
+      "epoch": 0.7404807797214241,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3292603559897274e-06,
+      "loss": 1.0395,
+      "step": 20361
+    },
+    {
+      "epoch": 0.7405171473251627,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3283829053380524e-06,
+      "loss": 0.8793,
+      "step": 20362
+    },
+    {
+      "epoch": 0.7405535149289013,
+      "grad_norm": 0.0,
+      "learning_rate": 3.327505547245601e-06,
+      "loss": 0.8748,
+      "step": 20363
+    },
+    {
+      "epoch": 0.7405898825326399,
+      "grad_norm": 0.0,
+      "learning_rate": 3.326628281724543e-06,
+      "loss": 0.8906,
+      "step": 20364
+    },
+    {
+      "epoch": 0.7406262501363785,
+      "grad_norm": 0.0,
+      "learning_rate": 3.325751108787051e-06,
+      "loss": 1.0131,
+      "step": 20365
+    },
+    {
+      "epoch": 0.7406626177401171,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3248740284452873e-06,
+      "loss": 0.8405,
+      "step": 20366
+    },
+    {
+      "epoch": 0.7406989853438557,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3239970407114307e-06,
+      "loss": 0.8749,
+      "step": 20367
+    },
+    {
+      "epoch": 0.7407353529475943,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3231201455976424e-06,
+      "loss": 0.9624,
+      "step": 20368
+    },
+    {
+      "epoch": 0.7407717205513329,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3222433431160896e-06,
+      "loss": 1.0559,
+      "step": 20369
+    },
+    {
+      "epoch": 0.7408080881550715,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3213666332789352e-06,
+      "loss": 0.9991,
+      "step": 20370
+    },
+    {
+      "epoch": 0.7408444557588101,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3204900160983435e-06,
+      "loss": 0.8195,
+      "step": 20371
+    },
+    {
+      "epoch": 0.7408808233625487,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3196134915864753e-06,
+      "loss": 0.9452,
+      "step": 20372
+    },
+    {
+      "epoch": 0.7409171909662873,
+      "grad_norm": 0.0,
+      "learning_rate": 3.318737059755488e-06,
+      "loss": 0.9557,
+      "step": 20373
+    },
+    {
+      "epoch": 0.7409535585700259,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3178607206175517e-06,
+      "loss": 1.0219,
+      "step": 20374
+    },
+    {
+      "epoch": 0.7409899261737644,
+      "grad_norm": 0.0,
+      "learning_rate": 3.316984474184811e-06,
+      "loss": 0.9103,
+      "step": 20375
+    },
+    {
+      "epoch": 0.741026293777503,
+      "grad_norm": 0.0,
+      "learning_rate": 3.316108320469431e-06,
+      "loss": 0.8888,
+      "step": 20376
+    },
+    {
+      "epoch": 0.7410626613812415,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3152322594835663e-06,
+      "loss": 0.9622,
+      "step": 20377
+    },
+    {
+      "epoch": 0.7410990289849801,
+      "grad_norm": 0.0,
+      "learning_rate": 3.314356291239368e-06,
+      "loss": 0.8923,
+      "step": 20378
+    },
+    {
+      "epoch": 0.7411353965887187,
+      "grad_norm": 0.0,
+      "learning_rate": 3.313480415748989e-06,
+      "loss": 0.8991,
+      "step": 20379
+    },
+    {
+      "epoch": 0.7411717641924573,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3126046330245843e-06,
+      "loss": 0.9657,
+      "step": 20380
+    },
+    {
+      "epoch": 0.7412081317961959,
+      "grad_norm": 0.0,
+      "learning_rate": 3.311728943078302e-06,
+      "loss": 0.9212,
+      "step": 20381
+    },
+    {
+      "epoch": 0.7412444993999345,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3108533459222914e-06,
+      "loss": 0.9401,
+      "step": 20382
+    },
+    {
+      "epoch": 0.7412808670036731,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3099778415687e-06,
+      "loss": 0.874,
+      "step": 20383
+    },
+    {
+      "epoch": 0.7413172346074117,
+      "grad_norm": 0.0,
+      "learning_rate": 3.309102430029675e-06,
+      "loss": 0.7735,
+      "step": 20384
+    },
+    {
+      "epoch": 0.7413536022111503,
+      "grad_norm": 0.0,
+      "learning_rate": 3.308227111317356e-06,
+      "loss": 0.986,
+      "step": 20385
+    },
+    {
+      "epoch": 0.7413899698148889,
+      "grad_norm": 0.0,
+      "learning_rate": 3.307351885443896e-06,
+      "loss": 0.8577,
+      "step": 20386
+    },
+    {
+      "epoch": 0.7414263374186275,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3064767524214315e-06,
+      "loss": 0.9726,
+      "step": 20387
+    },
+    {
+      "epoch": 0.7414627050223661,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3056017122621063e-06,
+      "loss": 0.8444,
+      "step": 20388
+    },
+    {
+      "epoch": 0.7414990726261047,
+      "grad_norm": 0.0,
+      "learning_rate": 3.304726764978059e-06,
+      "loss": 0.9423,
+      "step": 20389
+    },
+    {
+      "epoch": 0.7415354402298433,
+      "grad_norm": 0.0,
+      "learning_rate": 3.303851910581428e-06,
+      "loss": 0.8769,
+      "step": 20390
+    },
+    {
+      "epoch": 0.7415718078335819,
+      "grad_norm": 0.0,
+      "learning_rate": 3.302977149084352e-06,
+      "loss": 0.9977,
+      "step": 20391
+    },
+    {
+      "epoch": 0.7416081754373204,
+      "grad_norm": 0.0,
+      "learning_rate": 3.302102480498962e-06,
+      "loss": 1.0092,
+      "step": 20392
+    },
+    {
+      "epoch": 0.741644543041059,
+      "grad_norm": 0.0,
+      "learning_rate": 3.301227904837405e-06,
+      "loss": 1.034,
+      "step": 20393
+    },
+    {
+      "epoch": 0.7416809106447976,
+      "grad_norm": 0.0,
+      "learning_rate": 3.300353422111798e-06,
+      "loss": 0.9593,
+      "step": 20394
+    },
+    {
+      "epoch": 0.7417172782485362,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2994790323342863e-06,
+      "loss": 0.9777,
+      "step": 20395
+    },
+    {
+      "epoch": 0.7417536458522748,
+      "grad_norm": 0.0,
+      "learning_rate": 3.298604735516996e-06,
+      "loss": 1.0052,
+      "step": 20396
+    },
+    {
+      "epoch": 0.7417900134560134,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2977305316720576e-06,
+      "loss": 0.9224,
+      "step": 20397
+    },
+    {
+      "epoch": 0.741826381059752,
+      "grad_norm": 0.0,
+      "learning_rate": 3.296856420811595e-06,
+      "loss": 0.9589,
+      "step": 20398
+    },
+    {
+      "epoch": 0.7418627486634906,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2959824029477415e-06,
+      "loss": 1.0235,
+      "step": 20399
+    },
+    {
+      "epoch": 0.7418991162672292,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2951084780926214e-06,
+      "loss": 0.9531,
+      "step": 20400
+    },
+    {
+      "epoch": 0.7419354838709677,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2942346462583576e-06,
+      "loss": 0.8881,
+      "step": 20401
+    },
+    {
+      "epoch": 0.7419718514747063,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2933609074570707e-06,
+      "loss": 0.9551,
+      "step": 20402
+    },
+    {
+      "epoch": 0.7420082190784449,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2924872617008927e-06,
+      "loss": 1.0339,
+      "step": 20403
+    },
+    {
+      "epoch": 0.7420445866821835,
+      "grad_norm": 0.0,
+      "learning_rate": 3.291613709001931e-06,
+      "loss": 0.9417,
+      "step": 20404
+    },
+    {
+      "epoch": 0.7420809542859221,
+      "grad_norm": 0.0,
+      "learning_rate": 3.290740249372313e-06,
+      "loss": 1.0396,
+      "step": 20405
+    },
+    {
+      "epoch": 0.7421173218896607,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2898668828241554e-06,
+      "loss": 0.9276,
+      "step": 20406
+    },
+    {
+      "epoch": 0.7421536894933993,
+      "grad_norm": 0.0,
+      "learning_rate": 3.288993609369575e-06,
+      "loss": 0.8987,
+      "step": 20407
+    },
+    {
+      "epoch": 0.7421900570971379,
+      "grad_norm": 0.0,
+      "learning_rate": 3.288120429020686e-06,
+      "loss": 0.8544,
+      "step": 20408
+    },
+    {
+      "epoch": 0.7422264247008764,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2872473417896034e-06,
+      "loss": 0.9859,
+      "step": 20409
+    },
+    {
+      "epoch": 0.742262792304615,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2863743476884402e-06,
+      "loss": 1.0426,
+      "step": 20410
+    },
+    {
+      "epoch": 0.7422991599083536,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2855014467293035e-06,
+      "loss": 0.9794,
+      "step": 20411
+    },
+    {
+      "epoch": 0.7423355275120922,
+      "grad_norm": 0.0,
+      "learning_rate": 3.284628638924311e-06,
+      "loss": 0.9064,
+      "step": 20412
+    },
+    {
+      "epoch": 0.7423718951158308,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2837559242855687e-06,
+      "loss": 0.8064,
+      "step": 20413
+    },
+    {
+      "epoch": 0.7424082627195694,
+      "grad_norm": 0.0,
+      "learning_rate": 3.282883302825184e-06,
+      "loss": 0.9005,
+      "step": 20414
+    },
+    {
+      "epoch": 0.742444630323308,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2820107745552623e-06,
+      "loss": 0.8863,
+      "step": 20415
+    },
+    {
+      "epoch": 0.7424809979270466,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2811383394879093e-06,
+      "loss": 0.9784,
+      "step": 20416
+    },
+    {
+      "epoch": 0.7425173655307852,
+      "grad_norm": 0.0,
+      "learning_rate": 3.280265997635227e-06,
+      "loss": 0.9214,
+      "step": 20417
+    },
+    {
+      "epoch": 0.7425537331345238,
+      "grad_norm": 0.0,
+      "learning_rate": 3.279393749009322e-06,
+      "loss": 0.9366,
+      "step": 20418
+    },
+    {
+      "epoch": 0.7425901007382624,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2785215936222947e-06,
+      "loss": 0.9594,
+      "step": 20419
+    },
+    {
+      "epoch": 0.742626468342001,
+      "grad_norm": 0.0,
+      "learning_rate": 3.277649531486242e-06,
+      "loss": 1.0512,
+      "step": 20420
+    },
+    {
+      "epoch": 0.7426628359457396,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2767775626132627e-06,
+      "loss": 0.8135,
+      "step": 20421
+    },
+    {
+      "epoch": 0.7426992035494782,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2759056870154616e-06,
+      "loss": 0.8995,
+      "step": 20422
+    },
+    {
+      "epoch": 0.7427355711532168,
+      "grad_norm": 0.0,
+      "learning_rate": 3.275033904704923e-06,
+      "loss": 0.9919,
+      "step": 20423
+    },
+    {
+      "epoch": 0.7427719387569554,
+      "grad_norm": 0.0,
+      "learning_rate": 3.274162215693749e-06,
+      "loss": 0.9411,
+      "step": 20424
+    },
+    {
+      "epoch": 0.742808306360694,
+      "grad_norm": 0.0,
+      "learning_rate": 3.273290619994033e-06,
+      "loss": 0.9568,
+      "step": 20425
+    },
+    {
+      "epoch": 0.7428446739644324,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2724191176178656e-06,
+      "loss": 0.9372,
+      "step": 20426
+    },
+    {
+      "epoch": 0.742881041568171,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2715477085773374e-06,
+      "loss": 0.9477,
+      "step": 20427
+    },
+    {
+      "epoch": 0.7429174091719096,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2706763928845385e-06,
+      "loss": 0.9695,
+      "step": 20428
+    },
+    {
+      "epoch": 0.7429537767756482,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2698051705515564e-06,
+      "loss": 0.9607,
+      "step": 20429
+    },
+    {
+      "epoch": 0.7429901443793868,
+      "grad_norm": 0.0,
+      "learning_rate": 3.268934041590476e-06,
+      "loss": 0.8402,
+      "step": 20430
+    },
+    {
+      "epoch": 0.7430265119831254,
+      "grad_norm": 0.0,
+      "learning_rate": 3.268063006013389e-06,
+      "loss": 0.9269,
+      "step": 20431
+    },
+    {
+      "epoch": 0.743062879586864,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2671920638323762e-06,
+      "loss": 1.0219,
+      "step": 20432
+    },
+    {
+      "epoch": 0.7430992471906026,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2663212150595214e-06,
+      "loss": 0.8905,
+      "step": 20433
+    },
+    {
+      "epoch": 0.7431356147943412,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2654504597069014e-06,
+      "loss": 0.8612,
+      "step": 20434
+    },
+    {
+      "epoch": 0.7431719823980798,
+      "grad_norm": 0.0,
+      "learning_rate": 3.264579797786609e-06,
+      "loss": 0.99,
+      "step": 20435
+    },
+    {
+      "epoch": 0.7432083500018184,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2637092293107087e-06,
+      "loss": 0.9308,
+      "step": 20436
+    },
+    {
+      "epoch": 0.743244717605557,
+      "grad_norm": 0.0,
+      "learning_rate": 3.262838754291289e-06,
+      "loss": 0.9876,
+      "step": 20437
+    },
+    {
+      "epoch": 0.7432810852092956,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2619683727404215e-06,
+      "loss": 0.9112,
+      "step": 20438
+    },
+    {
+      "epoch": 0.7433174528130342,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2610980846701833e-06,
+      "loss": 0.9272,
+      "step": 20439
+    },
+    {
+      "epoch": 0.7433538204167728,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2602278900926454e-06,
+      "loss": 0.9305,
+      "step": 20440
+    },
+    {
+      "epoch": 0.7433901880205114,
+      "grad_norm": 0.0,
+      "learning_rate": 3.259357789019889e-06,
+      "loss": 1.0204,
+      "step": 20441
+    },
+    {
+      "epoch": 0.7434265556242499,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2584877814639727e-06,
+      "loss": 1.0114,
+      "step": 20442
+    },
+    {
+      "epoch": 0.7434629232279885,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2576178674369776e-06,
+      "loss": 1.019,
+      "step": 20443
+    },
+    {
+      "epoch": 0.743499290831727,
+      "grad_norm": 0.0,
+      "learning_rate": 3.256748046950967e-06,
+      "loss": 0.7906,
+      "step": 20444
+    },
+    {
+      "epoch": 0.7435356584354657,
+      "grad_norm": 0.0,
+      "learning_rate": 3.255878320018011e-06,
+      "loss": 0.8553,
+      "step": 20445
+    },
+    {
+      "epoch": 0.7435720260392042,
+      "grad_norm": 0.0,
+      "learning_rate": 3.255008686650175e-06,
+      "loss": 1.0556,
+      "step": 20446
+    },
+    {
+      "epoch": 0.7436083936429428,
+      "grad_norm": 0.0,
+      "learning_rate": 3.254139146859523e-06,
+      "loss": 0.918,
+      "step": 20447
+    },
+    {
+      "epoch": 0.7436447612466814,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2532697006581192e-06,
+      "loss": 0.8453,
+      "step": 20448
+    },
+    {
+      "epoch": 0.74368112885042,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2524003480580226e-06,
+      "loss": 1.0515,
+      "step": 20449
+    },
+    {
+      "epoch": 0.7437174964541586,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2515310890713004e-06,
+      "loss": 0.9144,
+      "step": 20450
+    },
+    {
+      "epoch": 0.7437538640578972,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2506619237100102e-06,
+      "loss": 0.858,
+      "step": 20451
+    },
+    {
+      "epoch": 0.7437902316616358,
+      "grad_norm": 0.0,
+      "learning_rate": 3.24979285198621e-06,
+      "loss": 0.9907,
+      "step": 20452
+    },
+    {
+      "epoch": 0.7438265992653744,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2489238739119534e-06,
+      "loss": 1.1033,
+      "step": 20453
+    },
+    {
+      "epoch": 0.743862966869113,
+      "grad_norm": 0.0,
+      "learning_rate": 3.248054989499305e-06,
+      "loss": 0.9327,
+      "step": 20454
+    },
+    {
+      "epoch": 0.7438993344728516,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2471861987603083e-06,
+      "loss": 0.8433,
+      "step": 20455
+    },
+    {
+      "epoch": 0.7439357020765902,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2463175017070248e-06,
+      "loss": 0.8036,
+      "step": 20456
+    },
+    {
+      "epoch": 0.7439720696803288,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2454488983515043e-06,
+      "loss": 0.9527,
+      "step": 20457
+    },
+    {
+      "epoch": 0.7440084372840674,
+      "grad_norm": 0.0,
+      "learning_rate": 3.244580388705796e-06,
+      "loss": 0.7455,
+      "step": 20458
+    },
+    {
+      "epoch": 0.7440448048878059,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2437119727819476e-06,
+      "loss": 0.8797,
+      "step": 20459
+    },
+    {
+      "epoch": 0.7440811724915445,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2428436505920157e-06,
+      "loss": 0.9147,
+      "step": 20460
+    },
+    {
+      "epoch": 0.7441175400952831,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2419754221480347e-06,
+      "loss": 0.9036,
+      "step": 20461
+    },
+    {
+      "epoch": 0.7441539076990217,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2411072874620608e-06,
+      "loss": 0.8331,
+      "step": 20462
+    },
+    {
+      "epoch": 0.7441902753027603,
+      "grad_norm": 0.0,
+      "learning_rate": 3.240239246546132e-06,
+      "loss": 0.8116,
+      "step": 20463
+    },
+    {
+      "epoch": 0.7442266429064989,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2393712994122937e-06,
+      "loss": 0.8994,
+      "step": 20464
+    },
+    {
+      "epoch": 0.7442630105102375,
+      "grad_norm": 0.0,
+      "learning_rate": 3.238503446072586e-06,
+      "loss": 0.8867,
+      "step": 20465
+    },
+    {
+      "epoch": 0.7442993781139761,
+      "grad_norm": 0.0,
+      "learning_rate": 3.237635686539047e-06,
+      "loss": 0.9133,
+      "step": 20466
+    },
+    {
+      "epoch": 0.7443357457177147,
+      "grad_norm": 0.0,
+      "learning_rate": 3.236768020823724e-06,
+      "loss": 0.8827,
+      "step": 20467
+    },
+    {
+      "epoch": 0.7443721133214533,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2359004489386427e-06,
+      "loss": 0.8868,
+      "step": 20468
+    },
+    {
+      "epoch": 0.7444084809251919,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2350329708958495e-06,
+      "loss": 0.9478,
+      "step": 20469
+    },
+    {
+      "epoch": 0.7444448485289304,
+      "grad_norm": 0.0,
+      "learning_rate": 3.234165586707375e-06,
+      "loss": 0.9149,
+      "step": 20470
+    },
+    {
+      "epoch": 0.744481216132669,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2332982963852545e-06,
+      "loss": 1.0623,
+      "step": 20471
+    },
+    {
+      "epoch": 0.7445175837364076,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2324310999415144e-06,
+      "loss": 0.8364,
+      "step": 20472
+    },
+    {
+      "epoch": 0.7445539513401462,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2315639973881984e-06,
+      "loss": 0.8826,
+      "step": 20473
+    },
+    {
+      "epoch": 0.7445903189438848,
+      "grad_norm": 0.0,
+      "learning_rate": 3.230696988737323e-06,
+      "loss": 0.8542,
+      "step": 20474
+    },
+    {
+      "epoch": 0.7446266865476234,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2298300740009246e-06,
+      "loss": 0.9786,
+      "step": 20475
+    },
+    {
+      "epoch": 0.7446630541513619,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2289632531910287e-06,
+      "loss": 0.851,
+      "step": 20476
+    },
+    {
+      "epoch": 0.7446994217551005,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2280965263196604e-06,
+      "loss": 0.9576,
+      "step": 20477
+    },
+    {
+      "epoch": 0.7447357893588391,
+      "grad_norm": 0.0,
+      "learning_rate": 3.227229893398842e-06,
+      "loss": 0.8922,
+      "step": 20478
+    },
+    {
+      "epoch": 0.7447721569625777,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2263633544406037e-06,
+      "loss": 0.9119,
+      "step": 20479
+    },
+    {
+      "epoch": 0.7448085245663163,
+      "grad_norm": 0.0,
+      "learning_rate": 3.225496909456962e-06,
+      "loss": 0.9528,
+      "step": 20480
+    },
+    {
+      "epoch": 0.7448448921700549,
+      "grad_norm": 0.0,
+      "learning_rate": 3.224630558459939e-06,
+      "loss": 0.8983,
+      "step": 20481
+    },
+    {
+      "epoch": 0.7448812597737935,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2237643014615537e-06,
+      "loss": 0.9214,
+      "step": 20482
+    },
+    {
+      "epoch": 0.7449176273775321,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2228981384738255e-06,
+      "loss": 0.813,
+      "step": 20483
+    },
+    {
+      "epoch": 0.7449539949812707,
+      "grad_norm": 0.0,
+      "learning_rate": 3.222032069508767e-06,
+      "loss": 0.947,
+      "step": 20484
+    },
+    {
+      "epoch": 0.7449903625850093,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2211660945783996e-06,
+      "loss": 0.9208,
+      "step": 20485
+    },
+    {
+      "epoch": 0.7450267301887479,
+      "grad_norm": 0.0,
+      "learning_rate": 3.220300213694737e-06,
+      "loss": 0.9656,
+      "step": 20486
+    },
+    {
+      "epoch": 0.7450630977924865,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2194344268697853e-06,
+      "loss": 1.0155,
+      "step": 20487
+    },
+    {
+      "epoch": 0.7450994653962251,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2185687341155627e-06,
+      "loss": 1.0086,
+      "step": 20488
+    },
+    {
+      "epoch": 0.7451358329999637,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2177031354440767e-06,
+      "loss": 0.9182,
+      "step": 20489
+    },
+    {
+      "epoch": 0.7451722006037023,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2168376308673378e-06,
+      "loss": 0.7913,
+      "step": 20490
+    },
+    {
+      "epoch": 0.7452085682074409,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2159722203973497e-06,
+      "loss": 0.9178,
+      "step": 20491
+    },
+    {
+      "epoch": 0.7452449358111795,
+      "grad_norm": 0.0,
+      "learning_rate": 3.215106904046127e-06,
+      "loss": 0.9852,
+      "step": 20492
+    },
+    {
+      "epoch": 0.7452813034149179,
+      "grad_norm": 0.0,
+      "learning_rate": 3.214241681825664e-06,
+      "loss": 0.9208,
+      "step": 20493
+    },
+    {
+      "epoch": 0.7453176710186565,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2133765537479733e-06,
+      "loss": 0.9645,
+      "step": 20494
+    },
+    {
+      "epoch": 0.7453540386223951,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2125115198250534e-06,
+      "loss": 0.9086,
+      "step": 20495
+    },
+    {
+      "epoch": 0.7453904062261337,
+      "grad_norm": 0.0,
+      "learning_rate": 3.211646580068907e-06,
+      "loss": 0.9082,
+      "step": 20496
+    },
+    {
+      "epoch": 0.7454267738298723,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2107817344915284e-06,
+      "loss": 0.9203,
+      "step": 20497
+    },
+    {
+      "epoch": 0.7454631414336109,
+      "grad_norm": 0.0,
+      "learning_rate": 3.209916983104924e-06,
+      "loss": 1.0704,
+      "step": 20498
+    },
+    {
+      "epoch": 0.7454995090373495,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2090523259210883e-06,
+      "loss": 0.8653,
+      "step": 20499
+    },
+    {
+      "epoch": 0.7455358766410881,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2081877629520154e-06,
+      "loss": 0.9928,
+      "step": 20500
+    },
+    {
+      "epoch": 0.7455722442448267,
+      "grad_norm": 0.0,
+      "learning_rate": 3.207323294209699e-06,
+      "loss": 0.9108,
+      "step": 20501
+    },
+    {
+      "epoch": 0.7456086118485653,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2064589197061403e-06,
+      "loss": 0.8534,
+      "step": 20502
+    },
+    {
+      "epoch": 0.7456449794523039,
+      "grad_norm": 0.0,
+      "learning_rate": 3.205594639453319e-06,
+      "loss": 0.9087,
+      "step": 20503
+    },
+    {
+      "epoch": 0.7456813470560425,
+      "grad_norm": 0.0,
+      "learning_rate": 3.204730453463235e-06,
+      "loss": 0.9245,
+      "step": 20504
+    },
+    {
+      "epoch": 0.7457177146597811,
+      "grad_norm": 0.0,
+      "learning_rate": 3.203866361747877e-06,
+      "loss": 0.9583,
+      "step": 20505
+    },
+    {
+      "epoch": 0.7457540822635197,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2030023643192266e-06,
+      "loss": 0.9996,
+      "step": 20506
+    },
+    {
+      "epoch": 0.7457904498672583,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2021384611892757e-06,
+      "loss": 0.9897,
+      "step": 20507
+    },
+    {
+      "epoch": 0.7458268174709969,
+      "grad_norm": 0.0,
+      "learning_rate": 3.20127465237001e-06,
+      "loss": 0.9825,
+      "step": 20508
+    },
+    {
+      "epoch": 0.7458631850747355,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2004109378734126e-06,
+      "loss": 0.9305,
+      "step": 20509
+    },
+    {
+      "epoch": 0.745899552678474,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1995473177114624e-06,
+      "loss": 0.9037,
+      "step": 20510
+    },
+    {
+      "epoch": 0.7459359202822126,
+      "grad_norm": 0.0,
+      "learning_rate": 3.198683791896148e-06,
+      "loss": 0.8993,
+      "step": 20511
+    },
+    {
+      "epoch": 0.7459722878859512,
+      "grad_norm": 0.0,
+      "learning_rate": 3.197820360439445e-06,
+      "loss": 1.0035,
+      "step": 20512
+    },
+    {
+      "epoch": 0.7460086554896898,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1969570233533352e-06,
+      "loss": 1.0902,
+      "step": 20513
+    },
+    {
+      "epoch": 0.7460450230934283,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1960937806497938e-06,
+      "loss": 0.9167,
+      "step": 20514
+    },
+    {
+      "epoch": 0.746081390697167,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1952306323407966e-06,
+      "loss": 0.8716,
+      "step": 20515
+    },
+    {
+      "epoch": 0.7461177583009055,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1943675784383177e-06,
+      "loss": 0.8929,
+      "step": 20516
+    },
+    {
+      "epoch": 0.7461541259046441,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1935046189543364e-06,
+      "loss": 0.9294,
+      "step": 20517
+    },
+    {
+      "epoch": 0.7461904935083827,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1926417539008203e-06,
+      "loss": 0.8869,
+      "step": 20518
+    },
+    {
+      "epoch": 0.7462268611121213,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1917789832897418e-06,
+      "loss": 1.0459,
+      "step": 20519
+    },
+    {
+      "epoch": 0.7462632287158599,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1909163071330674e-06,
+      "loss": 0.9019,
+      "step": 20520
+    },
+    {
+      "epoch": 0.7462995963195985,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1900537254427745e-06,
+      "loss": 1.0529,
+      "step": 20521
+    },
+    {
+      "epoch": 0.7463359639233371,
+      "grad_norm": 0.0,
+      "learning_rate": 3.189191238230818e-06,
+      "loss": 0.9226,
+      "step": 20522
+    },
+    {
+      "epoch": 0.7463723315270757,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1883288455091722e-06,
+      "loss": 1.0706,
+      "step": 20523
+    },
+    {
+      "epoch": 0.7464086991308143,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1874665472898003e-06,
+      "loss": 0.8379,
+      "step": 20524
+    },
+    {
+      "epoch": 0.7464450667345529,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1866043435846627e-06,
+      "loss": 0.9715,
+      "step": 20525
+    },
+    {
+      "epoch": 0.7464814343382915,
+      "grad_norm": 0.0,
+      "learning_rate": 3.185742234405724e-06,
+      "loss": 0.9315,
+      "step": 20526
+    },
+    {
+      "epoch": 0.74651780194203,
+      "grad_norm": 0.0,
+      "learning_rate": 3.184880219764944e-06,
+      "loss": 0.9931,
+      "step": 20527
+    },
+    {
+      "epoch": 0.7465541695457686,
+      "grad_norm": 0.0,
+      "learning_rate": 3.184018299674281e-06,
+      "loss": 0.8549,
+      "step": 20528
+    },
+    {
+      "epoch": 0.7465905371495072,
+      "grad_norm": 0.0,
+      "learning_rate": 3.183156474145691e-06,
+      "loss": 0.8293,
+      "step": 20529
+    },
+    {
+      "epoch": 0.7466269047532458,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1822947431911356e-06,
+      "loss": 0.8891,
+      "step": 20530
+    },
+    {
+      "epoch": 0.7466632723569844,
+      "grad_norm": 0.0,
+      "learning_rate": 3.181433106822568e-06,
+      "loss": 0.9364,
+      "step": 20531
+    },
+    {
+      "epoch": 0.746699639960723,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1805715650519407e-06,
+      "loss": 0.9619,
+      "step": 20532
+    },
+    {
+      "epoch": 0.7467360075644616,
+      "grad_norm": 0.0,
+      "learning_rate": 3.179710117891205e-06,
+      "loss": 0.9257,
+      "step": 20533
+    },
+    {
+      "epoch": 0.7467723751682002,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1788487653523214e-06,
+      "loss": 0.7638,
+      "step": 20534
+    },
+    {
+      "epoch": 0.7468087427719388,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1779875074472266e-06,
+      "loss": 0.881,
+      "step": 20535
+    },
+    {
+      "epoch": 0.7468451103756774,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1771263441878786e-06,
+      "loss": 0.8301,
+      "step": 20536
+    },
+    {
+      "epoch": 0.746881477979416,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1762652755862234e-06,
+      "loss": 0.9211,
+      "step": 20537
+    },
+    {
+      "epoch": 0.7469178455831545,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1754043016542046e-06,
+      "loss": 0.8965,
+      "step": 20538
+    },
+    {
+      "epoch": 0.7469542131868931,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1745434224037654e-06,
+      "loss": 0.8993,
+      "step": 20539
+    },
+    {
+      "epoch": 0.7469905807906317,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1736826378468587e-06,
+      "loss": 0.9676,
+      "step": 20540
+    },
+    {
+      "epoch": 0.7470269483943703,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1728219479954137e-06,
+      "loss": 0.9308,
+      "step": 20541
+    },
+    {
+      "epoch": 0.7470633159981089,
+      "grad_norm": 0.0,
+      "learning_rate": 3.171961352861381e-06,
+      "loss": 0.9702,
+      "step": 20542
+    },
+    {
+      "epoch": 0.7470996836018474,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1711008524566966e-06,
+      "loss": 0.9481,
+      "step": 20543
+    },
+    {
+      "epoch": 0.747136051205586,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1702404467933003e-06,
+      "loss": 0.9204,
+      "step": 20544
+    },
+    {
+      "epoch": 0.7471724188093246,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1693801358831256e-06,
+      "loss": 0.8881,
+      "step": 20545
+    },
+    {
+      "epoch": 0.7472087864130632,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1685199197381077e-06,
+      "loss": 0.8921,
+      "step": 20546
+    },
+    {
+      "epoch": 0.7472451540168018,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1676597983701896e-06,
+      "loss": 0.8666,
+      "step": 20547
+    },
+    {
+      "epoch": 0.7472815216205404,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1667997717912923e-06,
+      "loss": 0.8937,
+      "step": 20548
+    },
+    {
+      "epoch": 0.747317889224279,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1659398400133557e-06,
+      "loss": 0.9157,
+      "step": 20549
+    },
+    {
+      "epoch": 0.7473542568280176,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1650800030483075e-06,
+      "loss": 0.9694,
+      "step": 20550
+    },
+    {
+      "epoch": 0.7473906244317562,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1642202609080773e-06,
+      "loss": 0.8479,
+      "step": 20551
+    },
+    {
+      "epoch": 0.7474269920354948,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1633606136045882e-06,
+      "loss": 0.9663,
+      "step": 20552
+    },
+    {
+      "epoch": 0.7474633596392334,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1625010611497776e-06,
+      "loss": 0.9253,
+      "step": 20553
+    },
+    {
+      "epoch": 0.747499727242972,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1616416035555577e-06,
+      "loss": 0.9582,
+      "step": 20554
+    },
+    {
+      "epoch": 0.7475360948467106,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1607822408338606e-06,
+      "loss": 0.9325,
+      "step": 20555
+    },
+    {
+      "epoch": 0.7475724624504492,
+      "grad_norm": 0.0,
+      "learning_rate": 3.159922972996606e-06,
+      "loss": 0.899,
+      "step": 20556
+    },
+    {
+      "epoch": 0.7476088300541878,
+      "grad_norm": 0.0,
+      "learning_rate": 3.159063800055716e-06,
+      "loss": 1.0268,
+      "step": 20557
+    },
+    {
+      "epoch": 0.7476451976579264,
+      "grad_norm": 0.0,
+      "learning_rate": 3.158204722023107e-06,
+      "loss": 0.9585,
+      "step": 20558
+    },
+    {
+      "epoch": 0.747681565261665,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1573457389107065e-06,
+      "loss": 0.9404,
+      "step": 20559
+    },
+    {
+      "epoch": 0.7477179328654034,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1564868507304182e-06,
+      "loss": 1.0256,
+      "step": 20560
+    },
+    {
+      "epoch": 0.747754300469142,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1556280574941687e-06,
+      "loss": 0.9638,
+      "step": 20561
+    },
+    {
+      "epoch": 0.7477906680728806,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1547693592138697e-06,
+      "loss": 0.9241,
+      "step": 20562
+    },
+    {
+      "epoch": 0.7478270356766192,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1539107559014325e-06,
+      "loss": 1.0658,
+      "step": 20563
+    },
+    {
+      "epoch": 0.7478634032803578,
+      "grad_norm": 0.0,
+      "learning_rate": 3.153052247568771e-06,
+      "loss": 0.9725,
+      "step": 20564
+    },
+    {
+      "epoch": 0.7478997708840964,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1521938342277925e-06,
+      "loss": 1.0198,
+      "step": 20565
+    },
+    {
+      "epoch": 0.747936138487835,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1513355158904137e-06,
+      "loss": 1.0074,
+      "step": 20566
+    },
+    {
+      "epoch": 0.7479725060915736,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1504772925685323e-06,
+      "loss": 0.926,
+      "step": 20567
+    },
+    {
+      "epoch": 0.7480088736953122,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1496191642740625e-06,
+      "loss": 0.9909,
+      "step": 20568
+    },
+    {
+      "epoch": 0.7480452412990508,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1487611310189083e-06,
+      "loss": 1.0168,
+      "step": 20569
+    },
+    {
+      "epoch": 0.7480816089027894,
+      "grad_norm": 0.0,
+      "learning_rate": 3.147903192814973e-06,
+      "loss": 1.0172,
+      "step": 20570
+    },
+    {
+      "epoch": 0.748117976506528,
+      "grad_norm": 0.0,
+      "learning_rate": 3.147045349674155e-06,
+      "loss": 0.8476,
+      "step": 20571
+    },
+    {
+      "epoch": 0.7481543441102666,
+      "grad_norm": 0.0,
+      "learning_rate": 3.146187601608367e-06,
+      "loss": 0.8465,
+      "step": 20572
+    },
+    {
+      "epoch": 0.7481907117140052,
+      "grad_norm": 0.0,
+      "learning_rate": 3.145329948629494e-06,
+      "loss": 0.9325,
+      "step": 20573
+    },
+    {
+      "epoch": 0.7482270793177438,
+      "grad_norm": 0.0,
+      "learning_rate": 3.144472390749447e-06,
+      "loss": 0.9589,
+      "step": 20574
+    },
+    {
+      "epoch": 0.7482634469214824,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1436149279801174e-06,
+      "loss": 0.9904,
+      "step": 20575
+    },
+    {
+      "epoch": 0.748299814525221,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1427575603334036e-06,
+      "loss": 0.9276,
+      "step": 20576
+    },
+    {
+      "epoch": 0.7483361821289595,
+      "grad_norm": 0.0,
+      "learning_rate": 3.141900287821197e-06,
+      "loss": 0.9304,
+      "step": 20577
+    },
+    {
+      "epoch": 0.7483725497326981,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1410431104553962e-06,
+      "loss": 0.8422,
+      "step": 20578
+    },
+    {
+      "epoch": 0.7484089173364367,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1401860282478903e-06,
+      "loss": 0.9282,
+      "step": 20579
+    },
+    {
+      "epoch": 0.7484452849401753,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1393290412105702e-06,
+      "loss": 0.9575,
+      "step": 20580
+    },
+    {
+      "epoch": 0.7484816525439139,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1384721493553263e-06,
+      "loss": 0.8759,
+      "step": 20581
+    },
+    {
+      "epoch": 0.7485180201476525,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1376153526940457e-06,
+      "loss": 0.945,
+      "step": 20582
+    },
+    {
+      "epoch": 0.748554387751391,
+      "grad_norm": 0.0,
+      "learning_rate": 3.136758651238616e-06,
+      "loss": 0.9789,
+      "step": 20583
+    },
+    {
+      "epoch": 0.7485907553551296,
+      "grad_norm": 0.0,
+      "learning_rate": 3.135902045000919e-06,
+      "loss": 0.9432,
+      "step": 20584
+    },
+    {
+      "epoch": 0.7486271229588682,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1350455339928488e-06,
+      "loss": 0.9883,
+      "step": 20585
+    },
+    {
+      "epoch": 0.7486634905626068,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1341891182262752e-06,
+      "loss": 0.8712,
+      "step": 20586
+    },
+    {
+      "epoch": 0.7486998581663454,
+      "grad_norm": 0.0,
+      "learning_rate": 3.133332797713089e-06,
+      "loss": 0.902,
+      "step": 20587
+    },
+    {
+      "epoch": 0.748736225770084,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1324765724651683e-06,
+      "loss": 1.0526,
+      "step": 20588
+    },
+    {
+      "epoch": 0.7487725933738226,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1316204424943907e-06,
+      "loss": 0.8917,
+      "step": 20589
+    },
+    {
+      "epoch": 0.7488089609775612,
+      "grad_norm": 0.0,
+      "learning_rate": 3.130764407812632e-06,
+      "loss": 1.0206,
+      "step": 20590
+    },
+    {
+      "epoch": 0.7488453285812998,
+      "grad_norm": 0.0,
+      "learning_rate": 3.129908468431774e-06,
+      "loss": 0.9532,
+      "step": 20591
+    },
+    {
+      "epoch": 0.7488816961850384,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1290526243636888e-06,
+      "loss": 1.0418,
+      "step": 20592
+    },
+    {
+      "epoch": 0.748918063788777,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1281968756202498e-06,
+      "loss": 0.9158,
+      "step": 20593
+    },
+    {
+      "epoch": 0.7489544313925155,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1273412222133294e-06,
+      "loss": 1.0583,
+      "step": 20594
+    },
+    {
+      "epoch": 0.7489907989962541,
+      "grad_norm": 0.0,
+      "learning_rate": 3.126485664154798e-06,
+      "loss": 0.989,
+      "step": 20595
+    },
+    {
+      "epoch": 0.7490271665999927,
+      "grad_norm": 0.0,
+      "learning_rate": 3.125630201456522e-06,
+      "loss": 0.8748,
+      "step": 20596
+    },
+    {
+      "epoch": 0.7490635342037313,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1247748341303784e-06,
+      "loss": 0.9221,
+      "step": 20597
+    },
+    {
+      "epoch": 0.7490999018074699,
+      "grad_norm": 0.0,
+      "learning_rate": 3.123919562188228e-06,
+      "loss": 0.785,
+      "step": 20598
+    },
+    {
+      "epoch": 0.7491362694112085,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1230643856419384e-06,
+      "loss": 0.8709,
+      "step": 20599
+    },
+    {
+      "epoch": 0.7491726370149471,
+      "grad_norm": 0.0,
+      "learning_rate": 3.122209304503373e-06,
+      "loss": 0.9883,
+      "step": 20600
+    },
+    {
+      "epoch": 0.7492090046186857,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1213543187843953e-06,
+      "loss": 0.9621,
+      "step": 20601
+    },
+    {
+      "epoch": 0.7492453722224243,
+      "grad_norm": 0.0,
+      "learning_rate": 3.120499428496867e-06,
+      "loss": 0.8604,
+      "step": 20602
+    },
+    {
+      "epoch": 0.7492817398261629,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1196446336526454e-06,
+      "loss": 0.8402,
+      "step": 20603
+    },
+    {
+      "epoch": 0.7493181074299015,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1187899342635986e-06,
+      "loss": 0.8595,
+      "step": 20604
+    },
+    {
+      "epoch": 0.74935447503364,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1179353303415727e-06,
+      "loss": 0.7889,
+      "step": 20605
+    },
+    {
+      "epoch": 0.7493908426373787,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1170808218984317e-06,
+      "loss": 0.8722,
+      "step": 20606
+    },
+    {
+      "epoch": 0.7494272102411172,
+      "grad_norm": 0.0,
+      "learning_rate": 3.116226408946029e-06,
+      "loss": 1.0094,
+      "step": 20607
+    },
+    {
+      "epoch": 0.7494635778448558,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1153720914962194e-06,
+      "loss": 0.8731,
+      "step": 20608
+    },
+    {
+      "epoch": 0.7494999454485944,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1145178695608492e-06,
+      "loss": 0.9584,
+      "step": 20609
+    },
+    {
+      "epoch": 0.749536313052333,
+      "grad_norm": 0.0,
+      "learning_rate": 3.113663743151778e-06,
+      "loss": 0.8247,
+      "step": 20610
+    },
+    {
+      "epoch": 0.7495726806560715,
+      "grad_norm": 0.0,
+      "learning_rate": 3.112809712280852e-06,
+      "loss": 0.9838,
+      "step": 20611
+    },
+    {
+      "epoch": 0.7496090482598101,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1119557769599197e-06,
+      "loss": 1.0145,
+      "step": 20612
+    },
+    {
+      "epoch": 0.7496454158635487,
+      "grad_norm": 0.0,
+      "learning_rate": 3.111101937200829e-06,
+      "loss": 0.9489,
+      "step": 20613
+    },
+    {
+      "epoch": 0.7496817834672873,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1102481930154237e-06,
+      "loss": 0.9632,
+      "step": 20614
+    },
+    {
+      "epoch": 0.7497181510710259,
+      "grad_norm": 0.0,
+      "learning_rate": 3.109394544415546e-06,
+      "loss": 0.7993,
+      "step": 20615
+    },
+    {
+      "epoch": 0.7497545186747645,
+      "grad_norm": 0.0,
+      "learning_rate": 3.108540991413046e-06,
+      "loss": 0.9608,
+      "step": 20616
+    },
+    {
+      "epoch": 0.7497908862785031,
+      "grad_norm": 0.0,
+      "learning_rate": 3.107687534019762e-06,
+      "loss": 0.9594,
+      "step": 20617
+    },
+    {
+      "epoch": 0.7498272538822417,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1068341722475347e-06,
+      "loss": 1.0194,
+      "step": 20618
+    },
+    {
+      "epoch": 0.7498636214859803,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1059809061082038e-06,
+      "loss": 1.0043,
+      "step": 20619
+    },
+    {
+      "epoch": 0.7498999890897189,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1051277356136055e-06,
+      "loss": 0.9184,
+      "step": 20620
+    },
+    {
+      "epoch": 0.7499363566934575,
+      "grad_norm": 0.0,
+      "learning_rate": 3.104274660775578e-06,
+      "loss": 0.8789,
+      "step": 20621
+    },
+    {
+      "epoch": 0.7499727242971961,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1034216816059525e-06,
+      "loss": 1.0495,
+      "step": 20622
+    },
+    {
+      "epoch": 0.7500090919009347,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1025687981165698e-06,
+      "loss": 0.9331,
+      "step": 20623
+    },
+    {
+      "epoch": 0.7500454595046733,
+      "grad_norm": 0.0,
+      "learning_rate": 3.101716010319258e-06,
+      "loss": 1.0111,
+      "step": 20624
+    },
+    {
+      "epoch": 0.7500818271084119,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1008633182258494e-06,
+      "loss": 0.8528,
+      "step": 20625
+    },
+    {
+      "epoch": 0.7501181947121505,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1000107218481733e-06,
+      "loss": 0.9412,
+      "step": 20626
+    },
+    {
+      "epoch": 0.750154562315889,
+      "grad_norm": 0.0,
+      "learning_rate": 3.09915822119806e-06,
+      "loss": 0.8402,
+      "step": 20627
+    },
+    {
+      "epoch": 0.7501909299196275,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0983058162873313e-06,
+      "loss": 0.889,
+      "step": 20628
+    },
+    {
+      "epoch": 0.7502272975233661,
+      "grad_norm": 0.0,
+      "learning_rate": 3.09745350712782e-06,
+      "loss": 0.9575,
+      "step": 20629
+    },
+    {
+      "epoch": 0.7502636651271047,
+      "grad_norm": 0.0,
+      "learning_rate": 3.096601293731348e-06,
+      "loss": 0.9312,
+      "step": 20630
+    },
+    {
+      "epoch": 0.7503000327308433,
+      "grad_norm": 0.0,
+      "learning_rate": 3.095749176109739e-06,
+      "loss": 0.9082,
+      "step": 20631
+    },
+    {
+      "epoch": 0.7503364003345819,
+      "grad_norm": 0.0,
+      "learning_rate": 3.094897154274811e-06,
+      "loss": 0.9769,
+      "step": 20632
+    },
+    {
+      "epoch": 0.7503727679383205,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0940452282383937e-06,
+      "loss": 0.9742,
+      "step": 20633
+    },
+    {
+      "epoch": 0.7504091355420591,
+      "grad_norm": 0.0,
+      "learning_rate": 3.093193398012294e-06,
+      "loss": 0.9655,
+      "step": 20634
+    },
+    {
+      "epoch": 0.7504455031457977,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0923416636083404e-06,
+      "loss": 0.9132,
+      "step": 20635
+    },
+    {
+      "epoch": 0.7504818707495363,
+      "grad_norm": 0.0,
+      "learning_rate": 3.091490025038345e-06,
+      "loss": 0.7933,
+      "step": 20636
+    },
+    {
+      "epoch": 0.7505182383532749,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0906384823141244e-06,
+      "loss": 0.8806,
+      "step": 20637
+    },
+    {
+      "epoch": 0.7505546059570135,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0897870354474914e-06,
+      "loss": 0.8793,
+      "step": 20638
+    },
+    {
+      "epoch": 0.7505909735607521,
+      "grad_norm": 0.0,
+      "learning_rate": 3.088935684450258e-06,
+      "loss": 1.0069,
+      "step": 20639
+    },
+    {
+      "epoch": 0.7506273411644907,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0880844293342382e-06,
+      "loss": 1.0171,
+      "step": 20640
+    },
+    {
+      "epoch": 0.7506637087682293,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0872332701112362e-06,
+      "loss": 1.0689,
+      "step": 20641
+    },
+    {
+      "epoch": 0.7507000763719679,
+      "grad_norm": 0.0,
+      "learning_rate": 3.086382206793067e-06,
+      "loss": 0.8835,
+      "step": 20642
+    },
+    {
+      "epoch": 0.7507364439757065,
+      "grad_norm": 0.0,
+      "learning_rate": 3.085531239391537e-06,
+      "loss": 0.938,
+      "step": 20643
+    },
+    {
+      "epoch": 0.750772811579445,
+      "grad_norm": 0.0,
+      "learning_rate": 3.08468036791845e-06,
+      "loss": 0.9736,
+      "step": 20644
+    },
+    {
+      "epoch": 0.7508091791831836,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0838295923856077e-06,
+      "loss": 0.8802,
+      "step": 20645
+    },
+    {
+      "epoch": 0.7508455467869222,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0829789128048237e-06,
+      "loss": 0.9658,
+      "step": 20646
+    },
+    {
+      "epoch": 0.7508819143906608,
+      "grad_norm": 0.0,
+      "learning_rate": 3.082128329187888e-06,
+      "loss": 0.9336,
+      "step": 20647
+    },
+    {
+      "epoch": 0.7509182819943994,
+      "grad_norm": 0.0,
+      "learning_rate": 3.081277841546608e-06,
+      "loss": 0.9456,
+      "step": 20648
+    },
+    {
+      "epoch": 0.750954649598138,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0804274498927834e-06,
+      "loss": 0.989,
+      "step": 20649
+    },
+    {
+      "epoch": 0.7509910172018766,
+      "grad_norm": 0.0,
+      "learning_rate": 3.079577154238209e-06,
+      "loss": 0.9147,
+      "step": 20650
+    },
+    {
+      "epoch": 0.7510273848056152,
+      "grad_norm": 0.0,
+      "learning_rate": 3.07872695459468e-06,
+      "loss": 0.8493,
+      "step": 20651
+    },
+    {
+      "epoch": 0.7510637524093537,
+      "grad_norm": 0.0,
+      "learning_rate": 3.077876850974001e-06,
+      "loss": 1.0227,
+      "step": 20652
+    },
+    {
+      "epoch": 0.7511001200130923,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0770268433879525e-06,
+      "loss": 0.9928,
+      "step": 20653
+    },
+    {
+      "epoch": 0.7511364876168309,
+      "grad_norm": 0.0,
+      "learning_rate": 3.076176931848338e-06,
+      "loss": 0.946,
+      "step": 20654
+    },
+    {
+      "epoch": 0.7511728552205695,
+      "grad_norm": 0.0,
+      "learning_rate": 3.075327116366944e-06,
+      "loss": 0.8071,
+      "step": 20655
+    },
+    {
+      "epoch": 0.7512092228243081,
+      "grad_norm": 0.0,
+      "learning_rate": 3.074477396955563e-06,
+      "loss": 0.8527,
+      "step": 20656
+    },
+    {
+      "epoch": 0.7512455904280467,
+      "grad_norm": 0.0,
+      "learning_rate": 3.073627773625981e-06,
+      "loss": 0.9382,
+      "step": 20657
+    },
+    {
+      "epoch": 0.7512819580317853,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0727782463899868e-06,
+      "loss": 1.0233,
+      "step": 20658
+    },
+    {
+      "epoch": 0.7513183256355239,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0719288152593664e-06,
+      "loss": 0.974,
+      "step": 20659
+    },
+    {
+      "epoch": 0.7513546932392625,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0710794802459e-06,
+      "loss": 0.8854,
+      "step": 20660
+    },
+    {
+      "epoch": 0.751391060843001,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0702302413613783e-06,
+      "loss": 0.9109,
+      "step": 20661
+    },
+    {
+      "epoch": 0.7514274284467396,
+      "grad_norm": 0.0,
+      "learning_rate": 3.06938109861758e-06,
+      "loss": 1.0875,
+      "step": 20662
+    },
+    {
+      "epoch": 0.7514637960504782,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0685320520262853e-06,
+      "loss": 0.8852,
+      "step": 20663
+    },
+    {
+      "epoch": 0.7515001636542168,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0676831015992714e-06,
+      "loss": 0.9568,
+      "step": 20664
+    },
+    {
+      "epoch": 0.7515365312579554,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0668342473483248e-06,
+      "loss": 0.9723,
+      "step": 20665
+    },
+    {
+      "epoch": 0.751572898861694,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0659854892852105e-06,
+      "loss": 0.921,
+      "step": 20666
+    },
+    {
+      "epoch": 0.7516092664654326,
+      "grad_norm": 0.0,
+      "learning_rate": 3.065136827421712e-06,
+      "loss": 0.9368,
+      "step": 20667
+    },
+    {
+      "epoch": 0.7516456340691712,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0642882617696003e-06,
+      "loss": 0.8766,
+      "step": 20668
+    },
+    {
+      "epoch": 0.7516820016729098,
+      "grad_norm": 0.0,
+      "learning_rate": 3.063439792340649e-06,
+      "loss": 0.9466,
+      "step": 20669
+    },
+    {
+      "epoch": 0.7517183692766484,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0625914191466265e-06,
+      "loss": 0.992,
+      "step": 20670
+    },
+    {
+      "epoch": 0.751754736880387,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0617431421993105e-06,
+      "loss": 0.8411,
+      "step": 20671
+    },
+    {
+      "epoch": 0.7517911044841256,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0608949615104577e-06,
+      "loss": 0.9146,
+      "step": 20672
+    },
+    {
+      "epoch": 0.7518274720878642,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0600468770918455e-06,
+      "loss": 0.8739,
+      "step": 20673
+    },
+    {
+      "epoch": 0.7518638396916028,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0591988889552372e-06,
+      "loss": 0.8735,
+      "step": 20674
+    },
+    {
+      "epoch": 0.7519002072953413,
+      "grad_norm": 0.0,
+      "learning_rate": 3.058350997112396e-06,
+      "loss": 0.8231,
+      "step": 20675
+    },
+    {
+      "epoch": 0.75193657489908,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0575032015750863e-06,
+      "loss": 0.9095,
+      "step": 20676
+    },
+    {
+      "epoch": 0.7519729425028185,
+      "grad_norm": 0.0,
+      "learning_rate": 3.056655502355065e-06,
+      "loss": 0.9323,
+      "step": 20677
+    },
+    {
+      "epoch": 0.752009310106557,
+      "grad_norm": 0.0,
+      "learning_rate": 3.055807899464105e-06,
+      "loss": 0.8051,
+      "step": 20678
+    },
+    {
+      "epoch": 0.7520456777102956,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0549603929139505e-06,
+      "loss": 0.9962,
+      "step": 20679
+    },
+    {
+      "epoch": 0.7520820453140342,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0541129827163705e-06,
+      "loss": 0.927,
+      "step": 20680
+    },
+    {
+      "epoch": 0.7521184129177728,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0532656688831174e-06,
+      "loss": 0.7365,
+      "step": 20681
+    },
+    {
+      "epoch": 0.7521547805215114,
+      "grad_norm": 0.0,
+      "learning_rate": 3.052418451425948e-06,
+      "loss": 0.8947,
+      "step": 20682
+    },
+    {
+      "epoch": 0.75219114812525,
+      "grad_norm": 0.0,
+      "learning_rate": 3.051571330356612e-06,
+      "loss": 0.9916,
+      "step": 20683
+    },
+    {
+      "epoch": 0.7522275157289886,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0507243056868707e-06,
+      "loss": 0.9441,
+      "step": 20684
+    },
+    {
+      "epoch": 0.7522638833327272,
+      "grad_norm": 0.0,
+      "learning_rate": 3.049877377428464e-06,
+      "loss": 0.9552,
+      "step": 20685
+    },
+    {
+      "epoch": 0.7523002509364658,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0490305455931514e-06,
+      "loss": 0.8384,
+      "step": 20686
+    },
+    {
+      "epoch": 0.7523366185402044,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0481838101926774e-06,
+      "loss": 0.858,
+      "step": 20687
+    },
+    {
+      "epoch": 0.752372986143943,
+      "grad_norm": 0.0,
+      "learning_rate": 3.04733717123879e-06,
+      "loss": 0.9662,
+      "step": 20688
+    },
+    {
+      "epoch": 0.7524093537476816,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0464906287432315e-06,
+      "loss": 0.8883,
+      "step": 20689
+    },
+    {
+      "epoch": 0.7524457213514202,
+      "grad_norm": 0.0,
+      "learning_rate": 3.045644182717753e-06,
+      "loss": 0.9151,
+      "step": 20690
+    },
+    {
+      "epoch": 0.7524820889551588,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0447978331740946e-06,
+      "loss": 0.9337,
+      "step": 20691
+    },
+    {
+      "epoch": 0.7525184565588974,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0439515801239973e-06,
+      "loss": 0.9256,
+      "step": 20692
+    },
+    {
+      "epoch": 0.752554824162636,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0431054235792033e-06,
+      "loss": 0.9241,
+      "step": 20693
+    },
+    {
+      "epoch": 0.7525911917663746,
+      "grad_norm": 0.0,
+      "learning_rate": 3.042259363551451e-06,
+      "loss": 1.0429,
+      "step": 20694
+    },
+    {
+      "epoch": 0.752627559370113,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0414134000524785e-06,
+      "loss": 0.8721,
+      "step": 20695
+    },
+    {
+      "epoch": 0.7526639269738516,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0405675330940186e-06,
+      "loss": 0.926,
+      "step": 20696
+    },
+    {
+      "epoch": 0.7527002945775902,
+      "grad_norm": 0.0,
+      "learning_rate": 3.039721762687816e-06,
+      "loss": 1.0288,
+      "step": 20697
+    },
+    {
+      "epoch": 0.7527366621813288,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0388760888455925e-06,
+      "loss": 0.7534,
+      "step": 20698
+    },
+    {
+      "epoch": 0.7527730297850674,
+      "grad_norm": 0.0,
+      "learning_rate": 3.03803051157909e-06,
+      "loss": 0.9089,
+      "step": 20699
+    },
+    {
+      "epoch": 0.752809397388806,
+      "grad_norm": 0.0,
+      "learning_rate": 3.037185030900036e-06,
+      "loss": 0.981,
+      "step": 20700
+    },
+    {
+      "epoch": 0.7528457649925446,
+      "grad_norm": 0.0,
+      "learning_rate": 3.036339646820162e-06,
+      "loss": 0.9549,
+      "step": 20701
+    },
+    {
+      "epoch": 0.7528821325962832,
+      "grad_norm": 0.0,
+      "learning_rate": 3.035494359351191e-06,
+      "loss": 0.968,
+      "step": 20702
+    },
+    {
+      "epoch": 0.7529185002000218,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0346491685048606e-06,
+      "loss": 0.9187,
+      "step": 20703
+    },
+    {
+      "epoch": 0.7529548678037604,
+      "grad_norm": 0.0,
+      "learning_rate": 3.033804074292885e-06,
+      "loss": 1.0003,
+      "step": 20704
+    },
+    {
+      "epoch": 0.752991235407499,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0329590767269967e-06,
+      "loss": 0.9484,
+      "step": 20705
+    },
+    {
+      "epoch": 0.7530276030112376,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0321141758189166e-06,
+      "loss": 0.968,
+      "step": 20706
+    },
+    {
+      "epoch": 0.7530639706149762,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0312693715803654e-06,
+      "loss": 0.891,
+      "step": 20707
+    },
+    {
+      "epoch": 0.7531003382187148,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0304246640230617e-06,
+      "loss": 0.9617,
+      "step": 20708
+    },
+    {
+      "epoch": 0.7531367058224534,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0295800531587303e-06,
+      "loss": 0.8734,
+      "step": 20709
+    },
+    {
+      "epoch": 0.753173073426192,
+      "grad_norm": 0.0,
+      "learning_rate": 3.028735538999087e-06,
+      "loss": 0.9656,
+      "step": 20710
+    },
+    {
+      "epoch": 0.7532094410299306,
+      "grad_norm": 0.0,
+      "learning_rate": 3.027891121555846e-06,
+      "loss": 0.9837,
+      "step": 20711
+    },
+    {
+      "epoch": 0.7532458086336691,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0270468008407204e-06,
+      "loss": 0.8257,
+      "step": 20712
+    },
+    {
+      "epoch": 0.7532821762374077,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0262025768654334e-06,
+      "loss": 0.9363,
+      "step": 20713
+    },
+    {
+      "epoch": 0.7533185438411463,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0253584496416853e-06,
+      "loss": 0.8837,
+      "step": 20714
+    },
+    {
+      "epoch": 0.7533549114448849,
+      "grad_norm": 0.0,
+      "learning_rate": 3.024514419181196e-06,
+      "loss": 0.8287,
+      "step": 20715
+    },
+    {
+      "epoch": 0.7533912790486235,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0236704854956756e-06,
+      "loss": 0.9128,
+      "step": 20716
+    },
+    {
+      "epoch": 0.7534276466523621,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0228266485968226e-06,
+      "loss": 1.0117,
+      "step": 20717
+    },
+    {
+      "epoch": 0.7534640142561007,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0219829084963536e-06,
+      "loss": 0.9537,
+      "step": 20718
+    },
+    {
+      "epoch": 0.7535003818598393,
+      "grad_norm": 0.0,
+      "learning_rate": 3.021139265205971e-06,
+      "loss": 0.9847,
+      "step": 20719
+    },
+    {
+      "epoch": 0.7535367494635778,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0202957187373795e-06,
+      "loss": 0.8368,
+      "step": 20720
+    },
+    {
+      "epoch": 0.7535731170673164,
+      "grad_norm": 0.0,
+      "learning_rate": 3.019452269102279e-06,
+      "loss": 0.9937,
+      "step": 20721
+    },
+    {
+      "epoch": 0.753609484671055,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0186089163123776e-06,
+      "loss": 0.9651,
+      "step": 20722
+    },
+    {
+      "epoch": 0.7536458522747936,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0177656603793713e-06,
+      "loss": 0.8748,
+      "step": 20723
+    },
+    {
+      "epoch": 0.7536822198785322,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0169225013149604e-06,
+      "loss": 0.9298,
+      "step": 20724
+    },
+    {
+      "epoch": 0.7537185874822708,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0160794391308423e-06,
+      "loss": 0.8968,
+      "step": 20725
+    },
+    {
+      "epoch": 0.7537549550860094,
+      "grad_norm": 0.0,
+      "learning_rate": 3.015236473838713e-06,
+      "loss": 0.8596,
+      "step": 20726
+    },
+    {
+      "epoch": 0.753791322689748,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0143936054502652e-06,
+      "loss": 0.8771,
+      "step": 20727
+    },
+    {
+      "epoch": 0.7538276902934865,
+      "grad_norm": 0.0,
+      "learning_rate": 3.013550833977197e-06,
+      "loss": 0.9296,
+      "step": 20728
+    },
+    {
+      "epoch": 0.7538640578972251,
+      "grad_norm": 0.0,
+      "learning_rate": 3.012708159431199e-06,
+      "loss": 0.8694,
+      "step": 20729
+    },
+    {
+      "epoch": 0.7539004255009637,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0118655818239605e-06,
+      "loss": 0.9604,
+      "step": 20730
+    },
+    {
+      "epoch": 0.7539367931047023,
+      "grad_norm": 0.0,
+      "learning_rate": 3.01102310116717e-06,
+      "loss": 0.884,
+      "step": 20731
+    },
+    {
+      "epoch": 0.7539731607084409,
+      "grad_norm": 0.0,
+      "learning_rate": 3.010180717472524e-06,
+      "loss": 0.9931,
+      "step": 20732
+    },
+    {
+      "epoch": 0.7540095283121795,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0093384307516973e-06,
+      "loss": 0.8304,
+      "step": 20733
+    },
+    {
+      "epoch": 0.7540458959159181,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0084962410163845e-06,
+      "loss": 0.9604,
+      "step": 20734
+    },
+    {
+      "epoch": 0.7540822635196567,
+      "grad_norm": 0.0,
+      "learning_rate": 3.007654148278266e-06,
+      "loss": 0.8394,
+      "step": 20735
+    },
+    {
+      "epoch": 0.7541186311233953,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0068121525490266e-06,
+      "loss": 0.9654,
+      "step": 20736
+    },
+    {
+      "epoch": 0.7541549987271339,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0059702538403457e-06,
+      "loss": 1.0178,
+      "step": 20737
+    },
+    {
+      "epoch": 0.7541913663308725,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0051284521639044e-06,
+      "loss": 0.9746,
+      "step": 20738
+    },
+    {
+      "epoch": 0.7542277339346111,
+      "grad_norm": 0.0,
+      "learning_rate": 3.004286747531382e-06,
+      "loss": 0.8052,
+      "step": 20739
+    },
+    {
+      "epoch": 0.7542641015383497,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0034451399544506e-06,
+      "loss": 0.9156,
+      "step": 20740
+    },
+    {
+      "epoch": 0.7543004691420883,
+      "grad_norm": 0.0,
+      "learning_rate": 3.002603629444796e-06,
+      "loss": 0.8201,
+      "step": 20741
+    },
+    {
+      "epoch": 0.7543368367458269,
+      "grad_norm": 0.0,
+      "learning_rate": 3.001762216014087e-06,
+      "loss": 0.9335,
+      "step": 20742
+    },
+    {
+      "epoch": 0.7543732043495655,
+      "grad_norm": 0.0,
+      "learning_rate": 3.000920899673998e-06,
+      "loss": 0.9889,
+      "step": 20743
+    },
+    {
+      "epoch": 0.754409571953304,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0000796804361985e-06,
+      "loss": 0.883,
+      "step": 20744
+    },
+    {
+      "epoch": 0.7544459395570425,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9992385583123683e-06,
+      "loss": 0.9471,
+      "step": 20745
+    },
+    {
+      "epoch": 0.7544823071607811,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9983975333141635e-06,
+      "loss": 0.937,
+      "step": 20746
+    },
+    {
+      "epoch": 0.7545186747645197,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9975566054532625e-06,
+      "loss": 1.0142,
+      "step": 20747
+    },
+    {
+      "epoch": 0.7545550423682583,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9967157747413278e-06,
+      "loss": 0.9361,
+      "step": 20748
+    },
+    {
+      "epoch": 0.7545914099719969,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9958750411900263e-06,
+      "loss": 0.8853,
+      "step": 20749
+    },
+    {
+      "epoch": 0.7546277775757355,
+      "grad_norm": 0.0,
+      "learning_rate": 2.995034404811017e-06,
+      "loss": 0.9792,
+      "step": 20750
+    },
+    {
+      "epoch": 0.7546641451794741,
+      "grad_norm": 0.0,
+      "learning_rate": 2.994193865615973e-06,
+      "loss": 1.1414,
+      "step": 20751
+    },
+    {
+      "epoch": 0.7547005127832127,
+      "grad_norm": 0.0,
+      "learning_rate": 2.993353423616544e-06,
+      "loss": 0.9147,
+      "step": 20752
+    },
+    {
+      "epoch": 0.7547368803869513,
+      "grad_norm": 0.0,
+      "learning_rate": 2.992513078824397e-06,
+      "loss": 1.0285,
+      "step": 20753
+    },
+    {
+      "epoch": 0.7547732479906899,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9916728312511888e-06,
+      "loss": 0.974,
+      "step": 20754
+    },
+    {
+      "epoch": 0.7548096155944285,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9908326809085773e-06,
+      "loss": 0.975,
+      "step": 20755
+    },
+    {
+      "epoch": 0.7548459831981671,
+      "grad_norm": 0.0,
+      "learning_rate": 2.989992627808217e-06,
+      "loss": 0.9678,
+      "step": 20756
+    },
+    {
+      "epoch": 0.7548823508019057,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9891526719617636e-06,
+      "loss": 0.942,
+      "step": 20757
+    },
+    {
+      "epoch": 0.7549187184056443,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9883128133808693e-06,
+      "loss": 0.9512,
+      "step": 20758
+    },
+    {
+      "epoch": 0.7549550860093829,
+      "grad_norm": 0.0,
+      "learning_rate": 2.987473052077183e-06,
+      "loss": 0.9172,
+      "step": 20759
+    },
+    {
+      "epoch": 0.7549914536131215,
+      "grad_norm": 0.0,
+      "learning_rate": 2.986633388062363e-06,
+      "loss": 0.9591,
+      "step": 20760
+    },
+    {
+      "epoch": 0.7550278212168601,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9857938213480532e-06,
+      "loss": 0.9643,
+      "step": 20761
+    },
+    {
+      "epoch": 0.7550641888205986,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9849543519459023e-06,
+      "loss": 0.9606,
+      "step": 20762
+    },
+    {
+      "epoch": 0.7551005564243372,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9841149798675538e-06,
+      "loss": 0.9391,
+      "step": 20763
+    },
+    {
+      "epoch": 0.7551369240280758,
+      "grad_norm": 0.0,
+      "learning_rate": 2.983275705124662e-06,
+      "loss": 1.0324,
+      "step": 20764
+    },
+    {
+      "epoch": 0.7551732916318143,
+      "grad_norm": 0.0,
+      "learning_rate": 2.982436527728858e-06,
+      "loss": 0.9953,
+      "step": 20765
+    },
+    {
+      "epoch": 0.7552096592355529,
+      "grad_norm": 0.0,
+      "learning_rate": 2.981597447691793e-06,
+      "loss": 0.9303,
+      "step": 20766
+    },
+    {
+      "epoch": 0.7552460268392915,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9807584650251064e-06,
+      "loss": 0.9371,
+      "step": 20767
+    },
+    {
+      "epoch": 0.7552823944430301,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9799195797404377e-06,
+      "loss": 1.0019,
+      "step": 20768
+    },
+    {
+      "epoch": 0.7553187620467687,
+      "grad_norm": 0.0,
+      "learning_rate": 2.97908079184942e-06,
+      "loss": 0.976,
+      "step": 20769
+    },
+    {
+      "epoch": 0.7553551296505073,
+      "grad_norm": 0.0,
+      "learning_rate": 2.978242101363702e-06,
+      "loss": 0.9875,
+      "step": 20770
+    },
+    {
+      "epoch": 0.7553914972542459,
+      "grad_norm": 0.0,
+      "learning_rate": 2.977403508294906e-06,
+      "loss": 0.9811,
+      "step": 20771
+    },
+    {
+      "epoch": 0.7554278648579845,
+      "grad_norm": 0.0,
+      "learning_rate": 2.976565012654676e-06,
+      "loss": 0.9164,
+      "step": 20772
+    },
+    {
+      "epoch": 0.7554642324617231,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9757266144546415e-06,
+      "loss": 0.8498,
+      "step": 20773
+    },
+    {
+      "epoch": 0.7555006000654617,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9748883137064335e-06,
+      "loss": 0.938,
+      "step": 20774
+    },
+    {
+      "epoch": 0.7555369676692003,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9740501104216836e-06,
+      "loss": 1.0243,
+      "step": 20775
+    },
+    {
+      "epoch": 0.7555733352729389,
+      "grad_norm": 0.0,
+      "learning_rate": 2.973212004612016e-06,
+      "loss": 0.8103,
+      "step": 20776
+    },
+    {
+      "epoch": 0.7556097028766775,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9723739962890684e-06,
+      "loss": 1.0167,
+      "step": 20777
+    },
+    {
+      "epoch": 0.7556460704804161,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9715360854644546e-06,
+      "loss": 0.9364,
+      "step": 20778
+    },
+    {
+      "epoch": 0.7556824380841546,
+      "grad_norm": 0.0,
+      "learning_rate": 2.970698272149809e-06,
+      "loss": 0.8066,
+      "step": 20779
+    },
+    {
+      "epoch": 0.7557188056878932,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9698605563567506e-06,
+      "loss": 0.9606,
+      "step": 20780
+    },
+    {
+      "epoch": 0.7557551732916318,
+      "grad_norm": 0.0,
+      "learning_rate": 2.969022938096904e-06,
+      "loss": 0.969,
+      "step": 20781
+    },
+    {
+      "epoch": 0.7557915408953704,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9681854173818845e-06,
+      "loss": 0.9645,
+      "step": 20782
+    },
+    {
+      "epoch": 0.755827908499109,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9673479942233217e-06,
+      "loss": 0.9747,
+      "step": 20783
+    },
+    {
+      "epoch": 0.7558642761028476,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9665106686328217e-06,
+      "loss": 0.9111,
+      "step": 20784
+    },
+    {
+      "epoch": 0.7559006437065862,
+      "grad_norm": 0.0,
+      "learning_rate": 2.965673440622009e-06,
+      "loss": 0.8961,
+      "step": 20785
+    },
+    {
+      "epoch": 0.7559370113103248,
+      "grad_norm": 0.0,
+      "learning_rate": 2.964836310202498e-06,
+      "loss": 0.969,
+      "step": 20786
+    },
+    {
+      "epoch": 0.7559733789140634,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9639992773859016e-06,
+      "loss": 0.9191,
+      "step": 20787
+    },
+    {
+      "epoch": 0.756009746517802,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9631623421838284e-06,
+      "loss": 0.8047,
+      "step": 20788
+    },
+    {
+      "epoch": 0.7560461141215405,
+      "grad_norm": 0.0,
+      "learning_rate": 2.962325504607897e-06,
+      "loss": 0.8845,
+      "step": 20789
+    },
+    {
+      "epoch": 0.7560824817252791,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9614887646697153e-06,
+      "loss": 0.8409,
+      "step": 20790
+    },
+    {
+      "epoch": 0.7561188493290177,
+      "grad_norm": 0.0,
+      "learning_rate": 2.960652122380889e-06,
+      "loss": 0.9049,
+      "step": 20791
+    },
+    {
+      "epoch": 0.7561552169327563,
+      "grad_norm": 0.0,
+      "learning_rate": 2.959815577753028e-06,
+      "loss": 0.7913,
+      "step": 20792
+    },
+    {
+      "epoch": 0.7561915845364949,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9589791307977376e-06,
+      "loss": 0.8509,
+      "step": 20793
+    },
+    {
+      "epoch": 0.7562279521402335,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9581427815266205e-06,
+      "loss": 0.9859,
+      "step": 20794
+    },
+    {
+      "epoch": 0.7562643197439721,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9573065299512793e-06,
+      "loss": 0.9842,
+      "step": 20795
+    },
+    {
+      "epoch": 0.7563006873477106,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9564703760833236e-06,
+      "loss": 0.6795,
+      "step": 20796
+    },
+    {
+      "epoch": 0.7563370549514492,
+      "grad_norm": 0.0,
+      "learning_rate": 2.955634319934342e-06,
+      "loss": 1.0229,
+      "step": 20797
+    },
+    {
+      "epoch": 0.7563734225551878,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9547983615159413e-06,
+      "loss": 0.8229,
+      "step": 20798
+    },
+    {
+      "epoch": 0.7564097901589264,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9539625008397187e-06,
+      "loss": 0.8724,
+      "step": 20799
+    },
+    {
+      "epoch": 0.756446157762665,
+      "grad_norm": 0.0,
+      "learning_rate": 2.953126737917269e-06,
+      "loss": 0.9715,
+      "step": 20800
+    },
+    {
+      "epoch": 0.7564825253664036,
+      "grad_norm": 0.0,
+      "learning_rate": 2.952291072760185e-06,
+      "loss": 0.868,
+      "step": 20801
+    },
+    {
+      "epoch": 0.7565188929701422,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9514555053800686e-06,
+      "loss": 0.9948,
+      "step": 20802
+    },
+    {
+      "epoch": 0.7565552605738808,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9506200357885005e-06,
+      "loss": 0.9958,
+      "step": 20803
+    },
+    {
+      "epoch": 0.7565916281776194,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9497846639970807e-06,
+      "loss": 1.0249,
+      "step": 20804
+    },
+    {
+      "epoch": 0.756627995781358,
+      "grad_norm": 0.0,
+      "learning_rate": 2.948949390017396e-06,
+      "loss": 0.87,
+      "step": 20805
+    },
+    {
+      "epoch": 0.7566643633850966,
+      "grad_norm": 0.0,
+      "learning_rate": 2.948114213861034e-06,
+      "loss": 1.0145,
+      "step": 20806
+    },
+    {
+      "epoch": 0.7567007309888352,
+      "grad_norm": 0.0,
+      "learning_rate": 2.947279135539579e-06,
+      "loss": 1.0168,
+      "step": 20807
+    },
+    {
+      "epoch": 0.7567370985925738,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9464441550646227e-06,
+      "loss": 0.8708,
+      "step": 20808
+    },
+    {
+      "epoch": 0.7567734661963124,
+      "grad_norm": 0.0,
+      "learning_rate": 2.945609272447746e-06,
+      "loss": 0.911,
+      "step": 20809
+    },
+    {
+      "epoch": 0.756809833800051,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9447744877005323e-06,
+      "loss": 1.0037,
+      "step": 20810
+    },
+    {
+      "epoch": 0.7568462014037896,
+      "grad_norm": 0.0,
+      "learning_rate": 2.943939800834562e-06,
+      "loss": 0.9606,
+      "step": 20811
+    },
+    {
+      "epoch": 0.756882569007528,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9431052118614158e-06,
+      "loss": 0.921,
+      "step": 20812
+    },
+    {
+      "epoch": 0.7569189366112666,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9422707207926715e-06,
+      "loss": 0.9567,
+      "step": 20813
+    },
+    {
+      "epoch": 0.7569553042150052,
+      "grad_norm": 0.0,
+      "learning_rate": 2.941436327639906e-06,
+      "loss": 0.9212,
+      "step": 20814
+    },
+    {
+      "epoch": 0.7569916718187438,
+      "grad_norm": 0.0,
+      "learning_rate": 2.940602032414701e-06,
+      "loss": 0.9997,
+      "step": 20815
+    },
+    {
+      "epoch": 0.7570280394224824,
+      "grad_norm": 0.0,
+      "learning_rate": 2.939767835128622e-06,
+      "loss": 0.9668,
+      "step": 20816
+    },
+    {
+      "epoch": 0.757064407026221,
+      "grad_norm": 0.0,
+      "learning_rate": 2.93893373579325e-06,
+      "loss": 0.9949,
+      "step": 20817
+    },
+    {
+      "epoch": 0.7571007746299596,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9380997344201533e-06,
+      "loss": 1.0105,
+      "step": 20818
+    },
+    {
+      "epoch": 0.7571371422336982,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9372658310209036e-06,
+      "loss": 0.879,
+      "step": 20819
+    },
+    {
+      "epoch": 0.7571735098374368,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9364320256070676e-06,
+      "loss": 0.9806,
+      "step": 20820
+    },
+    {
+      "epoch": 0.7572098774411754,
+      "grad_norm": 0.0,
+      "learning_rate": 2.935598318190217e-06,
+      "loss": 0.9562,
+      "step": 20821
+    },
+    {
+      "epoch": 0.757246245044914,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9347647087819174e-06,
+      "loss": 0.8928,
+      "step": 20822
+    },
+    {
+      "epoch": 0.7572826126486526,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9339311973937335e-06,
+      "loss": 0.9253,
+      "step": 20823
+    },
+    {
+      "epoch": 0.7573189802523912,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9330977840372287e-06,
+      "loss": 0.8772,
+      "step": 20824
+    },
+    {
+      "epoch": 0.7573553478561298,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9322644687239656e-06,
+      "loss": 0.8585,
+      "step": 20825
+    },
+    {
+      "epoch": 0.7573917154598684,
+      "grad_norm": 0.0,
+      "learning_rate": 2.931431251465501e-06,
+      "loss": 0.8008,
+      "step": 20826
+    },
+    {
+      "epoch": 0.757428083063607,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9305981322734035e-06,
+      "loss": 0.9042,
+      "step": 20827
+    },
+    {
+      "epoch": 0.7574644506673456,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9297651111592253e-06,
+      "loss": 0.7787,
+      "step": 20828
+    },
+    {
+      "epoch": 0.7575008182710841,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.8986,
+      "step": 20829
+    },
+    {
+      "epoch": 0.7575371858748227,
+      "grad_norm": 0.0,
+      "learning_rate": 2.928099363210859e-06,
+      "loss": 0.9421,
+      "step": 20830
+    },
+    {
+      "epoch": 0.7575735534785613,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9272666363997805e-06,
+      "loss": 0.8911,
+      "step": 20831
+    },
+    {
+      "epoch": 0.7576099210822999,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9264340077128416e-06,
+      "loss": 0.8977,
+      "step": 20832
+    },
+    {
+      "epoch": 0.7576462886860384,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9256014771615924e-06,
+      "loss": 0.8315,
+      "step": 20833
+    },
+    {
+      "epoch": 0.757682656289777,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9247690447575883e-06,
+      "loss": 0.8582,
+      "step": 20834
+    },
+    {
+      "epoch": 0.7577190238935156,
+      "grad_norm": 0.0,
+      "learning_rate": 2.923936710512376e-06,
+      "loss": 0.9637,
+      "step": 20835
+    },
+    {
+      "epoch": 0.7577553914972542,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9231044744375014e-06,
+      "loss": 0.8845,
+      "step": 20836
+    },
+    {
+      "epoch": 0.7577917591009928,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9222723365445117e-06,
+      "loss": 0.8793,
+      "step": 20837
+    },
+    {
+      "epoch": 0.7578281267047314,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9214402968449507e-06,
+      "loss": 0.8732,
+      "step": 20838
+    },
+    {
+      "epoch": 0.75786449430847,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9206083553503607e-06,
+      "loss": 0.8658,
+      "step": 20839
+    },
+    {
+      "epoch": 0.7579008619122086,
+      "grad_norm": 0.0,
+      "learning_rate": 2.919776512072288e-06,
+      "loss": 0.9332,
+      "step": 20840
+    },
+    {
+      "epoch": 0.7579372295159472,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9189447670222693e-06,
+      "loss": 1.0034,
+      "step": 20841
+    },
+    {
+      "epoch": 0.7579735971196858,
+      "grad_norm": 0.0,
+      "learning_rate": 2.918113120211846e-06,
+      "loss": 1.0078,
+      "step": 20842
+    },
+    {
+      "epoch": 0.7580099647234244,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9172815716525527e-06,
+      "loss": 0.9834,
+      "step": 20843
+    },
+    {
+      "epoch": 0.758046332327163,
+      "grad_norm": 0.0,
+      "learning_rate": 2.916450121355935e-06,
+      "loss": 0.9444,
+      "step": 20844
+    },
+    {
+      "epoch": 0.7580826999309016,
+      "grad_norm": 0.0,
+      "learning_rate": 2.915618769333515e-06,
+      "loss": 0.8881,
+      "step": 20845
+    },
+    {
+      "epoch": 0.7581190675346401,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9147875155968365e-06,
+      "loss": 0.9544,
+      "step": 20846
+    },
+    {
+      "epoch": 0.7581554351383787,
+      "grad_norm": 0.0,
+      "learning_rate": 2.913956360157427e-06,
+      "loss": 1.0448,
+      "step": 20847
+    },
+    {
+      "epoch": 0.7581918027421173,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9131253030268214e-06,
+      "loss": 0.8656,
+      "step": 20848
+    },
+    {
+      "epoch": 0.7582281703458559,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9122943442165464e-06,
+      "loss": 1.0125,
+      "step": 20849
+    },
+    {
+      "epoch": 0.7582645379495945,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9114634837381305e-06,
+      "loss": 0.8646,
+      "step": 20850
+    },
+    {
+      "epoch": 0.7583009055533331,
+      "grad_norm": 0.0,
+      "learning_rate": 2.910632721603103e-06,
+      "loss": 0.9778,
+      "step": 20851
+    },
+    {
+      "epoch": 0.7583372731570717,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9098020578229834e-06,
+      "loss": 0.8292,
+      "step": 20852
+    },
+    {
+      "epoch": 0.7583736407608103,
+      "grad_norm": 0.0,
+      "learning_rate": 2.908971492409305e-06,
+      "loss": 0.8803,
+      "step": 20853
+    },
+    {
+      "epoch": 0.7584100083645489,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9081410253735854e-06,
+      "loss": 0.928,
+      "step": 20854
+    },
+    {
+      "epoch": 0.7584463759682875,
+      "grad_norm": 0.0,
+      "learning_rate": 2.907310656727347e-06,
+      "loss": 0.9742,
+      "step": 20855
+    },
+    {
+      "epoch": 0.758482743572026,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9064803864821077e-06,
+      "loss": 0.9789,
+      "step": 20856
+    },
+    {
+      "epoch": 0.7585191111757646,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9056502146493935e-06,
+      "loss": 0.8655,
+      "step": 20857
+    },
+    {
+      "epoch": 0.7585554787795032,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9048201412407118e-06,
+      "loss": 0.9164,
+      "step": 20858
+    },
+    {
+      "epoch": 0.7585918463832418,
+      "grad_norm": 0.0,
+      "learning_rate": 2.903990166267587e-06,
+      "loss": 0.9026,
+      "step": 20859
+    },
+    {
+      "epoch": 0.7586282139869804,
+      "grad_norm": 0.0,
+      "learning_rate": 2.903160289741531e-06,
+      "loss": 0.7886,
+      "step": 20860
+    },
+    {
+      "epoch": 0.758664581590719,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9023305116740565e-06,
+      "loss": 0.9239,
+      "step": 20861
+    },
+    {
+      "epoch": 0.7587009491944576,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9015008320766723e-06,
+      "loss": 0.8679,
+      "step": 20862
+    },
+    {
+      "epoch": 0.7587373167981961,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9006712509608993e-06,
+      "loss": 0.9283,
+      "step": 20863
+    },
+    {
+      "epoch": 0.7587736844019347,
+      "grad_norm": 0.0,
+      "learning_rate": 2.899841768338234e-06,
+      "loss": 0.8408,
+      "step": 20864
+    },
+    {
+      "epoch": 0.7588100520056733,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8990123842201936e-06,
+      "loss": 0.9613,
+      "step": 20865
+    },
+    {
+      "epoch": 0.7588464196094119,
+      "grad_norm": 0.0,
+      "learning_rate": 2.89818309861828e-06,
+      "loss": 0.748,
+      "step": 20866
+    },
+    {
+      "epoch": 0.7588827872131505,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8973539115440008e-06,
+      "loss": 0.9861,
+      "step": 20867
+    },
+    {
+      "epoch": 0.7589191548168891,
+      "grad_norm": 0.0,
+      "learning_rate": 2.896524823008859e-06,
+      "loss": 0.8799,
+      "step": 20868
+    },
+    {
+      "epoch": 0.7589555224206277,
+      "grad_norm": 0.0,
+      "learning_rate": 2.895695833024357e-06,
+      "loss": 0.9351,
+      "step": 20869
+    },
+    {
+      "epoch": 0.7589918900243663,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8948669416019946e-06,
+      "loss": 0.8915,
+      "step": 20870
+    },
+    {
+      "epoch": 0.7590282576281049,
+      "grad_norm": 0.0,
+      "learning_rate": 2.89403814875327e-06,
+      "loss": 0.9189,
+      "step": 20871
+    },
+    {
+      "epoch": 0.7590646252318435,
+      "grad_norm": 0.0,
+      "learning_rate": 2.893209454489687e-06,
+      "loss": 0.7711,
+      "step": 20872
+    },
+    {
+      "epoch": 0.7591009928355821,
+      "grad_norm": 0.0,
+      "learning_rate": 2.89238085882274e-06,
+      "loss": 0.942,
+      "step": 20873
+    },
+    {
+      "epoch": 0.7591373604393207,
+      "grad_norm": 0.0,
+      "learning_rate": 2.891552361763924e-06,
+      "loss": 0.9746,
+      "step": 20874
+    },
+    {
+      "epoch": 0.7591737280430593,
+      "grad_norm": 0.0,
+      "learning_rate": 2.89072396332473e-06,
+      "loss": 1.0233,
+      "step": 20875
+    },
+    {
+      "epoch": 0.7592100956467979,
+      "grad_norm": 0.0,
+      "learning_rate": 2.889895663516661e-06,
+      "loss": 0.9484,
+      "step": 20876
+    },
+    {
+      "epoch": 0.7592464632505365,
+      "grad_norm": 0.0,
+      "learning_rate": 2.889067462351195e-06,
+      "loss": 0.9587,
+      "step": 20877
+    },
+    {
+      "epoch": 0.7592828308542751,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8882393598398318e-06,
+      "loss": 0.972,
+      "step": 20878
+    },
+    {
+      "epoch": 0.7593191984580137,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8874113559940574e-06,
+      "loss": 0.8692,
+      "step": 20879
+    },
+    {
+      "epoch": 0.7593555660617521,
+      "grad_norm": 0.0,
+      "learning_rate": 2.88658345082536e-06,
+      "loss": 0.8566,
+      "step": 20880
+    },
+    {
+      "epoch": 0.7593919336654907,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8857556443452206e-06,
+      "loss": 0.9306,
+      "step": 20881
+    },
+    {
+      "epoch": 0.7594283012692293,
+      "grad_norm": 0.0,
+      "learning_rate": 2.884927936565134e-06,
+      "loss": 0.8481,
+      "step": 20882
+    },
+    {
+      "epoch": 0.7594646688729679,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8841003274965706e-06,
+      "loss": 0.8986,
+      "step": 20883
+    },
+    {
+      "epoch": 0.7595010364767065,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8832728171510225e-06,
+      "loss": 1.0155,
+      "step": 20884
+    },
+    {
+      "epoch": 0.7595374040804451,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8824454055399674e-06,
+      "loss": 0.9453,
+      "step": 20885
+    },
+    {
+      "epoch": 0.7595737716841837,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8816180926748837e-06,
+      "loss": 0.8939,
+      "step": 20886
+    },
+    {
+      "epoch": 0.7596101392879223,
+      "grad_norm": 0.0,
+      "learning_rate": 2.880790878567249e-06,
+      "loss": 0.9379,
+      "step": 20887
+    },
+    {
+      "epoch": 0.7596465068916609,
+      "grad_norm": 0.0,
+      "learning_rate": 2.879963763228537e-06,
+      "loss": 0.8991,
+      "step": 20888
+    },
+    {
+      "epoch": 0.7596828744953995,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8791367466702315e-06,
+      "loss": 0.9778,
+      "step": 20889
+    },
+    {
+      "epoch": 0.7597192420991381,
+      "grad_norm": 0.0,
+      "learning_rate": 2.878309828903795e-06,
+      "loss": 1.0554,
+      "step": 20890
+    },
+    {
+      "epoch": 0.7597556097028767,
+      "grad_norm": 0.0,
+      "learning_rate": 2.877483009940708e-06,
+      "loss": 0.8911,
+      "step": 20891
+    },
+    {
+      "epoch": 0.7597919773066153,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8766562897924388e-06,
+      "loss": 0.9245,
+      "step": 20892
+    },
+    {
+      "epoch": 0.7598283449103539,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8758296684704566e-06,
+      "loss": 0.8726,
+      "step": 20893
+    },
+    {
+      "epoch": 0.7598647125140925,
+      "grad_norm": 0.0,
+      "learning_rate": 2.875003145986227e-06,
+      "loss": 0.8863,
+      "step": 20894
+    },
+    {
+      "epoch": 0.7599010801178311,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8741767223512264e-06,
+      "loss": 0.9593,
+      "step": 20895
+    },
+    {
+      "epoch": 0.7599374477215697,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8733503975769064e-06,
+      "loss": 0.9747,
+      "step": 20896
+    },
+    {
+      "epoch": 0.7599738153253082,
+      "grad_norm": 0.0,
+      "learning_rate": 2.872524171674741e-06,
+      "loss": 0.8793,
+      "step": 20897
+    },
+    {
+      "epoch": 0.7600101829290468,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8716980446561914e-06,
+      "loss": 0.9919,
+      "step": 20898
+    },
+    {
+      "epoch": 0.7600465505327854,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8708720165327177e-06,
+      "loss": 1.0724,
+      "step": 20899
+    },
+    {
+      "epoch": 0.760082918136524,
+      "grad_norm": 0.0,
+      "learning_rate": 2.870046087315775e-06,
+      "loss": 0.9176,
+      "step": 20900
+    },
+    {
+      "epoch": 0.7601192857402626,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8692202570168315e-06,
+      "loss": 0.8625,
+      "step": 20901
+    },
+    {
+      "epoch": 0.7601556533440011,
+      "grad_norm": 0.0,
+      "learning_rate": 2.86839452564734e-06,
+      "loss": 0.9212,
+      "step": 20902
+    },
+    {
+      "epoch": 0.7601920209477397,
+      "grad_norm": 0.0,
+      "learning_rate": 2.867568893218755e-06,
+      "loss": 0.8934,
+      "step": 20903
+    },
+    {
+      "epoch": 0.7602283885514783,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8667433597425322e-06,
+      "loss": 0.9298,
+      "step": 20904
+    },
+    {
+      "epoch": 0.7602647561552169,
+      "grad_norm": 0.0,
+      "learning_rate": 2.865917925230125e-06,
+      "loss": 0.9517,
+      "step": 20905
+    },
+    {
+      "epoch": 0.7603011237589555,
+      "grad_norm": 0.0,
+      "learning_rate": 2.865092589692984e-06,
+      "loss": 0.9083,
+      "step": 20906
+    },
+    {
+      "epoch": 0.7603374913626941,
+      "grad_norm": 0.0,
+      "learning_rate": 2.864267353142558e-06,
+      "loss": 1.1031,
+      "step": 20907
+    },
+    {
+      "epoch": 0.7603738589664327,
+      "grad_norm": 0.0,
+      "learning_rate": 2.863442215590303e-06,
+      "loss": 1.0082,
+      "step": 20908
+    },
+    {
+      "epoch": 0.7604102265701713,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8626171770476573e-06,
+      "loss": 0.9998,
+      "step": 20909
+    },
+    {
+      "epoch": 0.7604465941739099,
+      "grad_norm": 0.0,
+      "learning_rate": 2.861792237526073e-06,
+      "loss": 1.0125,
+      "step": 20910
+    },
+    {
+      "epoch": 0.7604829617776485,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8609673970369954e-06,
+      "loss": 0.9193,
+      "step": 20911
+    },
+    {
+      "epoch": 0.7605193293813871,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8601426555918653e-06,
+      "loss": 0.8475,
+      "step": 20912
+    },
+    {
+      "epoch": 0.7605556969851256,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8593180132021227e-06,
+      "loss": 0.8268,
+      "step": 20913
+    },
+    {
+      "epoch": 0.7605920645888642,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8584934698792166e-06,
+      "loss": 0.9121,
+      "step": 20914
+    },
+    {
+      "epoch": 0.7606284321926028,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8576690256345763e-06,
+      "loss": 0.8159,
+      "step": 20915
+    },
+    {
+      "epoch": 0.7606647997963414,
+      "grad_norm": 0.0,
+      "learning_rate": 2.856844680479647e-06,
+      "loss": 0.8714,
+      "step": 20916
+    },
+    {
+      "epoch": 0.76070116740008,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8560204344258637e-06,
+      "loss": 1.0047,
+      "step": 20917
+    },
+    {
+      "epoch": 0.7607375350038186,
+      "grad_norm": 0.0,
+      "learning_rate": 2.855196287484661e-06,
+      "loss": 0.8956,
+      "step": 20918
+    },
+    {
+      "epoch": 0.7607739026075572,
+      "grad_norm": 0.0,
+      "learning_rate": 2.854372239667468e-06,
+      "loss": 0.9442,
+      "step": 20919
+    },
+    {
+      "epoch": 0.7608102702112958,
+      "grad_norm": 0.0,
+      "learning_rate": 2.853548290985726e-06,
+      "loss": 1.0433,
+      "step": 20920
+    },
+    {
+      "epoch": 0.7608466378150344,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8527244414508615e-06,
+      "loss": 0.9765,
+      "step": 20921
+    },
+    {
+      "epoch": 0.760883005418773,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8519006910743054e-06,
+      "loss": 0.9102,
+      "step": 20922
+    },
+    {
+      "epoch": 0.7609193730225116,
+      "grad_norm": 0.0,
+      "learning_rate": 2.851077039867485e-06,
+      "loss": 1.0269,
+      "step": 20923
+    },
+    {
+      "epoch": 0.7609557406262502,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8502534878418275e-06,
+      "loss": 1.0673,
+      "step": 20924
+    },
+    {
+      "epoch": 0.7609921082299888,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8494300350087587e-06,
+      "loss": 0.8964,
+      "step": 20925
+    },
+    {
+      "epoch": 0.7610284758337273,
+      "grad_norm": 0.0,
+      "learning_rate": 2.848606681379701e-06,
+      "loss": 0.8707,
+      "step": 20926
+    },
+    {
+      "epoch": 0.7610648434374659,
+      "grad_norm": 0.0,
+      "learning_rate": 2.847783426966084e-06,
+      "loss": 0.9087,
+      "step": 20927
+    },
+    {
+      "epoch": 0.7611012110412045,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8469602717793177e-06,
+      "loss": 0.9114,
+      "step": 20928
+    },
+    {
+      "epoch": 0.7611375786449431,
+      "grad_norm": 0.0,
+      "learning_rate": 2.846137215830833e-06,
+      "loss": 0.9383,
+      "step": 20929
+    },
+    {
+      "epoch": 0.7611739462486816,
+      "grad_norm": 0.0,
+      "learning_rate": 2.845314259132044e-06,
+      "loss": 0.8619,
+      "step": 20930
+    },
+    {
+      "epoch": 0.7612103138524202,
+      "grad_norm": 0.0,
+      "learning_rate": 2.844491401694368e-06,
+      "loss": 0.9823,
+      "step": 20931
+    },
+    {
+      "epoch": 0.7612466814561588,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8436686435292193e-06,
+      "loss": 0.9603,
+      "step": 20932
+    },
+    {
+      "epoch": 0.7612830490598974,
+      "grad_norm": 0.0,
+      "learning_rate": 2.842845984648017e-06,
+      "loss": 1.1066,
+      "step": 20933
+    },
+    {
+      "epoch": 0.761319416663636,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8420234250621724e-06,
+      "loss": 0.9663,
+      "step": 20934
+    },
+    {
+      "epoch": 0.7613557842673746,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8412009647830973e-06,
+      "loss": 0.9205,
+      "step": 20935
+    },
+    {
+      "epoch": 0.7613921518711132,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8403786038222013e-06,
+      "loss": 1.0416,
+      "step": 20936
+    },
+    {
+      "epoch": 0.7614285194748518,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8395563421908934e-06,
+      "loss": 0.9689,
+      "step": 20937
+    },
+    {
+      "epoch": 0.7614648870785904,
+      "grad_norm": 0.0,
+      "learning_rate": 2.83873417990058e-06,
+      "loss": 0.865,
+      "step": 20938
+    },
+    {
+      "epoch": 0.761501254682329,
+      "grad_norm": 0.0,
+      "learning_rate": 2.837912116962671e-06,
+      "loss": 1.0687,
+      "step": 20939
+    },
+    {
+      "epoch": 0.7615376222860676,
+      "grad_norm": 0.0,
+      "learning_rate": 2.837090153388571e-06,
+      "loss": 0.8638,
+      "step": 20940
+    },
+    {
+      "epoch": 0.7615739898898062,
+      "grad_norm": 0.0,
+      "learning_rate": 2.836268289189682e-06,
+      "loss": 0.9209,
+      "step": 20941
+    },
+    {
+      "epoch": 0.7616103574935448,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8354465243774034e-06,
+      "loss": 0.8421,
+      "step": 20942
+    },
+    {
+      "epoch": 0.7616467250972834,
+      "grad_norm": 0.0,
+      "learning_rate": 2.834624858963145e-06,
+      "loss": 1.0044,
+      "step": 20943
+    },
+    {
+      "epoch": 0.761683092701022,
+      "grad_norm": 0.0,
+      "learning_rate": 2.833803292958297e-06,
+      "loss": 0.9409,
+      "step": 20944
+    },
+    {
+      "epoch": 0.7617194603047606,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8329818263742583e-06,
+      "loss": 0.9158,
+      "step": 20945
+    },
+    {
+      "epoch": 0.7617558279084992,
+      "grad_norm": 0.0,
+      "learning_rate": 2.832160459222434e-06,
+      "loss": 0.8454,
+      "step": 20946
+    },
+    {
+      "epoch": 0.7617921955122376,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8313391915142073e-06,
+      "loss": 0.8889,
+      "step": 20947
+    },
+    {
+      "epoch": 0.7618285631159762,
+      "grad_norm": 0.0,
+      "learning_rate": 2.830518023260982e-06,
+      "loss": 1.1169,
+      "step": 20948
+    },
+    {
+      "epoch": 0.7618649307197148,
+      "grad_norm": 0.0,
+      "learning_rate": 2.829696954474146e-06,
+      "loss": 0.9454,
+      "step": 20949
+    },
+    {
+      "epoch": 0.7619012983234534,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8288759851650915e-06,
+      "loss": 0.8946,
+      "step": 20950
+    },
+    {
+      "epoch": 0.761937665927192,
+      "grad_norm": 0.0,
+      "learning_rate": 2.828055115345206e-06,
+      "loss": 0.8238,
+      "step": 20951
+    },
+    {
+      "epoch": 0.7619740335309306,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8272343450258834e-06,
+      "loss": 0.892,
+      "step": 20952
+    },
+    {
+      "epoch": 0.7620104011346692,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8264136742185066e-06,
+      "loss": 0.9767,
+      "step": 20953
+    },
+    {
+      "epoch": 0.7620467687384078,
+      "grad_norm": 0.0,
+      "learning_rate": 2.825593102934462e-06,
+      "loss": 0.7605,
+      "step": 20954
+    },
+    {
+      "epoch": 0.7620831363421464,
+      "grad_norm": 0.0,
+      "learning_rate": 2.824772631185131e-06,
+      "loss": 0.8927,
+      "step": 20955
+    },
+    {
+      "epoch": 0.762119503945885,
+      "grad_norm": 0.0,
+      "learning_rate": 2.823952258981906e-06,
+      "loss": 0.8352,
+      "step": 20956
+    },
+    {
+      "epoch": 0.7621558715496236,
+      "grad_norm": 0.0,
+      "learning_rate": 2.823131986336155e-06,
+      "loss": 0.8779,
+      "step": 20957
+    },
+    {
+      "epoch": 0.7621922391533622,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8223118132592688e-06,
+      "loss": 0.9928,
+      "step": 20958
+    },
+    {
+      "epoch": 0.7622286067571008,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8214917397626217e-06,
+      "loss": 0.8423,
+      "step": 20959
+    },
+    {
+      "epoch": 0.7622649743608394,
+      "grad_norm": 0.0,
+      "learning_rate": 2.820671765857592e-06,
+      "loss": 0.9901,
+      "step": 20960
+    },
+    {
+      "epoch": 0.762301341964578,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8198518915555528e-06,
+      "loss": 0.8984,
+      "step": 20961
+    },
+    {
+      "epoch": 0.7623377095683166,
+      "grad_norm": 0.0,
+      "learning_rate": 2.819032116867887e-06,
+      "loss": 0.8453,
+      "step": 20962
+    },
+    {
+      "epoch": 0.7623740771720552,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8182124418059566e-06,
+      "loss": 0.8625,
+      "step": 20963
+    },
+    {
+      "epoch": 0.7624104447757937,
+      "grad_norm": 0.0,
+      "learning_rate": 2.817392866381141e-06,
+      "loss": 0.8928,
+      "step": 20964
+    },
+    {
+      "epoch": 0.7624468123795323,
+      "grad_norm": 0.0,
+      "learning_rate": 2.81657339060481e-06,
+      "loss": 0.8618,
+      "step": 20965
+    },
+    {
+      "epoch": 0.7624831799832709,
+      "grad_norm": 0.0,
+      "learning_rate": 2.81575401448833e-06,
+      "loss": 0.9974,
+      "step": 20966
+    },
+    {
+      "epoch": 0.7625195475870095,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8149347380430703e-06,
+      "loss": 0.9768,
+      "step": 20967
+    },
+    {
+      "epoch": 0.7625559151907481,
+      "grad_norm": 0.0,
+      "learning_rate": 2.814115561280397e-06,
+      "loss": 0.9373,
+      "step": 20968
+    },
+    {
+      "epoch": 0.7625922827944867,
+      "grad_norm": 0.0,
+      "learning_rate": 2.813296484211675e-06,
+      "loss": 0.9065,
+      "step": 20969
+    },
+    {
+      "epoch": 0.7626286503982252,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8124775068482646e-06,
+      "loss": 1.0383,
+      "step": 20970
+    },
+    {
+      "epoch": 0.7626650180019638,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8116586292015336e-06,
+      "loss": 0.816,
+      "step": 20971
+    },
+    {
+      "epoch": 0.7627013856057024,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8108398512828405e-06,
+      "loss": 0.9438,
+      "step": 20972
+    },
+    {
+      "epoch": 0.762737753209441,
+      "grad_norm": 0.0,
+      "learning_rate": 2.810021173103544e-06,
+      "loss": 0.9583,
+      "step": 20973
+    },
+    {
+      "epoch": 0.7627741208131796,
+      "grad_norm": 0.0,
+      "learning_rate": 2.809202594675e-06,
+      "loss": 0.9051,
+      "step": 20974
+    },
+    {
+      "epoch": 0.7628104884169182,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8083841160085733e-06,
+      "loss": 0.9322,
+      "step": 20975
+    },
+    {
+      "epoch": 0.7628468560206568,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8075657371156074e-06,
+      "loss": 0.8768,
+      "step": 20976
+    },
+    {
+      "epoch": 0.7628832236243954,
+      "grad_norm": 0.0,
+      "learning_rate": 2.806747458007465e-06,
+      "loss": 0.9616,
+      "step": 20977
+    },
+    {
+      "epoch": 0.762919591228134,
+      "grad_norm": 0.0,
+      "learning_rate": 2.805929278695496e-06,
+      "loss": 1.0795,
+      "step": 20978
+    },
+    {
+      "epoch": 0.7629559588318726,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8051111991910518e-06,
+      "loss": 0.8879,
+      "step": 20979
+    },
+    {
+      "epoch": 0.7629923264356112,
+      "grad_norm": 0.0,
+      "learning_rate": 2.804293219505477e-06,
+      "loss": 0.922,
+      "step": 20980
+    },
+    {
+      "epoch": 0.7630286940393497,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8034753396501325e-06,
+      "loss": 1.0168,
+      "step": 20981
+    },
+    {
+      "epoch": 0.7630650616430883,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8026575596363493e-06,
+      "loss": 0.8431,
+      "step": 20982
+    },
+    {
+      "epoch": 0.7631014292468269,
+      "grad_norm": 0.0,
+      "learning_rate": 2.801839879475484e-06,
+      "loss": 0.8685,
+      "step": 20983
+    },
+    {
+      "epoch": 0.7631377968505655,
+      "grad_norm": 0.0,
+      "learning_rate": 2.801022299178878e-06,
+      "loss": 0.8055,
+      "step": 20984
+    },
+    {
+      "epoch": 0.7631741644543041,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8002048187578745e-06,
+      "loss": 0.8707,
+      "step": 20985
+    },
+    {
+      "epoch": 0.7632105320580427,
+      "grad_norm": 0.0,
+      "learning_rate": 2.799387438223813e-06,
+      "loss": 0.8522,
+      "step": 20986
+    },
+    {
+      "epoch": 0.7632468996617813,
+      "grad_norm": 0.0,
+      "learning_rate": 2.798570157588031e-06,
+      "loss": 0.9264,
+      "step": 20987
+    },
+    {
+      "epoch": 0.7632832672655199,
+      "grad_norm": 0.0,
+      "learning_rate": 2.797752976861877e-06,
+      "loss": 0.8596,
+      "step": 20988
+    },
+    {
+      "epoch": 0.7633196348692585,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7969358960566763e-06,
+      "loss": 0.9783,
+      "step": 20989
+    },
+    {
+      "epoch": 0.7633560024729971,
+      "grad_norm": 0.0,
+      "learning_rate": 2.796118915183772e-06,
+      "loss": 0.9118,
+      "step": 20990
+    },
+    {
+      "epoch": 0.7633923700767357,
+      "grad_norm": 0.0,
+      "learning_rate": 2.795302034254498e-06,
+      "loss": 1.0095,
+      "step": 20991
+    },
+    {
+      "epoch": 0.7634287376804743,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7944852532801857e-06,
+      "loss": 1.0288,
+      "step": 20992
+    },
+    {
+      "epoch": 0.7634651052842129,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7936685722721648e-06,
+      "loss": 1.0791,
+      "step": 20993
+    },
+    {
+      "epoch": 0.7635014728879514,
+      "grad_norm": 0.0,
+      "learning_rate": 2.792851991241773e-06,
+      "loss": 0.8515,
+      "step": 20994
+    },
+    {
+      "epoch": 0.76353784049169,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7920355102003283e-06,
+      "loss": 0.847,
+      "step": 20995
+    },
+    {
+      "epoch": 0.7635742080954286,
+      "grad_norm": 0.0,
+      "learning_rate": 2.791219129159167e-06,
+      "loss": 0.9037,
+      "step": 20996
+    },
+    {
+      "epoch": 0.7636105756991671,
+      "grad_norm": 0.0,
+      "learning_rate": 2.790402848129613e-06,
+      "loss": 1.0186,
+      "step": 20997
+    },
+    {
+      "epoch": 0.7636469433029057,
+      "grad_norm": 0.0,
+      "learning_rate": 2.789586667122991e-06,
+      "loss": 0.9398,
+      "step": 20998
+    },
+    {
+      "epoch": 0.7636833109066443,
+      "grad_norm": 0.0,
+      "learning_rate": 2.78877058615062e-06,
+      "loss": 0.9288,
+      "step": 20999
+    },
+    {
+      "epoch": 0.7637196785103829,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7879546052238282e-06,
+      "loss": 0.8651,
+      "step": 21000
+    },
+    {
+      "epoch": 0.7637560461141215,
+      "grad_norm": 0.0,
+      "learning_rate": 2.787138724353934e-06,
+      "loss": 0.9226,
+      "step": 21001
+    },
+    {
+      "epoch": 0.7637924137178601,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7863229435522564e-06,
+      "loss": 0.8513,
+      "step": 21002
+    },
+    {
+      "epoch": 0.7638287813215987,
+      "grad_norm": 0.0,
+      "learning_rate": 2.785507262830114e-06,
+      "loss": 1.0446,
+      "step": 21003
+    },
+    {
+      "epoch": 0.7638651489253373,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7846916821988214e-06,
+      "loss": 0.9453,
+      "step": 21004
+    },
+    {
+      "epoch": 0.7639015165290759,
+      "grad_norm": 0.0,
+      "learning_rate": 2.783876201669694e-06,
+      "loss": 0.9479,
+      "step": 21005
+    },
+    {
+      "epoch": 0.7639378841328145,
+      "grad_norm": 0.0,
+      "learning_rate": 2.783060821254043e-06,
+      "loss": 0.9435,
+      "step": 21006
+    },
+    {
+      "epoch": 0.7639742517365531,
+      "grad_norm": 0.0,
+      "learning_rate": 2.782245540963191e-06,
+      "loss": 1.0334,
+      "step": 21007
+    },
+    {
+      "epoch": 0.7640106193402917,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7814303608084347e-06,
+      "loss": 0.9644,
+      "step": 21008
+    },
+    {
+      "epoch": 0.7640469869440303,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7806152808010933e-06,
+      "loss": 0.9646,
+      "step": 21009
+    },
+    {
+      "epoch": 0.7640833545477689,
+      "grad_norm": 0.0,
+      "learning_rate": 2.779800300952472e-06,
+      "loss": 0.9643,
+      "step": 21010
+    },
+    {
+      "epoch": 0.7641197221515075,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7789854212738775e-06,
+      "loss": 0.809,
+      "step": 21011
+    },
+    {
+      "epoch": 0.7641560897552461,
+      "grad_norm": 0.0,
+      "learning_rate": 2.778170641776612e-06,
+      "loss": 0.8319,
+      "step": 21012
+    },
+    {
+      "epoch": 0.7641924573589847,
+      "grad_norm": 0.0,
+      "learning_rate": 2.777355962471988e-06,
+      "loss": 1.0091,
+      "step": 21013
+    },
+    {
+      "epoch": 0.7642288249627232,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7765413833712964e-06,
+      "loss": 0.9503,
+      "step": 21014
+    },
+    {
+      "epoch": 0.7642651925664617,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7757269044858472e-06,
+      "loss": 0.9501,
+      "step": 21015
+    },
+    {
+      "epoch": 0.7643015601702003,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7749125258269383e-06,
+      "loss": 0.9702,
+      "step": 21016
+    },
+    {
+      "epoch": 0.7643379277739389,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7740982474058664e-06,
+      "loss": 0.9293,
+      "step": 21017
+    },
+    {
+      "epoch": 0.7643742953776775,
+      "grad_norm": 0.0,
+      "learning_rate": 2.773284069233926e-06,
+      "loss": 0.9301,
+      "step": 21018
+    },
+    {
+      "epoch": 0.7644106629814161,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7724699913224187e-06,
+      "loss": 0.9186,
+      "step": 21019
+    },
+    {
+      "epoch": 0.7644470305851547,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7716560136826354e-06,
+      "loss": 1.0328,
+      "step": 21020
+    },
+    {
+      "epoch": 0.7644833981888933,
+      "grad_norm": 0.0,
+      "learning_rate": 2.77084213632587e-06,
+      "loss": 0.8985,
+      "step": 21021
+    },
+    {
+      "epoch": 0.7645197657926319,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7700283592634135e-06,
+      "loss": 0.897,
+      "step": 21022
+    },
+    {
+      "epoch": 0.7645561333963705,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7692146825065547e-06,
+      "loss": 0.9903,
+      "step": 21023
+    },
+    {
+      "epoch": 0.7645925010001091,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7684011060665827e-06,
+      "loss": 0.8599,
+      "step": 21024
+    },
+    {
+      "epoch": 0.7646288686038477,
+      "grad_norm": 0.0,
+      "learning_rate": 2.767587629954783e-06,
+      "loss": 0.8423,
+      "step": 21025
+    },
+    {
+      "epoch": 0.7646652362075863,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7667742541824495e-06,
+      "loss": 1.0261,
+      "step": 21026
+    },
+    {
+      "epoch": 0.7647016038113249,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7659609787608545e-06,
+      "loss": 0.9184,
+      "step": 21027
+    },
+    {
+      "epoch": 0.7647379714150635,
+      "grad_norm": 0.0,
+      "learning_rate": 2.765147803701289e-06,
+      "loss": 0.7076,
+      "step": 21028
+    },
+    {
+      "epoch": 0.7647743390188021,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7643347290150336e-06,
+      "loss": 0.9356,
+      "step": 21029
+    },
+    {
+      "epoch": 0.7648107066225407,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7635217547133684e-06,
+      "loss": 0.8973,
+      "step": 21030
+    },
+    {
+      "epoch": 0.7648470742262792,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7627088808075674e-06,
+      "loss": 1.0377,
+      "step": 21031
+    },
+    {
+      "epoch": 0.7648834418300178,
+      "grad_norm": 0.0,
+      "learning_rate": 2.761896107308917e-06,
+      "loss": 0.8867,
+      "step": 21032
+    },
+    {
+      "epoch": 0.7649198094337564,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7610834342286874e-06,
+      "loss": 1.0725,
+      "step": 21033
+    },
+    {
+      "epoch": 0.764956177037495,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7602708615781548e-06,
+      "loss": 0.9678,
+      "step": 21034
+    },
+    {
+      "epoch": 0.7649925446412336,
+      "grad_norm": 0.0,
+      "learning_rate": 2.759458389368591e-06,
+      "loss": 0.892,
+      "step": 21035
+    },
+    {
+      "epoch": 0.7650289122449722,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7586460176112707e-06,
+      "loss": 0.8616,
+      "step": 21036
+    },
+    {
+      "epoch": 0.7650652798487108,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7578337463174585e-06,
+      "loss": 0.7832,
+      "step": 21037
+    },
+    {
+      "epoch": 0.7651016474524494,
+      "grad_norm": 0.0,
+      "learning_rate": 2.757021575498432e-06,
+      "loss": 0.9665,
+      "step": 21038
+    },
+    {
+      "epoch": 0.765138015056188,
+      "grad_norm": 0.0,
+      "learning_rate": 2.756209505165455e-06,
+      "loss": 0.9219,
+      "step": 21039
+    },
+    {
+      "epoch": 0.7651743826599265,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7553975353297922e-06,
+      "loss": 0.8863,
+      "step": 21040
+    },
+    {
+      "epoch": 0.7652107502636651,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7545856660027102e-06,
+      "loss": 1.0389,
+      "step": 21041
+    },
+    {
+      "epoch": 0.7652471178674037,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7537738971954718e-06,
+      "loss": 0.8667,
+      "step": 21042
+    },
+    {
+      "epoch": 0.7652834854711423,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7529622289193403e-06,
+      "loss": 1.025,
+      "step": 21043
+    },
+    {
+      "epoch": 0.7653198530748809,
+      "grad_norm": 0.0,
+      "learning_rate": 2.752150661185572e-06,
+      "loss": 0.9981,
+      "step": 21044
+    },
+    {
+      "epoch": 0.7653562206786195,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7513391940054336e-06,
+      "loss": 0.9119,
+      "step": 21045
+    },
+    {
+      "epoch": 0.7653925882823581,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7505278273901794e-06,
+      "loss": 0.988,
+      "step": 21046
+    },
+    {
+      "epoch": 0.7654289558860967,
+      "grad_norm": 0.0,
+      "learning_rate": 2.749716561351066e-06,
+      "loss": 0.9198,
+      "step": 21047
+    },
+    {
+      "epoch": 0.7654653234898352,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7489053958993472e-06,
+      "loss": 0.9791,
+      "step": 21048
+    },
+    {
+      "epoch": 0.7655016910935738,
+      "grad_norm": 0.0,
+      "learning_rate": 2.748094331046279e-06,
+      "loss": 0.8991,
+      "step": 21049
+    },
+    {
+      "epoch": 0.7655380586973124,
+      "grad_norm": 0.0,
+      "learning_rate": 2.747283366803111e-06,
+      "loss": 0.8491,
+      "step": 21050
+    },
+    {
+      "epoch": 0.765574426301051,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7464725031810967e-06,
+      "loss": 0.9466,
+      "step": 21051
+    },
+    {
+      "epoch": 0.7656107939047896,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7456617401914866e-06,
+      "loss": 0.9447,
+      "step": 21052
+    },
+    {
+      "epoch": 0.7656471615085282,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7448510778455274e-06,
+      "loss": 0.8088,
+      "step": 21053
+    },
+    {
+      "epoch": 0.7656835291122668,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7440405161544615e-06,
+      "loss": 0.9634,
+      "step": 21054
+    },
+    {
+      "epoch": 0.7657198967160054,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7432300551295453e-06,
+      "loss": 0.9424,
+      "step": 21055
+    },
+    {
+      "epoch": 0.765756264319744,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7424196947820103e-06,
+      "loss": 0.8967,
+      "step": 21056
+    },
+    {
+      "epoch": 0.7657926319234826,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7416094351231084e-06,
+      "loss": 0.9652,
+      "step": 21057
+    },
+    {
+      "epoch": 0.7658289995272212,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7407992761640758e-06,
+      "loss": 0.8772,
+      "step": 21058
+    },
+    {
+      "epoch": 0.7658653671309598,
+      "grad_norm": 0.0,
+      "learning_rate": 2.739989217916155e-06,
+      "loss": 0.8736,
+      "step": 21059
+    },
+    {
+      "epoch": 0.7659017347346984,
+      "grad_norm": 0.0,
+      "learning_rate": 2.739179260390582e-06,
+      "loss": 0.9609,
+      "step": 21060
+    },
+    {
+      "epoch": 0.765938102338437,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7383694035985965e-06,
+      "loss": 0.949,
+      "step": 21061
+    },
+    {
+      "epoch": 0.7659744699421756,
+      "grad_norm": 0.0,
+      "learning_rate": 2.737559647551432e-06,
+      "loss": 0.9572,
+      "step": 21062
+    },
+    {
+      "epoch": 0.7660108375459141,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7367499922603204e-06,
+      "loss": 0.9864,
+      "step": 21063
+    },
+    {
+      "epoch": 0.7660472051496527,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7359404377364994e-06,
+      "loss": 0.8816,
+      "step": 21064
+    },
+    {
+      "epoch": 0.7660835727533912,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7351309839911998e-06,
+      "loss": 0.8037,
+      "step": 21065
+    },
+    {
+      "epoch": 0.7661199403571298,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7343216310356504e-06,
+      "loss": 0.9222,
+      "step": 21066
+    },
+    {
+      "epoch": 0.7661563079608684,
+      "grad_norm": 0.0,
+      "learning_rate": 2.733512378881078e-06,
+      "loss": 1.0497,
+      "step": 21067
+    },
+    {
+      "epoch": 0.766192675564607,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7327032275387134e-06,
+      "loss": 0.8813,
+      "step": 21068
+    },
+    {
+      "epoch": 0.7662290431683456,
+      "grad_norm": 0.0,
+      "learning_rate": 2.731894177019777e-06,
+      "loss": 0.8572,
+      "step": 21069
+    },
+    {
+      "epoch": 0.7662654107720842,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7310852273354995e-06,
+      "loss": 0.9928,
+      "step": 21070
+    },
+    {
+      "epoch": 0.7663017783758228,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7302763784971008e-06,
+      "loss": 0.9198,
+      "step": 21071
+    },
+    {
+      "epoch": 0.7663381459795614,
+      "grad_norm": 0.0,
+      "learning_rate": 2.729467630515804e-06,
+      "loss": 0.8712,
+      "step": 21072
+    },
+    {
+      "epoch": 0.7663745135833,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7286589834028245e-06,
+      "loss": 0.8719,
+      "step": 21073
+    },
+    {
+      "epoch": 0.7664108811870386,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7278504371693916e-06,
+      "loss": 0.891,
+      "step": 21074
+    },
+    {
+      "epoch": 0.7664472487907772,
+      "grad_norm": 0.0,
+      "learning_rate": 2.72704199182671e-06,
+      "loss": 0.8058,
+      "step": 21075
+    },
+    {
+      "epoch": 0.7664836163945158,
+      "grad_norm": 0.0,
+      "learning_rate": 2.726233647386004e-06,
+      "loss": 1.0787,
+      "step": 21076
+    },
+    {
+      "epoch": 0.7665199839982544,
+      "grad_norm": 0.0,
+      "learning_rate": 2.725425403858487e-06,
+      "loss": 0.9303,
+      "step": 21077
+    },
+    {
+      "epoch": 0.766556351601993,
+      "grad_norm": 0.0,
+      "learning_rate": 2.724617261255371e-06,
+      "loss": 0.864,
+      "step": 21078
+    },
+    {
+      "epoch": 0.7665927192057316,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7238092195878673e-06,
+      "loss": 0.9641,
+      "step": 21079
+    },
+    {
+      "epoch": 0.7666290868094702,
+      "grad_norm": 0.0,
+      "learning_rate": 2.723001278867188e-06,
+      "loss": 0.787,
+      "step": 21080
+    },
+    {
+      "epoch": 0.7666654544132088,
+      "grad_norm": 0.0,
+      "learning_rate": 2.722193439104541e-06,
+      "loss": 0.8746,
+      "step": 21081
+    },
+    {
+      "epoch": 0.7667018220169473,
+      "grad_norm": 0.0,
+      "learning_rate": 2.721385700311131e-06,
+      "loss": 0.8481,
+      "step": 21082
+    },
+    {
+      "epoch": 0.7667381896206859,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7205780624981702e-06,
+      "loss": 1.0175,
+      "step": 21083
+    },
+    {
+      "epoch": 0.7667745572244244,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7197705256768614e-06,
+      "loss": 0.9461,
+      "step": 21084
+    },
+    {
+      "epoch": 0.766810924828163,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7189630898584065e-06,
+      "loss": 0.9457,
+      "step": 21085
+    },
+    {
+      "epoch": 0.7668472924319016,
+      "grad_norm": 0.0,
+      "learning_rate": 2.718155755054006e-06,
+      "loss": 0.8571,
+      "step": 21086
+    },
+    {
+      "epoch": 0.7668836600356402,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7173485212748685e-06,
+      "loss": 0.8687,
+      "step": 21087
+    },
+    {
+      "epoch": 0.7669200276393788,
+      "grad_norm": 0.0,
+      "learning_rate": 2.716541388532181e-06,
+      "loss": 0.9164,
+      "step": 21088
+    },
+    {
+      "epoch": 0.7669563952431174,
+      "grad_norm": 0.0,
+      "learning_rate": 2.715734356837151e-06,
+      "loss": 0.9015,
+      "step": 21089
+    },
+    {
+      "epoch": 0.766992762846856,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7149274262009728e-06,
+      "loss": 0.9009,
+      "step": 21090
+    },
+    {
+      "epoch": 0.7670291304505946,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7141205966348395e-06,
+      "loss": 0.9072,
+      "step": 21091
+    },
+    {
+      "epoch": 0.7670654980543332,
+      "grad_norm": 0.0,
+      "learning_rate": 2.713313868149943e-06,
+      "loss": 0.976,
+      "step": 21092
+    },
+    {
+      "epoch": 0.7671018656580718,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7125072407574836e-06,
+      "loss": 0.871,
+      "step": 21093
+    },
+    {
+      "epoch": 0.7671382332618104,
+      "grad_norm": 0.0,
+      "learning_rate": 2.71170071446864e-06,
+      "loss": 0.8413,
+      "step": 21094
+    },
+    {
+      "epoch": 0.767174600865549,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7108942892946134e-06,
+      "loss": 0.9197,
+      "step": 21095
+    },
+    {
+      "epoch": 0.7672109684692876,
+      "grad_norm": 0.0,
+      "learning_rate": 2.710087965246585e-06,
+      "loss": 0.8271,
+      "step": 21096
+    },
+    {
+      "epoch": 0.7672473360730262,
+      "grad_norm": 0.0,
+      "learning_rate": 2.709281742335743e-06,
+      "loss": 0.8465,
+      "step": 21097
+    },
+    {
+      "epoch": 0.7672837036767647,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7084756205732732e-06,
+      "loss": 0.9874,
+      "step": 21098
+    },
+    {
+      "epoch": 0.7673200712805033,
+      "grad_norm": 0.0,
+      "learning_rate": 2.707669599970356e-06,
+      "loss": 1.0464,
+      "step": 21099
+    },
+    {
+      "epoch": 0.7673564388842419,
+      "grad_norm": 0.0,
+      "learning_rate": 2.706863680538182e-06,
+      "loss": 0.9522,
+      "step": 21100
+    },
+    {
+      "epoch": 0.7673928064879805,
+      "grad_norm": 0.0,
+      "learning_rate": 2.706057862287922e-06,
+      "loss": 0.9954,
+      "step": 21101
+    },
+    {
+      "epoch": 0.7674291740917191,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7052521452307624e-06,
+      "loss": 0.9912,
+      "step": 21102
+    },
+    {
+      "epoch": 0.7674655416954577,
+      "grad_norm": 0.0,
+      "learning_rate": 2.70444652937788e-06,
+      "loss": 0.9375,
+      "step": 21103
+    },
+    {
+      "epoch": 0.7675019092991963,
+      "grad_norm": 0.0,
+      "learning_rate": 2.70364101474045e-06,
+      "loss": 0.8646,
+      "step": 21104
+    },
+    {
+      "epoch": 0.7675382769029349,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7028356013296462e-06,
+      "loss": 0.9216,
+      "step": 21105
+    },
+    {
+      "epoch": 0.7675746445066735,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7020302891566518e-06,
+      "loss": 0.9343,
+      "step": 21106
+    },
+    {
+      "epoch": 0.767611012110412,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7012250782326255e-06,
+      "loss": 0.9112,
+      "step": 21107
+    },
+    {
+      "epoch": 0.7676473797141506,
+      "grad_norm": 0.0,
+      "learning_rate": 2.700419968568749e-06,
+      "loss": 0.9059,
+      "step": 21108
+    },
+    {
+      "epoch": 0.7676837473178892,
+      "grad_norm": 0.0,
+      "learning_rate": 2.69961496017619e-06,
+      "loss": 0.9522,
+      "step": 21109
+    },
+    {
+      "epoch": 0.7677201149216278,
+      "grad_norm": 0.0,
+      "learning_rate": 2.698810053066114e-06,
+      "loss": 0.9074,
+      "step": 21110
+    },
+    {
+      "epoch": 0.7677564825253664,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6980052472496874e-06,
+      "loss": 0.896,
+      "step": 21111
+    },
+    {
+      "epoch": 0.767792850129105,
+      "grad_norm": 0.0,
+      "learning_rate": 2.697200542738083e-06,
+      "loss": 0.894,
+      "step": 21112
+    },
+    {
+      "epoch": 0.7678292177328436,
+      "grad_norm": 0.0,
+      "learning_rate": 2.696395939542454e-06,
+      "loss": 0.9574,
+      "step": 21113
+    },
+    {
+      "epoch": 0.7678655853365822,
+      "grad_norm": 0.0,
+      "learning_rate": 2.695591437673972e-06,
+      "loss": 0.9001,
+      "step": 21114
+    },
+    {
+      "epoch": 0.7679019529403207,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6947870371437956e-06,
+      "loss": 0.8353,
+      "step": 21115
+    },
+    {
+      "epoch": 0.7679383205440593,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6939827379630836e-06,
+      "loss": 0.9998,
+      "step": 21116
+    },
+    {
+      "epoch": 0.7679746881477979,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6931785401429957e-06,
+      "loss": 0.777,
+      "step": 21117
+    },
+    {
+      "epoch": 0.7680110557515365,
+      "grad_norm": 0.0,
+      "learning_rate": 2.692374443694685e-06,
+      "loss": 0.8143,
+      "step": 21118
+    },
+    {
+      "epoch": 0.7680474233552751,
+      "grad_norm": 0.0,
+      "learning_rate": 2.691570448629318e-06,
+      "loss": 0.8949,
+      "step": 21119
+    },
+    {
+      "epoch": 0.7680837909590137,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6907665549580342e-06,
+      "loss": 0.8905,
+      "step": 21120
+    },
+    {
+      "epoch": 0.7681201585627523,
+      "grad_norm": 0.0,
+      "learning_rate": 2.689962762691998e-06,
+      "loss": 0.8966,
+      "step": 21121
+    },
+    {
+      "epoch": 0.7681565261664909,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6891590718423568e-06,
+      "loss": 0.8594,
+      "step": 21122
+    },
+    {
+      "epoch": 0.7681928937702295,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6883554824202605e-06,
+      "loss": 0.9389,
+      "step": 21123
+    },
+    {
+      "epoch": 0.7682292613739681,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6875519944368543e-06,
+      "loss": 0.932,
+      "step": 21124
+    },
+    {
+      "epoch": 0.7682656289777067,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6867486079032966e-06,
+      "loss": 0.9311,
+      "step": 21125
+    },
+    {
+      "epoch": 0.7683019965814453,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6859453228307185e-06,
+      "loss": 0.9479,
+      "step": 21126
+    },
+    {
+      "epoch": 0.7683383641851839,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6851421392302756e-06,
+      "loss": 0.9288,
+      "step": 21127
+    },
+    {
+      "epoch": 0.7683747317889225,
+      "grad_norm": 0.0,
+      "learning_rate": 2.684339057113107e-06,
+      "loss": 0.9405,
+      "step": 21128
+    },
+    {
+      "epoch": 0.7684110993926611,
+      "grad_norm": 0.0,
+      "learning_rate": 2.683536076490354e-06,
+      "loss": 0.968,
+      "step": 21129
+    },
+    {
+      "epoch": 0.7684474669963997,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6827331973731542e-06,
+      "loss": 0.9748,
+      "step": 21130
+    },
+    {
+      "epoch": 0.7684838346001382,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6819304197726525e-06,
+      "loss": 0.9505,
+      "step": 21131
+    },
+    {
+      "epoch": 0.7685202022038767,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6811277436999838e-06,
+      "loss": 1.0551,
+      "step": 21132
+    },
+    {
+      "epoch": 0.7685565698076153,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6803251691662833e-06,
+      "loss": 0.9363,
+      "step": 21133
+    },
+    {
+      "epoch": 0.7685929374113539,
+      "grad_norm": 0.0,
+      "learning_rate": 2.679522696182686e-06,
+      "loss": 0.994,
+      "step": 21134
+    },
+    {
+      "epoch": 0.7686293050150925,
+      "grad_norm": 0.0,
+      "learning_rate": 2.678720324760324e-06,
+      "loss": 0.8397,
+      "step": 21135
+    },
+    {
+      "epoch": 0.7686656726188311,
+      "grad_norm": 0.0,
+      "learning_rate": 2.677918054910331e-06,
+      "loss": 0.9556,
+      "step": 21136
+    },
+    {
+      "epoch": 0.7687020402225697,
+      "grad_norm": 0.0,
+      "learning_rate": 2.677115886643833e-06,
+      "loss": 0.9902,
+      "step": 21137
+    },
+    {
+      "epoch": 0.7687384078263083,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6763138199719686e-06,
+      "loss": 0.9605,
+      "step": 21138
+    },
+    {
+      "epoch": 0.7687747754300469,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6755118549058522e-06,
+      "loss": 0.8311,
+      "step": 21139
+    },
+    {
+      "epoch": 0.7688111430337855,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6747099914566197e-06,
+      "loss": 0.9358,
+      "step": 21140
+    },
+    {
+      "epoch": 0.7688475106375241,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6739082296353937e-06,
+      "loss": 0.9233,
+      "step": 21141
+    },
+    {
+      "epoch": 0.7688838782412627,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6731065694532955e-06,
+      "loss": 0.9007,
+      "step": 21142
+    },
+    {
+      "epoch": 0.7689202458450013,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6723050109214465e-06,
+      "loss": 0.9472,
+      "step": 21143
+    },
+    {
+      "epoch": 0.7689566134487399,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6715035540509704e-06,
+      "loss": 0.8226,
+      "step": 21144
+    },
+    {
+      "epoch": 0.7689929810524785,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6707021988529857e-06,
+      "loss": 0.9351,
+      "step": 21145
+    },
+    {
+      "epoch": 0.7690293486562171,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6699009453386095e-06,
+      "loss": 0.9193,
+      "step": 21146
+    },
+    {
+      "epoch": 0.7690657162599557,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6690997935189567e-06,
+      "loss": 0.9939,
+      "step": 21147
+    },
+    {
+      "epoch": 0.7691020838636943,
+      "grad_norm": 0.0,
+      "learning_rate": 2.668298743405143e-06,
+      "loss": 0.8551,
+      "step": 21148
+    },
+    {
+      "epoch": 0.7691384514674328,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6674977950082793e-06,
+      "loss": 0.8491,
+      "step": 21149
+    },
+    {
+      "epoch": 0.7691748190711714,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6666969483394834e-06,
+      "loss": 0.8317,
+      "step": 21150
+    },
+    {
+      "epoch": 0.76921118667491,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6658962034098624e-06,
+      "loss": 0.9314,
+      "step": 21151
+    },
+    {
+      "epoch": 0.7692475542786485,
+      "grad_norm": 0.0,
+      "learning_rate": 2.665095560230526e-06,
+      "loss": 1.0211,
+      "step": 21152
+    },
+    {
+      "epoch": 0.7692839218823871,
+      "grad_norm": 0.0,
+      "learning_rate": 2.664295018812578e-06,
+      "loss": 0.9377,
+      "step": 21153
+    },
+    {
+      "epoch": 0.7693202894861257,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6634945791671365e-06,
+      "loss": 0.9137,
+      "step": 21154
+    },
+    {
+      "epoch": 0.7693566570898643,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6626942413052937e-06,
+      "loss": 0.9029,
+      "step": 21155
+    },
+    {
+      "epoch": 0.7693930246936029,
+      "grad_norm": 0.0,
+      "learning_rate": 2.661894005238156e-06,
+      "loss": 0.8963,
+      "step": 21156
+    },
+    {
+      "epoch": 0.7694293922973415,
+      "grad_norm": 0.0,
+      "learning_rate": 2.661093870976834e-06,
+      "loss": 0.8735,
+      "step": 21157
+    },
+    {
+      "epoch": 0.7694657599010801,
+      "grad_norm": 0.0,
+      "learning_rate": 2.660293838532415e-06,
+      "loss": 0.9251,
+      "step": 21158
+    },
+    {
+      "epoch": 0.7695021275048187,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6594939079160087e-06,
+      "loss": 0.7839,
+      "step": 21159
+    },
+    {
+      "epoch": 0.7695384951085573,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6586940791387096e-06,
+      "loss": 0.9417,
+      "step": 21160
+    },
+    {
+      "epoch": 0.7695748627122959,
+      "grad_norm": 0.0,
+      "learning_rate": 2.657894352211614e-06,
+      "loss": 0.8615,
+      "step": 21161
+    },
+    {
+      "epoch": 0.7696112303160345,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6570947271458146e-06,
+      "loss": 1.0541,
+      "step": 21162
+    },
+    {
+      "epoch": 0.7696475979197731,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6562952039524106e-06,
+      "loss": 0.9362,
+      "step": 21163
+    },
+    {
+      "epoch": 0.7696839655235117,
+      "grad_norm": 0.0,
+      "learning_rate": 2.655495782642491e-06,
+      "loss": 0.8414,
+      "step": 21164
+    },
+    {
+      "epoch": 0.7697203331272503,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6546964632271464e-06,
+      "loss": 0.9415,
+      "step": 21165
+    },
+    {
+      "epoch": 0.7697567007309888,
+      "grad_norm": 0.0,
+      "learning_rate": 2.653897245717465e-06,
+      "loss": 0.8997,
+      "step": 21166
+    },
+    {
+      "epoch": 0.7697930683347274,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6530981301245416e-06,
+      "loss": 0.9639,
+      "step": 21167
+    },
+    {
+      "epoch": 0.769829435938466,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6522991164594525e-06,
+      "loss": 0.9041,
+      "step": 21168
+    },
+    {
+      "epoch": 0.7698658035422046,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6515002047332907e-06,
+      "loss": 0.896,
+      "step": 21169
+    },
+    {
+      "epoch": 0.7699021711459432,
+      "grad_norm": 0.0,
+      "learning_rate": 2.650701394957137e-06,
+      "loss": 0.9554,
+      "step": 21170
+    },
+    {
+      "epoch": 0.7699385387496818,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6499026871420753e-06,
+      "loss": 0.9366,
+      "step": 21171
+    },
+    {
+      "epoch": 0.7699749063534204,
+      "grad_norm": 0.0,
+      "learning_rate": 2.649104081299181e-06,
+      "loss": 0.9597,
+      "step": 21172
+    },
+    {
+      "epoch": 0.770011273957159,
+      "grad_norm": 0.0,
+      "learning_rate": 2.648305577439545e-06,
+      "loss": 0.913,
+      "step": 21173
+    },
+    {
+      "epoch": 0.7700476415608976,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6475071755742353e-06,
+      "loss": 0.9078,
+      "step": 21174
+    },
+    {
+      "epoch": 0.7700840091646362,
+      "grad_norm": 0.0,
+      "learning_rate": 2.646708875714329e-06,
+      "loss": 1.0119,
+      "step": 21175
+    },
+    {
+      "epoch": 0.7701203767683747,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6459106778709066e-06,
+      "loss": 0.8259,
+      "step": 21176
+    },
+    {
+      "epoch": 0.7701567443721133,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6451125820550393e-06,
+      "loss": 0.8812,
+      "step": 21177
+    },
+    {
+      "epoch": 0.7701931119758519,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6443145882778e-06,
+      "loss": 0.9417,
+      "step": 21178
+    },
+    {
+      "epoch": 0.7702294795795905,
+      "grad_norm": 0.0,
+      "learning_rate": 2.643516696550259e-06,
+      "loss": 0.9733,
+      "step": 21179
+    },
+    {
+      "epoch": 0.7702658471833291,
+      "grad_norm": 0.0,
+      "learning_rate": 2.642718906883487e-06,
+      "loss": 0.9636,
+      "step": 21180
+    },
+    {
+      "epoch": 0.7703022147870677,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6419212192885478e-06,
+      "loss": 0.8073,
+      "step": 21181
+    },
+    {
+      "epoch": 0.7703385823908062,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6411236337765154e-06,
+      "loss": 0.8747,
+      "step": 21182
+    },
+    {
+      "epoch": 0.7703749499945448,
+      "grad_norm": 0.0,
+      "learning_rate": 2.640326150358451e-06,
+      "loss": 0.8821,
+      "step": 21183
+    },
+    {
+      "epoch": 0.7704113175982834,
+      "grad_norm": 0.0,
+      "learning_rate": 2.639528769045421e-06,
+      "loss": 0.9201,
+      "step": 21184
+    },
+    {
+      "epoch": 0.770447685202022,
+      "grad_norm": 0.0,
+      "learning_rate": 2.638731489848482e-06,
+      "loss": 0.9432,
+      "step": 21185
+    },
+    {
+      "epoch": 0.7704840528057606,
+      "grad_norm": 0.0,
+      "learning_rate": 2.637934312778705e-06,
+      "loss": 0.904,
+      "step": 21186
+    },
+    {
+      "epoch": 0.7705204204094992,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6371372378471384e-06,
+      "loss": 0.9154,
+      "step": 21187
+    },
+    {
+      "epoch": 0.7705567880132378,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6363402650648496e-06,
+      "loss": 0.9856,
+      "step": 21188
+    },
+    {
+      "epoch": 0.7705931556169764,
+      "grad_norm": 0.0,
+      "learning_rate": 2.635543394442891e-06,
+      "loss": 0.9088,
+      "step": 21189
+    },
+    {
+      "epoch": 0.770629523220715,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6347466259923194e-06,
+      "loss": 0.9746,
+      "step": 21190
+    },
+    {
+      "epoch": 0.7706658908244536,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6339499597241856e-06,
+      "loss": 0.8769,
+      "step": 21191
+    },
+    {
+      "epoch": 0.7707022584281922,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6331533956495503e-06,
+      "loss": 0.9261,
+      "step": 21192
+    },
+    {
+      "epoch": 0.7707386260319308,
+      "grad_norm": 0.0,
+      "learning_rate": 2.632356933779454e-06,
+      "loss": 0.9677,
+      "step": 21193
+    },
+    {
+      "epoch": 0.7707749936356694,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6315605741249563e-06,
+      "loss": 0.9266,
+      "step": 21194
+    },
+    {
+      "epoch": 0.770811361239408,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6307643166970997e-06,
+      "loss": 0.8445,
+      "step": 21195
+    },
+    {
+      "epoch": 0.7708477288431466,
+      "grad_norm": 0.0,
+      "learning_rate": 2.629968161506933e-06,
+      "loss": 1.0532,
+      "step": 21196
+    },
+    {
+      "epoch": 0.7708840964468852,
+      "grad_norm": 0.0,
+      "learning_rate": 2.629172108565501e-06,
+      "loss": 1.0942,
+      "step": 21197
+    },
+    {
+      "epoch": 0.7709204640506238,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6283761578838453e-06,
+      "loss": 0.9442,
+      "step": 21198
+    },
+    {
+      "epoch": 0.7709568316543622,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6275803094730167e-06,
+      "loss": 0.9951,
+      "step": 21199
+    },
+    {
+      "epoch": 0.7709931992581008,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6267845633440457e-06,
+      "loss": 0.9469,
+      "step": 21200
+    },
+    {
+      "epoch": 0.7710295668618394,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6259889195079814e-06,
+      "loss": 0.9791,
+      "step": 21201
+    },
+    {
+      "epoch": 0.771065934465578,
+      "grad_norm": 0.0,
+      "learning_rate": 2.625193377975859e-06,
+      "loss": 0.9099,
+      "step": 21202
+    },
+    {
+      "epoch": 0.7711023020693166,
+      "grad_norm": 0.0,
+      "learning_rate": 2.624397938758714e-06,
+      "loss": 0.7955,
+      "step": 21203
+    },
+    {
+      "epoch": 0.7711386696730552,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6236026018675796e-06,
+      "loss": 0.7471,
+      "step": 21204
+    },
+    {
+      "epoch": 0.7711750372767938,
+      "grad_norm": 0.0,
+      "learning_rate": 2.622807367313499e-06,
+      "loss": 0.9168,
+      "step": 21205
+    },
+    {
+      "epoch": 0.7712114048805324,
+      "grad_norm": 0.0,
+      "learning_rate": 2.622012235107495e-06,
+      "loss": 0.9465,
+      "step": 21206
+    },
+    {
+      "epoch": 0.771247772484271,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6212172052606054e-06,
+      "loss": 1.0209,
+      "step": 21207
+    },
+    {
+      "epoch": 0.7712841400880096,
+      "grad_norm": 0.0,
+      "learning_rate": 2.620422277783857e-06,
+      "loss": 0.8026,
+      "step": 21208
+    },
+    {
+      "epoch": 0.7713205076917482,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6196274526882803e-06,
+      "loss": 0.8606,
+      "step": 21209
+    },
+    {
+      "epoch": 0.7713568752954868,
+      "grad_norm": 0.0,
+      "learning_rate": 2.618832729984898e-06,
+      "loss": 0.9023,
+      "step": 21210
+    },
+    {
+      "epoch": 0.7713932428992254,
+      "grad_norm": 0.0,
+      "learning_rate": 2.618038109684743e-06,
+      "loss": 0.8819,
+      "step": 21211
+    },
+    {
+      "epoch": 0.771429610502964,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6172435917988346e-06,
+      "loss": 0.9924,
+      "step": 21212
+    },
+    {
+      "epoch": 0.7714659781067026,
+      "grad_norm": 0.0,
+      "learning_rate": 2.616449176338197e-06,
+      "loss": 0.9771,
+      "step": 21213
+    },
+    {
+      "epoch": 0.7715023457104412,
+      "grad_norm": 0.0,
+      "learning_rate": 2.615654863313851e-06,
+      "loss": 0.893,
+      "step": 21214
+    },
+    {
+      "epoch": 0.7715387133141798,
+      "grad_norm": 0.0,
+      "learning_rate": 2.614860652736817e-06,
+      "loss": 0.8523,
+      "step": 21215
+    },
+    {
+      "epoch": 0.7715750809179183,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6140665446181124e-06,
+      "loss": 1.0018,
+      "step": 21216
+    },
+    {
+      "epoch": 0.7716114485216569,
+      "grad_norm": 0.0,
+      "learning_rate": 2.613272538968753e-06,
+      "loss": 0.8614,
+      "step": 21217
+    },
+    {
+      "epoch": 0.7716478161253955,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6124786357997622e-06,
+      "loss": 0.965,
+      "step": 21218
+    },
+    {
+      "epoch": 0.771684183729134,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6116848351221435e-06,
+      "loss": 0.88,
+      "step": 21219
+    },
+    {
+      "epoch": 0.7717205513328727,
+      "grad_norm": 0.0,
+      "learning_rate": 2.610891136946918e-06,
+      "loss": 1.0029,
+      "step": 21220
+    },
+    {
+      "epoch": 0.7717569189366112,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6100975412850936e-06,
+      "loss": 0.8466,
+      "step": 21221
+    },
+    {
+      "epoch": 0.7717932865403498,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6093040481476815e-06,
+      "loss": 0.8487,
+      "step": 21222
+    },
+    {
+      "epoch": 0.7718296541440884,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6085106575456863e-06,
+      "loss": 0.924,
+      "step": 21223
+    },
+    {
+      "epoch": 0.771866021747827,
+      "grad_norm": 0.0,
+      "learning_rate": 2.607717369490125e-06,
+      "loss": 1.0053,
+      "step": 21224
+    },
+    {
+      "epoch": 0.7719023893515656,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6069241839919913e-06,
+      "loss": 0.9155,
+      "step": 21225
+    },
+    {
+      "epoch": 0.7719387569553042,
+      "grad_norm": 0.0,
+      "learning_rate": 2.606131101062297e-06,
+      "loss": 0.8096,
+      "step": 21226
+    },
+    {
+      "epoch": 0.7719751245590428,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6053381207120443e-06,
+      "loss": 0.9017,
+      "step": 21227
+    },
+    {
+      "epoch": 0.7720114921627814,
+      "grad_norm": 0.0,
+      "learning_rate": 2.604545242952232e-06,
+      "loss": 1.0504,
+      "step": 21228
+    },
+    {
+      "epoch": 0.77204785976652,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6037524677938596e-06,
+      "loss": 0.959,
+      "step": 21229
+    },
+    {
+      "epoch": 0.7720842273702586,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6029597952479315e-06,
+      "loss": 0.9313,
+      "step": 21230
+    },
+    {
+      "epoch": 0.7721205949739972,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6021672253254405e-06,
+      "loss": 0.896,
+      "step": 21231
+    },
+    {
+      "epoch": 0.7721569625777358,
+      "grad_norm": 0.0,
+      "learning_rate": 2.601374758037383e-06,
+      "loss": 0.9904,
+      "step": 21232
+    },
+    {
+      "epoch": 0.7721933301814743,
+      "grad_norm": 0.0,
+      "learning_rate": 2.600582393394754e-06,
+      "loss": 0.8594,
+      "step": 21233
+    },
+    {
+      "epoch": 0.7722296977852129,
+      "grad_norm": 0.0,
+      "learning_rate": 2.599790131408545e-06,
+      "loss": 0.8782,
+      "step": 21234
+    },
+    {
+      "epoch": 0.7722660653889515,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5989979720897485e-06,
+      "loss": 0.948,
+      "step": 21235
+    },
+    {
+      "epoch": 0.7723024329926901,
+      "grad_norm": 0.0,
+      "learning_rate": 2.598205915449352e-06,
+      "loss": 0.9629,
+      "step": 21236
+    },
+    {
+      "epoch": 0.7723388005964287,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5974139614983527e-06,
+      "loss": 0.8595,
+      "step": 21237
+    },
+    {
+      "epoch": 0.7723751682001673,
+      "grad_norm": 0.0,
+      "learning_rate": 2.596622110247725e-06,
+      "loss": 0.937,
+      "step": 21238
+    },
+    {
+      "epoch": 0.7724115358039059,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5958303617084657e-06,
+      "loss": 0.8865,
+      "step": 21239
+    },
+    {
+      "epoch": 0.7724479034076445,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5950387158915548e-06,
+      "loss": 0.9175,
+      "step": 21240
+    },
+    {
+      "epoch": 0.7724842710113831,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5942471728079742e-06,
+      "loss": 0.7971,
+      "step": 21241
+    },
+    {
+      "epoch": 0.7725206386151217,
+      "grad_norm": 0.0,
+      "learning_rate": 2.593455732468705e-06,
+      "loss": 0.9703,
+      "step": 21242
+    },
+    {
+      "epoch": 0.7725570062188603,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5926643948847306e-06,
+      "loss": 0.9009,
+      "step": 21243
+    },
+    {
+      "epoch": 0.7725933738225989,
+      "grad_norm": 0.0,
+      "learning_rate": 2.59187316006703e-06,
+      "loss": 0.8734,
+      "step": 21244
+    },
+    {
+      "epoch": 0.7726297414263374,
+      "grad_norm": 0.0,
+      "learning_rate": 2.591082028026577e-06,
+      "loss": 1.0015,
+      "step": 21245
+    },
+    {
+      "epoch": 0.772666109030076,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5902909987743496e-06,
+      "loss": 1.0032,
+      "step": 21246
+    },
+    {
+      "epoch": 0.7727024766338146,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5895000723213213e-06,
+      "loss": 1.0797,
+      "step": 21247
+    },
+    {
+      "epoch": 0.7727388442375532,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5887092486784628e-06,
+      "loss": 0.8314,
+      "step": 21248
+    },
+    {
+      "epoch": 0.7727752118412918,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5879185278567507e-06,
+      "loss": 0.9535,
+      "step": 21249
+    },
+    {
+      "epoch": 0.7728115794450303,
+      "grad_norm": 0.0,
+      "learning_rate": 2.587127909867153e-06,
+      "loss": 0.9522,
+      "step": 21250
+    },
+    {
+      "epoch": 0.7728479470487689,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5863373947206383e-06,
+      "loss": 0.9723,
+      "step": 21251
+    },
+    {
+      "epoch": 0.7728843146525075,
+      "grad_norm": 0.0,
+      "learning_rate": 2.585546982428173e-06,
+      "loss": 0.9593,
+      "step": 21252
+    },
+    {
+      "epoch": 0.7729206822562461,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5847566730007247e-06,
+      "loss": 0.8677,
+      "step": 21253
+    },
+    {
+      "epoch": 0.7729570498599847,
+      "grad_norm": 0.0,
+      "learning_rate": 2.583966466449256e-06,
+      "loss": 0.9624,
+      "step": 21254
+    },
+    {
+      "epoch": 0.7729934174637233,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5831763627847274e-06,
+      "loss": 0.8837,
+      "step": 21255
+    },
+    {
+      "epoch": 0.7730297850674619,
+      "grad_norm": 0.0,
+      "learning_rate": 2.58238636201811e-06,
+      "loss": 0.8483,
+      "step": 21256
+    },
+    {
+      "epoch": 0.7730661526712005,
+      "grad_norm": 0.0,
+      "learning_rate": 2.581596464160351e-06,
+      "loss": 0.8486,
+      "step": 21257
+    },
+    {
+      "epoch": 0.7731025202749391,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5808066692224197e-06,
+      "loss": 0.8612,
+      "step": 21258
+    },
+    {
+      "epoch": 0.7731388878786777,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5800169772152683e-06,
+      "loss": 1.0367,
+      "step": 21259
+    },
+    {
+      "epoch": 0.7731752554824163,
+      "grad_norm": 0.0,
+      "learning_rate": 2.579227388149854e-06,
+      "loss": 0.8385,
+      "step": 21260
+    },
+    {
+      "epoch": 0.7732116230861549,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5784379020371286e-06,
+      "loss": 0.8978,
+      "step": 21261
+    },
+    {
+      "epoch": 0.7732479906898935,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5776485188880507e-06,
+      "loss": 0.9298,
+      "step": 21262
+    },
+    {
+      "epoch": 0.7732843582936321,
+      "grad_norm": 0.0,
+      "learning_rate": 2.576859238713568e-06,
+      "loss": 0.9149,
+      "step": 21263
+    },
+    {
+      "epoch": 0.7733207258973707,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5760700615246314e-06,
+      "loss": 0.9606,
+      "step": 21264
+    },
+    {
+      "epoch": 0.7733570935011093,
+      "grad_norm": 0.0,
+      "learning_rate": 2.575280987332187e-06,
+      "loss": 0.9141,
+      "step": 21265
+    },
+    {
+      "epoch": 0.7733934611048479,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5744920161471897e-06,
+      "loss": 0.8919,
+      "step": 21266
+    },
+    {
+      "epoch": 0.7734298287085863,
+      "grad_norm": 0.0,
+      "learning_rate": 2.573703147980575e-06,
+      "loss": 1.0292,
+      "step": 21267
+    },
+    {
+      "epoch": 0.7734661963123249,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5729143828432957e-06,
+      "loss": 0.9159,
+      "step": 21268
+    },
+    {
+      "epoch": 0.7735025639160635,
+      "grad_norm": 0.0,
+      "learning_rate": 2.572125720746291e-06,
+      "loss": 0.8994,
+      "step": 21269
+    },
+    {
+      "epoch": 0.7735389315198021,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5713371617005033e-06,
+      "loss": 0.84,
+      "step": 21270
+    },
+    {
+      "epoch": 0.7735752991235407,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5705487057168734e-06,
+      "loss": 0.8601,
+      "step": 21271
+    },
+    {
+      "epoch": 0.7736116667272793,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5697603528063386e-06,
+      "loss": 0.9171,
+      "step": 21272
+    },
+    {
+      "epoch": 0.7736480343310179,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5689721029798364e-06,
+      "loss": 0.8755,
+      "step": 21273
+    },
+    {
+      "epoch": 0.7736844019347565,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5681839562483e-06,
+      "loss": 0.8729,
+      "step": 21274
+    },
+    {
+      "epoch": 0.7737207695384951,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5673959126226698e-06,
+      "loss": 0.9581,
+      "step": 21275
+    },
+    {
+      "epoch": 0.7737571371422337,
+      "grad_norm": 0.0,
+      "learning_rate": 2.566607972113877e-06,
+      "loss": 0.892,
+      "step": 21276
+    },
+    {
+      "epoch": 0.7737935047459723,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5658201347328517e-06,
+      "loss": 0.8925,
+      "step": 21277
+    },
+    {
+      "epoch": 0.7738298723497109,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5650324004905238e-06,
+      "loss": 0.8842,
+      "step": 21278
+    },
+    {
+      "epoch": 0.7738662399534495,
+      "grad_norm": 0.0,
+      "learning_rate": 2.564244769397822e-06,
+      "loss": 0.98,
+      "step": 21279
+    },
+    {
+      "epoch": 0.7739026075571881,
+      "grad_norm": 0.0,
+      "learning_rate": 2.563457241465671e-06,
+      "loss": 0.9344,
+      "step": 21280
+    },
+    {
+      "epoch": 0.7739389751609267,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5626698167050036e-06,
+      "loss": 0.9566,
+      "step": 21281
+    },
+    {
+      "epoch": 0.7739753427646653,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5618824951267395e-06,
+      "loss": 0.8754,
+      "step": 21282
+    },
+    {
+      "epoch": 0.7740117103684038,
+      "grad_norm": 0.0,
+      "learning_rate": 2.561095276741804e-06,
+      "loss": 0.843,
+      "step": 21283
+    },
+    {
+      "epoch": 0.7740480779721424,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5603081615611125e-06,
+      "loss": 0.9312,
+      "step": 21284
+    },
+    {
+      "epoch": 0.774084445575881,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5595211495955965e-06,
+      "loss": 0.9865,
+      "step": 21285
+    },
+    {
+      "epoch": 0.7741208131796196,
+      "grad_norm": 0.0,
+      "learning_rate": 2.558734240856161e-06,
+      "loss": 1.0009,
+      "step": 21286
+    },
+    {
+      "epoch": 0.7741571807833582,
+      "grad_norm": 0.0,
+      "learning_rate": 2.557947435353735e-06,
+      "loss": 0.8861,
+      "step": 21287
+    },
+    {
+      "epoch": 0.7741935483870968,
+      "grad_norm": 0.0,
+      "learning_rate": 2.557160733099228e-06,
+      "loss": 0.9938,
+      "step": 21288
+    },
+    {
+      "epoch": 0.7742299159908353,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5563741341035555e-06,
+      "loss": 1.0057,
+      "step": 21289
+    },
+    {
+      "epoch": 0.774266283594574,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5555876383776314e-06,
+      "loss": 0.8494,
+      "step": 21290
+    },
+    {
+      "epoch": 0.7743026511983125,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5548012459323667e-06,
+      "loss": 1.048,
+      "step": 21291
+    },
+    {
+      "epoch": 0.7743390188020511,
+      "grad_norm": 0.0,
+      "learning_rate": 2.554014956778671e-06,
+      "loss": 0.8708,
+      "step": 21292
+    },
+    {
+      "epoch": 0.7743753864057897,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5532287709274506e-06,
+      "loss": 1.0171,
+      "step": 21293
+    },
+    {
+      "epoch": 0.7744117540095283,
+      "grad_norm": 0.0,
+      "learning_rate": 2.552442688389618e-06,
+      "loss": 1.0085,
+      "step": 21294
+    },
+    {
+      "epoch": 0.7744481216132669,
+      "grad_norm": 0.0,
+      "learning_rate": 2.551656709176077e-06,
+      "loss": 0.9725,
+      "step": 21295
+    },
+    {
+      "epoch": 0.7744844892170055,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5508708332977304e-06,
+      "loss": 0.7603,
+      "step": 21296
+    },
+    {
+      "epoch": 0.7745208568207441,
+      "grad_norm": 0.0,
+      "learning_rate": 2.55008506076548e-06,
+      "loss": 0.9304,
+      "step": 21297
+    },
+    {
+      "epoch": 0.7745572244244827,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5492993915902353e-06,
+      "loss": 0.9462,
+      "step": 21298
+    },
+    {
+      "epoch": 0.7745935920282213,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5485138257828845e-06,
+      "loss": 0.8461,
+      "step": 21299
+    },
+    {
+      "epoch": 0.7746299596319598,
+      "grad_norm": 0.0,
+      "learning_rate": 2.547728363354335e-06,
+      "loss": 0.9446,
+      "step": 21300
+    },
+    {
+      "epoch": 0.7746663272356984,
+      "grad_norm": 0.0,
+      "learning_rate": 2.54694300431548e-06,
+      "loss": 0.743,
+      "step": 21301
+    },
+    {
+      "epoch": 0.774702694839437,
+      "grad_norm": 0.0,
+      "learning_rate": 2.546157748677217e-06,
+      "loss": 0.8807,
+      "step": 21302
+    },
+    {
+      "epoch": 0.7747390624431756,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5453725964504372e-06,
+      "loss": 0.9233,
+      "step": 21303
+    },
+    {
+      "epoch": 0.7747754300469142,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5445875476460413e-06,
+      "loss": 0.8587,
+      "step": 21304
+    },
+    {
+      "epoch": 0.7748117976506528,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5438026022749096e-06,
+      "loss": 1.0034,
+      "step": 21305
+    },
+    {
+      "epoch": 0.7748481652543914,
+      "grad_norm": 0.0,
+      "learning_rate": 2.54301776034794e-06,
+      "loss": 0.9637,
+      "step": 21306
+    },
+    {
+      "epoch": 0.77488453285813,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5422330218760195e-06,
+      "loss": 0.9385,
+      "step": 21307
+    },
+    {
+      "epoch": 0.7749209004618686,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5414483868700344e-06,
+      "loss": 1.04,
+      "step": 21308
+    },
+    {
+      "epoch": 0.7749572680656072,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5406638553408714e-06,
+      "loss": 0.8936,
+      "step": 21309
+    },
+    {
+      "epoch": 0.7749936356693458,
+      "grad_norm": 0.0,
+      "learning_rate": 2.539879427299411e-06,
+      "loss": 0.986,
+      "step": 21310
+    },
+    {
+      "epoch": 0.7750300032730844,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5390951027565438e-06,
+      "loss": 0.9554,
+      "step": 21311
+    },
+    {
+      "epoch": 0.775066370876823,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5383108817231416e-06,
+      "loss": 0.8548,
+      "step": 21312
+    },
+    {
+      "epoch": 0.7751027384805615,
+      "grad_norm": 0.0,
+      "learning_rate": 2.537526764210092e-06,
+      "loss": 0.9447,
+      "step": 21313
+    },
+    {
+      "epoch": 0.7751391060843001,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5367427502282704e-06,
+      "loss": 0.7362,
+      "step": 21314
+    },
+    {
+      "epoch": 0.7751754736880387,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5359588397885536e-06,
+      "loss": 0.8801,
+      "step": 21315
+    },
+    {
+      "epoch": 0.7752118412917773,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5351750329018156e-06,
+      "loss": 0.7502,
+      "step": 21316
+    },
+    {
+      "epoch": 0.7752482088955158,
+      "grad_norm": 0.0,
+      "learning_rate": 2.534391329578938e-06,
+      "loss": 0.9288,
+      "step": 21317
+    },
+    {
+      "epoch": 0.7752845764992544,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5336077298307836e-06,
+      "loss": 0.9137,
+      "step": 21318
+    },
+    {
+      "epoch": 0.775320944102993,
+      "grad_norm": 0.0,
+      "learning_rate": 2.532824233668231e-06,
+      "loss": 0.9331,
+      "step": 21319
+    },
+    {
+      "epoch": 0.7753573117067316,
+      "grad_norm": 0.0,
+      "learning_rate": 2.532040841102147e-06,
+      "loss": 1.0427,
+      "step": 21320
+    },
+    {
+      "epoch": 0.7753936793104702,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5312575521434014e-06,
+      "loss": 1.0776,
+      "step": 21321
+    },
+    {
+      "epoch": 0.7754300469142088,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5304743668028574e-06,
+      "loss": 0.9618,
+      "step": 21322
+    },
+    {
+      "epoch": 0.7754664145179474,
+      "grad_norm": 0.0,
+      "learning_rate": 2.529691285091389e-06,
+      "loss": 0.9666,
+      "step": 21323
+    },
+    {
+      "epoch": 0.775502782121686,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5289083070198496e-06,
+      "loss": 0.9042,
+      "step": 21324
+    },
+    {
+      "epoch": 0.7755391497254246,
+      "grad_norm": 0.0,
+      "learning_rate": 2.528125432599111e-06,
+      "loss": 0.9223,
+      "step": 21325
+    },
+    {
+      "epoch": 0.7755755173291632,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5273426618400297e-06,
+      "loss": 0.9129,
+      "step": 21326
+    },
+    {
+      "epoch": 0.7756118849329018,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5265599947534667e-06,
+      "loss": 0.9731,
+      "step": 21327
+    },
+    {
+      "epoch": 0.7756482525366404,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5257774313502816e-06,
+      "loss": 0.9987,
+      "step": 21328
+    },
+    {
+      "epoch": 0.775684620140379,
+      "grad_norm": 0.0,
+      "learning_rate": 2.524994971641326e-06,
+      "loss": 0.9102,
+      "step": 21329
+    },
+    {
+      "epoch": 0.7757209877441176,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5242126156374656e-06,
+      "loss": 0.9521,
+      "step": 21330
+    },
+    {
+      "epoch": 0.7757573553478562,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5234303633495425e-06,
+      "loss": 0.9716,
+      "step": 21331
+    },
+    {
+      "epoch": 0.7757937229515948,
+      "grad_norm": 0.0,
+      "learning_rate": 2.522648214788419e-06,
+      "loss": 0.8254,
+      "step": 21332
+    },
+    {
+      "epoch": 0.7758300905553334,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5218661699649416e-06,
+      "loss": 0.8835,
+      "step": 21333
+    },
+    {
+      "epoch": 0.7758664581590718,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5210842288899617e-06,
+      "loss": 0.9515,
+      "step": 21334
+    },
+    {
+      "epoch": 0.7759028257628104,
+      "grad_norm": 0.0,
+      "learning_rate": 2.520302391574323e-06,
+      "loss": 0.9249,
+      "step": 21335
+    },
+    {
+      "epoch": 0.775939193366549,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5195206580288833e-06,
+      "loss": 0.9109,
+      "step": 21336
+    },
+    {
+      "epoch": 0.7759755609702876,
+      "grad_norm": 0.0,
+      "learning_rate": 2.518739028264474e-06,
+      "loss": 0.8594,
+      "step": 21337
+    },
+    {
+      "epoch": 0.7760119285740262,
+      "grad_norm": 0.0,
+      "learning_rate": 2.51795750229195e-06,
+      "loss": 0.9431,
+      "step": 21338
+    },
+    {
+      "epoch": 0.7760482961777648,
+      "grad_norm": 0.0,
+      "learning_rate": 2.51717608012215e-06,
+      "loss": 0.9266,
+      "step": 21339
+    },
+    {
+      "epoch": 0.7760846637815034,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5163947617659144e-06,
+      "loss": 0.8559,
+      "step": 21340
+    },
+    {
+      "epoch": 0.776121031385242,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5156135472340816e-06,
+      "loss": 0.9613,
+      "step": 21341
+    },
+    {
+      "epoch": 0.7761573989889806,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5148324365374945e-06,
+      "loss": 0.9669,
+      "step": 21342
+    },
+    {
+      "epoch": 0.7761937665927192,
+      "grad_norm": 0.0,
+      "learning_rate": 2.514051429686988e-06,
+      "loss": 0.931,
+      "step": 21343
+    },
+    {
+      "epoch": 0.7762301341964578,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5132705266933953e-06,
+      "loss": 0.9491,
+      "step": 21344
+    },
+    {
+      "epoch": 0.7762665018001964,
+      "grad_norm": 0.0,
+      "learning_rate": 2.512489727567553e-06,
+      "loss": 0.861,
+      "step": 21345
+    },
+    {
+      "epoch": 0.776302869403935,
+      "grad_norm": 0.0,
+      "learning_rate": 2.511709032320291e-06,
+      "loss": 0.7902,
+      "step": 21346
+    },
+    {
+      "epoch": 0.7763392370076736,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5109284409624422e-06,
+      "loss": 0.9202,
+      "step": 21347
+    },
+    {
+      "epoch": 0.7763756046114122,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5101479535048334e-06,
+      "loss": 0.8725,
+      "step": 21348
+    },
+    {
+      "epoch": 0.7764119722151508,
+      "grad_norm": 0.0,
+      "learning_rate": 2.509367569958301e-06,
+      "loss": 0.9066,
+      "step": 21349
+    },
+    {
+      "epoch": 0.7764483398188894,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5085872903336593e-06,
+      "loss": 0.9559,
+      "step": 21350
+    },
+    {
+      "epoch": 0.7764847074226279,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5078071146417428e-06,
+      "loss": 0.9629,
+      "step": 21351
+    },
+    {
+      "epoch": 0.7765210750263665,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5070270428933728e-06,
+      "loss": 0.8799,
+      "step": 21352
+    },
+    {
+      "epoch": 0.7765574426301051,
+      "grad_norm": 0.0,
+      "learning_rate": 2.506247075099372e-06,
+      "loss": 0.9967,
+      "step": 21353
+    },
+    {
+      "epoch": 0.7765938102338437,
+      "grad_norm": 0.0,
+      "learning_rate": 2.505467211270557e-06,
+      "loss": 0.971,
+      "step": 21354
+    },
+    {
+      "epoch": 0.7766301778375823,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5046874514177544e-06,
+      "loss": 1.0092,
+      "step": 21355
+    },
+    {
+      "epoch": 0.7766665454413209,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5039077955517787e-06,
+      "loss": 0.7874,
+      "step": 21356
+    },
+    {
+      "epoch": 0.7767029130450595,
+      "grad_norm": 0.0,
+      "learning_rate": 2.503128243683447e-06,
+      "loss": 0.9591,
+      "step": 21357
+    },
+    {
+      "epoch": 0.776739280648798,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5023487958235738e-06,
+      "loss": 0.8321,
+      "step": 21358
+    },
+    {
+      "epoch": 0.7767756482525366,
+      "grad_norm": 0.0,
+      "learning_rate": 2.501569451982974e-06,
+      "loss": 0.9495,
+      "step": 21359
+    },
+    {
+      "epoch": 0.7768120158562752,
+      "grad_norm": 0.0,
+      "learning_rate": 2.500790212172456e-06,
+      "loss": 1.0772,
+      "step": 21360
+    },
+    {
+      "epoch": 0.7768483834600138,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5000110764028363e-06,
+      "loss": 0.9366,
+      "step": 21361
+    },
+    {
+      "epoch": 0.7768847510637524,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4992320446849226e-06,
+      "loss": 0.9794,
+      "step": 21362
+    },
+    {
+      "epoch": 0.776921118667491,
+      "grad_norm": 0.0,
+      "learning_rate": 2.498453117029521e-06,
+      "loss": 0.9245,
+      "step": 21363
+    },
+    {
+      "epoch": 0.7769574862712296,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4976742934474376e-06,
+      "loss": 0.9425,
+      "step": 21364
+    },
+    {
+      "epoch": 0.7769938538749682,
+      "grad_norm": 0.0,
+      "learning_rate": 2.496895573949484e-06,
+      "loss": 0.936,
+      "step": 21365
+    },
+    {
+      "epoch": 0.7770302214787068,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4961169585464553e-06,
+      "loss": 0.9159,
+      "step": 21366
+    },
+    {
+      "epoch": 0.7770665890824453,
+      "grad_norm": 0.0,
+      "learning_rate": 2.495338447249155e-06,
+      "loss": 0.9012,
+      "step": 21367
+    },
+    {
+      "epoch": 0.7771029566861839,
+      "grad_norm": 0.0,
+      "learning_rate": 2.494560040068392e-06,
+      "loss": 0.9128,
+      "step": 21368
+    },
+    {
+      "epoch": 0.7771393242899225,
+      "grad_norm": 0.0,
+      "learning_rate": 2.493781737014953e-06,
+      "loss": 0.814,
+      "step": 21369
+    },
+    {
+      "epoch": 0.7771756918936611,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4930035380996467e-06,
+      "loss": 0.8683,
+      "step": 21370
+    },
+    {
+      "epoch": 0.7772120594973997,
+      "grad_norm": 0.0,
+      "learning_rate": 2.492225443333264e-06,
+      "loss": 1.0154,
+      "step": 21371
+    },
+    {
+      "epoch": 0.7772484271011383,
+      "grad_norm": 0.0,
+      "learning_rate": 2.491447452726602e-06,
+      "loss": 0.9175,
+      "step": 21372
+    },
+    {
+      "epoch": 0.7772847947048769,
+      "grad_norm": 0.0,
+      "learning_rate": 2.49066956629045e-06,
+      "loss": 0.9283,
+      "step": 21373
+    },
+    {
+      "epoch": 0.7773211623086155,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4898917840356062e-06,
+      "loss": 0.9959,
+      "step": 21374
+    },
+    {
+      "epoch": 0.7773575299123541,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4891141059728587e-06,
+      "loss": 0.9511,
+      "step": 21375
+    },
+    {
+      "epoch": 0.7773938975160927,
+      "grad_norm": 0.0,
+      "learning_rate": 2.488336532112996e-06,
+      "loss": 1.1024,
+      "step": 21376
+    },
+    {
+      "epoch": 0.7774302651198313,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4875590624668046e-06,
+      "loss": 1.0422,
+      "step": 21377
+    },
+    {
+      "epoch": 0.7774666327235699,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4867816970450775e-06,
+      "loss": 0.8788,
+      "step": 21378
+    },
+    {
+      "epoch": 0.7775030003273085,
+      "grad_norm": 0.0,
+      "learning_rate": 2.486004435858588e-06,
+      "loss": 0.9668,
+      "step": 21379
+    },
+    {
+      "epoch": 0.777539367931047,
+      "grad_norm": 0.0,
+      "learning_rate": 2.48522727891813e-06,
+      "loss": 1.0003,
+      "step": 21380
+    },
+    {
+      "epoch": 0.7775757355347857,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4844502262344805e-06,
+      "loss": 0.9378,
+      "step": 21381
+    },
+    {
+      "epoch": 0.7776121031385242,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4836732778184214e-06,
+      "loss": 0.9008,
+      "step": 21382
+    },
+    {
+      "epoch": 0.7776484707422628,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4828964336807283e-06,
+      "loss": 0.9405,
+      "step": 21383
+    },
+    {
+      "epoch": 0.7776848383460013,
+      "grad_norm": 0.0,
+      "learning_rate": 2.482119693832188e-06,
+      "loss": 0.8477,
+      "step": 21384
+    },
+    {
+      "epoch": 0.7777212059497399,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4813430582835676e-06,
+      "loss": 0.915,
+      "step": 21385
+    },
+    {
+      "epoch": 0.7777575735534785,
+      "grad_norm": 0.0,
+      "learning_rate": 2.48056652704564e-06,
+      "loss": 0.9733,
+      "step": 21386
+    },
+    {
+      "epoch": 0.7777939411572171,
+      "grad_norm": 0.0,
+      "learning_rate": 2.479790100129188e-06,
+      "loss": 0.9846,
+      "step": 21387
+    },
+    {
+      "epoch": 0.7778303087609557,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4790137775449787e-06,
+      "loss": 0.8174,
+      "step": 21388
+    },
+    {
+      "epoch": 0.7778666763646943,
+      "grad_norm": 0.0,
+      "learning_rate": 2.478237559303781e-06,
+      "loss": 0.881,
+      "step": 21389
+    },
+    {
+      "epoch": 0.7779030439684329,
+      "grad_norm": 0.0,
+      "learning_rate": 2.477461445416366e-06,
+      "loss": 0.9554,
+      "step": 21390
+    },
+    {
+      "epoch": 0.7779394115721715,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4766854358935e-06,
+      "loss": 1.0024,
+      "step": 21391
+    },
+    {
+      "epoch": 0.7779757791759101,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4759095307459467e-06,
+      "loss": 0.924,
+      "step": 21392
+    },
+    {
+      "epoch": 0.7780121467796487,
+      "grad_norm": 0.0,
+      "learning_rate": 2.475133729984477e-06,
+      "loss": 0.8909,
+      "step": 21393
+    },
+    {
+      "epoch": 0.7780485143833873,
+      "grad_norm": 0.0,
+      "learning_rate": 2.474358033619849e-06,
+      "loss": 1.0704,
+      "step": 21394
+    },
+    {
+      "epoch": 0.7780848819871259,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4735824416628262e-06,
+      "loss": 0.775,
+      "step": 21395
+    },
+    {
+      "epoch": 0.7781212495908645,
+      "grad_norm": 0.0,
+      "learning_rate": 2.472806954124166e-06,
+      "loss": 0.9695,
+      "step": 21396
+    },
+    {
+      "epoch": 0.7781576171946031,
+      "grad_norm": 0.0,
+      "learning_rate": 2.472031571014636e-06,
+      "loss": 0.9709,
+      "step": 21397
+    },
+    {
+      "epoch": 0.7781939847983417,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4712562923449802e-06,
+      "loss": 0.9777,
+      "step": 21398
+    },
+    {
+      "epoch": 0.7782303524020803,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4704811181259646e-06,
+      "loss": 0.9809,
+      "step": 21399
+    },
+    {
+      "epoch": 0.7782667200058189,
+      "grad_norm": 0.0,
+      "learning_rate": 2.469706048368341e-06,
+      "loss": 1.017,
+      "step": 21400
+    },
+    {
+      "epoch": 0.7783030876095574,
+      "grad_norm": 0.0,
+      "learning_rate": 2.468931083082862e-06,
+      "loss": 1.1557,
+      "step": 21401
+    },
+    {
+      "epoch": 0.778339455213296,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4681562222802745e-06,
+      "loss": 0.9846,
+      "step": 21402
+    },
+    {
+      "epoch": 0.7783758228170345,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4673814659713404e-06,
+      "loss": 0.8781,
+      "step": 21403
+    },
+    {
+      "epoch": 0.7784121904207731,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4666068141667988e-06,
+      "loss": 0.9492,
+      "step": 21404
+    },
+    {
+      "epoch": 0.7784485580245117,
+      "grad_norm": 0.0,
+      "learning_rate": 2.465832266877395e-06,
+      "loss": 0.9056,
+      "step": 21405
+    },
+    {
+      "epoch": 0.7784849256282503,
+      "grad_norm": 0.0,
+      "learning_rate": 2.465057824113882e-06,
+      "loss": 0.9131,
+      "step": 21406
+    },
+    {
+      "epoch": 0.7785212932319889,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4642834858870014e-06,
+      "loss": 0.9657,
+      "step": 21407
+    },
+    {
+      "epoch": 0.7785576608357275,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4635092522074954e-06,
+      "loss": 0.9865,
+      "step": 21408
+    },
+    {
+      "epoch": 0.7785940284394661,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4627351230861028e-06,
+      "loss": 0.9728,
+      "step": 21409
+    },
+    {
+      "epoch": 0.7786303960432047,
+      "grad_norm": 0.0,
+      "learning_rate": 2.461961098533572e-06,
+      "loss": 0.8122,
+      "step": 21410
+    },
+    {
+      "epoch": 0.7786667636469433,
+      "grad_norm": 0.0,
+      "learning_rate": 2.46118717856063e-06,
+      "loss": 0.9299,
+      "step": 21411
+    },
+    {
+      "epoch": 0.7787031312506819,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4604133631780227e-06,
+      "loss": 0.9193,
+      "step": 21412
+    },
+    {
+      "epoch": 0.7787394988544205,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4596396523964825e-06,
+      "loss": 0.8056,
+      "step": 21413
+    },
+    {
+      "epoch": 0.7787758664581591,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4588660462267445e-06,
+      "loss": 0.9866,
+      "step": 21414
+    },
+    {
+      "epoch": 0.7788122340618977,
+      "grad_norm": 0.0,
+      "learning_rate": 2.458092544679538e-06,
+      "loss": 0.9157,
+      "step": 21415
+    },
+    {
+      "epoch": 0.7788486016656363,
+      "grad_norm": 0.0,
+      "learning_rate": 2.457319147765601e-06,
+      "loss": 0.9184,
+      "step": 21416
+    },
+    {
+      "epoch": 0.7788849692693749,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4565458554956546e-06,
+      "loss": 0.8783,
+      "step": 21417
+    },
+    {
+      "epoch": 0.7789213368731134,
+      "grad_norm": 0.0,
+      "learning_rate": 2.455772667880435e-06,
+      "loss": 0.967,
+      "step": 21418
+    },
+    {
+      "epoch": 0.778957704476852,
+      "grad_norm": 0.0,
+      "learning_rate": 2.454999584930665e-06,
+      "loss": 0.8494,
+      "step": 21419
+    },
+    {
+      "epoch": 0.7789940720805906,
+      "grad_norm": 0.0,
+      "learning_rate": 2.454226606657071e-06,
+      "loss": 0.764,
+      "step": 21420
+    },
+    {
+      "epoch": 0.7790304396843292,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4534537330703733e-06,
+      "loss": 0.8729,
+      "step": 21421
+    },
+    {
+      "epoch": 0.7790668072880678,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4526809641813042e-06,
+      "loss": 1.0094,
+      "step": 21422
+    },
+    {
+      "epoch": 0.7791031748918064,
+      "grad_norm": 0.0,
+      "learning_rate": 2.451908300000576e-06,
+      "loss": 0.9357,
+      "step": 21423
+    },
+    {
+      "epoch": 0.779139542495545,
+      "grad_norm": 0.0,
+      "learning_rate": 2.451135740538908e-06,
+      "loss": 0.7926,
+      "step": 21424
+    },
+    {
+      "epoch": 0.7791759100992836,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4503632858070225e-06,
+      "loss": 0.862,
+      "step": 21425
+    },
+    {
+      "epoch": 0.7792122777030221,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4495909358156365e-06,
+      "loss": 1.0212,
+      "step": 21426
+    },
+    {
+      "epoch": 0.7792486453067607,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4488186905754627e-06,
+      "loss": 1.0055,
+      "step": 21427
+    },
+    {
+      "epoch": 0.7792850129104993,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4480465500972137e-06,
+      "loss": 0.7401,
+      "step": 21428
+    },
+    {
+      "epoch": 0.7793213805142379,
+      "grad_norm": 0.0,
+      "learning_rate": 2.44727451439161e-06,
+      "loss": 0.9335,
+      "step": 21429
+    },
+    {
+      "epoch": 0.7793577481179765,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4465025834693502e-06,
+      "loss": 1.0292,
+      "step": 21430
+    },
+    {
+      "epoch": 0.7793941157217151,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4457307573411525e-06,
+      "loss": 0.8903,
+      "step": 21431
+    },
+    {
+      "epoch": 0.7794304833254537,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4449590360177224e-06,
+      "loss": 0.8019,
+      "step": 21432
+    },
+    {
+      "epoch": 0.7794668509291923,
+      "grad_norm": 0.0,
+      "learning_rate": 2.444187419509767e-06,
+      "loss": 1.0025,
+      "step": 21433
+    },
+    {
+      "epoch": 0.7795032185329309,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4434159078279882e-06,
+      "loss": 0.9872,
+      "step": 21434
+    },
+    {
+      "epoch": 0.7795395861366694,
+      "grad_norm": 0.0,
+      "learning_rate": 2.442644500983098e-06,
+      "loss": 0.905,
+      "step": 21435
+    },
+    {
+      "epoch": 0.779575953740408,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4418731989857857e-06,
+      "loss": 0.8866,
+      "step": 21436
+    },
+    {
+      "epoch": 0.7796123213441466,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4411020018467633e-06,
+      "loss": 0.9566,
+      "step": 21437
+    },
+    {
+      "epoch": 0.7796486889478852,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4403309095767256e-06,
+      "loss": 0.8624,
+      "step": 21438
+    },
+    {
+      "epoch": 0.7796850565516238,
+      "grad_norm": 0.0,
+      "learning_rate": 2.439559922186372e-06,
+      "loss": 0.9023,
+      "step": 21439
+    },
+    {
+      "epoch": 0.7797214241553624,
+      "grad_norm": 0.0,
+      "learning_rate": 2.438789039686393e-06,
+      "loss": 0.9361,
+      "step": 21440
+    },
+    {
+      "epoch": 0.779757791759101,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4380182620874915e-06,
+      "loss": 0.9751,
+      "step": 21441
+    },
+    {
+      "epoch": 0.7797941593628396,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4372475894003576e-06,
+      "loss": 0.9108,
+      "step": 21442
+    },
+    {
+      "epoch": 0.7798305269665782,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4364770216356837e-06,
+      "loss": 0.9844,
+      "step": 21443
+    },
+    {
+      "epoch": 0.7798668945703168,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4357065588041605e-06,
+      "loss": 0.9136,
+      "step": 21444
+    },
+    {
+      "epoch": 0.7799032621740554,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4349362009164755e-06,
+      "loss": 0.9391,
+      "step": 21445
+    },
+    {
+      "epoch": 0.779939629777794,
+      "grad_norm": 0.0,
+      "learning_rate": 2.434165947983317e-06,
+      "loss": 0.9948,
+      "step": 21446
+    },
+    {
+      "epoch": 0.7799759973815326,
+      "grad_norm": 0.0,
+      "learning_rate": 2.433395800015369e-06,
+      "loss": 0.8397,
+      "step": 21447
+    },
+    {
+      "epoch": 0.7800123649852712,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4326257570233247e-06,
+      "loss": 0.8697,
+      "step": 21448
+    },
+    {
+      "epoch": 0.7800487325890098,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4318558190178553e-06,
+      "loss": 1.034,
+      "step": 21449
+    },
+    {
+      "epoch": 0.7800851001927483,
+      "grad_norm": 0.0,
+      "learning_rate": 2.431085986009651e-06,
+      "loss": 0.9841,
+      "step": 21450
+    },
+    {
+      "epoch": 0.780121467796487,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4303162580093908e-06,
+      "loss": 0.9383,
+      "step": 21451
+    },
+    {
+      "epoch": 0.7801578354002254,
+      "grad_norm": 0.0,
+      "learning_rate": 2.429546635027752e-06,
+      "loss": 0.951,
+      "step": 21452
+    },
+    {
+      "epoch": 0.780194203003964,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4287771170754095e-06,
+      "loss": 0.9469,
+      "step": 21453
+    },
+    {
+      "epoch": 0.7802305706077026,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4280077041630444e-06,
+      "loss": 0.8036,
+      "step": 21454
+    },
+    {
+      "epoch": 0.7802669382114412,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4272383963013313e-06,
+      "loss": 1.0093,
+      "step": 21455
+    },
+    {
+      "epoch": 0.7803033058151798,
+      "grad_norm": 0.0,
+      "learning_rate": 2.42646919350094e-06,
+      "loss": 0.885,
+      "step": 21456
+    },
+    {
+      "epoch": 0.7803396734189184,
+      "grad_norm": 0.0,
+      "learning_rate": 2.425700095772543e-06,
+      "loss": 0.8958,
+      "step": 21457
+    },
+    {
+      "epoch": 0.780376041022657,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4249311031268107e-06,
+      "loss": 0.9601,
+      "step": 21458
+    },
+    {
+      "epoch": 0.7804124086263956,
+      "grad_norm": 0.0,
+      "learning_rate": 2.424162215574408e-06,
+      "loss": 0.8616,
+      "step": 21459
+    },
+    {
+      "epoch": 0.7804487762301342,
+      "grad_norm": 0.0,
+      "learning_rate": 2.42339343312601e-06,
+      "loss": 0.9383,
+      "step": 21460
+    },
+    {
+      "epoch": 0.7804851438338728,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4226247557922778e-06,
+      "loss": 0.9913,
+      "step": 21461
+    },
+    {
+      "epoch": 0.7805215114376114,
+      "grad_norm": 0.0,
+      "learning_rate": 2.421856183583877e-06,
+      "loss": 0.962,
+      "step": 21462
+    },
+    {
+      "epoch": 0.78055787904135,
+      "grad_norm": 0.0,
+      "learning_rate": 2.421087716511469e-06,
+      "loss": 0.9213,
+      "step": 21463
+    },
+    {
+      "epoch": 0.7805942466450886,
+      "grad_norm": 0.0,
+      "learning_rate": 2.420319354585716e-06,
+      "loss": 0.9395,
+      "step": 21464
+    },
+    {
+      "epoch": 0.7806306142488272,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4195510978172776e-06,
+      "loss": 0.8999,
+      "step": 21465
+    },
+    {
+      "epoch": 0.7806669818525658,
+      "grad_norm": 0.0,
+      "learning_rate": 2.41878294621681e-06,
+      "loss": 0.8699,
+      "step": 21466
+    },
+    {
+      "epoch": 0.7807033494563044,
+      "grad_norm": 0.0,
+      "learning_rate": 2.418014899794978e-06,
+      "loss": 0.9971,
+      "step": 21467
+    },
+    {
+      "epoch": 0.7807397170600429,
+      "grad_norm": 0.0,
+      "learning_rate": 2.417246958562426e-06,
+      "loss": 0.8093,
+      "step": 21468
+    },
+    {
+      "epoch": 0.7807760846637815,
+      "grad_norm": 0.0,
+      "learning_rate": 2.416479122529817e-06,
+      "loss": 0.9081,
+      "step": 21469
+    },
+    {
+      "epoch": 0.78081245226752,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4157113917078e-06,
+      "loss": 0.9088,
+      "step": 21470
+    },
+    {
+      "epoch": 0.7808488198712586,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4149437661070286e-06,
+      "loss": 0.8002,
+      "step": 21471
+    },
+    {
+      "epoch": 0.7808851874749972,
+      "grad_norm": 0.0,
+      "learning_rate": 2.414176245738146e-06,
+      "loss": 0.8662,
+      "step": 21472
+    },
+    {
+      "epoch": 0.7809215550787358,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4134088306118086e-06,
+      "loss": 0.8649,
+      "step": 21473
+    },
+    {
+      "epoch": 0.7809579226824744,
+      "grad_norm": 0.0,
+      "learning_rate": 2.41264152073866e-06,
+      "loss": 0.8834,
+      "step": 21474
+    },
+    {
+      "epoch": 0.780994290286213,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4118743161293446e-06,
+      "loss": 0.9417,
+      "step": 21475
+    },
+    {
+      "epoch": 0.7810306578899516,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4111072167945038e-06,
+      "loss": 1.0101,
+      "step": 21476
+    },
+    {
+      "epoch": 0.7810670254936902,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4103402227447893e-06,
+      "loss": 0.9941,
+      "step": 21477
+    },
+    {
+      "epoch": 0.7811033930974288,
+      "grad_norm": 0.0,
+      "learning_rate": 2.409573333990831e-06,
+      "loss": 0.8755,
+      "step": 21478
+    },
+    {
+      "epoch": 0.7811397607011674,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4088065505432746e-06,
+      "loss": 1.0389,
+      "step": 21479
+    },
+    {
+      "epoch": 0.781176128304906,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4080398724127586e-06,
+      "loss": 0.9371,
+      "step": 21480
+    },
+    {
+      "epoch": 0.7812124959086446,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4072732996099167e-06,
+      "loss": 1.0305,
+      "step": 21481
+    },
+    {
+      "epoch": 0.7812488635123832,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4065068321453843e-06,
+      "loss": 0.8307,
+      "step": 21482
+    },
+    {
+      "epoch": 0.7812852311161218,
+      "grad_norm": 0.0,
+      "learning_rate": 2.405740470029797e-06,
+      "loss": 0.9506,
+      "step": 21483
+    },
+    {
+      "epoch": 0.7813215987198604,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4049742132737863e-06,
+      "loss": 0.9511,
+      "step": 21484
+    },
+    {
+      "epoch": 0.7813579663235989,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4042080618879782e-06,
+      "loss": 0.9476,
+      "step": 21485
+    },
+    {
+      "epoch": 0.7813943339273375,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4034420158830096e-06,
+      "loss": 1.0329,
+      "step": 21486
+    },
+    {
+      "epoch": 0.7814307015310761,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4026760752695056e-06,
+      "loss": 0.9907,
+      "step": 21487
+    },
+    {
+      "epoch": 0.7814670691348147,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4019102400580918e-06,
+      "loss": 0.9012,
+      "step": 21488
+    },
+    {
+      "epoch": 0.7815034367385533,
+      "grad_norm": 0.0,
+      "learning_rate": 2.401144510259392e-06,
+      "loss": 0.9182,
+      "step": 21489
+    },
+    {
+      "epoch": 0.7815398043422919,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4003788858840316e-06,
+      "loss": 0.9058,
+      "step": 21490
+    },
+    {
+      "epoch": 0.7815761719460305,
+      "grad_norm": 0.0,
+      "learning_rate": 2.399613366942629e-06,
+      "loss": 1.0362,
+      "step": 21491
+    },
+    {
+      "epoch": 0.7816125395497691,
+      "grad_norm": 0.0,
+      "learning_rate": 2.39884795344581e-06,
+      "loss": 0.9767,
+      "step": 21492
+    },
+    {
+      "epoch": 0.7816489071535077,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3980826454041906e-06,
+      "loss": 1.0153,
+      "step": 21493
+    },
+    {
+      "epoch": 0.7816852747572463,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3973174428283896e-06,
+      "loss": 0.8841,
+      "step": 21494
+    },
+    {
+      "epoch": 0.7817216423609848,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3965523457290187e-06,
+      "loss": 0.9471,
+      "step": 21495
+    },
+    {
+      "epoch": 0.7817580099647234,
+      "grad_norm": 0.0,
+      "learning_rate": 2.395787354116702e-06,
+      "loss": 0.8932,
+      "step": 21496
+    },
+    {
+      "epoch": 0.781794377568462,
+      "grad_norm": 0.0,
+      "learning_rate": 2.395022468002042e-06,
+      "loss": 0.875,
+      "step": 21497
+    },
+    {
+      "epoch": 0.7818307451722006,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3942576873956567e-06,
+      "loss": 0.8966,
+      "step": 21498
+    },
+    {
+      "epoch": 0.7818671127759392,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3934930123081558e-06,
+      "loss": 0.9702,
+      "step": 21499
+    },
+    {
+      "epoch": 0.7819034803796778,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3927284427501464e-06,
+      "loss": 0.9387,
+      "step": 21500
+    },
+    {
+      "epoch": 0.7819398479834164,
+      "grad_norm": 0.0,
+      "learning_rate": 2.391963978732237e-06,
+      "loss": 0.8653,
+      "step": 21501
+    },
+    {
+      "epoch": 0.7819762155871549,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3911996202650323e-06,
+      "loss": 0.9785,
+      "step": 21502
+    },
+    {
+      "epoch": 0.7820125831908935,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3904353673591387e-06,
+      "loss": 1.0053,
+      "step": 21503
+    },
+    {
+      "epoch": 0.7820489507946321,
+      "grad_norm": 0.0,
+      "learning_rate": 2.389671220025154e-06,
+      "loss": 0.8545,
+      "step": 21504
+    },
+    {
+      "epoch": 0.7820853183983707,
+      "grad_norm": 0.0,
+      "learning_rate": 2.388907178273686e-06,
+      "loss": 0.8354,
+      "step": 21505
+    },
+    {
+      "epoch": 0.7821216860021093,
+      "grad_norm": 0.0,
+      "learning_rate": 2.388143242115333e-06,
+      "loss": 0.9629,
+      "step": 21506
+    },
+    {
+      "epoch": 0.7821580536058479,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3873794115606928e-06,
+      "loss": 0.9171,
+      "step": 21507
+    },
+    {
+      "epoch": 0.7821944212095865,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3866156866203594e-06,
+      "loss": 0.9032,
+      "step": 21508
+    },
+    {
+      "epoch": 0.7822307888133251,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3858520673049367e-06,
+      "loss": 0.9602,
+      "step": 21509
+    },
+    {
+      "epoch": 0.7822671564170637,
+      "grad_norm": 0.0,
+      "learning_rate": 2.385088553625008e-06,
+      "loss": 0.9922,
+      "step": 21510
+    },
+    {
+      "epoch": 0.7823035240208023,
+      "grad_norm": 0.0,
+      "learning_rate": 2.384325145591174e-06,
+      "loss": 0.7253,
+      "step": 21511
+    },
+    {
+      "epoch": 0.7823398916245409,
+      "grad_norm": 0.0,
+      "learning_rate": 2.383561843214024e-06,
+      "loss": 0.9954,
+      "step": 21512
+    },
+    {
+      "epoch": 0.7823762592282795,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3827986465041462e-06,
+      "loss": 0.949,
+      "step": 21513
+    },
+    {
+      "epoch": 0.7824126268320181,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3820355554721274e-06,
+      "loss": 0.9135,
+      "step": 21514
+    },
+    {
+      "epoch": 0.7824489944357567,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3812725701285633e-06,
+      "loss": 0.9582,
+      "step": 21515
+    },
+    {
+      "epoch": 0.7824853620394953,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3805096904840265e-06,
+      "loss": 0.9325,
+      "step": 21516
+    },
+    {
+      "epoch": 0.7825217296432339,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3797469165491092e-06,
+      "loss": 0.9098,
+      "step": 21517
+    },
+    {
+      "epoch": 0.7825580972469725,
+      "grad_norm": 0.0,
+      "learning_rate": 2.378984248334393e-06,
+      "loss": 0.9665,
+      "step": 21518
+    },
+    {
+      "epoch": 0.7825944648507109,
+      "grad_norm": 0.0,
+      "learning_rate": 2.378221685850456e-06,
+      "loss": 0.8582,
+      "step": 21519
+    },
+    {
+      "epoch": 0.7826308324544495,
+      "grad_norm": 0.0,
+      "learning_rate": 2.37745922910788e-06,
+      "loss": 0.9399,
+      "step": 21520
+    },
+    {
+      "epoch": 0.7826672000581881,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3766968781172395e-06,
+      "loss": 0.9594,
+      "step": 21521
+    },
+    {
+      "epoch": 0.7827035676619267,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3759346328891197e-06,
+      "loss": 0.9446,
+      "step": 21522
+    },
+    {
+      "epoch": 0.7827399352656653,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3751724934340834e-06,
+      "loss": 0.8414,
+      "step": 21523
+    },
+    {
+      "epoch": 0.7827763028694039,
+      "grad_norm": 0.0,
+      "learning_rate": 2.374410459762714e-06,
+      "loss": 0.8629,
+      "step": 21524
+    },
+    {
+      "epoch": 0.7828126704731425,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3736485318855797e-06,
+      "loss": 0.8506,
+      "step": 21525
+    },
+    {
+      "epoch": 0.7828490380768811,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3728867098132513e-06,
+      "loss": 0.9547,
+      "step": 21526
+    },
+    {
+      "epoch": 0.7828854056806197,
+      "grad_norm": 0.0,
+      "learning_rate": 2.372124993556296e-06,
+      "loss": 0.9278,
+      "step": 21527
+    },
+    {
+      "epoch": 0.7829217732843583,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3713633831252903e-06,
+      "loss": 0.9762,
+      "step": 21528
+    },
+    {
+      "epoch": 0.7829581408880969,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3706018785307873e-06,
+      "loss": 0.9521,
+      "step": 21529
+    },
+    {
+      "epoch": 0.7829945084918355,
+      "grad_norm": 0.0,
+      "learning_rate": 2.369840479783362e-06,
+      "loss": 0.968,
+      "step": 21530
+    },
+    {
+      "epoch": 0.7830308760955741,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3690791868935736e-06,
+      "loss": 0.9695,
+      "step": 21531
+    },
+    {
+      "epoch": 0.7830672436993127,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3683179998719853e-06,
+      "loss": 0.9187,
+      "step": 21532
+    },
+    {
+      "epoch": 0.7831036113030513,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3675569187291537e-06,
+      "loss": 0.9464,
+      "step": 21533
+    },
+    {
+      "epoch": 0.7831399789067899,
+      "grad_norm": 0.0,
+      "learning_rate": 2.366795943475646e-06,
+      "loss": 0.9254,
+      "step": 21534
+    },
+    {
+      "epoch": 0.7831763465105285,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3660350741220094e-06,
+      "loss": 0.8059,
+      "step": 21535
+    },
+    {
+      "epoch": 0.783212714114267,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3652743106788078e-06,
+      "loss": 0.8625,
+      "step": 21536
+    },
+    {
+      "epoch": 0.7832490817180056,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3645136531565925e-06,
+      "loss": 1.045,
+      "step": 21537
+    },
+    {
+      "epoch": 0.7832854493217442,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3637531015659177e-06,
+      "loss": 0.86,
+      "step": 21538
+    },
+    {
+      "epoch": 0.7833218169254828,
+      "grad_norm": 0.0,
+      "learning_rate": 2.362992655917333e-06,
+      "loss": 0.9126,
+      "step": 21539
+    },
+    {
+      "epoch": 0.7833581845292213,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3622323162213877e-06,
+      "loss": 0.8692,
+      "step": 21540
+    },
+    {
+      "epoch": 0.7833945521329599,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3614720824886384e-06,
+      "loss": 0.8983,
+      "step": 21541
+    },
+    {
+      "epoch": 0.7834309197366985,
+      "grad_norm": 0.0,
+      "learning_rate": 2.36071195472962e-06,
+      "loss": 0.8876,
+      "step": 21542
+    },
+    {
+      "epoch": 0.7834672873404371,
+      "grad_norm": 0.0,
+      "learning_rate": 2.359951932954888e-06,
+      "loss": 0.9224,
+      "step": 21543
+    },
+    {
+      "epoch": 0.7835036549441757,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3591920171749826e-06,
+      "loss": 0.8934,
+      "step": 21544
+    },
+    {
+      "epoch": 0.7835400225479143,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3584322074004473e-06,
+      "loss": 0.9601,
+      "step": 21545
+    },
+    {
+      "epoch": 0.7835763901516529,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3576725036418204e-06,
+      "loss": 0.9846,
+      "step": 21546
+    },
+    {
+      "epoch": 0.7836127577553915,
+      "grad_norm": 0.0,
+      "learning_rate": 2.35691290590965e-06,
+      "loss": 0.8868,
+      "step": 21547
+    },
+    {
+      "epoch": 0.7836491253591301,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3561534142144626e-06,
+      "loss": 0.9131,
+      "step": 21548
+    },
+    {
+      "epoch": 0.7836854929628687,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3553940285668042e-06,
+      "loss": 0.8978,
+      "step": 21549
+    },
+    {
+      "epoch": 0.7837218605666073,
+      "grad_norm": 0.0,
+      "learning_rate": 2.354634748977208e-06,
+      "loss": 0.8775,
+      "step": 21550
+    },
+    {
+      "epoch": 0.7837582281703459,
+      "grad_norm": 0.0,
+      "learning_rate": 2.353875575456207e-06,
+      "loss": 0.8594,
+      "step": 21551
+    },
+    {
+      "epoch": 0.7837945957740844,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3531165080143315e-06,
+      "loss": 0.9433,
+      "step": 21552
+    },
+    {
+      "epoch": 0.783830963377823,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3523575466621185e-06,
+      "loss": 0.8955,
+      "step": 21553
+    },
+    {
+      "epoch": 0.7838673309815616,
+      "grad_norm": 0.0,
+      "learning_rate": 2.351598691410093e-06,
+      "loss": 0.9605,
+      "step": 21554
+    },
+    {
+      "epoch": 0.7839036985853002,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3508399422687833e-06,
+      "loss": 0.8895,
+      "step": 21555
+    },
+    {
+      "epoch": 0.7839400661890388,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3500812992487174e-06,
+      "loss": 0.885,
+      "step": 21556
+    },
+    {
+      "epoch": 0.7839764337927774,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3493227623604196e-06,
+      "loss": 0.9449,
+      "step": 21557
+    },
+    {
+      "epoch": 0.784012801396516,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3485643316144146e-06,
+      "loss": 0.8487,
+      "step": 21558
+    },
+    {
+      "epoch": 0.7840491690002546,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3478060070212195e-06,
+      "loss": 0.9685,
+      "step": 21559
+    },
+    {
+      "epoch": 0.7840855366039932,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3470477885913644e-06,
+      "loss": 0.9274,
+      "step": 21560
+    },
+    {
+      "epoch": 0.7841219042077318,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3462896763353583e-06,
+      "loss": 0.8949,
+      "step": 21561
+    },
+    {
+      "epoch": 0.7841582718114704,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3455316702637264e-06,
+      "loss": 1.0157,
+      "step": 21562
+    },
+    {
+      "epoch": 0.784194639415209,
+      "grad_norm": 0.0,
+      "learning_rate": 2.344773770386982e-06,
+      "loss": 0.9741,
+      "step": 21563
+    },
+    {
+      "epoch": 0.7842310070189475,
+      "grad_norm": 0.0,
+      "learning_rate": 2.34401597671564e-06,
+      "loss": 0.8824,
+      "step": 21564
+    },
+    {
+      "epoch": 0.7842673746226861,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3432582892602118e-06,
+      "loss": 0.9446,
+      "step": 21565
+    },
+    {
+      "epoch": 0.7843037422264247,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3425007080312166e-06,
+      "loss": 0.9467,
+      "step": 21566
+    },
+    {
+      "epoch": 0.7843401098301633,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3417432330391545e-06,
+      "loss": 0.852,
+      "step": 21567
+    },
+    {
+      "epoch": 0.7843764774339019,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3409858642945414e-06,
+      "loss": 0.8723,
+      "step": 21568
+    },
+    {
+      "epoch": 0.7844128450376404,
+      "grad_norm": 0.0,
+      "learning_rate": 2.340228601807882e-06,
+      "loss": 0.9161,
+      "step": 21569
+    },
+    {
+      "epoch": 0.784449212641379,
+      "grad_norm": 0.0,
+      "learning_rate": 2.339471445589685e-06,
+      "loss": 0.8417,
+      "step": 21570
+    },
+    {
+      "epoch": 0.7844855802451176,
+      "grad_norm": 0.0,
+      "learning_rate": 2.338714395650449e-06,
+      "loss": 0.9031,
+      "step": 21571
+    },
+    {
+      "epoch": 0.7845219478488562,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3379574520006833e-06,
+      "loss": 0.9321,
+      "step": 21572
+    },
+    {
+      "epoch": 0.7845583154525948,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3372006146508865e-06,
+      "loss": 0.7922,
+      "step": 21573
+    },
+    {
+      "epoch": 0.7845946830563334,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3364438836115597e-06,
+      "loss": 0.8131,
+      "step": 21574
+    },
+    {
+      "epoch": 0.784631050660072,
+      "grad_norm": 0.0,
+      "learning_rate": 2.335687258893199e-06,
+      "loss": 0.935,
+      "step": 21575
+    },
+    {
+      "epoch": 0.7846674182638106,
+      "grad_norm": 0.0,
+      "learning_rate": 2.334930740506307e-06,
+      "loss": 1.0227,
+      "step": 21576
+    },
+    {
+      "epoch": 0.7847037858675492,
+      "grad_norm": 0.0,
+      "learning_rate": 2.334174328461374e-06,
+      "loss": 0.8486,
+      "step": 21577
+    },
+    {
+      "epoch": 0.7847401534712878,
+      "grad_norm": 0.0,
+      "learning_rate": 2.333418022768893e-06,
+      "loss": 0.8169,
+      "step": 21578
+    },
+    {
+      "epoch": 0.7847765210750264,
+      "grad_norm": 0.0,
+      "learning_rate": 2.332661823439365e-06,
+      "loss": 0.8375,
+      "step": 21579
+    },
+    {
+      "epoch": 0.784812888678765,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3319057304832703e-06,
+      "loss": 0.9265,
+      "step": 21580
+    },
+    {
+      "epoch": 0.7848492562825036,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3311497439111065e-06,
+      "loss": 0.9651,
+      "step": 21581
+    },
+    {
+      "epoch": 0.7848856238862422,
+      "grad_norm": 0.0,
+      "learning_rate": 2.330393863733359e-06,
+      "loss": 1.0294,
+      "step": 21582
+    },
+    {
+      "epoch": 0.7849219914899808,
+      "grad_norm": 0.0,
+      "learning_rate": 2.329638089960515e-06,
+      "loss": 0.9235,
+      "step": 21583
+    },
+    {
+      "epoch": 0.7849583590937194,
+      "grad_norm": 0.0,
+      "learning_rate": 2.328882422603058e-06,
+      "loss": 0.9881,
+      "step": 21584
+    },
+    {
+      "epoch": 0.784994726697458,
+      "grad_norm": 0.0,
+      "learning_rate": 2.328126861671475e-06,
+      "loss": 0.924,
+      "step": 21585
+    },
+    {
+      "epoch": 0.7850310943011964,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3273714071762477e-06,
+      "loss": 1.0416,
+      "step": 21586
+    },
+    {
+      "epoch": 0.785067461904935,
+      "grad_norm": 0.0,
+      "learning_rate": 2.326616059127855e-06,
+      "loss": 1.0313,
+      "step": 21587
+    },
+    {
+      "epoch": 0.7851038295086736,
+      "grad_norm": 0.0,
+      "learning_rate": 2.325860817536778e-06,
+      "loss": 0.9885,
+      "step": 21588
+    },
+    {
+      "epoch": 0.7851401971124122,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3251056824134944e-06,
+      "loss": 0.8637,
+      "step": 21589
+    },
+    {
+      "epoch": 0.7851765647161508,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3243506537684767e-06,
+      "loss": 0.9125,
+      "step": 21590
+    },
+    {
+      "epoch": 0.7852129323198894,
+      "grad_norm": 0.0,
+      "learning_rate": 2.323595731612207e-06,
+      "loss": 0.9739,
+      "step": 21591
+    },
+    {
+      "epoch": 0.785249299923628,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3228409159551545e-06,
+      "loss": 0.9474,
+      "step": 21592
+    },
+    {
+      "epoch": 0.7852856675273666,
+      "grad_norm": 0.0,
+      "learning_rate": 2.322086206807792e-06,
+      "loss": 0.9924,
+      "step": 21593
+    },
+    {
+      "epoch": 0.7853220351311052,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3213316041805866e-06,
+      "loss": 0.8522,
+      "step": 21594
+    },
+    {
+      "epoch": 0.7853584027348438,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3205771080840166e-06,
+      "loss": 0.9796,
+      "step": 21595
+    },
+    {
+      "epoch": 0.7853947703385824,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3198227185285403e-06,
+      "loss": 1.0143,
+      "step": 21596
+    },
+    {
+      "epoch": 0.785431137942321,
+      "grad_norm": 0.0,
+      "learning_rate": 2.319068435524625e-06,
+      "loss": 1.0812,
+      "step": 21597
+    },
+    {
+      "epoch": 0.7854675055460596,
+      "grad_norm": 0.0,
+      "learning_rate": 2.31831425908274e-06,
+      "loss": 0.909,
+      "step": 21598
+    },
+    {
+      "epoch": 0.7855038731497982,
+      "grad_norm": 0.0,
+      "learning_rate": 2.317560189213346e-06,
+      "loss": 1.0135,
+      "step": 21599
+    },
+    {
+      "epoch": 0.7855402407535368,
+      "grad_norm": 0.0,
+      "learning_rate": 2.316806225926903e-06,
+      "loss": 0.842,
+      "step": 21600
+    },
+    {
+      "epoch": 0.7855766083572754,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3160523692338743e-06,
+      "loss": 0.8113,
+      "step": 21601
+    },
+    {
+      "epoch": 0.785612975961014,
+      "grad_norm": 0.0,
+      "learning_rate": 2.315298619144716e-06,
+      "loss": 0.9668,
+      "step": 21602
+    },
+    {
+      "epoch": 0.7856493435647525,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3145449756698834e-06,
+      "loss": 0.8446,
+      "step": 21603
+    },
+    {
+      "epoch": 0.7856857111684911,
+      "grad_norm": 0.0,
+      "learning_rate": 2.313791438819838e-06,
+      "loss": 0.9207,
+      "step": 21604
+    },
+    {
+      "epoch": 0.7857220787722297,
+      "grad_norm": 0.0,
+      "learning_rate": 2.313038008605031e-06,
+      "loss": 0.9302,
+      "step": 21605
+    },
+    {
+      "epoch": 0.7857584463759683,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3122846850359147e-06,
+      "loss": 0.941,
+      "step": 21606
+    },
+    {
+      "epoch": 0.7857948139797069,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3115314681229397e-06,
+      "loss": 0.9307,
+      "step": 21607
+    },
+    {
+      "epoch": 0.7858311815834454,
+      "grad_norm": 0.0,
+      "learning_rate": 2.310778357876562e-06,
+      "loss": 0.8461,
+      "step": 21608
+    },
+    {
+      "epoch": 0.785867549187184,
+      "grad_norm": 0.0,
+      "learning_rate": 2.310025354307218e-06,
+      "loss": 0.8576,
+      "step": 21609
+    },
+    {
+      "epoch": 0.7859039167909226,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3092724574253655e-06,
+      "loss": 1.0212,
+      "step": 21610
+    },
+    {
+      "epoch": 0.7859402843946612,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3085196672414466e-06,
+      "loss": 0.9239,
+      "step": 21611
+    },
+    {
+      "epoch": 0.7859766519983998,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3077669837659032e-06,
+      "loss": 0.8894,
+      "step": 21612
+    },
+    {
+      "epoch": 0.7860130196021384,
+      "grad_norm": 0.0,
+      "learning_rate": 2.307014407009177e-06,
+      "loss": 0.994,
+      "step": 21613
+    },
+    {
+      "epoch": 0.786049387205877,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3062619369817174e-06,
+      "loss": 0.8363,
+      "step": 21614
+    },
+    {
+      "epoch": 0.7860857548096156,
+      "grad_norm": 0.0,
+      "learning_rate": 2.305509573693954e-06,
+      "loss": 1.0604,
+      "step": 21615
+    },
+    {
+      "epoch": 0.7861221224133542,
+      "grad_norm": 0.0,
+      "learning_rate": 2.304757317156324e-06,
+      "loss": 0.9288,
+      "step": 21616
+    },
+    {
+      "epoch": 0.7861584900170928,
+      "grad_norm": 0.0,
+      "learning_rate": 2.304005167379273e-06,
+      "loss": 0.9679,
+      "step": 21617
+    },
+    {
+      "epoch": 0.7861948576208314,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3032531243732292e-06,
+      "loss": 0.9582,
+      "step": 21618
+    },
+    {
+      "epoch": 0.78623122522457,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3025011881486293e-06,
+      "loss": 0.9457,
+      "step": 21619
+    },
+    {
+      "epoch": 0.7862675928283085,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3017493587159013e-06,
+      "loss": 0.9337,
+      "step": 21620
+    },
+    {
+      "epoch": 0.7863039604320471,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3009976360854834e-06,
+      "loss": 1.0881,
+      "step": 21621
+    },
+    {
+      "epoch": 0.7863403280357857,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3002460202677946e-06,
+      "loss": 0.9387,
+      "step": 21622
+    },
+    {
+      "epoch": 0.7863766956395243,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2994945112732713e-06,
+      "loss": 0.9277,
+      "step": 21623
+    },
+    {
+      "epoch": 0.7864130632432629,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2987431091123345e-06,
+      "loss": 0.8602,
+      "step": 21624
+    },
+    {
+      "epoch": 0.7864494308470015,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2979918137954115e-06,
+      "loss": 0.8708,
+      "step": 21625
+    },
+    {
+      "epoch": 0.7864857984507401,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2972406253329217e-06,
+      "loss": 0.8749,
+      "step": 21626
+    },
+    {
+      "epoch": 0.7865221660544787,
+      "grad_norm": 0.0,
+      "learning_rate": 2.296489543735295e-06,
+      "loss": 0.8898,
+      "step": 21627
+    },
+    {
+      "epoch": 0.7865585336582173,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2957385690129396e-06,
+      "loss": 0.9283,
+      "step": 21628
+    },
+    {
+      "epoch": 0.7865949012619559,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2949877011762843e-06,
+      "loss": 0.8871,
+      "step": 21629
+    },
+    {
+      "epoch": 0.7866312688656945,
+      "grad_norm": 0.0,
+      "learning_rate": 2.294236940235742e-06,
+      "loss": 0.91,
+      "step": 21630
+    },
+    {
+      "epoch": 0.786667636469433,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2934862862017304e-06,
+      "loss": 0.918,
+      "step": 21631
+    },
+    {
+      "epoch": 0.7867040040731716,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2927357390846582e-06,
+      "loss": 0.9223,
+      "step": 21632
+    },
+    {
+      "epoch": 0.7867403716769102,
+      "grad_norm": 0.0,
+      "learning_rate": 2.291985298894949e-06,
+      "loss": 0.9799,
+      "step": 21633
+    },
+    {
+      "epoch": 0.7867767392806488,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2912349656430046e-06,
+      "loss": 0.863,
+      "step": 21634
+    },
+    {
+      "epoch": 0.7868131068843874,
+      "grad_norm": 0.0,
+      "learning_rate": 2.290484739339234e-06,
+      "loss": 0.9418,
+      "step": 21635
+    },
+    {
+      "epoch": 0.786849474488126,
+      "grad_norm": 0.0,
+      "learning_rate": 2.289734619994053e-06,
+      "loss": 0.8655,
+      "step": 21636
+    },
+    {
+      "epoch": 0.7868858420918645,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2889846076178646e-06,
+      "loss": 0.976,
+      "step": 21637
+    },
+    {
+      "epoch": 0.7869222096956031,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2882347022210738e-06,
+      "loss": 1.0062,
+      "step": 21638
+    },
+    {
+      "epoch": 0.7869585772993417,
+      "grad_norm": 0.0,
+      "learning_rate": 2.287484903814082e-06,
+      "loss": 0.8473,
+      "step": 21639
+    },
+    {
+      "epoch": 0.7869949449030803,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2867352124073008e-06,
+      "loss": 0.9396,
+      "step": 21640
+    },
+    {
+      "epoch": 0.7870313125068189,
+      "grad_norm": 0.0,
+      "learning_rate": 2.285985628011119e-06,
+      "loss": 0.9022,
+      "step": 21641
+    },
+    {
+      "epoch": 0.7870676801105575,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2852361506359445e-06,
+      "loss": 0.9507,
+      "step": 21642
+    },
+    {
+      "epoch": 0.7871040477142961,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2844867802921732e-06,
+      "loss": 0.9609,
+      "step": 21643
+    },
+    {
+      "epoch": 0.7871404153180347,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2837375169901997e-06,
+      "loss": 0.9723,
+      "step": 21644
+    },
+    {
+      "epoch": 0.7871767829217733,
+      "grad_norm": 0.0,
+      "learning_rate": 2.282988360740419e-06,
+      "loss": 0.8448,
+      "step": 21645
+    },
+    {
+      "epoch": 0.7872131505255119,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2822393115532306e-06,
+      "loss": 0.961,
+      "step": 21646
+    },
+    {
+      "epoch": 0.7872495181292505,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2814903694390157e-06,
+      "loss": 0.9864,
+      "step": 21647
+    },
+    {
+      "epoch": 0.7872858857329891,
+      "grad_norm": 0.0,
+      "learning_rate": 2.280741534408174e-06,
+      "loss": 0.9455,
+      "step": 21648
+    },
+    {
+      "epoch": 0.7873222533367277,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2799928064710906e-06,
+      "loss": 0.9678,
+      "step": 21649
+    },
+    {
+      "epoch": 0.7873586209404663,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2792441856381553e-06,
+      "loss": 0.9937,
+      "step": 21650
+    },
+    {
+      "epoch": 0.7873949885442049,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2784956719197494e-06,
+      "loss": 0.9019,
+      "step": 21651
+    },
+    {
+      "epoch": 0.7874313561479435,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2777472653262623e-06,
+      "loss": 0.9837,
+      "step": 21652
+    },
+    {
+      "epoch": 0.787467723751682,
+      "grad_norm": 0.0,
+      "learning_rate": 2.276998965868079e-06,
+      "loss": 0.9704,
+      "step": 21653
+    },
+    {
+      "epoch": 0.7875040913554205,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2762507735555716e-06,
+      "loss": 0.8695,
+      "step": 21654
+    },
+    {
+      "epoch": 0.7875404589591591,
+      "grad_norm": 0.0,
+      "learning_rate": 2.275502688399129e-06,
+      "loss": 0.93,
+      "step": 21655
+    },
+    {
+      "epoch": 0.7875768265628977,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2747547104091284e-06,
+      "loss": 0.912,
+      "step": 21656
+    },
+    {
+      "epoch": 0.7876131941666363,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2740068395959437e-06,
+      "loss": 1.0151,
+      "step": 21657
+    },
+    {
+      "epoch": 0.7876495617703749,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2732590759699503e-06,
+      "loss": 0.9066,
+      "step": 21658
+    },
+    {
+      "epoch": 0.7876859293741135,
+      "grad_norm": 0.0,
+      "learning_rate": 2.27251141954153e-06,
+      "loss": 0.9469,
+      "step": 21659
+    },
+    {
+      "epoch": 0.7877222969778521,
+      "grad_norm": 0.0,
+      "learning_rate": 2.271763870321044e-06,
+      "loss": 0.8296,
+      "step": 21660
+    },
+    {
+      "epoch": 0.7877586645815907,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2710164283188717e-06,
+      "loss": 0.9608,
+      "step": 21661
+    },
+    {
+      "epoch": 0.7877950321853293,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2702690935453818e-06,
+      "loss": 0.9054,
+      "step": 21662
+    },
+    {
+      "epoch": 0.7878313997890679,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2695218660109407e-06,
+      "loss": 0.8604,
+      "step": 21663
+    },
+    {
+      "epoch": 0.7878677673928065,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2687747457259123e-06,
+      "loss": 0.8444,
+      "step": 21664
+    },
+    {
+      "epoch": 0.7879041349965451,
+      "grad_norm": 0.0,
+      "learning_rate": 2.268027732700668e-06,
+      "loss": 0.9578,
+      "step": 21665
+    },
+    {
+      "epoch": 0.7879405026002837,
+      "grad_norm": 0.0,
+      "learning_rate": 2.267280826945569e-06,
+      "loss": 0.9408,
+      "step": 21666
+    },
+    {
+      "epoch": 0.7879768702040223,
+      "grad_norm": 0.0,
+      "learning_rate": 2.266534028470977e-06,
+      "loss": 0.959,
+      "step": 21667
+    },
+    {
+      "epoch": 0.7880132378077609,
+      "grad_norm": 0.0,
+      "learning_rate": 2.265787337287253e-06,
+      "loss": 0.9878,
+      "step": 21668
+    },
+    {
+      "epoch": 0.7880496054114995,
+      "grad_norm": 0.0,
+      "learning_rate": 2.265040753404757e-06,
+      "loss": 0.7703,
+      "step": 21669
+    },
+    {
+      "epoch": 0.788085973015238,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2642942768338426e-06,
+      "loss": 0.9568,
+      "step": 21670
+    },
+    {
+      "epoch": 0.7881223406189766,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2635479075848723e-06,
+      "loss": 1.0372,
+      "step": 21671
+    },
+    {
+      "epoch": 0.7881587082227152,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2628016456681977e-06,
+      "loss": 0.8566,
+      "step": 21672
+    },
+    {
+      "epoch": 0.7881950758264538,
+      "grad_norm": 0.0,
+      "learning_rate": 2.262055491094174e-06,
+      "loss": 0.8837,
+      "step": 21673
+    },
+    {
+      "epoch": 0.7882314434301924,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2613094438731507e-06,
+      "loss": 0.9675,
+      "step": 21674
+    },
+    {
+      "epoch": 0.788267811033931,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2605635040154793e-06,
+      "loss": 1.0148,
+      "step": 21675
+    },
+    {
+      "epoch": 0.7883041786376696,
+      "grad_norm": 0.0,
+      "learning_rate": 2.259817671531508e-06,
+      "loss": 0.9328,
+      "step": 21676
+    },
+    {
+      "epoch": 0.7883405462414081,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2590719464315825e-06,
+      "loss": 0.8014,
+      "step": 21677
+    },
+    {
+      "epoch": 0.7883769138451467,
+      "grad_norm": 0.0,
+      "learning_rate": 2.258326328726056e-06,
+      "loss": 0.7459,
+      "step": 21678
+    },
+    {
+      "epoch": 0.7884132814488853,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2575808184252625e-06,
+      "loss": 0.983,
+      "step": 21679
+    },
+    {
+      "epoch": 0.7884496490526239,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2568354155395523e-06,
+      "loss": 0.8815,
+      "step": 21680
+    },
+    {
+      "epoch": 0.7884860166563625,
+      "grad_norm": 0.0,
+      "learning_rate": 2.256090120079265e-06,
+      "loss": 0.9377,
+      "step": 21681
+    },
+    {
+      "epoch": 0.7885223842601011,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2553449320547405e-06,
+      "loss": 0.9801,
+      "step": 21682
+    },
+    {
+      "epoch": 0.7885587518638397,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2545998514763133e-06,
+      "loss": 0.9579,
+      "step": 21683
+    },
+    {
+      "epoch": 0.7885951194675783,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2538548783543267e-06,
+      "loss": 0.9807,
+      "step": 21684
+    },
+    {
+      "epoch": 0.7886314870713169,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2531100126991147e-06,
+      "loss": 0.9767,
+      "step": 21685
+    },
+    {
+      "epoch": 0.7886678546750555,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2523652545210085e-06,
+      "loss": 0.9072,
+      "step": 21686
+    },
+    {
+      "epoch": 0.788704222278794,
+      "grad_norm": 0.0,
+      "learning_rate": 2.251620603830339e-06,
+      "loss": 0.949,
+      "step": 21687
+    },
+    {
+      "epoch": 0.7887405898825326,
+      "grad_norm": 0.0,
+      "learning_rate": 2.250876060637446e-06,
+      "loss": 0.8445,
+      "step": 21688
+    },
+    {
+      "epoch": 0.7887769574862712,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2501316249526484e-06,
+      "loss": 0.9485,
+      "step": 21689
+    },
+    {
+      "epoch": 0.7888133250900098,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2493872967862804e-06,
+      "loss": 0.9016,
+      "step": 21690
+    },
+    {
+      "epoch": 0.7888496926937484,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2486430761486678e-06,
+      "loss": 1.0113,
+      "step": 21691
+    },
+    {
+      "epoch": 0.788886060297487,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2478989630501345e-06,
+      "loss": 0.8289,
+      "step": 21692
+    },
+    {
+      "epoch": 0.7889224279012256,
+      "grad_norm": 0.0,
+      "learning_rate": 2.247154957501004e-06,
+      "loss": 0.8381,
+      "step": 21693
+    },
+    {
+      "epoch": 0.7889587955049642,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2464110595115984e-06,
+      "loss": 0.9359,
+      "step": 21694
+    },
+    {
+      "epoch": 0.7889951631087028,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2456672690922386e-06,
+      "loss": 0.9175,
+      "step": 21695
+    },
+    {
+      "epoch": 0.7890315307124414,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2449235862532414e-06,
+      "loss": 0.9451,
+      "step": 21696
+    },
+    {
+      "epoch": 0.78906789831618,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2441800110049294e-06,
+      "loss": 0.8429,
+      "step": 21697
+    },
+    {
+      "epoch": 0.7891042659199186,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2434365433576144e-06,
+      "loss": 1.0368,
+      "step": 21698
+    },
+    {
+      "epoch": 0.7891406335236572,
+      "grad_norm": 0.0,
+      "learning_rate": 2.242693183321614e-06,
+      "loss": 0.959,
+      "step": 21699
+    },
+    {
+      "epoch": 0.7891770011273957,
+      "grad_norm": 0.0,
+      "learning_rate": 2.241949930907238e-06,
+      "loss": 1.0001,
+      "step": 21700
+    },
+    {
+      "epoch": 0.7892133687311343,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2412067861248e-06,
+      "loss": 0.991,
+      "step": 21701
+    },
+    {
+      "epoch": 0.7892497363348729,
+      "grad_norm": 0.0,
+      "learning_rate": 2.240463748984606e-06,
+      "loss": 0.909,
+      "step": 21702
+    },
+    {
+      "epoch": 0.7892861039386115,
+      "grad_norm": 0.0,
+      "learning_rate": 2.239720819496972e-06,
+      "loss": 1.0084,
+      "step": 21703
+    },
+    {
+      "epoch": 0.78932247154235,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2389779976722005e-06,
+      "loss": 0.9397,
+      "step": 21704
+    },
+    {
+      "epoch": 0.7893588391460886,
+      "grad_norm": 0.0,
+      "learning_rate": 2.238235283520599e-06,
+      "loss": 0.911,
+      "step": 21705
+    },
+    {
+      "epoch": 0.7893952067498272,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2374926770524663e-06,
+      "loss": 0.8157,
+      "step": 21706
+    },
+    {
+      "epoch": 0.7894315743535658,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2367501782781155e-06,
+      "loss": 0.8903,
+      "step": 21707
+    },
+    {
+      "epoch": 0.7894679419573044,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2360077872078355e-06,
+      "loss": 0.9135,
+      "step": 21708
+    },
+    {
+      "epoch": 0.789504309561043,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2352655038519342e-06,
+      "loss": 0.915,
+      "step": 21709
+    },
+    {
+      "epoch": 0.7895406771647816,
+      "grad_norm": 0.0,
+      "learning_rate": 2.234523328220708e-06,
+      "loss": 0.8907,
+      "step": 21710
+    },
+    {
+      "epoch": 0.7895770447685202,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2337812603244525e-06,
+      "loss": 0.9847,
+      "step": 21711
+    },
+    {
+      "epoch": 0.7896134123722588,
+      "grad_norm": 0.0,
+      "learning_rate": 2.233039300173463e-06,
+      "loss": 0.984,
+      "step": 21712
+    },
+    {
+      "epoch": 0.7896497799759974,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2322974477780336e-06,
+      "loss": 0.8285,
+      "step": 21713
+    },
+    {
+      "epoch": 0.789686147579736,
+      "grad_norm": 0.0,
+      "learning_rate": 2.231555703148457e-06,
+      "loss": 1.0425,
+      "step": 21714
+    },
+    {
+      "epoch": 0.7897225151834746,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2308140662950185e-06,
+      "loss": 0.9276,
+      "step": 21715
+    },
+    {
+      "epoch": 0.7897588827872132,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2300725372280165e-06,
+      "loss": 0.918,
+      "step": 21716
+    },
+    {
+      "epoch": 0.7897952503909518,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2293311159577324e-06,
+      "loss": 0.9025,
+      "step": 21717
+    },
+    {
+      "epoch": 0.7898316179946904,
+      "grad_norm": 0.0,
+      "learning_rate": 2.228589802494454e-06,
+      "loss": 0.9926,
+      "step": 21718
+    },
+    {
+      "epoch": 0.789867985598429,
+      "grad_norm": 0.0,
+      "learning_rate": 2.227848596848464e-06,
+      "loss": 0.8835,
+      "step": 21719
+    },
+    {
+      "epoch": 0.7899043532021676,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2271074990300523e-06,
+      "loss": 0.8216,
+      "step": 21720
+    },
+    {
+      "epoch": 0.789940720805906,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2263665090494912e-06,
+      "loss": 0.8401,
+      "step": 21721
+    },
+    {
+      "epoch": 0.7899770884096446,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2256256269170672e-06,
+      "loss": 0.8734,
+      "step": 21722
+    },
+    {
+      "epoch": 0.7900134560133832,
+      "grad_norm": 0.0,
+      "learning_rate": 2.224884852643058e-06,
+      "loss": 1.0615,
+      "step": 21723
+    },
+    {
+      "epoch": 0.7900498236171218,
+      "grad_norm": 0.0,
+      "learning_rate": 2.22414418623774e-06,
+      "loss": 0.9585,
+      "step": 21724
+    },
+    {
+      "epoch": 0.7900861912208604,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2234036277113856e-06,
+      "loss": 0.9868,
+      "step": 21725
+    },
+    {
+      "epoch": 0.790122558824599,
+      "grad_norm": 0.0,
+      "learning_rate": 2.222663177074278e-06,
+      "loss": 0.8703,
+      "step": 21726
+    },
+    {
+      "epoch": 0.7901589264283376,
+      "grad_norm": 0.0,
+      "learning_rate": 2.221922834336678e-06,
+      "loss": 0.887,
+      "step": 21727
+    },
+    {
+      "epoch": 0.7901952940320762,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2211825995088674e-06,
+      "loss": 0.7835,
+      "step": 21728
+    },
+    {
+      "epoch": 0.7902316616358148,
+      "grad_norm": 0.0,
+      "learning_rate": 2.22044247260111e-06,
+      "loss": 0.9024,
+      "step": 21729
+    },
+    {
+      "epoch": 0.7902680292395534,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2197024536236767e-06,
+      "loss": 0.8448,
+      "step": 21730
+    },
+    {
+      "epoch": 0.790304396843292,
+      "grad_norm": 0.0,
+      "learning_rate": 2.218962542586832e-06,
+      "loss": 0.8725,
+      "step": 21731
+    },
+    {
+      "epoch": 0.7903407644470306,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2182227395008427e-06,
+      "loss": 0.8752,
+      "step": 21732
+    },
+    {
+      "epoch": 0.7903771320507692,
+      "grad_norm": 0.0,
+      "learning_rate": 2.217483044375972e-06,
+      "loss": 0.93,
+      "step": 21733
+    },
+    {
+      "epoch": 0.7904134996545078,
+      "grad_norm": 0.0,
+      "learning_rate": 2.21674345722248e-06,
+      "loss": 0.8483,
+      "step": 21734
+    },
+    {
+      "epoch": 0.7904498672582464,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2160039780506317e-06,
+      "loss": 0.8933,
+      "step": 21735
+    },
+    {
+      "epoch": 0.790486234861985,
+      "grad_norm": 0.0,
+      "learning_rate": 2.215264606870684e-06,
+      "loss": 0.9789,
+      "step": 21736
+    },
+    {
+      "epoch": 0.7905226024657235,
+      "grad_norm": 0.0,
+      "learning_rate": 2.214525343692895e-06,
+      "loss": 0.8587,
+      "step": 21737
+    },
+    {
+      "epoch": 0.7905589700694621,
+      "grad_norm": 0.0,
+      "learning_rate": 2.213786188527518e-06,
+      "loss": 0.8547,
+      "step": 21738
+    },
+    {
+      "epoch": 0.7905953376732007,
+      "grad_norm": 0.0,
+      "learning_rate": 2.213047141384815e-06,
+      "loss": 0.9081,
+      "step": 21739
+    },
+    {
+      "epoch": 0.7906317052769393,
+      "grad_norm": 0.0,
+      "learning_rate": 2.212308202275031e-06,
+      "loss": 0.9084,
+      "step": 21740
+    },
+    {
+      "epoch": 0.7906680728806779,
+      "grad_norm": 0.0,
+      "learning_rate": 2.211569371208423e-06,
+      "loss": 0.8595,
+      "step": 21741
+    },
+    {
+      "epoch": 0.7907044404844165,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2108306481952393e-06,
+      "loss": 1.0127,
+      "step": 21742
+    },
+    {
+      "epoch": 0.7907408080881551,
+      "grad_norm": 0.0,
+      "learning_rate": 2.21009203324573e-06,
+      "loss": 0.8756,
+      "step": 21743
+    },
+    {
+      "epoch": 0.7907771756918937,
+      "grad_norm": 0.0,
+      "learning_rate": 2.209353526370137e-06,
+      "loss": 0.9493,
+      "step": 21744
+    },
+    {
+      "epoch": 0.7908135432956322,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2086151275787173e-06,
+      "loss": 0.8628,
+      "step": 21745
+    },
+    {
+      "epoch": 0.7908499108993708,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2078768368817028e-06,
+      "loss": 0.9945,
+      "step": 21746
+    },
+    {
+      "epoch": 0.7908862785031094,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2071386542893437e-06,
+      "loss": 0.8397,
+      "step": 21747
+    },
+    {
+      "epoch": 0.790922646106848,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2064005798118792e-06,
+      "loss": 0.9719,
+      "step": 21748
+    },
+    {
+      "epoch": 0.7909590137105866,
+      "grad_norm": 0.0,
+      "learning_rate": 2.205662613459548e-06,
+      "loss": 1.0347,
+      "step": 21749
+    },
+    {
+      "epoch": 0.7909953813143252,
+      "grad_norm": 0.0,
+      "learning_rate": 2.204924755242591e-06,
+      "loss": 0.998,
+      "step": 21750
+    },
+    {
+      "epoch": 0.7910317489180638,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2041870051712398e-06,
+      "loss": 0.9466,
+      "step": 21751
+    },
+    {
+      "epoch": 0.7910681165218024,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2034493632557387e-06,
+      "loss": 0.8742,
+      "step": 21752
+    },
+    {
+      "epoch": 0.791104484125541,
+      "grad_norm": 0.0,
+      "learning_rate": 2.20271182950631e-06,
+      "loss": 0.9157,
+      "step": 21753
+    },
+    {
+      "epoch": 0.7911408517292795,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2019744039331947e-06,
+      "loss": 0.8421,
+      "step": 21754
+    },
+    {
+      "epoch": 0.7911772193330181,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2012370865466215e-06,
+      "loss": 0.8558,
+      "step": 21755
+    },
+    {
+      "epoch": 0.7912135869367567,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2004998773568175e-06,
+      "loss": 0.8556,
+      "step": 21756
+    },
+    {
+      "epoch": 0.7912499545404953,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1997627763740093e-06,
+      "loss": 0.9286,
+      "step": 21757
+    },
+    {
+      "epoch": 0.7912863221442339,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1990257836084306e-06,
+      "loss": 0.9208,
+      "step": 21758
+    },
+    {
+      "epoch": 0.7913226897479725,
+      "grad_norm": 0.0,
+      "learning_rate": 2.198288899070297e-06,
+      "loss": 0.933,
+      "step": 21759
+    },
+    {
+      "epoch": 0.7913590573517111,
+      "grad_norm": 0.0,
+      "learning_rate": 2.197552122769837e-06,
+      "loss": 0.9615,
+      "step": 21760
+    },
+    {
+      "epoch": 0.7913954249554497,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1968154547172716e-06,
+      "loss": 0.9431,
+      "step": 21761
+    },
+    {
+      "epoch": 0.7914317925591883,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1960788949228194e-06,
+      "loss": 0.9916,
+      "step": 21762
+    },
+    {
+      "epoch": 0.7914681601629269,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1953424433966984e-06,
+      "loss": 0.9511,
+      "step": 21763
+    },
+    {
+      "epoch": 0.7915045277666655,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1946061001491314e-06,
+      "loss": 0.8983,
+      "step": 21764
+    },
+    {
+      "epoch": 0.7915408953704041,
+      "grad_norm": 0.0,
+      "learning_rate": 2.193869865190329e-06,
+      "loss": 0.9755,
+      "step": 21765
+    },
+    {
+      "epoch": 0.7915772629741427,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1931337385305073e-06,
+      "loss": 0.988,
+      "step": 21766
+    },
+    {
+      "epoch": 0.7916136305778813,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1923977201798786e-06,
+      "loss": 0.9383,
+      "step": 21767
+    },
+    {
+      "epoch": 0.7916499981816199,
+      "grad_norm": 0.0,
+      "learning_rate": 2.191661810148653e-06,
+      "loss": 1.0011,
+      "step": 21768
+    },
+    {
+      "epoch": 0.7916863657853584,
+      "grad_norm": 0.0,
+      "learning_rate": 2.190926008447042e-06,
+      "loss": 0.9984,
+      "step": 21769
+    },
+    {
+      "epoch": 0.791722733389097,
+      "grad_norm": 0.0,
+      "learning_rate": 2.19019031508525e-06,
+      "loss": 0.7405,
+      "step": 21770
+    },
+    {
+      "epoch": 0.7917591009928355,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1894547300734927e-06,
+      "loss": 0.9569,
+      "step": 21771
+    },
+    {
+      "epoch": 0.7917954685965741,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1887192534219636e-06,
+      "loss": 0.8977,
+      "step": 21772
+    },
+    {
+      "epoch": 0.7918318362003127,
+      "grad_norm": 0.0,
+      "learning_rate": 2.187983885140875e-06,
+      "loss": 0.8615,
+      "step": 21773
+    },
+    {
+      "epoch": 0.7918682038040513,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1872486252404257e-06,
+      "loss": 0.9468,
+      "step": 21774
+    },
+    {
+      "epoch": 0.7919045714077899,
+      "grad_norm": 0.0,
+      "learning_rate": 2.186513473730817e-06,
+      "loss": 1.0649,
+      "step": 21775
+    },
+    {
+      "epoch": 0.7919409390115285,
+      "grad_norm": 0.0,
+      "learning_rate": 2.185778430622245e-06,
+      "loss": 0.9582,
+      "step": 21776
+    },
+    {
+      "epoch": 0.7919773066152671,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1850434959249165e-06,
+      "loss": 0.8906,
+      "step": 21777
+    },
+    {
+      "epoch": 0.7920136742190057,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1843086696490157e-06,
+      "loss": 1.0139,
+      "step": 21778
+    },
+    {
+      "epoch": 0.7920500418227443,
+      "grad_norm": 0.0,
+      "learning_rate": 2.183573951804746e-06,
+      "loss": 0.8526,
+      "step": 21779
+    },
+    {
+      "epoch": 0.7920864094264829,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1828393424022977e-06,
+      "loss": 0.9163,
+      "step": 21780
+    },
+    {
+      "epoch": 0.7921227770302215,
+      "grad_norm": 0.0,
+      "learning_rate": 2.182104841451862e-06,
+      "loss": 0.8989,
+      "step": 21781
+    },
+    {
+      "epoch": 0.7921591446339601,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1813704489636265e-06,
+      "loss": 0.8386,
+      "step": 21782
+    },
+    {
+      "epoch": 0.7921955122376987,
+      "grad_norm": 0.0,
+      "learning_rate": 2.180636164947787e-06,
+      "loss": 0.9111,
+      "step": 21783
+    },
+    {
+      "epoch": 0.7922318798414373,
+      "grad_norm": 0.0,
+      "learning_rate": 2.179901989414525e-06,
+      "loss": 0.8981,
+      "step": 21784
+    },
+    {
+      "epoch": 0.7922682474451759,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1791679223740283e-06,
+      "loss": 0.8998,
+      "step": 21785
+    },
+    {
+      "epoch": 0.7923046150489145,
+      "grad_norm": 0.0,
+      "learning_rate": 2.178433963836477e-06,
+      "loss": 0.9511,
+      "step": 21786
+    },
+    {
+      "epoch": 0.7923409826526531,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1777001138120634e-06,
+      "loss": 0.926,
+      "step": 21787
+    },
+    {
+      "epoch": 0.7923773502563916,
+      "grad_norm": 0.0,
+      "learning_rate": 2.176966372310959e-06,
+      "loss": 0.8699,
+      "step": 21788
+    },
+    {
+      "epoch": 0.7924137178601302,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1762327393433447e-06,
+      "loss": 0.9288,
+      "step": 21789
+    },
+    {
+      "epoch": 0.7924500854638687,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1754992149194052e-06,
+      "loss": 0.9362,
+      "step": 21790
+    },
+    {
+      "epoch": 0.7924864530676073,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1747657990493075e-06,
+      "loss": 0.9278,
+      "step": 21791
+    },
+    {
+      "epoch": 0.7925228206713459,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1740324917432355e-06,
+      "loss": 0.9388,
+      "step": 21792
+    },
+    {
+      "epoch": 0.7925591882750845,
+      "grad_norm": 0.0,
+      "learning_rate": 2.173299293011358e-06,
+      "loss": 0.8415,
+      "step": 21793
+    },
+    {
+      "epoch": 0.7925955558788231,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1725662028638484e-06,
+      "loss": 0.9153,
+      "step": 21794
+    },
+    {
+      "epoch": 0.7926319234825617,
+      "grad_norm": 0.0,
+      "learning_rate": 2.171833221310874e-06,
+      "loss": 1.0606,
+      "step": 21795
+    },
+    {
+      "epoch": 0.7926682910863003,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1711003483626104e-06,
+      "loss": 0.8551,
+      "step": 21796
+    },
+    {
+      "epoch": 0.7927046586900389,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1703675840292215e-06,
+      "loss": 0.8794,
+      "step": 21797
+    },
+    {
+      "epoch": 0.7927410262937775,
+      "grad_norm": 0.0,
+      "learning_rate": 2.169634928320873e-06,
+      "loss": 0.9512,
+      "step": 21798
+    },
+    {
+      "epoch": 0.7927773938975161,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1689023812477306e-06,
+      "loss": 0.8013,
+      "step": 21799
+    },
+    {
+      "epoch": 0.7928137615012547,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1681699428199566e-06,
+      "loss": 0.9993,
+      "step": 21800
+    },
+    {
+      "epoch": 0.7928501291049933,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1674376130477093e-06,
+      "loss": 0.9354,
+      "step": 21801
+    },
+    {
+      "epoch": 0.7928864967087319,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1667053919411562e-06,
+      "loss": 0.7774,
+      "step": 21802
+    },
+    {
+      "epoch": 0.7929228643124705,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1659732795104503e-06,
+      "loss": 0.8705,
+      "step": 21803
+    },
+    {
+      "epoch": 0.7929592319162091,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1652412757657516e-06,
+      "loss": 0.9265,
+      "step": 21804
+    },
+    {
+      "epoch": 0.7929955995199476,
+      "grad_norm": 0.0,
+      "learning_rate": 2.164509380717211e-06,
+      "loss": 0.7665,
+      "step": 21805
+    },
+    {
+      "epoch": 0.7930319671236862,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1637775943749906e-06,
+      "loss": 0.8874,
+      "step": 21806
+    },
+    {
+      "epoch": 0.7930683347274248,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1630459167492358e-06,
+      "loss": 0.9364,
+      "step": 21807
+    },
+    {
+      "epoch": 0.7931047023311634,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1623143478500963e-06,
+      "loss": 1.0329,
+      "step": 21808
+    },
+    {
+      "epoch": 0.793141069934902,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1615828876877286e-06,
+      "loss": 0.9309,
+      "step": 21809
+    },
+    {
+      "epoch": 0.7931774375386406,
+      "grad_norm": 0.0,
+      "learning_rate": 2.160851536272276e-06,
+      "loss": 0.8992,
+      "step": 21810
+    },
+    {
+      "epoch": 0.7932138051423792,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1601202936138864e-06,
+      "loss": 1.0742,
+      "step": 21811
+    },
+    {
+      "epoch": 0.7932501727461178,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1593891597227056e-06,
+      "loss": 0.8952,
+      "step": 21812
+    },
+    {
+      "epoch": 0.7932865403498564,
+      "grad_norm": 0.0,
+      "learning_rate": 2.158658134608874e-06,
+      "loss": 0.868,
+      "step": 21813
+    },
+    {
+      "epoch": 0.793322907953595,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1579272182825337e-06,
+      "loss": 0.8952,
+      "step": 21814
+    },
+    {
+      "epoch": 0.7933592755573335,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1571964107538303e-06,
+      "loss": 0.8689,
+      "step": 21815
+    },
+    {
+      "epoch": 0.7933956431610721,
+      "grad_norm": 0.0,
+      "learning_rate": 2.156465712032898e-06,
+      "loss": 0.8938,
+      "step": 21816
+    },
+    {
+      "epoch": 0.7934320107648107,
+      "grad_norm": 0.0,
+      "learning_rate": 2.155735122129875e-06,
+      "loss": 0.9476,
+      "step": 21817
+    },
+    {
+      "epoch": 0.7934683783685493,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1550046410548965e-06,
+      "loss": 0.8575,
+      "step": 21818
+    },
+    {
+      "epoch": 0.7935047459722879,
+      "grad_norm": 0.0,
+      "learning_rate": 2.154274268818102e-06,
+      "loss": 0.9497,
+      "step": 21819
+    },
+    {
+      "epoch": 0.7935411135760265,
+      "grad_norm": 0.0,
+      "learning_rate": 2.153544005429615e-06,
+      "loss": 0.9663,
+      "step": 21820
+    },
+    {
+      "epoch": 0.7935774811797651,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1528138508995757e-06,
+      "loss": 0.9473,
+      "step": 21821
+    },
+    {
+      "epoch": 0.7936138487835036,
+      "grad_norm": 0.0,
+      "learning_rate": 2.15208380523811e-06,
+      "loss": 0.9461,
+      "step": 21822
+    },
+    {
+      "epoch": 0.7936502163872422,
+      "grad_norm": 0.0,
+      "learning_rate": 2.151353868455347e-06,
+      "loss": 0.8913,
+      "step": 21823
+    },
+    {
+      "epoch": 0.7936865839909808,
+      "grad_norm": 0.0,
+      "learning_rate": 2.150624040561411e-06,
+      "loss": 0.8947,
+      "step": 21824
+    },
+    {
+      "epoch": 0.7937229515947194,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1498943215664346e-06,
+      "loss": 0.9284,
+      "step": 21825
+    },
+    {
+      "epoch": 0.793759319198458,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1491647114805346e-06,
+      "loss": 0.894,
+      "step": 21826
+    },
+    {
+      "epoch": 0.7937956868021966,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1484352103138317e-06,
+      "loss": 1.0173,
+      "step": 21827
+    },
+    {
+      "epoch": 0.7938320544059352,
+      "grad_norm": 0.0,
+      "learning_rate": 2.147705818076453e-06,
+      "loss": 0.9186,
+      "step": 21828
+    },
+    {
+      "epoch": 0.7938684220096738,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1469765347785153e-06,
+      "loss": 0.9262,
+      "step": 21829
+    },
+    {
+      "epoch": 0.7939047896134124,
+      "grad_norm": 0.0,
+      "learning_rate": 2.146247360430137e-06,
+      "loss": 0.8426,
+      "step": 21830
+    },
+    {
+      "epoch": 0.793941157217151,
+      "grad_norm": 0.0,
+      "learning_rate": 2.14551829504143e-06,
+      "loss": 0.9403,
+      "step": 21831
+    },
+    {
+      "epoch": 0.7939775248208896,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1447893386225182e-06,
+      "loss": 0.8805,
+      "step": 21832
+    },
+    {
+      "epoch": 0.7940138924246282,
+      "grad_norm": 0.0,
+      "learning_rate": 2.144060491183504e-06,
+      "loss": 0.9551,
+      "step": 21833
+    },
+    {
+      "epoch": 0.7940502600283668,
+      "grad_norm": 0.0,
+      "learning_rate": 2.143331752734508e-06,
+      "loss": 0.969,
+      "step": 21834
+    },
+    {
+      "epoch": 0.7940866276321054,
+      "grad_norm": 0.0,
+      "learning_rate": 2.142603123285635e-06,
+      "loss": 0.8956,
+      "step": 21835
+    },
+    {
+      "epoch": 0.794122995235844,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1418746028469963e-06,
+      "loss": 0.9844,
+      "step": 21836
+    },
+    {
+      "epoch": 0.7941593628395825,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1411461914286957e-06,
+      "loss": 0.8654,
+      "step": 21837
+    },
+    {
+      "epoch": 0.794195730443321,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1404178890408468e-06,
+      "loss": 0.8969,
+      "step": 21838
+    },
+    {
+      "epoch": 0.7942320980470596,
+      "grad_norm": 0.0,
+      "learning_rate": 2.139689695693543e-06,
+      "loss": 0.9093,
+      "step": 21839
+    },
+    {
+      "epoch": 0.7942684656507982,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1389616113968946e-06,
+      "loss": 0.9493,
+      "step": 21840
+    },
+    {
+      "epoch": 0.7943048332545368,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1382336361610012e-06,
+      "loss": 0.9024,
+      "step": 21841
+    },
+    {
+      "epoch": 0.7943412008582754,
+      "grad_norm": 0.0,
+      "learning_rate": 2.137505769995961e-06,
+      "loss": 0.9007,
+      "step": 21842
+    },
+    {
+      "epoch": 0.794377568462014,
+      "grad_norm": 0.0,
+      "learning_rate": 2.136778012911871e-06,
+      "loss": 0.9106,
+      "step": 21843
+    },
+    {
+      "epoch": 0.7944139360657526,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1360503649188335e-06,
+      "loss": 1.0453,
+      "step": 21844
+    },
+    {
+      "epoch": 0.7944503036694912,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1353228260269387e-06,
+      "loss": 0.9579,
+      "step": 21845
+    },
+    {
+      "epoch": 0.7944866712732298,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1345953962462775e-06,
+      "loss": 0.8684,
+      "step": 21846
+    },
+    {
+      "epoch": 0.7945230388769684,
+      "grad_norm": 0.0,
+      "learning_rate": 2.133868075586948e-06,
+      "loss": 0.9469,
+      "step": 21847
+    },
+    {
+      "epoch": 0.794559406480707,
+      "grad_norm": 0.0,
+      "learning_rate": 2.133140864059039e-06,
+      "loss": 1.0541,
+      "step": 21848
+    },
+    {
+      "epoch": 0.7945957740844456,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1324137616726394e-06,
+      "loss": 0.9527,
+      "step": 21849
+    },
+    {
+      "epoch": 0.7946321416881842,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1316867684378327e-06,
+      "loss": 0.9544,
+      "step": 21850
+    },
+    {
+      "epoch": 0.7946685092919228,
+      "grad_norm": 0.0,
+      "learning_rate": 2.130959884364714e-06,
+      "loss": 1.0522,
+      "step": 21851
+    },
+    {
+      "epoch": 0.7947048768956614,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1302331094633568e-06,
+      "loss": 0.8671,
+      "step": 21852
+    },
+    {
+      "epoch": 0.7947412444994,
+      "grad_norm": 0.0,
+      "learning_rate": 2.129506443743852e-06,
+      "loss": 0.8506,
+      "step": 21853
+    },
+    {
+      "epoch": 0.7947776121031386,
+      "grad_norm": 0.0,
+      "learning_rate": 2.128779887216279e-06,
+      "loss": 1.0339,
+      "step": 21854
+    },
+    {
+      "epoch": 0.7948139797068771,
+      "grad_norm": 0.0,
+      "learning_rate": 2.128053439890717e-06,
+      "loss": 0.9453,
+      "step": 21855
+    },
+    {
+      "epoch": 0.7948503473106157,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1273271017772424e-06,
+      "loss": 0.8101,
+      "step": 21856
+    },
+    {
+      "epoch": 0.7948867149143543,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1266008728859387e-06,
+      "loss": 0.9525,
+      "step": 21857
+    },
+    {
+      "epoch": 0.7949230825180928,
+      "grad_norm": 0.0,
+      "learning_rate": 2.125874753226872e-06,
+      "loss": 0.8817,
+      "step": 21858
+    },
+    {
+      "epoch": 0.7949594501218314,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1251487428101245e-06,
+      "loss": 0.9889,
+      "step": 21859
+    },
+    {
+      "epoch": 0.79499581772557,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1244228416457645e-06,
+      "loss": 0.9259,
+      "step": 21860
+    },
+    {
+      "epoch": 0.7950321853293086,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1236970497438646e-06,
+      "loss": 1.0343,
+      "step": 21861
+    },
+    {
+      "epoch": 0.7950685529330472,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1229713671144903e-06,
+      "loss": 0.9912,
+      "step": 21862
+    },
+    {
+      "epoch": 0.7951049205367858,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1222457937677144e-06,
+      "loss": 0.8308,
+      "step": 21863
+    },
+    {
+      "epoch": 0.7951412881405244,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1215203297136035e-06,
+      "loss": 0.7321,
+      "step": 21864
+    },
+    {
+      "epoch": 0.795177655744263,
+      "grad_norm": 0.0,
+      "learning_rate": 2.120794974962215e-06,
+      "loss": 0.9942,
+      "step": 21865
+    },
+    {
+      "epoch": 0.7952140233480016,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1200697295236193e-06,
+      "loss": 0.9755,
+      "step": 21866
+    },
+    {
+      "epoch": 0.7952503909517402,
+      "grad_norm": 0.0,
+      "learning_rate": 2.119344593407876e-06,
+      "loss": 0.9194,
+      "step": 21867
+    },
+    {
+      "epoch": 0.7952867585554788,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1186195666250455e-06,
+      "loss": 0.9899,
+      "step": 21868
+    },
+    {
+      "epoch": 0.7953231261592174,
+      "grad_norm": 0.0,
+      "learning_rate": 2.117894649185184e-06,
+      "loss": 0.8727,
+      "step": 21869
+    },
+    {
+      "epoch": 0.795359493762956,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1171698410983554e-06,
+      "loss": 0.8492,
+      "step": 21870
+    },
+    {
+      "epoch": 0.7953958613666946,
+      "grad_norm": 0.0,
+      "learning_rate": 2.116445142374607e-06,
+      "loss": 0.9449,
+      "step": 21871
+    },
+    {
+      "epoch": 0.7954322289704331,
+      "grad_norm": 0.0,
+      "learning_rate": 2.115720553023999e-06,
+      "loss": 0.7985,
+      "step": 21872
+    },
+    {
+      "epoch": 0.7954685965741717,
+      "grad_norm": 0.0,
+      "learning_rate": 2.114996073056581e-06,
+      "loss": 0.9419,
+      "step": 21873
+    },
+    {
+      "epoch": 0.7955049641779103,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1142717024824067e-06,
+      "loss": 0.9064,
+      "step": 21874
+    },
+    {
+      "epoch": 0.7955413317816489,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1135474413115207e-06,
+      "loss": 0.8603,
+      "step": 21875
+    },
+    {
+      "epoch": 0.7955776993853875,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1128232895539803e-06,
+      "loss": 0.8576,
+      "step": 21876
+    },
+    {
+      "epoch": 0.7956140669891261,
+      "grad_norm": 0.0,
+      "learning_rate": 2.11209924721982e-06,
+      "loss": 0.8944,
+      "step": 21877
+    },
+    {
+      "epoch": 0.7956504345928647,
+      "grad_norm": 0.0,
+      "learning_rate": 2.111375314319094e-06,
+      "loss": 0.8578,
+      "step": 21878
+    },
+    {
+      "epoch": 0.7956868021966033,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1106514908618437e-06,
+      "loss": 0.8231,
+      "step": 21879
+    },
+    {
+      "epoch": 0.7957231698003419,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1099277768581107e-06,
+      "loss": 0.9519,
+      "step": 21880
+    },
+    {
+      "epoch": 0.7957595374040805,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1092041723179312e-06,
+      "loss": 0.848,
+      "step": 21881
+    },
+    {
+      "epoch": 0.795795905007819,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1084806772513523e-06,
+      "loss": 0.8626,
+      "step": 21882
+    },
+    {
+      "epoch": 0.7958322726115576,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1077572916684095e-06,
+      "loss": 0.8843,
+      "step": 21883
+    },
+    {
+      "epoch": 0.7958686402152962,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1070340155791314e-06,
+      "loss": 0.7781,
+      "step": 21884
+    },
+    {
+      "epoch": 0.7959050078190348,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1063108489935604e-06,
+      "loss": 1.0359,
+      "step": 21885
+    },
+    {
+      "epoch": 0.7959413754227734,
+      "grad_norm": 0.0,
+      "learning_rate": 2.105587791921727e-06,
+      "loss": 0.9545,
+      "step": 21886
+    },
+    {
+      "epoch": 0.795977743026512,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1048648443736617e-06,
+      "loss": 0.8549,
+      "step": 21887
+    },
+    {
+      "epoch": 0.7960141106302506,
+      "grad_norm": 0.0,
+      "learning_rate": 2.104142006359392e-06,
+      "loss": 0.9326,
+      "step": 21888
+    },
+    {
+      "epoch": 0.7960504782339891,
+      "grad_norm": 0.0,
+      "learning_rate": 2.103419277888955e-06,
+      "loss": 0.9393,
+      "step": 21889
+    },
+    {
+      "epoch": 0.7960868458377277,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1026966589723665e-06,
+      "loss": 0.977,
+      "step": 21890
+    },
+    {
+      "epoch": 0.7961232134414663,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1019741496196597e-06,
+      "loss": 0.8289,
+      "step": 21891
+    },
+    {
+      "epoch": 0.7961595810452049,
+      "grad_norm": 0.0,
+      "learning_rate": 2.101251749840856e-06,
+      "loss": 1.0054,
+      "step": 21892
+    },
+    {
+      "epoch": 0.7961959486489435,
+      "grad_norm": 0.0,
+      "learning_rate": 2.100529459645977e-06,
+      "loss": 0.8886,
+      "step": 21893
+    },
+    {
+      "epoch": 0.7962323162526821,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0998072790450406e-06,
+      "loss": 0.9286,
+      "step": 21894
+    },
+    {
+      "epoch": 0.7962686838564207,
+      "grad_norm": 0.0,
+      "learning_rate": 2.099085208048073e-06,
+      "loss": 0.7812,
+      "step": 21895
+    },
+    {
+      "epoch": 0.7963050514601593,
+      "grad_norm": 0.0,
+      "learning_rate": 2.098363246665088e-06,
+      "loss": 0.8846,
+      "step": 21896
+    },
+    {
+      "epoch": 0.7963414190638979,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0976413949061025e-06,
+      "loss": 0.941,
+      "step": 21897
+    },
+    {
+      "epoch": 0.7963777866676365,
+      "grad_norm": 0.0,
+      "learning_rate": 2.09691965278113e-06,
+      "loss": 0.8743,
+      "step": 21898
+    },
+    {
+      "epoch": 0.7964141542713751,
+      "grad_norm": 0.0,
+      "learning_rate": 2.096198020300184e-06,
+      "loss": 0.9103,
+      "step": 21899
+    },
+    {
+      "epoch": 0.7964505218751137,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0954764974732744e-06,
+      "loss": 0.8059,
+      "step": 21900
+    },
+    {
+      "epoch": 0.7964868894788523,
+      "grad_norm": 0.0,
+      "learning_rate": 2.094755084310416e-06,
+      "loss": 0.9975,
+      "step": 21901
+    },
+    {
+      "epoch": 0.7965232570825909,
+      "grad_norm": 0.0,
+      "learning_rate": 2.094033780821614e-06,
+      "loss": 0.8342,
+      "step": 21902
+    },
+    {
+      "epoch": 0.7965596246863295,
+      "grad_norm": 0.0,
+      "learning_rate": 2.093312587016877e-06,
+      "loss": 0.8574,
+      "step": 21903
+    },
+    {
+      "epoch": 0.7965959922900681,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0925915029062083e-06,
+      "loss": 0.9816,
+      "step": 21904
+    },
+    {
+      "epoch": 0.7966323598938067,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0918705284996157e-06,
+      "loss": 0.942,
+      "step": 21905
+    },
+    {
+      "epoch": 0.7966687274975451,
+      "grad_norm": 0.0,
+      "learning_rate": 2.091149663807097e-06,
+      "loss": 0.972,
+      "step": 21906
+    },
+    {
+      "epoch": 0.7967050951012837,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0904289088386543e-06,
+      "loss": 0.8167,
+      "step": 21907
+    },
+    {
+      "epoch": 0.7967414627050223,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0897082636042908e-06,
+      "loss": 1.0217,
+      "step": 21908
+    },
+    {
+      "epoch": 0.7967778303087609,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0889877281140015e-06,
+      "loss": 0.8448,
+      "step": 21909
+    },
+    {
+      "epoch": 0.7968141979124995,
+      "grad_norm": 0.0,
+      "learning_rate": 2.088267302377783e-06,
+      "loss": 0.9821,
+      "step": 21910
+    },
+    {
+      "epoch": 0.7968505655162381,
+      "grad_norm": 0.0,
+      "learning_rate": 2.087546986405631e-06,
+      "loss": 0.8462,
+      "step": 21911
+    },
+    {
+      "epoch": 0.7968869331199767,
+      "grad_norm": 0.0,
+      "learning_rate": 2.086826780207538e-06,
+      "loss": 0.8656,
+      "step": 21912
+    },
+    {
+      "epoch": 0.7969233007237153,
+      "grad_norm": 0.0,
+      "learning_rate": 2.086106683793494e-06,
+      "loss": 0.9814,
+      "step": 21913
+    },
+    {
+      "epoch": 0.7969596683274539,
+      "grad_norm": 0.0,
+      "learning_rate": 2.085386697173494e-06,
+      "loss": 0.9269,
+      "step": 21914
+    },
+    {
+      "epoch": 0.7969960359311925,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0846668203575225e-06,
+      "loss": 0.8346,
+      "step": 21915
+    },
+    {
+      "epoch": 0.7970324035349311,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0839470533555704e-06,
+      "loss": 0.979,
+      "step": 21916
+    },
+    {
+      "epoch": 0.7970687711386697,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0832273961776182e-06,
+      "loss": 0.9812,
+      "step": 21917
+    },
+    {
+      "epoch": 0.7971051387424083,
+      "grad_norm": 0.0,
+      "learning_rate": 2.082507848833659e-06,
+      "loss": 0.8688,
+      "step": 21918
+    },
+    {
+      "epoch": 0.7971415063461469,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0817884113336638e-06,
+      "loss": 0.9521,
+      "step": 21919
+    },
+    {
+      "epoch": 0.7971778739498855,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0810690836876235e-06,
+      "loss": 0.9637,
+      "step": 21920
+    },
+    {
+      "epoch": 0.7972142415536241,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0803498659055144e-06,
+      "loss": 0.8723,
+      "step": 21921
+    },
+    {
+      "epoch": 0.7972506091573626,
+      "grad_norm": 0.0,
+      "learning_rate": 2.079630757997313e-06,
+      "loss": 0.8508,
+      "step": 21922
+    },
+    {
+      "epoch": 0.7972869767611012,
+      "grad_norm": 0.0,
+      "learning_rate": 2.078911759972998e-06,
+      "loss": 0.9281,
+      "step": 21923
+    },
+    {
+      "epoch": 0.7973233443648398,
+      "grad_norm": 0.0,
+      "learning_rate": 2.078192871842545e-06,
+      "loss": 0.9857,
+      "step": 21924
+    },
+    {
+      "epoch": 0.7973597119685784,
+      "grad_norm": 0.0,
+      "learning_rate": 2.077474093615924e-06,
+      "loss": 0.9072,
+      "step": 21925
+    },
+    {
+      "epoch": 0.797396079572317,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0767554253031088e-06,
+      "loss": 0.9382,
+      "step": 21926
+    },
+    {
+      "epoch": 0.7974324471760555,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0760368669140716e-06,
+      "loss": 0.9536,
+      "step": 21927
+    },
+    {
+      "epoch": 0.7974688147797941,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0753184184587814e-06,
+      "loss": 0.9529,
+      "step": 21928
+    },
+    {
+      "epoch": 0.7975051823835327,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0746000799472043e-06,
+      "loss": 0.9702,
+      "step": 21929
+    },
+    {
+      "epoch": 0.7975415499872713,
+      "grad_norm": 0.0,
+      "learning_rate": 2.073881851389303e-06,
+      "loss": 0.9762,
+      "step": 21930
+    },
+    {
+      "epoch": 0.7975779175910099,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0731637327950514e-06,
+      "loss": 0.9319,
+      "step": 21931
+    },
+    {
+      "epoch": 0.7976142851947485,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0724457241744012e-06,
+      "loss": 0.9115,
+      "step": 21932
+    },
+    {
+      "epoch": 0.7976506527984871,
+      "grad_norm": 0.0,
+      "learning_rate": 2.071727825537321e-06,
+      "loss": 0.855,
+      "step": 21933
+    },
+    {
+      "epoch": 0.7976870204022257,
+      "grad_norm": 0.0,
+      "learning_rate": 2.071010036893768e-06,
+      "loss": 0.9869,
+      "step": 21934
+    },
+    {
+      "epoch": 0.7977233880059643,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0702923582537015e-06,
+      "loss": 1.1268,
+      "step": 21935
+    },
+    {
+      "epoch": 0.7977597556097029,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0695747896270746e-06,
+      "loss": 0.8749,
+      "step": 21936
+    },
+    {
+      "epoch": 0.7977961232134415,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0688573310238513e-06,
+      "loss": 0.8574,
+      "step": 21937
+    },
+    {
+      "epoch": 0.7978324908171801,
+      "grad_norm": 0.0,
+      "learning_rate": 2.068139982453975e-06,
+      "loss": 0.9543,
+      "step": 21938
+    },
+    {
+      "epoch": 0.7978688584209186,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0674227439274053e-06,
+      "loss": 0.939,
+      "step": 21939
+    },
+    {
+      "epoch": 0.7979052260246572,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0667056154540886e-06,
+      "loss": 0.8941,
+      "step": 21940
+    },
+    {
+      "epoch": 0.7979415936283958,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0659885970439773e-06,
+      "loss": 1.0139,
+      "step": 21941
+    },
+    {
+      "epoch": 0.7979779612321344,
+      "grad_norm": 0.0,
+      "learning_rate": 2.065271688707017e-06,
+      "loss": 0.921,
+      "step": 21942
+    },
+    {
+      "epoch": 0.798014328835873,
+      "grad_norm": 0.0,
+      "learning_rate": 2.064554890453153e-06,
+      "loss": 1.0212,
+      "step": 21943
+    },
+    {
+      "epoch": 0.7980506964396116,
+      "grad_norm": 0.0,
+      "learning_rate": 2.063838202292332e-06,
+      "loss": 0.9726,
+      "step": 21944
+    },
+    {
+      "epoch": 0.7980870640433502,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0631216242344933e-06,
+      "loss": 0.9361,
+      "step": 21945
+    },
+    {
+      "epoch": 0.7981234316470888,
+      "grad_norm": 0.0,
+      "learning_rate": 2.062405156289584e-06,
+      "loss": 1.0103,
+      "step": 21946
+    },
+    {
+      "epoch": 0.7981597992508274,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0616887984675405e-06,
+      "loss": 0.9943,
+      "step": 21947
+    },
+    {
+      "epoch": 0.798196166854566,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0609725507783017e-06,
+      "loss": 0.8796,
+      "step": 21948
+    },
+    {
+      "epoch": 0.7982325344583046,
+      "grad_norm": 0.0,
+      "learning_rate": 2.060256413231804e-06,
+      "loss": 0.7577,
+      "step": 21949
+    },
+    {
+      "epoch": 0.7982689020620432,
+      "grad_norm": 0.0,
+      "learning_rate": 2.059540385837987e-06,
+      "loss": 0.8953,
+      "step": 21950
+    },
+    {
+      "epoch": 0.7983052696657817,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0588244686067758e-06,
+      "loss": 1.0214,
+      "step": 21951
+    },
+    {
+      "epoch": 0.7983416372695203,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0581086615481114e-06,
+      "loss": 0.915,
+      "step": 21952
+    },
+    {
+      "epoch": 0.7983780048732589,
+      "grad_norm": 0.0,
+      "learning_rate": 2.05739296467192e-06,
+      "loss": 0.9493,
+      "step": 21953
+    },
+    {
+      "epoch": 0.7984143724769975,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0566773779881344e-06,
+      "loss": 0.8482,
+      "step": 21954
+    },
+    {
+      "epoch": 0.7984507400807361,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0559619015066757e-06,
+      "loss": 0.9231,
+      "step": 21955
+    },
+    {
+      "epoch": 0.7984871076844746,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0552465352374807e-06,
+      "loss": 0.9642,
+      "step": 21956
+    },
+    {
+      "epoch": 0.7985234752882132,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0545312791904613e-06,
+      "loss": 0.9138,
+      "step": 21957
+    },
+    {
+      "epoch": 0.7985598428919518,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0538161333755512e-06,
+      "loss": 0.8792,
+      "step": 21958
+    },
+    {
+      "epoch": 0.7985962104956904,
+      "grad_norm": 0.0,
+      "learning_rate": 2.053101097802668e-06,
+      "loss": 0.9166,
+      "step": 21959
+    },
+    {
+      "epoch": 0.798632578099429,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0523861724817316e-06,
+      "loss": 0.8183,
+      "step": 21960
+    },
+    {
+      "epoch": 0.7986689457031676,
+      "grad_norm": 0.0,
+      "learning_rate": 2.051671357422661e-06,
+      "loss": 1.009,
+      "step": 21961
+    },
+    {
+      "epoch": 0.7987053133069062,
+      "grad_norm": 0.0,
+      "learning_rate": 2.05095665263537e-06,
+      "loss": 0.9553,
+      "step": 21962
+    },
+    {
+      "epoch": 0.7987416809106448,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0502420581297834e-06,
+      "loss": 0.9251,
+      "step": 21963
+    },
+    {
+      "epoch": 0.7987780485143834,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0495275739158027e-06,
+      "loss": 0.9401,
+      "step": 21964
+    },
+    {
+      "epoch": 0.798814416118122,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0488132000033502e-06,
+      "loss": 0.9219,
+      "step": 21965
+    },
+    {
+      "epoch": 0.7988507837218606,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0480989364023316e-06,
+      "loss": 0.9991,
+      "step": 21966
+    },
+    {
+      "epoch": 0.7988871513255992,
+      "grad_norm": 0.0,
+      "learning_rate": 2.04738478312266e-06,
+      "loss": 1.0208,
+      "step": 21967
+    },
+    {
+      "epoch": 0.7989235189293378,
+      "grad_norm": 0.0,
+      "learning_rate": 2.046670740174238e-06,
+      "loss": 0.8323,
+      "step": 21968
+    },
+    {
+      "epoch": 0.7989598865330764,
+      "grad_norm": 0.0,
+      "learning_rate": 2.04595680756698e-06,
+      "loss": 0.8963,
+      "step": 21969
+    },
+    {
+      "epoch": 0.798996254136815,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0452429853107814e-06,
+      "loss": 0.9397,
+      "step": 21970
+    },
+    {
+      "epoch": 0.7990326217405536,
+      "grad_norm": 0.0,
+      "learning_rate": 2.044529273415552e-06,
+      "loss": 1.0255,
+      "step": 21971
+    },
+    {
+      "epoch": 0.7990689893442922,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0438156718911918e-06,
+      "loss": 0.9735,
+      "step": 21972
+    },
+    {
+      "epoch": 0.7991053569480306,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0431021807476004e-06,
+      "loss": 0.8517,
+      "step": 21973
+    },
+    {
+      "epoch": 0.7991417245517692,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0423887999946744e-06,
+      "loss": 0.812,
+      "step": 21974
+    },
+    {
+      "epoch": 0.7991780921555078,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0416755296423173e-06,
+      "loss": 0.8586,
+      "step": 21975
+    },
+    {
+      "epoch": 0.7992144597592464,
+      "grad_norm": 0.0,
+      "learning_rate": 2.04096236970042e-06,
+      "loss": 0.8696,
+      "step": 21976
+    },
+    {
+      "epoch": 0.799250827362985,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0402493201788774e-06,
+      "loss": 0.9016,
+      "step": 21977
+    },
+    {
+      "epoch": 0.7992871949667236,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0395363810875814e-06,
+      "loss": 0.926,
+      "step": 21978
+    },
+    {
+      "epoch": 0.7993235625704622,
+      "grad_norm": 0.0,
+      "learning_rate": 2.038823552436424e-06,
+      "loss": 0.9564,
+      "step": 21979
+    },
+    {
+      "epoch": 0.7993599301742008,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0381108342352953e-06,
+      "loss": 0.8918,
+      "step": 21980
+    },
+    {
+      "epoch": 0.7993962977779394,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0373982264940793e-06,
+      "loss": 0.9379,
+      "step": 21981
+    },
+    {
+      "epoch": 0.799432665381678,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0366857292226707e-06,
+      "loss": 0.8865,
+      "step": 21982
+    },
+    {
+      "epoch": 0.7994690329854166,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0359733424309434e-06,
+      "loss": 0.944,
+      "step": 21983
+    },
+    {
+      "epoch": 0.7995054005891552,
+      "grad_norm": 0.0,
+      "learning_rate": 2.03526106612879e-06,
+      "loss": 0.8616,
+      "step": 21984
+    },
+    {
+      "epoch": 0.7995417681928938,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0345489003260875e-06,
+      "loss": 0.8138,
+      "step": 21985
+    },
+    {
+      "epoch": 0.7995781357966324,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0338368450327184e-06,
+      "loss": 0.8689,
+      "step": 21986
+    },
+    {
+      "epoch": 0.799614503400371,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0331249002585574e-06,
+      "loss": 0.8373,
+      "step": 21987
+    },
+    {
+      "epoch": 0.7996508710041096,
+      "grad_norm": 0.0,
+      "learning_rate": 2.032413066013491e-06,
+      "loss": 0.7431,
+      "step": 21988
+    },
+    {
+      "epoch": 0.7996872386078482,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0317013423073827e-06,
+      "loss": 0.9981,
+      "step": 21989
+    },
+    {
+      "epoch": 0.7997236062115867,
+      "grad_norm": 0.0,
+      "learning_rate": 2.030989729150116e-06,
+      "loss": 0.8794,
+      "step": 21990
+    },
+    {
+      "epoch": 0.7997599738153253,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0302782265515598e-06,
+      "loss": 0.9719,
+      "step": 21991
+    },
+    {
+      "epoch": 0.7997963414190639,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0295668345215857e-06,
+      "loss": 0.9428,
+      "step": 21992
+    },
+    {
+      "epoch": 0.7998327090228025,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0288555530700604e-06,
+      "loss": 0.9736,
+      "step": 21993
+    },
+    {
+      "epoch": 0.799869076626541,
+      "grad_norm": 0.0,
+      "learning_rate": 2.028144382206858e-06,
+      "loss": 0.8923,
+      "step": 21994
+    },
+    {
+      "epoch": 0.7999054442302796,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0274333219418417e-06,
+      "loss": 0.9297,
+      "step": 21995
+    },
+    {
+      "epoch": 0.7999418118340182,
+      "grad_norm": 0.0,
+      "learning_rate": 2.026722372284876e-06,
+      "loss": 0.9047,
+      "step": 21996
+    },
+    {
+      "epoch": 0.7999781794377568,
+      "grad_norm": 0.0,
+      "learning_rate": 2.026011533245822e-06,
+      "loss": 0.9587,
+      "step": 21997
+    },
+    {
+      "epoch": 0.8000145470414954,
+      "grad_norm": 0.0,
+      "learning_rate": 2.02530080483455e-06,
+      "loss": 0.9637,
+      "step": 21998
+    },
+    {
+      "epoch": 0.800050914645234,
+      "grad_norm": 0.0,
+      "learning_rate": 2.024590187060912e-06,
+      "loss": 0.8966,
+      "step": 21999
+    },
+    {
+      "epoch": 0.8000872822489726,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0238796799347672e-06,
+      "loss": 1.0383,
+      "step": 22000
+    },
+    {
+      "epoch": 0.8001236498527112,
+      "grad_norm": 0.0,
+      "learning_rate": 2.02316928346598e-06,
+      "loss": 0.8988,
+      "step": 22001
+    },
+    {
+      "epoch": 0.8001600174564498,
+      "grad_norm": 0.0,
+      "learning_rate": 2.022458997664396e-06,
+      "loss": 0.9222,
+      "step": 22002
+    },
+    {
+      "epoch": 0.8001963850601884,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0217488225398784e-06,
+      "loss": 0.9954,
+      "step": 22003
+    },
+    {
+      "epoch": 0.800232752663927,
+      "grad_norm": 0.0,
+      "learning_rate": 2.021038758102275e-06,
+      "loss": 0.8245,
+      "step": 22004
+    },
+    {
+      "epoch": 0.8002691202676656,
+      "grad_norm": 0.0,
+      "learning_rate": 2.020328804361439e-06,
+      "loss": 1.0023,
+      "step": 22005
+    },
+    {
+      "epoch": 0.8003054878714042,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0196189613272166e-06,
+      "loss": 0.8895,
+      "step": 22006
+    },
+    {
+      "epoch": 0.8003418554751427,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0189092290094603e-06,
+      "loss": 0.9216,
+      "step": 22007
+    },
+    {
+      "epoch": 0.8003782230788813,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0181996074180154e-06,
+      "loss": 0.9464,
+      "step": 22008
+    },
+    {
+      "epoch": 0.8004145906826199,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0174900965627253e-06,
+      "loss": 0.9384,
+      "step": 22009
+    },
+    {
+      "epoch": 0.8004509582863585,
+      "grad_norm": 0.0,
+      "learning_rate": 2.016780696453435e-06,
+      "loss": 1.0318,
+      "step": 22010
+    },
+    {
+      "epoch": 0.8004873258900971,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0160714070999865e-06,
+      "loss": 0.8488,
+      "step": 22011
+    },
+    {
+      "epoch": 0.8005236934938357,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0153622285122165e-06,
+      "loss": 0.916,
+      "step": 22012
+    },
+    {
+      "epoch": 0.8005600610975743,
+      "grad_norm": 0.0,
+      "learning_rate": 2.014653160699969e-06,
+      "loss": 0.8714,
+      "step": 22013
+    },
+    {
+      "epoch": 0.8005964287013129,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0139442036730793e-06,
+      "loss": 1.019,
+      "step": 22014
+    },
+    {
+      "epoch": 0.8006327963050515,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0132353574413832e-06,
+      "loss": 0.989,
+      "step": 22015
+    },
+    {
+      "epoch": 0.8006691639087901,
+      "grad_norm": 0.0,
+      "learning_rate": 2.012526622014712e-06,
+      "loss": 1.0143,
+      "step": 22016
+    },
+    {
+      "epoch": 0.8007055315125287,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0118179974029072e-06,
+      "loss": 0.8315,
+      "step": 22017
+    },
+    {
+      "epoch": 0.8007418991162673,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0111094836157907e-06,
+      "loss": 0.8727,
+      "step": 22018
+    },
+    {
+      "epoch": 0.8007782667200058,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0104010806631924e-06,
+      "loss": 0.7829,
+      "step": 22019
+    },
+    {
+      "epoch": 0.8008146343237444,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0096927885549464e-06,
+      "loss": 1.0119,
+      "step": 22020
+    },
+    {
+      "epoch": 0.800851001927483,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0089846073008758e-06,
+      "loss": 1.03,
+      "step": 22021
+    },
+    {
+      "epoch": 0.8008873695312216,
+      "grad_norm": 0.0,
+      "learning_rate": 2.008276536910806e-06,
+      "loss": 0.9501,
+      "step": 22022
+    },
+    {
+      "epoch": 0.8009237371349601,
+      "grad_norm": 0.0,
+      "learning_rate": 2.007568577394561e-06,
+      "loss": 0.9032,
+      "step": 22023
+    },
+    {
+      "epoch": 0.8009601047386987,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0068607287619625e-06,
+      "loss": 0.9504,
+      "step": 22024
+    },
+    {
+      "epoch": 0.8009964723424373,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0061529910228273e-06,
+      "loss": 0.9126,
+      "step": 22025
+    },
+    {
+      "epoch": 0.8010328399461759,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0054453641869797e-06,
+      "loss": 0.976,
+      "step": 22026
+    },
+    {
+      "epoch": 0.8010692075499145,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0047378482642354e-06,
+      "loss": 0.942,
+      "step": 22027
+    },
+    {
+      "epoch": 0.8011055751536531,
+      "grad_norm": 0.0,
+      "learning_rate": 2.004030443264411e-06,
+      "loss": 0.8538,
+      "step": 22028
+    },
+    {
+      "epoch": 0.8011419427573917,
+      "grad_norm": 0.0,
+      "learning_rate": 2.003323149197315e-06,
+      "loss": 0.9695,
+      "step": 22029
+    },
+    {
+      "epoch": 0.8011783103611303,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0026159660727705e-06,
+      "loss": 0.9694,
+      "step": 22030
+    },
+    {
+      "epoch": 0.8012146779648689,
+      "grad_norm": 0.0,
+      "learning_rate": 2.001908893900577e-06,
+      "loss": 0.9787,
+      "step": 22031
+    },
+    {
+      "epoch": 0.8012510455686075,
+      "grad_norm": 0.0,
+      "learning_rate": 2.001201932690553e-06,
+      "loss": 0.8979,
+      "step": 22032
+    },
+    {
+      "epoch": 0.8012874131723461,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0004950824525027e-06,
+      "loss": 0.8864,
+      "step": 22033
+    },
+    {
+      "epoch": 0.8013237807760847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999788343196234e-06,
+      "loss": 0.9651,
+      "step": 22034
+    },
+    {
+      "epoch": 0.8013601483798233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9990817149315478e-06,
+      "loss": 0.9056,
+      "step": 22035
+    },
+    {
+      "epoch": 0.8013965159835619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9983751976682564e-06,
+      "loss": 0.9054,
+      "step": 22036
+    },
+    {
+      "epoch": 0.8014328835873005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.997668791416154e-06,
+      "loss": 1.0699,
+      "step": 22037
+    },
+    {
+      "epoch": 0.8014692511910391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9969624961850407e-06,
+      "loss": 0.9366,
+      "step": 22038
+    },
+    {
+      "epoch": 0.8015056187947777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.996256311984721e-06,
+      "loss": 0.9616,
+      "step": 22039
+    },
+    {
+      "epoch": 0.8015419863985161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.995550238824988e-06,
+      "loss": 0.8398,
+      "step": 22040
+    },
+    {
+      "epoch": 0.8015783540022547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9948442767156395e-06,
+      "loss": 0.933,
+      "step": 22041
+    },
+    {
+      "epoch": 0.8016147216059933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9941384256664684e-06,
+      "loss": 0.9401,
+      "step": 22042
+    },
+    {
+      "epoch": 0.8016510892097319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9934326856872687e-06,
+      "loss": 0.7217,
+      "step": 22043
+    },
+    {
+      "epoch": 0.8016874568134705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9927270567878278e-06,
+      "loss": 0.9085,
+      "step": 22044
+    },
+    {
+      "epoch": 0.8017238244172091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9920215389779395e-06,
+      "loss": 1.0951,
+      "step": 22045
+    },
+    {
+      "epoch": 0.8017601920209477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9913161322673913e-06,
+      "loss": 0.8758,
+      "step": 22046
+    },
+    {
+      "epoch": 0.8017965596246863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9906108366659695e-06,
+      "loss": 0.939,
+      "step": 22047
+    },
+    {
+      "epoch": 0.8018329272284249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9899056521834546e-06,
+      "loss": 0.8808,
+      "step": 22048
+    },
+    {
+      "epoch": 0.8018692948321635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9892005788296397e-06,
+      "loss": 0.8891,
+      "step": 22049
+    },
+    {
+      "epoch": 0.8019056624359021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9884956166142965e-06,
+      "loss": 0.8394,
+      "step": 22050
+    },
+    {
+      "epoch": 0.8019420300396407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9877907655472118e-06,
+      "loss": 0.9352,
+      "step": 22051
+    },
+    {
+      "epoch": 0.8019783976433793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.987086025638162e-06,
+      "loss": 0.9728,
+      "step": 22052
+    },
+    {
+      "epoch": 0.8020147652471179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9863813968969257e-06,
+      "loss": 0.9672,
+      "step": 22053
+    },
+    {
+      "epoch": 0.8020511328508565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.985676879333275e-06,
+      "loss": 0.9255,
+      "step": 22054
+    },
+    {
+      "epoch": 0.8020875004545951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9849724729569907e-06,
+      "loss": 1.0044,
+      "step": 22055
+    },
+    {
+      "epoch": 0.8021238680583337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9842681777778395e-06,
+      "loss": 0.9767,
+      "step": 22056
+    },
+    {
+      "epoch": 0.8021602356620722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.983563993805593e-06,
+      "loss": 0.8689,
+      "step": 22057
+    },
+    {
+      "epoch": 0.8021966032658108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.982859921050023e-06,
+      "loss": 0.901,
+      "step": 22058
+    },
+    {
+      "epoch": 0.8022329708695494,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9821559595208984e-06,
+      "loss": 0.9022,
+      "step": 22059
+    },
+    {
+      "epoch": 0.802269338473288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9814521092279825e-06,
+      "loss": 0.7721,
+      "step": 22060
+    },
+    {
+      "epoch": 0.8023057060770266,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9807483701810404e-06,
+      "loss": 1.0125,
+      "step": 22061
+    },
+    {
+      "epoch": 0.8023420736807652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.980044742389842e-06,
+      "loss": 0.8597,
+      "step": 22062
+    },
+    {
+      "epoch": 0.8023784412845038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9793412258641386e-06,
+      "loss": 0.9204,
+      "step": 22063
+    },
+    {
+      "epoch": 0.8024148088882423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.978637820613698e-06,
+      "loss": 0.962,
+      "step": 22064
+    },
+    {
+      "epoch": 0.8024511764919809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9779345266482773e-06,
+      "loss": 0.8163,
+      "step": 22065
+    },
+    {
+      "epoch": 0.8024875440957195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9772313439776325e-06,
+      "loss": 0.8181,
+      "step": 22066
+    },
+    {
+      "epoch": 0.8025239116994581,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9765282726115177e-06,
+      "loss": 0.9783,
+      "step": 22067
+    },
+    {
+      "epoch": 0.8025602793031967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.975825312559695e-06,
+      "loss": 1.0231,
+      "step": 22068
+    },
+    {
+      "epoch": 0.8025966469069353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9751224638319056e-06,
+      "loss": 0.9389,
+      "step": 22069
+    },
+    {
+      "epoch": 0.8026330145106739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9744197264379096e-06,
+      "loss": 0.8087,
+      "step": 22070
+    },
+    {
+      "epoch": 0.8026693821144125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9737171003874524e-06,
+      "loss": 0.8395,
+      "step": 22071
+    },
+    {
+      "epoch": 0.8027057497181511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9730145856902826e-06,
+      "loss": 0.9224,
+      "step": 22072
+    },
+    {
+      "epoch": 0.8027421173218897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.972312182356143e-06,
+      "loss": 0.7701,
+      "step": 22073
+    },
+    {
+      "epoch": 0.8027784849256282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.971609890394787e-06,
+      "loss": 1.0229,
+      "step": 22074
+    },
+    {
+      "epoch": 0.8028148525293668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9709077098159536e-06,
+      "loss": 0.9551,
+      "step": 22075
+    },
+    {
+      "epoch": 0.8028512201331054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.97020564062938e-06,
+      "loss": 0.8215,
+      "step": 22076
+    },
+    {
+      "epoch": 0.802887587736844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.969503682844812e-06,
+      "loss": 1.0133,
+      "step": 22077
+    },
+    {
+      "epoch": 0.8029239553405826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9688018364719875e-06,
+      "loss": 0.9271,
+      "step": 22078
+    },
+    {
+      "epoch": 0.8029603229443212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.968100101520641e-06,
+      "loss": 0.8787,
+      "step": 22079
+    },
+    {
+      "epoch": 0.8029966905480598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9673984780005097e-06,
+      "loss": 0.8975,
+      "step": 22080
+    },
+    {
+      "epoch": 0.8030330581517984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.966696965921331e-06,
+      "loss": 1.0227,
+      "step": 22081
+    },
+    {
+      "epoch": 0.803069425755537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9659955652928297e-06,
+      "loss": 1.018,
+      "step": 22082
+    },
+    {
+      "epoch": 0.8031057933592756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.965294276124744e-06,
+      "loss": 0.9549,
+      "step": 22083
+    },
+    {
+      "epoch": 0.8031421609630142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9645930984268004e-06,
+      "loss": 0.943,
+      "step": 22084
+    },
+    {
+      "epoch": 0.8031785285667528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963892032208726e-06,
+      "loss": 0.8537,
+      "step": 22085
+    },
+    {
+      "epoch": 0.8032148961704914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9631910774802463e-06,
+      "loss": 0.796,
+      "step": 22086
+    },
+    {
+      "epoch": 0.80325126377423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.962490234251092e-06,
+      "loss": 1.0357,
+      "step": 22087
+    },
+    {
+      "epoch": 0.8032876313779685,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9617895025309773e-06,
+      "loss": 0.9637,
+      "step": 22088
+    },
+    {
+      "epoch": 0.8033239989817071,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9610888823296316e-06,
+      "loss": 0.9063,
+      "step": 22089
+    },
+    {
+      "epoch": 0.8033603665854457,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9603883736567707e-06,
+      "loss": 0.8612,
+      "step": 22090
+    },
+    {
+      "epoch": 0.8033967341891842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9596879765221154e-06,
+      "loss": 1.0107,
+      "step": 22091
+    },
+    {
+      "epoch": 0.8034331017929228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9589876909353787e-06,
+      "loss": 1.0258,
+      "step": 22092
+    },
+    {
+      "epoch": 0.8034694693966614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958287516906282e-06,
+      "loss": 1.0261,
+      "step": 22093
+    },
+    {
+      "epoch": 0.8035058370004,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9575874544445383e-06,
+      "loss": 0.9045,
+      "step": 22094
+    },
+    {
+      "epoch": 0.8035422046041386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9568875035598536e-06,
+      "loss": 0.9241,
+      "step": 22095
+    },
+    {
+      "epoch": 0.8035785722078772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9561876642619447e-06,
+      "loss": 0.8983,
+      "step": 22096
+    },
+    {
+      "epoch": 0.8036149398116158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9554879365605207e-06,
+      "loss": 0.9586,
+      "step": 22097
+    },
+    {
+      "epoch": 0.8036513074153544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9547883204652874e-06,
+      "loss": 0.9342,
+      "step": 22098
+    },
+    {
+      "epoch": 0.803687675019093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9540888159859483e-06,
+      "loss": 1.0783,
+      "step": 22099
+    },
+    {
+      "epoch": 0.8037240426228316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.953389423132217e-06,
+      "loss": 0.8887,
+      "step": 22100
+    },
+    {
+      "epoch": 0.8037604102265702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9526901419137843e-06,
+      "loss": 0.9576,
+      "step": 22101
+    },
+    {
+      "epoch": 0.8037967778303088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951990972340362e-06,
+      "loss": 0.9742,
+      "step": 22102
+    },
+    {
+      "epoch": 0.8038331454340474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951291914421647e-06,
+      "loss": 0.9305,
+      "step": 22103
+    },
+    {
+      "epoch": 0.803869513037786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.950592968167335e-06,
+      "loss": 0.9158,
+      "step": 22104
+    },
+    {
+      "epoch": 0.8039058806415246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9498941335871247e-06,
+      "loss": 0.8664,
+      "step": 22105
+    },
+    {
+      "epoch": 0.8039422482452632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.949195410690713e-06,
+      "loss": 0.8995,
+      "step": 22106
+    },
+    {
+      "epoch": 0.8039786158490017,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9484967994877923e-06,
+      "loss": 0.8542,
+      "step": 22107
+    },
+    {
+      "epoch": 0.8040149834527403,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947798299988055e-06,
+      "loss": 0.9789,
+      "step": 22108
+    },
+    {
+      "epoch": 0.8040513510564788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9470999122011924e-06,
+      "loss": 1.0012,
+      "step": 22109
+    },
+    {
+      "epoch": 0.8040877186602174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.946401636136892e-06,
+      "loss": 0.8874,
+      "step": 22110
+    },
+    {
+      "epoch": 0.804124086263956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.945703471804841e-06,
+      "loss": 0.9059,
+      "step": 22111
+    },
+    {
+      "epoch": 0.8041604538676946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9450054192147274e-06,
+      "loss": 0.9995,
+      "step": 22112
+    },
+    {
+      "epoch": 0.8041968214714332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9443074783762393e-06,
+      "loss": 0.7942,
+      "step": 22113
+    },
+    {
+      "epoch": 0.8042331890751718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9436096492990497e-06,
+      "loss": 0.9159,
+      "step": 22114
+    },
+    {
+      "epoch": 0.8042695566789104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942911931992848e-06,
+      "loss": 0.8744,
+      "step": 22115
+    },
+    {
+      "epoch": 0.804305924282649,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942214326467312e-06,
+      "loss": 0.9815,
+      "step": 22116
+    },
+    {
+      "epoch": 0.8043422918863876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.94151683273212e-06,
+      "loss": 0.9428,
+      "step": 22117
+    },
+    {
+      "epoch": 0.8043786594901262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.940819450796946e-06,
+      "loss": 0.9904,
+      "step": 22118
+    },
+    {
+      "epoch": 0.8044150270938648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9401221806714687e-06,
+      "loss": 0.8122,
+      "step": 22119
+    },
+    {
+      "epoch": 0.8044513946976034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.939425022365362e-06,
+      "loss": 1.0673,
+      "step": 22120
+    },
+    {
+      "epoch": 0.804487762301342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.938727975888297e-06,
+      "loss": 0.9651,
+      "step": 22121
+    },
+    {
+      "epoch": 0.8045241299050806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9380310412499435e-06,
+      "loss": 1.0231,
+      "step": 22122
+    },
+    {
+      "epoch": 0.8045604975088192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.937334218459971e-06,
+      "loss": 0.9416,
+      "step": 22123
+    },
+    {
+      "epoch": 0.8045968651125577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.936637507528043e-06,
+      "loss": 1.0132,
+      "step": 22124
+    },
+    {
+      "epoch": 0.8046332327162963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9359409084638324e-06,
+      "loss": 0.8939,
+      "step": 22125
+    },
+    {
+      "epoch": 0.8046696003200349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9352444212770016e-06,
+      "loss": 0.7944,
+      "step": 22126
+    },
+    {
+      "epoch": 0.8047059679237735,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9345480459772104e-06,
+      "loss": 0.9618,
+      "step": 22127
+    },
+    {
+      "epoch": 0.8047423355275121,
+      "grad_norm": 0.0,
+      "learning_rate": 1.93385178257412e-06,
+      "loss": 0.9212,
+      "step": 22128
+    },
+    {
+      "epoch": 0.8047787031312507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933155631077397e-06,
+      "loss": 0.8856,
+      "step": 22129
+    },
+    {
+      "epoch": 0.8048150707349893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932459591496688e-06,
+      "loss": 0.8454,
+      "step": 22130
+    },
+    {
+      "epoch": 0.8048514383387279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9317636638416592e-06,
+      "loss": 0.8179,
+      "step": 22131
+    },
+    {
+      "epoch": 0.8048878059424664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9310678481219646e-06,
+      "loss": 0.9119,
+      "step": 22132
+    },
+    {
+      "epoch": 0.804924173546205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9303721443472513e-06,
+      "loss": 0.9597,
+      "step": 22133
+    },
+    {
+      "epoch": 0.8049605411499436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9296765525271764e-06,
+      "loss": 0.8786,
+      "step": 22134
+    },
+    {
+      "epoch": 0.8049969087536822,
+      "grad_norm": 0.0,
+      "learning_rate": 1.928981072671389e-06,
+      "loss": 0.8598,
+      "step": 22135
+    },
+    {
+      "epoch": 0.8050332763574208,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9282857047895385e-06,
+      "loss": 0.9019,
+      "step": 22136
+    },
+    {
+      "epoch": 0.8050696439611594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.927590448891269e-06,
+      "loss": 0.9102,
+      "step": 22137
+    },
+    {
+      "epoch": 0.805106011564898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9268953049862306e-06,
+      "loss": 0.7827,
+      "step": 22138
+    },
+    {
+      "epoch": 0.8051423791686366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9262002730840667e-06,
+      "loss": 0.9942,
+      "step": 22139
+    },
+    {
+      "epoch": 0.8051787467723752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.925505353194418e-06,
+      "loss": 0.9593,
+      "step": 22140
+    },
+    {
+      "epoch": 0.8052151143761137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.924810545326924e-06,
+      "loss": 0.8234,
+      "step": 22141
+    },
+    {
+      "epoch": 0.8052514819798523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9241158494912314e-06,
+      "loss": 0.8654,
+      "step": 22142
+    },
+    {
+      "epoch": 0.8052878495835909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.923421265696969e-06,
+      "loss": 0.9197,
+      "step": 22143
+    },
+    {
+      "epoch": 0.8053242171873295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9227267939537785e-06,
+      "loss": 1.0064,
+      "step": 22144
+    },
+    {
+      "epoch": 0.8053605847910681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.922032434271295e-06,
+      "loss": 0.8891,
+      "step": 22145
+    },
+    {
+      "epoch": 0.8053969523948067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9213381866591496e-06,
+      "loss": 0.9455,
+      "step": 22146
+    },
+    {
+      "epoch": 0.8054333199985453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.920644051126972e-06,
+      "loss": 0.8861,
+      "step": 22147
+    },
+    {
+      "epoch": 0.8054696876022839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919950027684401e-06,
+      "loss": 1.0316,
+      "step": 22148
+    },
+    {
+      "epoch": 0.8055060552060225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9192561163410528e-06,
+      "loss": 0.9145,
+      "step": 22149
+    },
+    {
+      "epoch": 0.8055424228097611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.918562317106565e-06,
+      "loss": 0.8868,
+      "step": 22150
+    },
+    {
+      "epoch": 0.8055787904134997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917868629990558e-06,
+      "loss": 0.8326,
+      "step": 22151
+    },
+    {
+      "epoch": 0.8056151580172383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.917175055002657e-06,
+      "loss": 0.9545,
+      "step": 22152
+    },
+    {
+      "epoch": 0.8056515256209769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9164815921524836e-06,
+      "loss": 0.907,
+      "step": 22153
+    },
+    {
+      "epoch": 0.8056878932247155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9157882414496597e-06,
+      "loss": 0.9837,
+      "step": 22154
+    },
+    {
+      "epoch": 0.805724260828454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.915095002903804e-06,
+      "loss": 0.8862,
+      "step": 22155
+    },
+    {
+      "epoch": 0.8057606284321926,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9144018765245322e-06,
+      "loss": 0.9127,
+      "step": 22156
+    },
+    {
+      "epoch": 0.8057969960359312,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913708862321464e-06,
+      "loss": 1.0256,
+      "step": 22157
+    },
+    {
+      "epoch": 0.8058333636396697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.913015960304212e-06,
+      "loss": 0.906,
+      "step": 22158
+    },
+    {
+      "epoch": 0.8058697312434083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9123231704823906e-06,
+      "loss": 0.9388,
+      "step": 22159
+    },
+    {
+      "epoch": 0.8059060988471469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9116304928656072e-06,
+      "loss": 0.8976,
+      "step": 22160
+    },
+    {
+      "epoch": 0.8059424664508855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9109379274634797e-06,
+      "loss": 0.9486,
+      "step": 22161
+    },
+    {
+      "epoch": 0.8059788340546241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9102454742856067e-06,
+      "loss": 0.8536,
+      "step": 22162
+    },
+    {
+      "epoch": 0.8060152016583627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.909553133341603e-06,
+      "loss": 0.847,
+      "step": 22163
+    },
+    {
+      "epoch": 0.8060515692621013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.90886090464107e-06,
+      "loss": 0.9523,
+      "step": 22164
+    },
+    {
+      "epoch": 0.8060879368658399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9081687881936127e-06,
+      "loss": 1.028,
+      "step": 22165
+    },
+    {
+      "epoch": 0.8061243044695785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9074767840088295e-06,
+      "loss": 0.8091,
+      "step": 22166
+    },
+    {
+      "epoch": 0.8061606720733171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9067848920963294e-06,
+      "loss": 0.9596,
+      "step": 22167
+    },
+    {
+      "epoch": 0.8061970396770557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9060931124657024e-06,
+      "loss": 0.946,
+      "step": 22168
+    },
+    {
+      "epoch": 0.8062334072807943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9054014451265502e-06,
+      "loss": 0.8868,
+      "step": 22169
+    },
+    {
+      "epoch": 0.8062697748845329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9047098900884697e-06,
+      "loss": 0.8558,
+      "step": 22170
+    },
+    {
+      "epoch": 0.8063061424882715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9040184473610522e-06,
+      "loss": 0.8805,
+      "step": 22171
+    },
+    {
+      "epoch": 0.8063425100920101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9033271169538935e-06,
+      "loss": 0.9002,
+      "step": 22172
+    },
+    {
+      "epoch": 0.8063788776957487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9026358988765802e-06,
+      "loss": 0.9982,
+      "step": 22173
+    },
+    {
+      "epoch": 0.8064152452994873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9019447931387102e-06,
+      "loss": 0.8455,
+      "step": 22174
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9012537997498615e-06,
+      "loss": 0.9258,
+      "step": 22175
+    },
+    {
+      "epoch": 0.8064879805069644,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9005629187196284e-06,
+      "loss": 0.9518,
+      "step": 22176
+    },
+    {
+      "epoch": 0.806524348110703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8998721500575934e-06,
+      "loss": 0.8731,
+      "step": 22177
+    },
+    {
+      "epoch": 0.8065607157144415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8991814937733399e-06,
+      "loss": 0.8341,
+      "step": 22178
+    },
+    {
+      "epoch": 0.8065970833181801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898490949876447e-06,
+      "loss": 0.8549,
+      "step": 22179
+    },
+    {
+      "epoch": 0.8066334509219187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8978005183765024e-06,
+      "loss": 0.9865,
+      "step": 22180
+    },
+    {
+      "epoch": 0.8066698185256573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8971101992830742e-06,
+      "loss": 0.944,
+      "step": 22181
+    },
+    {
+      "epoch": 0.8067061861293959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8964199926057491e-06,
+      "loss": 0.9229,
+      "step": 22182
+    },
+    {
+      "epoch": 0.8067425537331345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8957298983540995e-06,
+      "loss": 0.785,
+      "step": 22183
+    },
+    {
+      "epoch": 0.8067789213368731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8950399165376976e-06,
+      "loss": 0.963,
+      "step": 22184
+    },
+    {
+      "epoch": 0.8068152889406117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8943500471661158e-06,
+      "loss": 0.9265,
+      "step": 22185
+    },
+    {
+      "epoch": 0.8068516565443503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.893660290248932e-06,
+      "loss": 0.9443,
+      "step": 22186
+    },
+    {
+      "epoch": 0.8068880241480889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8929706457957053e-06,
+      "loss": 1.0178,
+      "step": 22187
+    },
+    {
+      "epoch": 0.8069243917518275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8922811138160102e-06,
+      "loss": 0.9008,
+      "step": 22188
+    },
+    {
+      "epoch": 0.8069607593555661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8915916943194123e-06,
+      "loss": 0.8394,
+      "step": 22189
+    },
+    {
+      "epoch": 0.8069971269593047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.890902387315474e-06,
+      "loss": 1.1004,
+      "step": 22190
+    },
+    {
+      "epoch": 0.8070334945630433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8902131928137612e-06,
+      "loss": 0.8911,
+      "step": 22191
+    },
+    {
+      "epoch": 0.8070698621667818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8895241108238305e-06,
+      "loss": 0.8896,
+      "step": 22192
+    },
+    {
+      "epoch": 0.8071062297705204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8888351413552519e-06,
+      "loss": 0.9903,
+      "step": 22193
+    },
+    {
+      "epoch": 0.807142597374259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8881462844175713e-06,
+      "loss": 0.9298,
+      "step": 22194
+    },
+    {
+      "epoch": 0.8071789649779976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8874575400203543e-06,
+      "loss": 0.8815,
+      "step": 22195
+    },
+    {
+      "epoch": 0.8072153325817362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8867689081731545e-06,
+      "loss": 1.0477,
+      "step": 22196
+    },
+    {
+      "epoch": 0.8072517001854748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8860803888855239e-06,
+      "loss": 0.8582,
+      "step": 22197
+    },
+    {
+      "epoch": 0.8072880677892134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.885391982167013e-06,
+      "loss": 0.8888,
+      "step": 22198
+    },
+    {
+      "epoch": 0.807324435392952,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8847036880271819e-06,
+      "loss": 1.014,
+      "step": 22199
+    },
+    {
+      "epoch": 0.8073608029966906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8840155064755673e-06,
+      "loss": 0.982,
+      "step": 22200
+    },
+    {
+      "epoch": 0.8073971706004291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8833274375217248e-06,
+      "loss": 1.0603,
+      "step": 22201
+    },
+    {
+      "epoch": 0.8074335382041677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8826394811751969e-06,
+      "loss": 0.8785,
+      "step": 22202
+    },
+    {
+      "epoch": 0.8074699058079063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8819516374455304e-06,
+      "loss": 0.8031,
+      "step": 22203
+    },
+    {
+      "epoch": 0.8075062734116449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8812639063422644e-06,
+      "loss": 0.9502,
+      "step": 22204
+    },
+    {
+      "epoch": 0.8075426410153835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8805762878749455e-06,
+      "loss": 0.9835,
+      "step": 22205
+    },
+    {
+      "epoch": 0.8075790086191221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8798887820531098e-06,
+      "loss": 0.9156,
+      "step": 22206
+    },
+    {
+      "epoch": 0.8076153762228607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.879201388886297e-06,
+      "loss": 1.0156,
+      "step": 22207
+    },
+    {
+      "epoch": 0.8076517438265992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.878514108384043e-06,
+      "loss": 0.8436,
+      "step": 22208
+    },
+    {
+      "epoch": 0.8076881114303378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8778269405558825e-06,
+      "loss": 0.9361,
+      "step": 22209
+    },
+    {
+      "epoch": 0.8077244790340764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8771398854113498e-06,
+      "loss": 0.8779,
+      "step": 22210
+    },
+    {
+      "epoch": 0.807760846637815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.876452942959973e-06,
+      "loss": 0.8195,
+      "step": 22211
+    },
+    {
+      "epoch": 0.8077972142415536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8757661132112925e-06,
+      "loss": 0.9333,
+      "step": 22212
+    },
+    {
+      "epoch": 0.8078335818452922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8750793961748236e-06,
+      "loss": 0.9361,
+      "step": 22213
+    },
+    {
+      "epoch": 0.8078699494490308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8743927918601035e-06,
+      "loss": 0.9186,
+      "step": 22214
+    },
+    {
+      "epoch": 0.8079063170527694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8737063002766553e-06,
+      "loss": 0.9589,
+      "step": 22215
+    },
+    {
+      "epoch": 0.807942684656508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8730199214340017e-06,
+      "loss": 0.8719,
+      "step": 22216
+    },
+    {
+      "epoch": 0.8079790522602466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8723336553416637e-06,
+      "loss": 0.8867,
+      "step": 22217
+    },
+    {
+      "epoch": 0.8080154198639852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8716475020091674e-06,
+      "loss": 0.8761,
+      "step": 22218
+    },
+    {
+      "epoch": 0.8080517874677238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870961461446028e-06,
+      "loss": 0.9332,
+      "step": 22219
+    },
+    {
+      "epoch": 0.8080881550714624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.870275533661766e-06,
+      "loss": 0.8031,
+      "step": 22220
+    },
+    {
+      "epoch": 0.808124522675201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8695897186658963e-06,
+      "loss": 0.8856,
+      "step": 22221
+    },
+    {
+      "epoch": 0.8081608902789396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868904016467934e-06,
+      "loss": 0.8925,
+      "step": 22222
+    },
+    {
+      "epoch": 0.8081972578826782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8682184270773885e-06,
+      "loss": 0.8489,
+      "step": 22223
+    },
+    {
+      "epoch": 0.8082336254864168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8675329505037788e-06,
+      "loss": 0.9685,
+      "step": 22224
+    },
+    {
+      "epoch": 0.8082699930901552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8668475867566093e-06,
+      "loss": 0.9181,
+      "step": 22225
+    },
+    {
+      "epoch": 0.8083063606938938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8661623358453906e-06,
+      "loss": 0.9319,
+      "step": 22226
+    },
+    {
+      "epoch": 0.8083427282976324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.865477197779627e-06,
+      "loss": 0.8898,
+      "step": 22227
+    },
+    {
+      "epoch": 0.808379095901371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8647921725688312e-06,
+      "loss": 0.8721,
+      "step": 22228
+    },
+    {
+      "epoch": 0.8084154635051096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8641072602224986e-06,
+      "loss": 0.9344,
+      "step": 22229
+    },
+    {
+      "epoch": 0.8084518311088482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8634224607501317e-06,
+      "loss": 0.9439,
+      "step": 22230
+    },
+    {
+      "epoch": 0.8084881987125868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8627377741612395e-06,
+      "loss": 0.8603,
+      "step": 22231
+    },
+    {
+      "epoch": 0.8085245663163254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8620532004653103e-06,
+      "loss": 0.9153,
+      "step": 22232
+    },
+    {
+      "epoch": 0.808560933920064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8613687396718482e-06,
+      "loss": 0.9547,
+      "step": 22233
+    },
+    {
+      "epoch": 0.8085973015238026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8606843917903493e-06,
+      "loss": 0.9375,
+      "step": 22234
+    },
+    {
+      "epoch": 0.8086336691275412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860000156830305e-06,
+      "loss": 0.9549,
+      "step": 22235
+    },
+    {
+      "epoch": 0.8086700367312798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8593160348012075e-06,
+      "loss": 0.9662,
+      "step": 22236
+    },
+    {
+      "epoch": 0.8087064043350184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8586320257125524e-06,
+      "loss": 0.8745,
+      "step": 22237
+    },
+    {
+      "epoch": 0.808742771938757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8579481295738267e-06,
+      "loss": 1.0101,
+      "step": 22238
+    },
+    {
+      "epoch": 0.8087791395424956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8572643463945183e-06,
+      "loss": 0.9153,
+      "step": 22239
+    },
+    {
+      "epoch": 0.8088155071462342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.856580676184112e-06,
+      "loss": 0.9693,
+      "step": 22240
+    },
+    {
+      "epoch": 0.8088518747499728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8558971189520991e-06,
+      "loss": 0.919,
+      "step": 22241
+    },
+    {
+      "epoch": 0.8088882423537113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.855213674707954e-06,
+      "loss": 0.9922,
+      "step": 22242
+    },
+    {
+      "epoch": 0.8089246099574499,
+      "grad_norm": 0.0,
+      "learning_rate": 1.854530343461166e-06,
+      "loss": 0.9007,
+      "step": 22243
+    },
+    {
+      "epoch": 0.8089609775611885,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8538471252212131e-06,
+      "loss": 0.6381,
+      "step": 22244
+    },
+    {
+      "epoch": 0.808997345164927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.853164019997573e-06,
+      "loss": 0.9329,
+      "step": 22245
+    },
+    {
+      "epoch": 0.8090337127686656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8524810277997208e-06,
+      "loss": 0.9241,
+      "step": 22246
+    },
+    {
+      "epoch": 0.8090700803724042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8517981486371395e-06,
+      "loss": 0.8926,
+      "step": 22247
+    },
+    {
+      "epoch": 0.8091064479761428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8511153825192952e-06,
+      "loss": 0.9337,
+      "step": 22248
+    },
+    {
+      "epoch": 0.8091428155798814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8504327294556613e-06,
+      "loss": 0.9039,
+      "step": 22249
+    },
+    {
+      "epoch": 0.80917918318362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8497501894557123e-06,
+      "loss": 0.9218,
+      "step": 22250
+    },
+    {
+      "epoch": 0.8092155507873586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8490677625289165e-06,
+      "loss": 0.9529,
+      "step": 22251
+    },
+    {
+      "epoch": 0.8092519183910972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8483854486847407e-06,
+      "loss": 0.8794,
+      "step": 22252
+    },
+    {
+      "epoch": 0.8092882859948358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8477032479326508e-06,
+      "loss": 1.0306,
+      "step": 22253
+    },
+    {
+      "epoch": 0.8093246535985744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.847021160282112e-06,
+      "loss": 0.9818,
+      "step": 22254
+    },
+    {
+      "epoch": 0.809361021202313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.846339185742585e-06,
+      "loss": 0.9476,
+      "step": 22255
+    },
+    {
+      "epoch": 0.8093973888060516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8456573243235354e-06,
+      "loss": 0.9295,
+      "step": 22256
+    },
+    {
+      "epoch": 0.8094337564097902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.84497557603442e-06,
+      "loss": 0.9013,
+      "step": 22257
+    },
+    {
+      "epoch": 0.8094701240135288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8442939408846993e-06,
+      "loss": 0.9412,
+      "step": 22258
+    },
+    {
+      "epoch": 0.8095064916172673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8436124188838257e-06,
+      "loss": 0.868,
+      "step": 22259
+    },
+    {
+      "epoch": 0.8095428592210059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8429310100412624e-06,
+      "loss": 1.0206,
+      "step": 22260
+    },
+    {
+      "epoch": 0.8095792268247445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.842249714366453e-06,
+      "loss": 0.8976,
+      "step": 22261
+    },
+    {
+      "epoch": 0.8096155944284831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841568531868857e-06,
+      "loss": 0.833,
+      "step": 22262
+    },
+    {
+      "epoch": 0.8096519620322217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.840887462557922e-06,
+      "loss": 0.8213,
+      "step": 22263
+    },
+    {
+      "epoch": 0.8096883296359603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8402065064430974e-06,
+      "loss": 0.937,
+      "step": 22264
+    },
+    {
+      "epoch": 0.8097246972396989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8395256635338287e-06,
+      "loss": 0.9004,
+      "step": 22265
+    },
+    {
+      "epoch": 0.8097610648434375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8388449338395675e-06,
+      "loss": 0.8931,
+      "step": 22266
+    },
+    {
+      "epoch": 0.8097974324471761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8381643173697505e-06,
+      "loss": 1.0997,
+      "step": 22267
+    },
+    {
+      "epoch": 0.8098338000509147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8374838141338214e-06,
+      "loss": 0.8648,
+      "step": 22268
+    },
+    {
+      "epoch": 0.8098701676546533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8368034241412259e-06,
+      "loss": 0.9296,
+      "step": 22269
+    },
+    {
+      "epoch": 0.8099065352583918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8361231474014007e-06,
+      "loss": 1.0317,
+      "step": 22270
+    },
+    {
+      "epoch": 0.8099429028621304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.835442983923784e-06,
+      "loss": 0.9825,
+      "step": 22271
+    },
+    {
+      "epoch": 0.809979270465869,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8347629337178097e-06,
+      "loss": 0.8875,
+      "step": 22272
+    },
+    {
+      "epoch": 0.8100156380696076,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8340829967929197e-06,
+      "loss": 0.9258,
+      "step": 22273
+    },
+    {
+      "epoch": 0.8100520056733462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8334031731585367e-06,
+      "loss": 0.9945,
+      "step": 22274
+    },
+    {
+      "epoch": 0.8100883732770848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8327234628241008e-06,
+      "loss": 0.7649,
+      "step": 22275
+    },
+    {
+      "epoch": 0.8101247408808233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8320438657990381e-06,
+      "loss": 0.8886,
+      "step": 22276
+    },
+    {
+      "epoch": 0.8101611084845619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8313643820927795e-06,
+      "loss": 0.9017,
+      "step": 22277
+    },
+    {
+      "epoch": 0.8101974760883005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8306850117147468e-06,
+      "loss": 0.9055,
+      "step": 22278
+    },
+    {
+      "epoch": 0.8102338436920391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8300057546743733e-06,
+      "loss": 0.9054,
+      "step": 22279
+    },
+    {
+      "epoch": 0.8102702112957777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8293266109810738e-06,
+      "loss": 0.8686,
+      "step": 22280
+    },
+    {
+      "epoch": 0.8103065788995163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8286475806442772e-06,
+      "loss": 0.8873,
+      "step": 22281
+    },
+    {
+      "epoch": 0.8103429465032549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8279686636734029e-06,
+      "loss": 0.8378,
+      "step": 22282
+    },
+    {
+      "epoch": 0.8103793141069935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8272898600778677e-06,
+      "loss": 0.9484,
+      "step": 22283
+    },
+    {
+      "epoch": 0.8104156817107321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8266111698670875e-06,
+      "loss": 1.0006,
+      "step": 22284
+    },
+    {
+      "epoch": 0.8104520493144707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8259325930504835e-06,
+      "loss": 0.9627,
+      "step": 22285
+    },
+    {
+      "epoch": 0.8104884169182093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8252541296374705e-06,
+      "loss": 0.9658,
+      "step": 22286
+    },
+    {
+      "epoch": 0.8105247845219479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.824575779637452e-06,
+      "loss": 0.7368,
+      "step": 22287
+    },
+    {
+      "epoch": 0.8105611521256865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8238975430598483e-06,
+      "loss": 1.0466,
+      "step": 22288
+    },
+    {
+      "epoch": 0.8105975197294251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8232194199140652e-06,
+      "loss": 0.7593,
+      "step": 22289
+    },
+    {
+      "epoch": 0.8106338873331637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.822541410209512e-06,
+      "loss": 0.8836,
+      "step": 22290
+    },
+    {
+      "epoch": 0.8106702549369023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8218635139555918e-06,
+      "loss": 0.7592,
+      "step": 22291
+    },
+    {
+      "epoch": 0.8107066225406407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8211857311617164e-06,
+      "loss": 0.8939,
+      "step": 22292
+    },
+    {
+      "epoch": 0.8107429901443793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8205080618372805e-06,
+      "loss": 0.9341,
+      "step": 22293
+    },
+    {
+      "epoch": 0.8107793577481179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8198305059916932e-06,
+      "loss": 0.9329,
+      "step": 22294
+    },
+    {
+      "epoch": 0.8108157253518565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8191530636343501e-06,
+      "loss": 0.948,
+      "step": 22295
+    },
+    {
+      "epoch": 0.8108520929555951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8184757347746518e-06,
+      "loss": 0.9129,
+      "step": 22296
+    },
+    {
+      "epoch": 0.8108884605593337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8177985194219916e-06,
+      "loss": 0.9307,
+      "step": 22297
+    },
+    {
+      "epoch": 0.8109248281630723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8171214175857743e-06,
+      "loss": 0.882,
+      "step": 22298
+    },
+    {
+      "epoch": 0.8109611957668109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8164444292753814e-06,
+      "loss": 0.8653,
+      "step": 22299
+    },
+    {
+      "epoch": 0.8109975633705495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8157675545002129e-06,
+      "loss": 0.9142,
+      "step": 22300
+    },
+    {
+      "epoch": 0.8110339309742881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8150907932696582e-06,
+      "loss": 0.9556,
+      "step": 22301
+    },
+    {
+      "epoch": 0.8110702985780267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8144141455931063e-06,
+      "loss": 0.8676,
+      "step": 22302
+    },
+    {
+      "epoch": 0.8111066661817653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.813737611479941e-06,
+      "loss": 0.8437,
+      "step": 22303
+    },
+    {
+      "epoch": 0.8111430337855039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8130611909395547e-06,
+      "loss": 0.8683,
+      "step": 22304
+    },
+    {
+      "epoch": 0.8111794013892425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8123848839813318e-06,
+      "loss": 0.8347,
+      "step": 22305
+    },
+    {
+      "epoch": 0.8112157689929811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8117086906146454e-06,
+      "loss": 0.7973,
+      "step": 22306
+    },
+    {
+      "epoch": 0.8112521365967197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8110326108488863e-06,
+      "loss": 0.8435,
+      "step": 22307
+    },
+    {
+      "epoch": 0.8112885042004583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8103566446934317e-06,
+      "loss": 0.864,
+      "step": 22308
+    },
+    {
+      "epoch": 0.8113248718041968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8096807921576587e-06,
+      "loss": 0.9197,
+      "step": 22309
+    },
+    {
+      "epoch": 0.8113612394079354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8090050532509417e-06,
+      "loss": 0.9023,
+      "step": 22310
+    },
+    {
+      "epoch": 0.811397607011674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8083294279826635e-06,
+      "loss": 1.0049,
+      "step": 22311
+    },
+    {
+      "epoch": 0.8114339746154126,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8076539163621864e-06,
+      "loss": 0.8767,
+      "step": 22312
+    },
+    {
+      "epoch": 0.8114703422191512,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8069785183988896e-06,
+      "loss": 0.954,
+      "step": 22313
+    },
+    {
+      "epoch": 0.8115067098228897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8063032341021425e-06,
+      "loss": 0.9387,
+      "step": 22314
+    },
+    {
+      "epoch": 0.8115430774266283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8056280634813129e-06,
+      "loss": 1.0141,
+      "step": 22315
+    },
+    {
+      "epoch": 0.8115794450303669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8049530065457644e-06,
+      "loss": 0.8915,
+      "step": 22316
+    },
+    {
+      "epoch": 0.8116158126341055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8042780633048685e-06,
+      "loss": 0.808,
+      "step": 22317
+    },
+    {
+      "epoch": 0.8116521802378441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8036032337679865e-06,
+      "loss": 0.9064,
+      "step": 22318
+    },
+    {
+      "epoch": 0.8116885478415827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.80292851794448e-06,
+      "loss": 0.8891,
+      "step": 22319
+    },
+    {
+      "epoch": 0.8117249154453213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8022539158437101e-06,
+      "loss": 0.8416,
+      "step": 22320
+    },
+    {
+      "epoch": 0.8117612830490599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8015794274750365e-06,
+      "loss": 0.9273,
+      "step": 22321
+    },
+    {
+      "epoch": 0.8117976506527985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8009050528478144e-06,
+      "loss": 0.9255,
+      "step": 22322
+    },
+    {
+      "epoch": 0.8118340182565371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8002307919714036e-06,
+      "loss": 0.8771,
+      "step": 22323
+    },
+    {
+      "epoch": 0.8118703858602757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7995566448551583e-06,
+      "loss": 0.7923,
+      "step": 22324
+    },
+    {
+      "epoch": 0.8119067534640143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7988826115084258e-06,
+      "loss": 0.9933,
+      "step": 22325
+    },
+    {
+      "epoch": 0.8119431210677528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7982086919405628e-06,
+      "loss": 0.8092,
+      "step": 22326
+    },
+    {
+      "epoch": 0.8119794886714914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7975348861609177e-06,
+      "loss": 0.9133,
+      "step": 22327
+    },
+    {
+      "epoch": 0.81201585627523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7968611941788382e-06,
+      "loss": 0.9361,
+      "step": 22328
+    },
+    {
+      "epoch": 0.8120522238789686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7961876160036685e-06,
+      "loss": 0.9387,
+      "step": 22329
+    },
+    {
+      "epoch": 0.8120885914827072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7955141516447593e-06,
+      "loss": 0.9546,
+      "step": 22330
+    },
+    {
+      "epoch": 0.8121249590864458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7948408011114493e-06,
+      "loss": 1.0421,
+      "step": 22331
+    },
+    {
+      "epoch": 0.8121613266901844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7941675644130819e-06,
+      "loss": 0.9211,
+      "step": 22332
+    },
+    {
+      "epoch": 0.812197694293923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7934944415589972e-06,
+      "loss": 0.9279,
+      "step": 22333
+    },
+    {
+      "epoch": 0.8122340618976616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7928214325585336e-06,
+      "loss": 0.8595,
+      "step": 22334
+    },
+    {
+      "epoch": 0.8122704295014002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7921485374210257e-06,
+      "loss": 0.8639,
+      "step": 22335
+    },
+    {
+      "epoch": 0.8123067971051388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7914757561558139e-06,
+      "loss": 0.9644,
+      "step": 22336
+    },
+    {
+      "epoch": 0.8123431647088774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7908030887722305e-06,
+      "loss": 0.8532,
+      "step": 22337
+    },
+    {
+      "epoch": 0.812379532312616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.790130535279605e-06,
+      "loss": 0.9075,
+      "step": 22338
+    },
+    {
+      "epoch": 0.8124158999163545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7894580956872687e-06,
+      "loss": 1.0184,
+      "step": 22339
+    },
+    {
+      "epoch": 0.8124522675200931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7887857700045574e-06,
+      "loss": 0.8733,
+      "step": 22340
+    },
+    {
+      "epoch": 0.8124886351238317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7881135582407883e-06,
+      "loss": 0.9463,
+      "step": 22341
+    },
+    {
+      "epoch": 0.8125250027275703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7874414604052937e-06,
+      "loss": 0.9259,
+      "step": 22342
+    },
+    {
+      "epoch": 0.8125613703313088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7867694765074007e-06,
+      "loss": 1.0082,
+      "step": 22343
+    },
+    {
+      "epoch": 0.8125977379350474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7860976065564229e-06,
+      "loss": 0.9811,
+      "step": 22344
+    },
+    {
+      "epoch": 0.812634105538786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7854258505616884e-06,
+      "loss": 0.9164,
+      "step": 22345
+    },
+    {
+      "epoch": 0.8126704731425246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7847542085325154e-06,
+      "loss": 0.8821,
+      "step": 22346
+    },
+    {
+      "epoch": 0.8127068407462632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.784082680478223e-06,
+      "loss": 0.8571,
+      "step": 22347
+    },
+    {
+      "epoch": 0.8127432083500018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7834112664081226e-06,
+      "loss": 0.956,
+      "step": 22348
+    },
+    {
+      "epoch": 0.8127795759537404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.782739966331537e-06,
+      "loss": 0.8379,
+      "step": 22349
+    },
+    {
+      "epoch": 0.812815943557479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.782068780257774e-06,
+      "loss": 0.9847,
+      "step": 22350
+    },
+    {
+      "epoch": 0.8128523111612176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7813977081961486e-06,
+      "loss": 0.8945,
+      "step": 22351
+    },
+    {
+      "epoch": 0.8128886787649562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7807267501559688e-06,
+      "loss": 0.925,
+      "step": 22352
+    },
+    {
+      "epoch": 0.8129250463686948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7800559061465427e-06,
+      "loss": 0.8504,
+      "step": 22353
+    },
+    {
+      "epoch": 0.8129614139724334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7793851761771773e-06,
+      "loss": 0.8423,
+      "step": 22354
+    },
+    {
+      "epoch": 0.812997781576172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.778714560257181e-06,
+      "loss": 0.8913,
+      "step": 22355
+    },
+    {
+      "epoch": 0.8130341491799106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7780440583958546e-06,
+      "loss": 0.8529,
+      "step": 22356
+    },
+    {
+      "epoch": 0.8130705167836492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7773736706025025e-06,
+      "loss": 0.9061,
+      "step": 22357
+    },
+    {
+      "epoch": 0.8131068843873878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7767033968864222e-06,
+      "loss": 0.7714,
+      "step": 22358
+    },
+    {
+      "epoch": 0.8131432519911264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7760332372569188e-06,
+      "loss": 0.9048,
+      "step": 22359
+    },
+    {
+      "epoch": 0.8131796195948648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7753631917232827e-06,
+      "loss": 0.8364,
+      "step": 22360
+    },
+    {
+      "epoch": 0.8132159871986034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774693260294814e-06,
+      "loss": 0.9847,
+      "step": 22361
+    },
+    {
+      "epoch": 0.813252354802342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7740234429808068e-06,
+      "loss": 0.9562,
+      "step": 22362
+    },
+    {
+      "epoch": 0.8132887224060806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.773353739790553e-06,
+      "loss": 0.971,
+      "step": 22363
+    },
+    {
+      "epoch": 0.8133250900098192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7726841507333436e-06,
+      "loss": 0.941,
+      "step": 22364
+    },
+    {
+      "epoch": 0.8133614576135578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7720146758184698e-06,
+      "loss": 0.9382,
+      "step": 22365
+    },
+    {
+      "epoch": 0.8133978252172964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7713453150552173e-06,
+      "loss": 0.9086,
+      "step": 22366
+    },
+    {
+      "epoch": 0.813434192821035,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7706760684528713e-06,
+      "loss": 0.8949,
+      "step": 22367
+    },
+    {
+      "epoch": 0.8134705604247736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7700069360207218e-06,
+      "loss": 1.0185,
+      "step": 22368
+    },
+    {
+      "epoch": 0.8135069280285122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7693379177680492e-06,
+      "loss": 0.9434,
+      "step": 22369
+    },
+    {
+      "epoch": 0.8135432956322508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.768669013704134e-06,
+      "loss": 0.816,
+      "step": 22370
+    },
+    {
+      "epoch": 0.8135796632359894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7680002238382565e-06,
+      "loss": 0.9627,
+      "step": 22371
+    },
+    {
+      "epoch": 0.813616030839728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7673315481797005e-06,
+      "loss": 0.8995,
+      "step": 22372
+    },
+    {
+      "epoch": 0.8136523984434666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7666629867377339e-06,
+      "loss": 0.8986,
+      "step": 22373
+    },
+    {
+      "epoch": 0.8136887660472052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7659945395216394e-06,
+      "loss": 0.8255,
+      "step": 22374
+    },
+    {
+      "epoch": 0.8137251336509438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7653262065406885e-06,
+      "loss": 0.8928,
+      "step": 22375
+    },
+    {
+      "epoch": 0.8137615012546823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7646579878041525e-06,
+      "loss": 0.8208,
+      "step": 22376
+    },
+    {
+      "epoch": 0.8137978688584209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7639898833213e-06,
+      "loss": 0.9195,
+      "step": 22377
+    },
+    {
+      "epoch": 0.8138342364621595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7633218931014072e-06,
+      "loss": 0.8802,
+      "step": 22378
+    },
+    {
+      "epoch": 0.8138706040658981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.762654017153732e-06,
+      "loss": 0.9759,
+      "step": 22379
+    },
+    {
+      "epoch": 0.8139069716696367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7619862554875478e-06,
+      "loss": 0.9332,
+      "step": 22380
+    },
+    {
+      "epoch": 0.8139433392733753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.761318608112116e-06,
+      "loss": 0.8953,
+      "step": 22381
+    },
+    {
+      "epoch": 0.8139797068771139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.760651075036699e-06,
+      "loss": 0.9735,
+      "step": 22382
+    },
+    {
+      "epoch": 0.8140160744808524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7599836562705585e-06,
+      "loss": 0.8899,
+      "step": 22383
+    },
+    {
+      "epoch": 0.814052442084591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7593163518229518e-06,
+      "loss": 0.8368,
+      "step": 22384
+    },
+    {
+      "epoch": 0.8140888096883296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7586491617031431e-06,
+      "loss": 0.9087,
+      "step": 22385
+    },
+    {
+      "epoch": 0.8141251772920682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7579820859203789e-06,
+      "loss": 1.0082,
+      "step": 22386
+    },
+    {
+      "epoch": 0.8141615448958068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7573151244839225e-06,
+      "loss": 0.9135,
+      "step": 22387
+    },
+    {
+      "epoch": 0.8141979124995454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7566482774030235e-06,
+      "loss": 1.0375,
+      "step": 22388
+    },
+    {
+      "epoch": 0.814234280103284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7559815446869344e-06,
+      "loss": 0.9592,
+      "step": 22389
+    },
+    {
+      "epoch": 0.8142706477070226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7553149263449022e-06,
+      "loss": 0.9256,
+      "step": 22390
+    },
+    {
+      "epoch": 0.8143070153107612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7546484223861826e-06,
+      "loss": 0.9467,
+      "step": 22391
+    },
+    {
+      "epoch": 0.8143433829144998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7539820328200119e-06,
+      "loss": 0.8842,
+      "step": 22392
+    },
+    {
+      "epoch": 0.8143797505182383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7533157576556436e-06,
+      "loss": 0.96,
+      "step": 22393
+    },
+    {
+      "epoch": 0.8144161181219769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7526495969023193e-06,
+      "loss": 0.9648,
+      "step": 22394
+    },
+    {
+      "epoch": 0.8144524857257155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7519835505692796e-06,
+      "loss": 0.9095,
+      "step": 22395
+    },
+    {
+      "epoch": 0.8144888533294541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7513176186657633e-06,
+      "loss": 0.9157,
+      "step": 22396
+    },
+    {
+      "epoch": 0.8145252209331927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7506518012010176e-06,
+      "loss": 0.8704,
+      "step": 22397
+    },
+    {
+      "epoch": 0.8145615885369313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7499860981842675e-06,
+      "loss": 0.8945,
+      "step": 22398
+    },
+    {
+      "epoch": 0.8145979561406699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7493205096247589e-06,
+      "loss": 0.8833,
+      "step": 22399
+    },
+    {
+      "epoch": 0.8146343237444085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.748655035531721e-06,
+      "loss": 0.9447,
+      "step": 22400
+    },
+    {
+      "epoch": 0.8146706913481471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7479896759143877e-06,
+      "loss": 0.8339,
+      "step": 22401
+    },
+    {
+      "epoch": 0.8147070589518857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7473244307819902e-06,
+      "loss": 0.9739,
+      "step": 22402
+    },
+    {
+      "epoch": 0.8147434265556243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7466593001437537e-06,
+      "loss": 0.9158,
+      "step": 22403
+    },
+    {
+      "epoch": 0.8147797941593629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.745994284008915e-06,
+      "loss": 0.896,
+      "step": 22404
+    },
+    {
+      "epoch": 0.8148161617631015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7453293823866901e-06,
+      "loss": 0.9065,
+      "step": 22405
+    },
+    {
+      "epoch": 0.81485252936684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7446645952863094e-06,
+      "loss": 0.8766,
+      "step": 22406
+    },
+    {
+      "epoch": 0.8148888969705786,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7439999227169957e-06,
+      "loss": 0.9672,
+      "step": 22407
+    },
+    {
+      "epoch": 0.8149252645743172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7433353646879691e-06,
+      "loss": 0.9918,
+      "step": 22408
+    },
+    {
+      "epoch": 0.8149616321780558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.742670921208447e-06,
+      "loss": 0.8561,
+      "step": 22409
+    },
+    {
+      "epoch": 0.8149979997817943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7420065922876562e-06,
+      "loss": 0.9146,
+      "step": 22410
+    },
+    {
+      "epoch": 0.8150343673855329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7413423779348015e-06,
+      "loss": 0.9313,
+      "step": 22411
+    },
+    {
+      "epoch": 0.8150707349892715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.740678278159106e-06,
+      "loss": 0.9891,
+      "step": 22412
+    },
+    {
+      "epoch": 0.8151071025930101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7400142929697816e-06,
+      "loss": 0.9814,
+      "step": 22413
+    },
+    {
+      "epoch": 0.8151434701967487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7393504223760394e-06,
+      "loss": 0.9201,
+      "step": 22414
+    },
+    {
+      "epoch": 0.8151798378004873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7386866663870872e-06,
+      "loss": 0.9159,
+      "step": 22415
+    },
+    {
+      "epoch": 0.8152162054042259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7380230250121377e-06,
+      "loss": 1.0245,
+      "step": 22416
+    },
+    {
+      "epoch": 0.8152525730079645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7373594982603982e-06,
+      "loss": 0.8786,
+      "step": 22417
+    },
+    {
+      "epoch": 0.8152889406117031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7366960861410708e-06,
+      "loss": 0.9072,
+      "step": 22418
+    },
+    {
+      "epoch": 0.8153253082154417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.736032788663362e-06,
+      "loss": 0.8966,
+      "step": 22419
+    },
+    {
+      "epoch": 0.8153616758191803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7353696058364733e-06,
+      "loss": 1.0028,
+      "step": 22420
+    },
+    {
+      "epoch": 0.8153980434229189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734706537669606e-06,
+      "loss": 0.8911,
+      "step": 22421
+    },
+    {
+      "epoch": 0.8154344110266575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7340435841719548e-06,
+      "loss": 1.0291,
+      "step": 22422
+    },
+    {
+      "epoch": 0.8154707786303961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7333807453527273e-06,
+      "loss": 0.8813,
+      "step": 22423
+    },
+    {
+      "epoch": 0.8155071462341347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.732718021221107e-06,
+      "loss": 0.8795,
+      "step": 22424
+    },
+    {
+      "epoch": 0.8155435138378733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7320554117862976e-06,
+      "loss": 0.8863,
+      "step": 22425
+    },
+    {
+      "epoch": 0.8155798814416119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7313929170574884e-06,
+      "loss": 0.8252,
+      "step": 22426
+    },
+    {
+      "epoch": 0.8156162490453503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.730730537043871e-06,
+      "loss": 0.7451,
+      "step": 22427
+    },
+    {
+      "epoch": 0.815652616649089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7300682717546324e-06,
+      "loss": 0.934,
+      "step": 22428
+    },
+    {
+      "epoch": 0.8156889842528275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7294061211989655e-06,
+      "loss": 0.8769,
+      "step": 22429
+    },
+    {
+      "epoch": 0.8157253518565661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7287440853860537e-06,
+      "loss": 0.9041,
+      "step": 22430
+    },
+    {
+      "epoch": 0.8157617194603047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7280821643250823e-06,
+      "loss": 0.9731,
+      "step": 22431
+    },
+    {
+      "epoch": 0.8157980870640433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7274203580252358e-06,
+      "loss": 0.8871,
+      "step": 22432
+    },
+    {
+      "epoch": 0.8158344546677819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7267586664956938e-06,
+      "loss": 0.8569,
+      "step": 22433
+    },
+    {
+      "epoch": 0.8158708222715205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7260970897456342e-06,
+      "loss": 0.8958,
+      "step": 22434
+    },
+    {
+      "epoch": 0.8159071898752591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.725435627784241e-06,
+      "loss": 0.9323,
+      "step": 22435
+    },
+    {
+      "epoch": 0.8159435574789977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.724774280620689e-06,
+      "loss": 0.9057,
+      "step": 22436
+    },
+    {
+      "epoch": 0.8159799250827363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7241130482641521e-06,
+      "loss": 0.8958,
+      "step": 22437
+    },
+    {
+      "epoch": 0.8160162926864749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.723451930723803e-06,
+      "loss": 1.0033,
+      "step": 22438
+    },
+    {
+      "epoch": 0.8160526602902135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7227909280088195e-06,
+      "loss": 0.9224,
+      "step": 22439
+    },
+    {
+      "epoch": 0.8160890278939521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7221300401283658e-06,
+      "loss": 0.8325,
+      "step": 22440
+    },
+    {
+      "epoch": 0.8161253954976907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.721469267091611e-06,
+      "loss": 0.8498,
+      "step": 22441
+    },
+    {
+      "epoch": 0.8161617631014293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7208086089077292e-06,
+      "loss": 1.0018,
+      "step": 22442
+    },
+    {
+      "epoch": 0.8161981307051679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.720148065585877e-06,
+      "loss": 0.9014,
+      "step": 22443
+    },
+    {
+      "epoch": 0.8162344983089064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7194876371352243e-06,
+      "loss": 0.902,
+      "step": 22444
+    },
+    {
+      "epoch": 0.816270865912645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7188273235649323e-06,
+      "loss": 0.8799,
+      "step": 22445
+    },
+    {
+      "epoch": 0.8163072335163836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7181671248841625e-06,
+      "loss": 0.9013,
+      "step": 22446
+    },
+    {
+      "epoch": 0.8163436011201222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7175070411020701e-06,
+      "loss": 0.9889,
+      "step": 22447
+    },
+    {
+      "epoch": 0.8163799687238608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7168470722278196e-06,
+      "loss": 0.9537,
+      "step": 22448
+    },
+    {
+      "epoch": 0.8164163363275994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.716187218270563e-06,
+      "loss": 0.9178,
+      "step": 22449
+    },
+    {
+      "epoch": 0.816452703931338,
+      "grad_norm": 0.0,
+      "learning_rate": 1.715527479239455e-06,
+      "loss": 0.933,
+      "step": 22450
+    },
+    {
+      "epoch": 0.8164890715350765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.714867855143647e-06,
+      "loss": 0.7662,
+      "step": 22451
+    },
+    {
+      "epoch": 0.8165254391388151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7142083459922986e-06,
+      "loss": 0.9973,
+      "step": 22452
+    },
+    {
+      "epoch": 0.8165618067425537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713548951794547e-06,
+      "loss": 1.0097,
+      "step": 22453
+    },
+    {
+      "epoch": 0.8165981743462923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7128896725595501e-06,
+      "loss": 0.9876,
+      "step": 22454
+    },
+    {
+      "epoch": 0.8166345419500309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7122305082964507e-06,
+      "loss": 0.9697,
+      "step": 22455
+    },
+    {
+      "epoch": 0.8166709095537695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711571459014394e-06,
+      "loss": 1.0492,
+      "step": 22456
+    },
+    {
+      "epoch": 0.8167072771575081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7109125247225211e-06,
+      "loss": 0.8593,
+      "step": 22457
+    },
+    {
+      "epoch": 0.8167436447612467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7102537054299817e-06,
+      "loss": 0.8729,
+      "step": 22458
+    },
+    {
+      "epoch": 0.8167800123649853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7095950011459083e-06,
+      "loss": 0.8953,
+      "step": 22459
+    },
+    {
+      "epoch": 0.8168163799687239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7089364118794382e-06,
+      "loss": 0.9728,
+      "step": 22460
+    },
+    {
+      "epoch": 0.8168527475724624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7082779376397152e-06,
+      "loss": 0.9648,
+      "step": 22461
+    },
+    {
+      "epoch": 0.816889115176201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7076195784358707e-06,
+      "loss": 1.0108,
+      "step": 22462
+    },
+    {
+      "epoch": 0.8169254827799396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7069613342770398e-06,
+      "loss": 0.9073,
+      "step": 22463
+    },
+    {
+      "epoch": 0.8169618503836782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7063032051723538e-06,
+      "loss": 0.831,
+      "step": 22464
+    },
+    {
+      "epoch": 0.8169982179874168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7056451911309434e-06,
+      "loss": 0.7777,
+      "step": 22465
+    },
+    {
+      "epoch": 0.8170345855911554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.704987292161936e-06,
+      "loss": 1.0228,
+      "step": 22466
+    },
+    {
+      "epoch": 0.817070953194894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7043295082744626e-06,
+      "loss": 0.9136,
+      "step": 22467
+    },
+    {
+      "epoch": 0.8171073207986326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7036718394776486e-06,
+      "loss": 0.7702,
+      "step": 22468
+    },
+    {
+      "epoch": 0.8171436884023712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7030142857806153e-06,
+      "loss": 0.8999,
+      "step": 22469
+    },
+    {
+      "epoch": 0.8171800560061098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7023568471924855e-06,
+      "loss": 0.9139,
+      "step": 22470
+    },
+    {
+      "epoch": 0.8172164236098484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7016995237223855e-06,
+      "loss": 0.8927,
+      "step": 22471
+    },
+    {
+      "epoch": 0.817252791213587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7010423153794265e-06,
+      "loss": 0.8345,
+      "step": 22472
+    },
+    {
+      "epoch": 0.8172891588173256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7003852221727335e-06,
+      "loss": 0.8403,
+      "step": 22473
+    },
+    {
+      "epoch": 0.8173255264210642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6997282441114193e-06,
+      "loss": 0.8814,
+      "step": 22474
+    },
+    {
+      "epoch": 0.8173618940248027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6990713812046e-06,
+      "loss": 0.9184,
+      "step": 22475
+    },
+    {
+      "epoch": 0.8173982616285413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.698414633461385e-06,
+      "loss": 0.9056,
+      "step": 22476
+    },
+    {
+      "epoch": 0.8174346292322798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6977580008908933e-06,
+      "loss": 0.8759,
+      "step": 22477
+    },
+    {
+      "epoch": 0.8174709968360184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.697101483502227e-06,
+      "loss": 0.819,
+      "step": 22478
+    },
+    {
+      "epoch": 0.817507364439757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6964450813044953e-06,
+      "loss": 0.8715,
+      "step": 22479
+    },
+    {
+      "epoch": 0.8175437320434956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6957887943068097e-06,
+      "loss": 0.9481,
+      "step": 22480
+    },
+    {
+      "epoch": 0.8175800996472342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6951326225182728e-06,
+      "loss": 1.0701,
+      "step": 22481
+    },
+    {
+      "epoch": 0.8176164672509728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6944765659479866e-06,
+      "loss": 0.9796,
+      "step": 22482
+    },
+    {
+      "epoch": 0.8176528348547114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6938206246050526e-06,
+      "loss": 0.8682,
+      "step": 22483
+    },
+    {
+      "epoch": 0.81768920245845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6931647984985767e-06,
+      "loss": 0.9401,
+      "step": 22484
+    },
+    {
+      "epoch": 0.8177255700621886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6925090876376482e-06,
+      "loss": 1.0071,
+      "step": 22485
+    },
+    {
+      "epoch": 0.8177619376659272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6918534920313723e-06,
+      "loss": 1.0037,
+      "step": 22486
+    },
+    {
+      "epoch": 0.8177983052696658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6911980116888416e-06,
+      "loss": 0.9665,
+      "step": 22487
+    },
+    {
+      "epoch": 0.8178346728734044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6905426466191498e-06,
+      "loss": 0.8989,
+      "step": 22488
+    },
+    {
+      "epoch": 0.817871040477143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6898873968313867e-06,
+      "loss": 0.9572,
+      "step": 22489
+    },
+    {
+      "epoch": 0.8179074080808816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6892322623346502e-06,
+      "loss": 1.0403,
+      "step": 22490
+    },
+    {
+      "epoch": 0.8179437756846202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.68857724313802e-06,
+      "loss": 0.9624,
+      "step": 22491
+    },
+    {
+      "epoch": 0.8179801432883588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6879223392505905e-06,
+      "loss": 1.004,
+      "step": 22492
+    },
+    {
+      "epoch": 0.8180165108920974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6872675506814452e-06,
+      "loss": 1.0173,
+      "step": 22493
+    },
+    {
+      "epoch": 0.8180528784958359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6866128774396684e-06,
+      "loss": 0.8483,
+      "step": 22494
+    },
+    {
+      "epoch": 0.8180892460995745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685958319534341e-06,
+      "loss": 0.9833,
+      "step": 22495
+    },
+    {
+      "epoch": 0.818125613703313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6853038769745466e-06,
+      "loss": 0.9567,
+      "step": 22496
+    },
+    {
+      "epoch": 0.8181619813070516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6846495497693683e-06,
+      "loss": 0.8878,
+      "step": 22497
+    },
+    {
+      "epoch": 0.8181983489107902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.683995337927873e-06,
+      "loss": 0.8443,
+      "step": 22498
+    },
+    {
+      "epoch": 0.8182347165145288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6833412414591465e-06,
+      "loss": 0.8036,
+      "step": 22499
+    },
+    {
+      "epoch": 0.8182710841182674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.682687260372261e-06,
+      "loss": 1.0839,
+      "step": 22500
+    },
+    {
+      "epoch": 0.818307451722006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.682033394676289e-06,
+      "loss": 0.9947,
+      "step": 22501
+    },
+    {
+      "epoch": 0.8183438193257446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6813796443802987e-06,
+      "loss": 0.962,
+      "step": 22502
+    },
+    {
+      "epoch": 0.8183801869294832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6807260094933697e-06,
+      "loss": 1.013,
+      "step": 22503
+    },
+    {
+      "epoch": 0.8184165545332218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.680072490024557e-06,
+      "loss": 0.887,
+      "step": 22504
+    },
+    {
+      "epoch": 0.8184529221369604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6794190859829373e-06,
+      "loss": 1.0378,
+      "step": 22505
+    },
+    {
+      "epoch": 0.818489289740699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6787657973775728e-06,
+      "loss": 0.9647,
+      "step": 22506
+    },
+    {
+      "epoch": 0.8185256573444376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.678112624217527e-06,
+      "loss": 0.9025,
+      "step": 22507
+    },
+    {
+      "epoch": 0.8185620249481762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6774595665118577e-06,
+      "loss": 0.841,
+      "step": 22508
+    },
+    {
+      "epoch": 0.8185983925519148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6768066242696346e-06,
+      "loss": 1.0228,
+      "step": 22509
+    },
+    {
+      "epoch": 0.8186347601556534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6761537974999054e-06,
+      "loss": 0.8729,
+      "step": 22510
+    },
+    {
+      "epoch": 0.8186711277593919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.675501086211734e-06,
+      "loss": 0.8593,
+      "step": 22511
+    },
+    {
+      "epoch": 0.8187074953631305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.674848490414175e-06,
+      "loss": 0.9412,
+      "step": 22512
+    },
+    {
+      "epoch": 0.8187438629668691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6741960101162812e-06,
+      "loss": 0.9083,
+      "step": 22513
+    },
+    {
+      "epoch": 0.8187802305706077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6735436453271014e-06,
+      "loss": 0.7897,
+      "step": 22514
+    },
+    {
+      "epoch": 0.8188165981743463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6728913960556937e-06,
+      "loss": 0.946,
+      "step": 22515
+    },
+    {
+      "epoch": 0.8188529657780849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672239262311105e-06,
+      "loss": 0.8717,
+      "step": 22516
+    },
+    {
+      "epoch": 0.8188893333818235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6715872441023761e-06,
+      "loss": 0.8879,
+      "step": 22517
+    },
+    {
+      "epoch": 0.818925700985562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.670935341438561e-06,
+      "loss": 0.8911,
+      "step": 22518
+    },
+    {
+      "epoch": 0.8189620685893007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6702835543287e-06,
+      "loss": 0.9661,
+      "step": 22519
+    },
+    {
+      "epoch": 0.8189984361930392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6696318827818359e-06,
+      "loss": 0.8029,
+      "step": 22520
+    },
+    {
+      "epoch": 0.8190348037967778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6689803268070081e-06,
+      "loss": 0.9355,
+      "step": 22521
+    },
+    {
+      "epoch": 0.8190711714005164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6683288864132629e-06,
+      "loss": 0.9291,
+      "step": 22522
+    },
+    {
+      "epoch": 0.819107539004255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6676775616096297e-06,
+      "loss": 0.9288,
+      "step": 22523
+    },
+    {
+      "epoch": 0.8191439066079936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.667026352405149e-06,
+      "loss": 0.9125,
+      "step": 22524
+    },
+    {
+      "epoch": 0.8191802742117322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6663752588088556e-06,
+      "loss": 0.9079,
+      "step": 22525
+    },
+    {
+      "epoch": 0.8192166418154708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6657242808297814e-06,
+      "loss": 0.7458,
+      "step": 22526
+    },
+    {
+      "epoch": 0.8192530094192094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6650734184769557e-06,
+      "loss": 1.0098,
+      "step": 22527
+    },
+    {
+      "epoch": 0.8192893770229479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6644226717594125e-06,
+      "loss": 0.9602,
+      "step": 22528
+    },
+    {
+      "epoch": 0.8193257446266865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6637720406861791e-06,
+      "loss": 0.8787,
+      "step": 22529
+    },
+    {
+      "epoch": 0.8193621122304251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6631215252662803e-06,
+      "loss": 0.8723,
+      "step": 22530
+    },
+    {
+      "epoch": 0.8193984798341637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6624711255087423e-06,
+      "loss": 0.9228,
+      "step": 22531
+    },
+    {
+      "epoch": 0.8194348474379023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6618208414225877e-06,
+      "loss": 0.9328,
+      "step": 22532
+    },
+    {
+      "epoch": 0.8194712150416409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6611706730168353e-06,
+      "loss": 0.9785,
+      "step": 22533
+    },
+    {
+      "epoch": 0.8195075826453795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.660520620300512e-06,
+      "loss": 0.9326,
+      "step": 22534
+    },
+    {
+      "epoch": 0.8195439502491181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659870683282636e-06,
+      "loss": 1.058,
+      "step": 22535
+    },
+    {
+      "epoch": 0.8195803178528567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.659220861972215e-06,
+      "loss": 1.0215,
+      "step": 22536
+    },
+    {
+      "epoch": 0.8196166854565953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6585711563782736e-06,
+      "loss": 0.9621,
+      "step": 22537
+    },
+    {
+      "epoch": 0.8196530530603339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6579215665098225e-06,
+      "loss": 0.8477,
+      "step": 22538
+    },
+    {
+      "epoch": 0.8196894206640725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6572720923758734e-06,
+      "loss": 0.974,
+      "step": 22539
+    },
+    {
+      "epoch": 0.8197257882678111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6566227339854357e-06,
+      "loss": 0.903,
+      "step": 22540
+    },
+    {
+      "epoch": 0.8197621558715497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6559734913475234e-06,
+      "loss": 1.0582,
+      "step": 22541
+    },
+    {
+      "epoch": 0.8197985234752883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.655324364471137e-06,
+      "loss": 1.0261,
+      "step": 22542
+    },
+    {
+      "epoch": 0.8198348910790269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.654675353365287e-06,
+      "loss": 1.015,
+      "step": 22543
+    },
+    {
+      "epoch": 0.8198712586827654,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6540264580389754e-06,
+      "loss": 0.9107,
+      "step": 22544
+    },
+    {
+      "epoch": 0.8199076262865039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6533776785012068e-06,
+      "loss": 0.8126,
+      "step": 22545
+    },
+    {
+      "epoch": 0.8199439938902425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6527290147609764e-06,
+      "loss": 0.8866,
+      "step": 22546
+    },
+    {
+      "epoch": 0.8199803614939811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6520804668272917e-06,
+      "loss": 0.9707,
+      "step": 22547
+    },
+    {
+      "epoch": 0.8200167290977197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.651432034709146e-06,
+      "loss": 0.8984,
+      "step": 22548
+    },
+    {
+      "epoch": 0.8200530967014583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.650783718415535e-06,
+      "loss": 0.9153,
+      "step": 22549
+    },
+    {
+      "epoch": 0.8200894643051969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6501355179554524e-06,
+      "loss": 0.8475,
+      "step": 22550
+    },
+    {
+      "epoch": 0.8201258319089355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6494874333378974e-06,
+      "loss": 0.8513,
+      "step": 22551
+    },
+    {
+      "epoch": 0.8201621995126741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6488394645718508e-06,
+      "loss": 0.9348,
+      "step": 22552
+    },
+    {
+      "epoch": 0.8201985671164127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.648191611666311e-06,
+      "loss": 0.9165,
+      "step": 22553
+    },
+    {
+      "epoch": 0.8202349347201513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6475438746302652e-06,
+      "loss": 0.9391,
+      "step": 22554
+    },
+    {
+      "epoch": 0.8202713023238899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6468962534726917e-06,
+      "loss": 0.9064,
+      "step": 22555
+    },
+    {
+      "epoch": 0.8203076699276285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6462487482025847e-06,
+      "loss": 1.016,
+      "step": 22556
+    },
+    {
+      "epoch": 0.8203440375313671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6456013588289232e-06,
+      "loss": 0.8633,
+      "step": 22557
+    },
+    {
+      "epoch": 0.8203804051351057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6449540853606905e-06,
+      "loss": 0.8232,
+      "step": 22558
+    },
+    {
+      "epoch": 0.8204167727388443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6443069278068623e-06,
+      "loss": 0.893,
+      "step": 22559
+    },
+    {
+      "epoch": 0.8204531403425829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.643659886176423e-06,
+      "loss": 1.0125,
+      "step": 22560
+    },
+    {
+      "epoch": 0.8204895079463214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6430129604783474e-06,
+      "loss": 0.8914,
+      "step": 22561
+    },
+    {
+      "epoch": 0.82052587555006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6423661507216094e-06,
+      "loss": 0.7298,
+      "step": 22562
+    },
+    {
+      "epoch": 0.8205622431537986,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6417194569151828e-06,
+      "loss": 1.0208,
+      "step": 22563
+    },
+    {
+      "epoch": 0.8205986107575372,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6410728790680396e-06,
+      "loss": 1.0534,
+      "step": 22564
+    },
+    {
+      "epoch": 0.8206349783612757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6404264171891492e-06,
+      "loss": 0.7995,
+      "step": 22565
+    },
+    {
+      "epoch": 0.8206713459650143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6397800712874834e-06,
+      "loss": 0.9001,
+      "step": 22566
+    },
+    {
+      "epoch": 0.8207077135687529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.639133841372007e-06,
+      "loss": 0.9852,
+      "step": 22567
+    },
+    {
+      "epoch": 0.8207440811724915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6384877274516864e-06,
+      "loss": 0.8685,
+      "step": 22568
+    },
+    {
+      "epoch": 0.8207804487762301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.637841729535483e-06,
+      "loss": 0.8158,
+      "step": 22569
+    },
+    {
+      "epoch": 0.8208168163799687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6371958476323656e-06,
+      "loss": 1.0321,
+      "step": 22570
+    },
+    {
+      "epoch": 0.8208531839837073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6365500817512869e-06,
+      "loss": 0.917,
+      "step": 22571
+    },
+    {
+      "epoch": 0.8208895515874459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6359044319012108e-06,
+      "loss": 0.901,
+      "step": 22572
+    },
+    {
+      "epoch": 0.8209259191911845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6352588980910944e-06,
+      "loss": 0.9191,
+      "step": 22573
+    },
+    {
+      "epoch": 0.8209622867949231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6346134803298918e-06,
+      "loss": 0.8514,
+      "step": 22574
+    },
+    {
+      "epoch": 0.8209986543986617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6339681786265593e-06,
+      "loss": 0.8891,
+      "step": 22575
+    },
+    {
+      "epoch": 0.8210350220024003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6333229929900473e-06,
+      "loss": 0.9625,
+      "step": 22576
+    },
+    {
+      "epoch": 0.8210713896061389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6326779234293088e-06,
+      "loss": 0.9227,
+      "step": 22577
+    },
+    {
+      "epoch": 0.8211077572098774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.63203296995329e-06,
+      "loss": 0.9952,
+      "step": 22578
+    },
+    {
+      "epoch": 0.821144124813616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6313881325709435e-06,
+      "loss": 0.9655,
+      "step": 22579
+    },
+    {
+      "epoch": 0.8211804924173546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6307434112912135e-06,
+      "loss": 0.9852,
+      "step": 22580
+    },
+    {
+      "epoch": 0.8212168600210932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6300988061230438e-06,
+      "loss": 0.9483,
+      "step": 22581
+    },
+    {
+      "epoch": 0.8212532276248318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6294543170753751e-06,
+      "loss": 1.0297,
+      "step": 22582
+    },
+    {
+      "epoch": 0.8212895952285704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.628809944157157e-06,
+      "loss": 0.8489,
+      "step": 22583
+    },
+    {
+      "epoch": 0.821325962832309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6281656873773187e-06,
+      "loss": 0.8208,
+      "step": 22584
+    },
+    {
+      "epoch": 0.8213623304360476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6275215467448057e-06,
+      "loss": 0.915,
+      "step": 22585
+    },
+    {
+      "epoch": 0.8213986980397862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6268775222685518e-06,
+      "loss": 0.9033,
+      "step": 22586
+    },
+    {
+      "epoch": 0.8214350656435248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.626233613957493e-06,
+      "loss": 1.0105,
+      "step": 22587
+    },
+    {
+      "epoch": 0.8214714332472633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6255898218205591e-06,
+      "loss": 0.916,
+      "step": 22588
+    },
+    {
+      "epoch": 0.821507800851002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6249461458666904e-06,
+      "loss": 1.0472,
+      "step": 22589
+    },
+    {
+      "epoch": 0.8215441684547405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6243025861048056e-06,
+      "loss": 0.8839,
+      "step": 22590
+    },
+    {
+      "epoch": 0.8215805360584791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.623659142543842e-06,
+      "loss": 0.9023,
+      "step": 22591
+    },
+    {
+      "epoch": 0.8216169036622177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6230158151927223e-06,
+      "loss": 0.9017,
+      "step": 22592
+    },
+    {
+      "epoch": 0.8216532712659563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6223726040603738e-06,
+      "loss": 0.8238,
+      "step": 22593
+    },
+    {
+      "epoch": 0.8216896388696949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6217295091557184e-06,
+      "loss": 0.9804,
+      "step": 22594
+    },
+    {
+      "epoch": 0.8217260064734334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6210865304876766e-06,
+      "loss": 0.8176,
+      "step": 22595
+    },
+    {
+      "epoch": 0.821762374077172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6204436680651768e-06,
+      "loss": 0.9801,
+      "step": 22596
+    },
+    {
+      "epoch": 0.8217987416809106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6198009218971256e-06,
+      "loss": 0.8365,
+      "step": 22597
+    },
+    {
+      "epoch": 0.8218351092846492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6191582919924508e-06,
+      "loss": 1.0164,
+      "step": 22598
+    },
+    {
+      "epoch": 0.8218714768883878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6185157783600624e-06,
+      "loss": 0.8704,
+      "step": 22599
+    },
+    {
+      "epoch": 0.8219078444921264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6178733810088764e-06,
+      "loss": 0.892,
+      "step": 22600
+    },
+    {
+      "epoch": 0.821944212095865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6172310999478026e-06,
+      "loss": 0.8641,
+      "step": 22601
+    },
+    {
+      "epoch": 0.8219805796996036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6165889351857578e-06,
+      "loss": 0.8602,
+      "step": 22602
+    },
+    {
+      "epoch": 0.8220169473033422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.615946886731642e-06,
+      "loss": 0.9936,
+      "step": 22603
+    },
+    {
+      "epoch": 0.8220533149070808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.615304954594371e-06,
+      "loss": 0.9542,
+      "step": 22604
+    },
+    {
+      "epoch": 0.8220896825108194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614663138782847e-06,
+      "loss": 1.1171,
+      "step": 22605
+    },
+    {
+      "epoch": 0.822126050114558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.614021439305975e-06,
+      "loss": 0.8815,
+      "step": 22606
+    },
+    {
+      "epoch": 0.8221624177182966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6133798561726544e-06,
+      "loss": 0.874,
+      "step": 22607
+    },
+    {
+      "epoch": 0.8221987853220352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6127383893917947e-06,
+      "loss": 1.0392,
+      "step": 22608
+    },
+    {
+      "epoch": 0.8222351529257738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6120970389722846e-06,
+      "loss": 0.9336,
+      "step": 22609
+    },
+    {
+      "epoch": 0.8222715205295124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.61145580492303e-06,
+      "loss": 1.0616,
+      "step": 22610
+    },
+    {
+      "epoch": 0.822307888133251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.610814687252924e-06,
+      "loss": 0.9201,
+      "step": 22611
+    },
+    {
+      "epoch": 0.8223442557369894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6101736859708618e-06,
+      "loss": 0.8688,
+      "step": 22612
+    },
+    {
+      "epoch": 0.822380623340728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.609532801085736e-06,
+      "loss": 1.0436,
+      "step": 22613
+    },
+    {
+      "epoch": 0.8224169909444666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.608892032606435e-06,
+      "loss": 0.9674,
+      "step": 22614
+    },
+    {
+      "epoch": 0.8224533585482052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6082513805418564e-06,
+      "loss": 0.9501,
+      "step": 22615
+    },
+    {
+      "epoch": 0.8224897261519438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.60761084490088e-06,
+      "loss": 0.9048,
+      "step": 22616
+    },
+    {
+      "epoch": 0.8225260937556824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.606970425692398e-06,
+      "loss": 0.877,
+      "step": 22617
+    },
+    {
+      "epoch": 0.822562461359421,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6063301229252926e-06,
+      "loss": 1.0315,
+      "step": 22618
+    },
+    {
+      "epoch": 0.8225988289631596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6056899366084478e-06,
+      "loss": 0.962,
+      "step": 22619
+    },
+    {
+      "epoch": 0.8226351965668982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.605049866750743e-06,
+      "loss": 0.9236,
+      "step": 22620
+    },
+    {
+      "epoch": 0.8226715641706368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6044099133610646e-06,
+      "loss": 0.9491,
+      "step": 22621
+    },
+    {
+      "epoch": 0.8227079317743754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6037700764482821e-06,
+      "loss": 0.8813,
+      "step": 22622
+    },
+    {
+      "epoch": 0.822744299378114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6031303560212808e-06,
+      "loss": 0.9009,
+      "step": 22623
+    },
+    {
+      "epoch": 0.8227806669818526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.602490752088931e-06,
+      "loss": 0.9127,
+      "step": 22624
+    },
+    {
+      "epoch": 0.8228170345855912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6018512646601081e-06,
+      "loss": 0.8782,
+      "step": 22625
+    },
+    {
+      "epoch": 0.8228534021893298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6012118937436804e-06,
+      "loss": 0.8023,
+      "step": 22626
+    },
+    {
+      "epoch": 0.8228897697930684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6005726393485243e-06,
+      "loss": 0.8925,
+      "step": 22627
+    },
+    {
+      "epoch": 0.822926137396807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.599933501483506e-06,
+      "loss": 0.8915,
+      "step": 22628
+    },
+    {
+      "epoch": 0.8229625050005455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5992944801574918e-06,
+      "loss": 0.8691,
+      "step": 22629
+    },
+    {
+      "epoch": 0.8229988726042841,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5986555753793464e-06,
+      "loss": 0.9514,
+      "step": 22630
+    },
+    {
+      "epoch": 0.8230352402080227,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5980167871579367e-06,
+      "loss": 0.8396,
+      "step": 22631
+    },
+    {
+      "epoch": 0.8230716078117613,
+      "grad_norm": 0.0,
+      "learning_rate": 1.597378115502123e-06,
+      "loss": 0.8419,
+      "step": 22632
+    },
+    {
+      "epoch": 0.8231079754154998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5967395604207626e-06,
+      "loss": 0.9079,
+      "step": 22633
+    },
+    {
+      "epoch": 0.8231443430192384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.596101121922724e-06,
+      "loss": 0.8857,
+      "step": 22634
+    },
+    {
+      "epoch": 0.823180710622977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5954628000168538e-06,
+      "loss": 0.9231,
+      "step": 22635
+    },
+    {
+      "epoch": 0.8232170782267156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5948245947120155e-06,
+      "loss": 0.9154,
+      "step": 22636
+    },
+    {
+      "epoch": 0.8232534458304542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5941865060170602e-06,
+      "loss": 0.948,
+      "step": 22637
+    },
+    {
+      "epoch": 0.8232898134341928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5935485339408408e-06,
+      "loss": 0.8604,
+      "step": 22638
+    },
+    {
+      "epoch": 0.8233261810379314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5929106784922056e-06,
+      "loss": 0.8417,
+      "step": 22639
+    },
+    {
+      "epoch": 0.82336254864167,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5922729396800097e-06,
+      "loss": 0.9343,
+      "step": 22640
+    },
+    {
+      "epoch": 0.8233989162454086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5916353175130972e-06,
+      "loss": 0.9222,
+      "step": 22641
+    },
+    {
+      "epoch": 0.8234352838491472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5909978120003145e-06,
+      "loss": 0.9507,
+      "step": 22642
+    },
+    {
+      "epoch": 0.8234716514528858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5903604231505065e-06,
+      "loss": 0.9357,
+      "step": 22643
+    },
+    {
+      "epoch": 0.8235080190566244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5897231509725163e-06,
+      "loss": 0.8554,
+      "step": 22644
+    },
+    {
+      "epoch": 0.823544386660363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5890859954751826e-06,
+      "loss": 0.9216,
+      "step": 22645
+    },
+    {
+      "epoch": 0.8235807542641015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.588448956667349e-06,
+      "loss": 1.0066,
+      "step": 22646
+    },
+    {
+      "epoch": 0.8236171218678401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5878120345578518e-06,
+      "loss": 0.9796,
+      "step": 22647
+    },
+    {
+      "epoch": 0.8236534894715787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5871752291555276e-06,
+      "loss": 0.8351,
+      "step": 22648
+    },
+    {
+      "epoch": 0.8236898570753173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.586538540469208e-06,
+      "loss": 0.9525,
+      "step": 22649
+    },
+    {
+      "epoch": 0.8237262246790559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5859019685077347e-06,
+      "loss": 0.9521,
+      "step": 22650
+    },
+    {
+      "epoch": 0.8237625922827945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.58526551327993e-06,
+      "loss": 0.8506,
+      "step": 22651
+    },
+    {
+      "epoch": 0.8237989598865331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5846291747946264e-06,
+      "loss": 0.9976,
+      "step": 22652
+    },
+    {
+      "epoch": 0.8238353274902717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5839929530606568e-06,
+      "loss": 1.0644,
+      "step": 22653
+    },
+    {
+      "epoch": 0.8238716950940103,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5833568480868388e-06,
+      "loss": 0.9397,
+      "step": 22654
+    },
+    {
+      "epoch": 0.8239080626977489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5827208598820066e-06,
+      "loss": 1.0385,
+      "step": 22655
+    },
+    {
+      "epoch": 0.8239444303014875,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5820849884549793e-06,
+      "loss": 0.7885,
+      "step": 22656
+    },
+    {
+      "epoch": 0.823980797905226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5814492338145793e-06,
+      "loss": 0.7619,
+      "step": 22657
+    },
+    {
+      "epoch": 0.8240171655089646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5808135959696247e-06,
+      "loss": 0.7964,
+      "step": 22658
+    },
+    {
+      "epoch": 0.8240535331127032,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5801780749289385e-06,
+      "loss": 0.9862,
+      "step": 22659
+    },
+    {
+      "epoch": 0.8240899007164418,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5795426707013362e-06,
+      "loss": 0.9889,
+      "step": 22660
+    },
+    {
+      "epoch": 0.8241262683201804,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5789073832956315e-06,
+      "loss": 0.8361,
+      "step": 22661
+    },
+    {
+      "epoch": 0.8241626359239189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5782722127206385e-06,
+      "loss": 0.9527,
+      "step": 22662
+    },
+    {
+      "epoch": 0.8241990035276575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5776371589851703e-06,
+      "loss": 0.9238,
+      "step": 22663
+    },
+    {
+      "epoch": 0.8242353711313961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577002222098033e-06,
+      "loss": 0.7934,
+      "step": 22664
+    },
+    {
+      "epoch": 0.8242717387351347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.576367402068043e-06,
+      "loss": 0.965,
+      "step": 22665
+    },
+    {
+      "epoch": 0.8243081063388733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5757326989040034e-06,
+      "loss": 0.9223,
+      "step": 22666
+    },
+    {
+      "epoch": 0.8243444739426119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5750981126147203e-06,
+      "loss": 0.8988,
+      "step": 22667
+    },
+    {
+      "epoch": 0.8243808415463505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5744636432089944e-06,
+      "loss": 0.8433,
+      "step": 22668
+    },
+    {
+      "epoch": 0.8244172091500891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5738292906956364e-06,
+      "loss": 0.9122,
+      "step": 22669
+    },
+    {
+      "epoch": 0.8244535767538277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5731950550834397e-06,
+      "loss": 0.8901,
+      "step": 22670
+    },
+    {
+      "epoch": 0.8244899443575663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5725609363812033e-06,
+      "loss": 0.9456,
+      "step": 22671
+    },
+    {
+      "epoch": 0.8245263119613049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5719269345977283e-06,
+      "loss": 0.7937,
+      "step": 22672
+    },
+    {
+      "epoch": 0.8245626795650435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.57129304974181e-06,
+      "loss": 0.9555,
+      "step": 22673
+    },
+    {
+      "epoch": 0.8245990471687821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5706592818222422e-06,
+      "loss": 1.0261,
+      "step": 22674
+    },
+    {
+      "epoch": 0.8246354147725207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5700256308478167e-06,
+      "loss": 0.9845,
+      "step": 22675
+    },
+    {
+      "epoch": 0.8246717823762593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5693920968273247e-06,
+      "loss": 0.996,
+      "step": 22676
+    },
+    {
+      "epoch": 0.8247081499799979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5687586797695543e-06,
+      "loss": 1.0005,
+      "step": 22677
+    },
+    {
+      "epoch": 0.8247445175837365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5681253796832963e-06,
+      "loss": 0.9427,
+      "step": 22678
+    },
+    {
+      "epoch": 0.8247808851874749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.567492196577336e-06,
+      "loss": 0.8465,
+      "step": 22679
+    },
+    {
+      "epoch": 0.8248172527912135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5668591304604574e-06,
+      "loss": 0.8279,
+      "step": 22680
+    },
+    {
+      "epoch": 0.8248536203949521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5662261813414404e-06,
+      "loss": 0.9371,
+      "step": 22681
+    },
+    {
+      "epoch": 0.8248899879986907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5655933492290743e-06,
+      "loss": 0.8424,
+      "step": 22682
+    },
+    {
+      "epoch": 0.8249263556024293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5649606341321289e-06,
+      "loss": 0.9098,
+      "step": 22683
+    },
+    {
+      "epoch": 0.8249627232061679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5643280360593893e-06,
+      "loss": 0.8306,
+      "step": 22684
+    },
+    {
+      "epoch": 0.8249990908099065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5636955550196298e-06,
+      "loss": 0.8433,
+      "step": 22685
+    },
+    {
+      "epoch": 0.8250354584136451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5630631910216242e-06,
+      "loss": 0.8072,
+      "step": 22686
+    },
+    {
+      "epoch": 0.8250718260173837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5624309440741447e-06,
+      "loss": 0.9363,
+      "step": 22687
+    },
+    {
+      "epoch": 0.8251081936211223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5617988141859686e-06,
+      "loss": 0.9094,
+      "step": 22688
+    },
+    {
+      "epoch": 0.8251445612248609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.56116680136586e-06,
+      "loss": 0.7696,
+      "step": 22689
+    },
+    {
+      "epoch": 0.8251809288285995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.560534905622586e-06,
+      "loss": 0.8549,
+      "step": 22690
+    },
+    {
+      "epoch": 0.8252172964323381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.559903126964919e-06,
+      "loss": 1.0563,
+      "step": 22691
+    },
+    {
+      "epoch": 0.8252536640360767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5592714654016207e-06,
+      "loss": 0.8431,
+      "step": 22692
+    },
+    {
+      "epoch": 0.8252900316398153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5586399209414548e-06,
+      "loss": 0.8897,
+      "step": 22693
+    },
+    {
+      "epoch": 0.8253263992435539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5580084935931816e-06,
+      "loss": 0.8616,
+      "step": 22694
+    },
+    {
+      "epoch": 0.8253627668472925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5573771833655672e-06,
+      "loss": 0.9268,
+      "step": 22695
+    },
+    {
+      "epoch": 0.825399134451031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5567459902673622e-06,
+      "loss": 0.8329,
+      "step": 22696
+    },
+    {
+      "epoch": 0.8254355020547696,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5561149143073284e-06,
+      "loss": 0.9373,
+      "step": 22697
+    },
+    {
+      "epoch": 0.8254718696585082,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5554839554942203e-06,
+      "loss": 0.9108,
+      "step": 22698
+    },
+    {
+      "epoch": 0.8255082372622468,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5548531138367906e-06,
+      "loss": 0.9755,
+      "step": 22699
+    },
+    {
+      "epoch": 0.8255446048659854,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5542223893437902e-06,
+      "loss": 0.8705,
+      "step": 22700
+    },
+    {
+      "epoch": 0.825580972469724,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5535917820239754e-06,
+      "loss": 0.8816,
+      "step": 22701
+    },
+    {
+      "epoch": 0.8256173400734625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.552961291886086e-06,
+      "loss": 0.884,
+      "step": 22702
+    },
+    {
+      "epoch": 0.8256537076772011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5523309189388757e-06,
+      "loss": 0.9147,
+      "step": 22703
+    },
+    {
+      "epoch": 0.8256900752809397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5517006631910892e-06,
+      "loss": 1.0471,
+      "step": 22704
+    },
+    {
+      "epoch": 0.8257264428846783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.551070524651469e-06,
+      "loss": 0.8603,
+      "step": 22705
+    },
+    {
+      "epoch": 0.8257628104884169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5504405033287562e-06,
+      "loss": 0.859,
+      "step": 22706
+    },
+    {
+      "epoch": 0.8257991780921555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.549810599231697e-06,
+      "loss": 1.0645,
+      "step": 22707
+    },
+    {
+      "epoch": 0.8258355456958941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5491808123690245e-06,
+      "loss": 0.8838,
+      "step": 22708
+    },
+    {
+      "epoch": 0.8258719132996327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5485511427494748e-06,
+      "loss": 0.8578,
+      "step": 22709
+    },
+    {
+      "epoch": 0.8259082809033713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5479215903817902e-06,
+      "loss": 0.9066,
+      "step": 22710
+    },
+    {
+      "epoch": 0.8259446485071099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.547292155274701e-06,
+      "loss": 0.9865,
+      "step": 22711
+    },
+    {
+      "epoch": 0.8259810161108485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5466628374369408e-06,
+      "loss": 0.848,
+      "step": 22712
+    },
+    {
+      "epoch": 0.826017383714587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5460336368772377e-06,
+      "loss": 1.0146,
+      "step": 22713
+    },
+    {
+      "epoch": 0.8260537513183256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5454045536043283e-06,
+      "loss": 0.9832,
+      "step": 22714
+    },
+    {
+      "epoch": 0.8260901189220642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5447755876269298e-06,
+      "loss": 0.9696,
+      "step": 22715
+    },
+    {
+      "epoch": 0.8261264865258028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5441467389537768e-06,
+      "loss": 0.9021,
+      "step": 22716
+    },
+    {
+      "epoch": 0.8261628541295414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5435180075935896e-06,
+      "loss": 1.09,
+      "step": 22717
+    },
+    {
+      "epoch": 0.82619922173328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5428893935550925e-06,
+      "loss": 0.9754,
+      "step": 22718
+    },
+    {
+      "epoch": 0.8262355893370186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5422608968470032e-06,
+      "loss": 0.7899,
+      "step": 22719
+    },
+    {
+      "epoch": 0.8262719569407572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.541632517478049e-06,
+      "loss": 0.8997,
+      "step": 22720
+    },
+    {
+      "epoch": 0.8263083245444958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.541004255456937e-06,
+      "loss": 0.9818,
+      "step": 22721
+    },
+    {
+      "epoch": 0.8263446921482344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.540376110792393e-06,
+      "loss": 0.9672,
+      "step": 22722
+    },
+    {
+      "epoch": 0.826381059751973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5397480834931279e-06,
+      "loss": 0.8997,
+      "step": 22723
+    },
+    {
+      "epoch": 0.8264174273557116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.539120173567854e-06,
+      "loss": 0.7148,
+      "step": 22724
+    },
+    {
+      "epoch": 0.8264537949594501,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5384923810252805e-06,
+      "loss": 0.912,
+      "step": 22725
+    },
+    {
+      "epoch": 0.8264901625631887,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5378647058741224e-06,
+      "loss": 0.8826,
+      "step": 22726
+    },
+    {
+      "epoch": 0.8265265301669273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5372371481230887e-06,
+      "loss": 0.8854,
+      "step": 22727
+    },
+    {
+      "epoch": 0.8265628977706659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5366097077808773e-06,
+      "loss": 0.9172,
+      "step": 22728
+    },
+    {
+      "epoch": 0.8265992653744045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5359823848561994e-06,
+      "loss": 0.8778,
+      "step": 22729
+    },
+    {
+      "epoch": 0.826635632978143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5353551793577582e-06,
+      "loss": 0.9955,
+      "step": 22730
+    },
+    {
+      "epoch": 0.8266720005818816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5347280912942542e-06,
+      "loss": 0.8674,
+      "step": 22731
+    },
+    {
+      "epoch": 0.8267083681856202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.534101120674385e-06,
+      "loss": 0.921,
+      "step": 22732
+    },
+    {
+      "epoch": 0.8267447357893588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5334742675068547e-06,
+      "loss": 0.9119,
+      "step": 22733
+    },
+    {
+      "epoch": 0.8267811033930974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5328475318003533e-06,
+      "loss": 0.8613,
+      "step": 22734
+    },
+    {
+      "epoch": 0.826817470996836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5322209135635802e-06,
+      "loss": 0.9979,
+      "step": 22735
+    },
+    {
+      "epoch": 0.8268538386005746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5315944128052284e-06,
+      "loss": 0.9843,
+      "step": 22736
+    },
+    {
+      "epoch": 0.8268902062043132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.530968029533989e-06,
+      "loss": 0.9276,
+      "step": 22737
+    },
+    {
+      "epoch": 0.8269265738080518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5303417637585506e-06,
+      "loss": 0.935,
+      "step": 22738
+    },
+    {
+      "epoch": 0.8269629414117904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5297156154876046e-06,
+      "loss": 0.8905,
+      "step": 22739
+    },
+    {
+      "epoch": 0.826999309015529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.529089584729837e-06,
+      "loss": 0.9836,
+      "step": 22740
+    },
+    {
+      "epoch": 0.8270356766192676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5284636714939327e-06,
+      "loss": 0.8876,
+      "step": 22741
+    },
+    {
+      "epoch": 0.8270720442230062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5278378757885748e-06,
+      "loss": 0.9084,
+      "step": 22742
+    },
+    {
+      "epoch": 0.8271084118267448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5272121976224462e-06,
+      "loss": 1.0023,
+      "step": 22743
+    },
+    {
+      "epoch": 0.8271447794304834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5265866370042259e-06,
+      "loss": 0.8872,
+      "step": 22744
+    },
+    {
+      "epoch": 0.827181147034222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5259611939425957e-06,
+      "loss": 0.9887,
+      "step": 22745
+    },
+    {
+      "epoch": 0.8272175146379604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5253358684462328e-06,
+      "loss": 0.8535,
+      "step": 22746
+    },
+    {
+      "epoch": 0.827253882241699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.524710660523806e-06,
+      "loss": 0.9109,
+      "step": 22747
+    },
+    {
+      "epoch": 0.8272902498454376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5240855701839962e-06,
+      "loss": 0.8377,
+      "step": 22748
+    },
+    {
+      "epoch": 0.8273266174491762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5234605974354732e-06,
+      "loss": 0.8506,
+      "step": 22749
+    },
+    {
+      "epoch": 0.8273629850529148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5228357422869077e-06,
+      "loss": 0.8545,
+      "step": 22750
+    },
+    {
+      "epoch": 0.8273993526566534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5222110047469673e-06,
+      "loss": 0.8281,
+      "step": 22751
+    },
+    {
+      "epoch": 0.827435720260392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5215863848243252e-06,
+      "loss": 1.0233,
+      "step": 22752
+    },
+    {
+      "epoch": 0.8274720878641306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5209618825276373e-06,
+      "loss": 0.9072,
+      "step": 22753
+    },
+    {
+      "epoch": 0.8275084554678692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.520337497865576e-06,
+      "loss": 0.8952,
+      "step": 22754
+    },
+    {
+      "epoch": 0.8275448230716078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.519713230846801e-06,
+      "loss": 0.9796,
+      "step": 22755
+    },
+    {
+      "epoch": 0.8275811906753464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.519089081479973e-06,
+      "loss": 0.9271,
+      "step": 22756
+    },
+    {
+      "epoch": 0.827617558279085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5184650497737486e-06,
+      "loss": 0.9564,
+      "step": 22757
+    },
+    {
+      "epoch": 0.8276539258828236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5178411357367907e-06,
+      "loss": 0.8584,
+      "step": 22758
+    },
+    {
+      "epoch": 0.8276902934865622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5172173393777512e-06,
+      "loss": 0.8729,
+      "step": 22759
+    },
+    {
+      "epoch": 0.8277266610903008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5165936607052867e-06,
+      "loss": 0.8673,
+      "step": 22760
+    },
+    {
+      "epoch": 0.8277630286940394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5159700997280458e-06,
+      "loss": 0.8717,
+      "step": 22761
+    },
+    {
+      "epoch": 0.827799396297778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.515346656454688e-06,
+      "loss": 0.8206,
+      "step": 22762
+    },
+    {
+      "epoch": 0.8278357639015165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5147233308938514e-06,
+      "loss": 0.9606,
+      "step": 22763
+    },
+    {
+      "epoch": 0.8278721315052551,
+      "grad_norm": 0.0,
+      "learning_rate": 1.514100123054193e-06,
+      "loss": 0.9742,
+      "step": 22764
+    },
+    {
+      "epoch": 0.8279084991089937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5134770329443571e-06,
+      "loss": 0.9714,
+      "step": 22765
+    },
+    {
+      "epoch": 0.8279448667127323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5128540605729824e-06,
+      "loss": 0.9019,
+      "step": 22766
+    },
+    {
+      "epoch": 0.8279812343164709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5122312059487177e-06,
+      "loss": 0.9278,
+      "step": 22767
+    },
+    {
+      "epoch": 0.8280176019202095,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5116084690802023e-06,
+      "loss": 0.8275,
+      "step": 22768
+    },
+    {
+      "epoch": 0.828053969523948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5109858499760765e-06,
+      "loss": 0.8601,
+      "step": 22769
+    },
+    {
+      "epoch": 0.8280903371276866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5103633486449754e-06,
+      "loss": 0.9848,
+      "step": 22770
+    },
+    {
+      "epoch": 0.8281267047314252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5097409650955398e-06,
+      "loss": 0.878,
+      "step": 22771
+    },
+    {
+      "epoch": 0.8281630723351638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5091186993364026e-06,
+      "loss": 0.9167,
+      "step": 22772
+    },
+    {
+      "epoch": 0.8281994399389024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5084965513761963e-06,
+      "loss": 0.8913,
+      "step": 22773
+    },
+    {
+      "epoch": 0.828235807542641,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5078745212235523e-06,
+      "loss": 0.889,
+      "step": 22774
+    },
+    {
+      "epoch": 0.8282721751463796,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5072526088870997e-06,
+      "loss": 0.9569,
+      "step": 22775
+    },
+    {
+      "epoch": 0.8283085427501182,
+      "grad_norm": 0.0,
+      "learning_rate": 1.506630814375466e-06,
+      "loss": 0.9146,
+      "step": 22776
+    },
+    {
+      "epoch": 0.8283449103538568,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5060091376972818e-06,
+      "loss": 1.0276,
+      "step": 22777
+    },
+    {
+      "epoch": 0.8283812779575954,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5053875788611694e-06,
+      "loss": 0.9315,
+      "step": 22778
+    },
+    {
+      "epoch": 0.828417645561334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5047661378757517e-06,
+      "loss": 0.8652,
+      "step": 22779
+    },
+    {
+      "epoch": 0.8284540131650725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5041448147496484e-06,
+      "loss": 0.8049,
+      "step": 22780
+    },
+    {
+      "epoch": 0.8284903807688111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5035236094914873e-06,
+      "loss": 0.9789,
+      "step": 22781
+    },
+    {
+      "epoch": 0.8285267483725497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5029025221098747e-06,
+      "loss": 0.877,
+      "step": 22782
+    },
+    {
+      "epoch": 0.8285631159762883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5022815526134382e-06,
+      "loss": 0.9496,
+      "step": 22783
+    },
+    {
+      "epoch": 0.8285994835800269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5016607010107876e-06,
+      "loss": 0.7916,
+      "step": 22784
+    },
+    {
+      "epoch": 0.8286358511837655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5010399673105368e-06,
+      "loss": 0.877,
+      "step": 22785
+    },
+    {
+      "epoch": 0.8286722187875041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5004193515212984e-06,
+      "loss": 0.8774,
+      "step": 22786
+    },
+    {
+      "epoch": 0.8287085863912427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4997988536516817e-06,
+      "loss": 0.8742,
+      "step": 22787
+    },
+    {
+      "epoch": 0.8287449539949813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4991784737102955e-06,
+      "loss": 0.9787,
+      "step": 22788
+    },
+    {
+      "epoch": 0.8287813215987199,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4985582117057452e-06,
+      "loss": 0.8526,
+      "step": 22789
+    },
+    {
+      "epoch": 0.8288176892024585,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4979380676466404e-06,
+      "loss": 0.9192,
+      "step": 22790
+    },
+    {
+      "epoch": 0.8288540568061971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4973180415415812e-06,
+      "loss": 0.9878,
+      "step": 22791
+    },
+    {
+      "epoch": 0.8288904244099357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4966981333991714e-06,
+      "loss": 1.0372,
+      "step": 22792
+    },
+    {
+      "epoch": 0.8289267920136743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4960783432280067e-06,
+      "loss": 0.886,
+      "step": 22793
+    },
+    {
+      "epoch": 0.8289631596174128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4954586710366958e-06,
+      "loss": 0.783,
+      "step": 22794
+    },
+    {
+      "epoch": 0.8289995272211514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.494839116833824e-06,
+      "loss": 0.894,
+      "step": 22795
+    },
+    {
+      "epoch": 0.82903589482489,
+      "grad_norm": 0.0,
+      "learning_rate": 1.494219680627994e-06,
+      "loss": 0.9021,
+      "step": 22796
+    },
+    {
+      "epoch": 0.8290722624286285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4936003624277985e-06,
+      "loss": 0.9138,
+      "step": 22797
+    },
+    {
+      "epoch": 0.8291086300323671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.492981162241829e-06,
+      "loss": 0.8622,
+      "step": 22798
+    },
+    {
+      "epoch": 0.8291449976361057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.492362080078673e-06,
+      "loss": 0.937,
+      "step": 22799
+    },
+    {
+      "epoch": 0.8291813652398443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4917431159469276e-06,
+      "loss": 0.9036,
+      "step": 22800
+    },
+    {
+      "epoch": 0.8292177328435829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4911242698551709e-06,
+      "loss": 0.883,
+      "step": 22801
+    },
+    {
+      "epoch": 0.8292541004473215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.490505541811994e-06,
+      "loss": 0.8064,
+      "step": 22802
+    },
+    {
+      "epoch": 0.8292904680510601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.489886931825979e-06,
+      "loss": 0.9761,
+      "step": 22803
+    },
+    {
+      "epoch": 0.8293268356547987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4892684399057089e-06,
+      "loss": 0.9637,
+      "step": 22804
+    },
+    {
+      "epoch": 0.8293632032585373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4886500660597636e-06,
+      "loss": 1.1024,
+      "step": 22805
+    },
+    {
+      "epoch": 0.8293995708622759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4880318102967207e-06,
+      "loss": 0.9066,
+      "step": 22806
+    },
+    {
+      "epoch": 0.8294359384660145,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4874136726251643e-06,
+      "loss": 0.8982,
+      "step": 22807
+    },
+    {
+      "epoch": 0.8294723060697531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.486795653053661e-06,
+      "loss": 0.8814,
+      "step": 22808
+    },
+    {
+      "epoch": 0.8295086736734917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4861777515907904e-06,
+      "loss": 0.8716,
+      "step": 22809
+    },
+    {
+      "epoch": 0.8295450412772303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4855599682451248e-06,
+      "loss": 0.83,
+      "step": 22810
+    },
+    {
+      "epoch": 0.8295814088809689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4849423030252352e-06,
+      "loss": 0.913,
+      "step": 22811
+    },
+    {
+      "epoch": 0.8296177764847075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4843247559396857e-06,
+      "loss": 0.9309,
+      "step": 22812
+    },
+    {
+      "epoch": 0.8296541440884461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4837073269970537e-06,
+      "loss": 0.8824,
+      "step": 22813
+    },
+    {
+      "epoch": 0.8296905116921846,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4830900162058947e-06,
+      "loss": 0.862,
+      "step": 22814
+    },
+    {
+      "epoch": 0.8297268792959231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4824728235747809e-06,
+      "loss": 0.9447,
+      "step": 22815
+    },
+    {
+      "epoch": 0.8297632468996617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4818557491122709e-06,
+      "loss": 0.7272,
+      "step": 22816
+    },
+    {
+      "epoch": 0.8297996145034003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4812387928269267e-06,
+      "loss": 0.9891,
+      "step": 22817
+    },
+    {
+      "epoch": 0.8298359821071389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.480621954727306e-06,
+      "loss": 0.8164,
+      "step": 22818
+    },
+    {
+      "epoch": 0.8298723497108775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4800052348219728e-06,
+      "loss": 0.9669,
+      "step": 22819
+    },
+    {
+      "epoch": 0.8299087173146161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.479388633119474e-06,
+      "loss": 0.8866,
+      "step": 22820
+    },
+    {
+      "epoch": 0.8299450849183547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4787721496283701e-06,
+      "loss": 0.8868,
+      "step": 22821
+    },
+    {
+      "epoch": 0.8299814525220933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4781557843572137e-06,
+      "loss": 0.8725,
+      "step": 22822
+    },
+    {
+      "epoch": 0.8300178201258319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.477539537314554e-06,
+      "loss": 0.954,
+      "step": 22823
+    },
+    {
+      "epoch": 0.8300541877295705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4769234085089423e-06,
+      "loss": 0.8989,
+      "step": 22824
+    },
+    {
+      "epoch": 0.8300905553333091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.476307397948923e-06,
+      "loss": 1.0165,
+      "step": 22825
+    },
+    {
+      "epoch": 0.8301269229370477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.47569150564305e-06,
+      "loss": 0.9177,
+      "step": 22826
+    },
+    {
+      "epoch": 0.8301632905407863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4750757315998575e-06,
+      "loss": 0.9469,
+      "step": 22827
+    },
+    {
+      "epoch": 0.8301996581445249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4744600758278971e-06,
+      "loss": 0.8392,
+      "step": 22828
+    },
+    {
+      "epoch": 0.8302360257482635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4738445383357069e-06,
+      "loss": 1.0119,
+      "step": 22829
+    },
+    {
+      "epoch": 0.8302723933520021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.473229119131826e-06,
+      "loss": 0.9535,
+      "step": 22830
+    },
+    {
+      "epoch": 0.8303087609557406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4726138182247907e-06,
+      "loss": 0.9048,
+      "step": 22831
+    },
+    {
+      "epoch": 0.8303451285594792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4719986356231453e-06,
+      "loss": 0.9497,
+      "step": 22832
+    },
+    {
+      "epoch": 0.8303814961632178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.471383571335414e-06,
+      "loss": 0.9336,
+      "step": 22833
+    },
+    {
+      "epoch": 0.8304178637669564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4707686253701381e-06,
+      "loss": 0.9189,
+      "step": 22834
+    },
+    {
+      "epoch": 0.830454231370695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.470153797735846e-06,
+      "loss": 0.8367,
+      "step": 22835
+    },
+    {
+      "epoch": 0.8304905989744336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4695390884410676e-06,
+      "loss": 0.9391,
+      "step": 22836
+    },
+    {
+      "epoch": 0.8305269665781722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4689244974943285e-06,
+      "loss": 0.8718,
+      "step": 22837
+    },
+    {
+      "epoch": 0.8305633341819108,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4683100249041604e-06,
+      "loss": 0.878,
+      "step": 22838
+    },
+    {
+      "epoch": 0.8305997017856493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4676956706790868e-06,
+      "loss": 0.8129,
+      "step": 22839
+    },
+    {
+      "epoch": 0.8306360693893879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4670814348276286e-06,
+      "loss": 0.9625,
+      "step": 22840
+    },
+    {
+      "epoch": 0.8306724369931265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.46646731735831e-06,
+      "loss": 1.0308,
+      "step": 22841
+    },
+    {
+      "epoch": 0.8307088045968651,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4658533182796486e-06,
+      "loss": 0.885,
+      "step": 22842
+    },
+    {
+      "epoch": 0.8307451722006037,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4652394376001656e-06,
+      "loss": 0.9964,
+      "step": 22843
+    },
+    {
+      "epoch": 0.8307815398043423,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4646256753283726e-06,
+      "loss": 0.8802,
+      "step": 22844
+    },
+    {
+      "epoch": 0.8308179074080809,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4640120314727935e-06,
+      "loss": 1.0019,
+      "step": 22845
+    },
+    {
+      "epoch": 0.8308542750118195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4633985060419308e-06,
+      "loss": 0.8484,
+      "step": 22846
+    },
+    {
+      "epoch": 0.830890642615558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.462785099044306e-06,
+      "loss": 0.8863,
+      "step": 22847
+    },
+    {
+      "epoch": 0.8309270102192966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4621718104884242e-06,
+      "loss": 0.9089,
+      "step": 22848
+    },
+    {
+      "epoch": 0.8309633778230352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4615586403827954e-06,
+      "loss": 0.9185,
+      "step": 22849
+    },
+    {
+      "epoch": 0.8309997454267738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4609455887359226e-06,
+      "loss": 0.864,
+      "step": 22850
+    },
+    {
+      "epoch": 0.8310361130305124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4603326555563202e-06,
+      "loss": 0.9168,
+      "step": 22851
+    },
+    {
+      "epoch": 0.831072480634251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4597198408524814e-06,
+      "loss": 0.8097,
+      "step": 22852
+    },
+    {
+      "epoch": 0.8311088482379896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.459107144632914e-06,
+      "loss": 0.9073,
+      "step": 22853
+    },
+    {
+      "epoch": 0.8311452158417282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4584945669061178e-06,
+      "loss": 0.8768,
+      "step": 22854
+    },
+    {
+      "epoch": 0.8311815834454668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4578821076805904e-06,
+      "loss": 0.9644,
+      "step": 22855
+    },
+    {
+      "epoch": 0.8312179510492054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4572697669648273e-06,
+      "loss": 0.8661,
+      "step": 22856
+    },
+    {
+      "epoch": 0.831254318652944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.456657544767327e-06,
+      "loss": 0.9243,
+      "step": 22857
+    },
+    {
+      "epoch": 0.8312906862566826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.456045441096583e-06,
+      "loss": 0.9026,
+      "step": 22858
+    },
+    {
+      "epoch": 0.8313270538604212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.455433455961085e-06,
+      "loss": 1.002,
+      "step": 22859
+    },
+    {
+      "epoch": 0.8313634214641598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.454821589369324e-06,
+      "loss": 0.9119,
+      "step": 22860
+    },
+    {
+      "epoch": 0.8313997890678984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4542098413297923e-06,
+      "loss": 0.916,
+      "step": 22861
+    },
+    {
+      "epoch": 0.831436156671637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4535982118509728e-06,
+      "loss": 0.9042,
+      "step": 22862
+    },
+    {
+      "epoch": 0.8314725242753755,
+      "grad_norm": 0.0,
+      "learning_rate": 1.45298670094135e-06,
+      "loss": 0.93,
+      "step": 22863
+    },
+    {
+      "epoch": 0.831508891879114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4523753086094138e-06,
+      "loss": 0.9096,
+      "step": 22864
+    },
+    {
+      "epoch": 0.8315452594828526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4517640348636397e-06,
+      "loss": 0.9404,
+      "step": 22865
+    },
+    {
+      "epoch": 0.8315816270865912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4511528797125119e-06,
+      "loss": 0.8991,
+      "step": 22866
+    },
+    {
+      "epoch": 0.8316179946903298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.45054184316451e-06,
+      "loss": 0.9161,
+      "step": 22867
+    },
+    {
+      "epoch": 0.8316543622940684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.449930925228109e-06,
+      "loss": 0.9686,
+      "step": 22868
+    },
+    {
+      "epoch": 0.831690729897807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.449320125911784e-06,
+      "loss": 0.9248,
+      "step": 22869
+    },
+    {
+      "epoch": 0.8317270975015456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4487094452240125e-06,
+      "loss": 0.8726,
+      "step": 22870
+    },
+    {
+      "epoch": 0.8317634651052842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4480988831732634e-06,
+      "loss": 0.8646,
+      "step": 22871
+    },
+    {
+      "epoch": 0.8317998327090228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.44748843976801e-06,
+      "loss": 0.9724,
+      "step": 22872
+    },
+    {
+      "epoch": 0.8318362003127614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.44687811501672e-06,
+      "loss": 1.0231,
+      "step": 22873
+    },
+    {
+      "epoch": 0.8318725679165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.446267908927861e-06,
+      "loss": 0.9297,
+      "step": 22874
+    },
+    {
+      "epoch": 0.8319089355202386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.445657821509896e-06,
+      "loss": 1.0146,
+      "step": 22875
+    },
+    {
+      "epoch": 0.8319453031239772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.445047852771294e-06,
+      "loss": 0.9217,
+      "step": 22876
+    },
+    {
+      "epoch": 0.8319816707277158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4444380027205151e-06,
+      "loss": 0.8847,
+      "step": 22877
+    },
+    {
+      "epoch": 0.8320180383314544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.44382827136602e-06,
+      "loss": 0.9056,
+      "step": 22878
+    },
+    {
+      "epoch": 0.832054405935193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4432186587162667e-06,
+      "loss": 1.0088,
+      "step": 22879
+    },
+    {
+      "epoch": 0.8320907735389316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.442609164779718e-06,
+      "loss": 0.9681,
+      "step": 22880
+    },
+    {
+      "epoch": 0.8321271411426701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.441999789564823e-06,
+      "loss": 0.8782,
+      "step": 22881
+    },
+    {
+      "epoch": 0.8321635087464087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4413905330800371e-06,
+      "loss": 0.9634,
+      "step": 22882
+    },
+    {
+      "epoch": 0.8321998763501472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4407813953338167e-06,
+      "loss": 0.9435,
+      "step": 22883
+    },
+    {
+      "epoch": 0.8322362439538858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4401723763346097e-06,
+      "loss": 0.9395,
+      "step": 22884
+    },
+    {
+      "epoch": 0.8322726115576244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4395634760908672e-06,
+      "loss": 0.9513,
+      "step": 22885
+    },
+    {
+      "epoch": 0.832308979161363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4389546946110355e-06,
+      "loss": 0.8212,
+      "step": 22886
+    },
+    {
+      "epoch": 0.8323453467651016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4383460319035602e-06,
+      "loss": 0.964,
+      "step": 22887
+    },
+    {
+      "epoch": 0.8323817143688402,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4377374879768845e-06,
+      "loss": 0.902,
+      "step": 22888
+    },
+    {
+      "epoch": 0.8324180819725788,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4371290628394551e-06,
+      "loss": 0.9109,
+      "step": 22889
+    },
+    {
+      "epoch": 0.8324544495763174,
+      "grad_norm": 0.0,
+      "learning_rate": 1.436520756499712e-06,
+      "loss": 0.8437,
+      "step": 22890
+    },
+    {
+      "epoch": 0.832490817180056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4359125689660914e-06,
+      "loss": 0.9376,
+      "step": 22891
+    },
+    {
+      "epoch": 0.8325271847837946,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4353045002470324e-06,
+      "loss": 0.8397,
+      "step": 22892
+    },
+    {
+      "epoch": 0.8325635523875332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4346965503509757e-06,
+      "loss": 0.7774,
+      "step": 22893
+    },
+    {
+      "epoch": 0.8325999199912718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4340887192863473e-06,
+      "loss": 1.0835,
+      "step": 22894
+    },
+    {
+      "epoch": 0.8326362875950104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4334810070615856e-06,
+      "loss": 0.946,
+      "step": 22895
+    },
+    {
+      "epoch": 0.832672655198749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4328734136851218e-06,
+      "loss": 0.9304,
+      "step": 22896
+    },
+    {
+      "epoch": 0.8327090228024876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.432265939165385e-06,
+      "loss": 0.9204,
+      "step": 22897
+    },
+    {
+      "epoch": 0.8327453904062261,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4316585835107988e-06,
+      "loss": 0.8768,
+      "step": 22898
+    },
+    {
+      "epoch": 0.8327817580099647,
+      "grad_norm": 0.0,
+      "learning_rate": 1.431051346729797e-06,
+      "loss": 0.8486,
+      "step": 22899
+    },
+    {
+      "epoch": 0.8328181256137033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4304442288307996e-06,
+      "loss": 0.8669,
+      "step": 22900
+    },
+    {
+      "epoch": 0.8328544932174419,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4298372298222262e-06,
+      "loss": 0.9553,
+      "step": 22901
+    },
+    {
+      "epoch": 0.8328908608211805,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4292303497125037e-06,
+      "loss": 0.9227,
+      "step": 22902
+    },
+    {
+      "epoch": 0.8329272284249191,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4286235885100508e-06,
+      "loss": 0.8994,
+      "step": 22903
+    },
+    {
+      "epoch": 0.8329635960286577,
+      "grad_norm": 0.0,
+      "learning_rate": 1.428016946223284e-06,
+      "loss": 0.8942,
+      "step": 22904
+    },
+    {
+      "epoch": 0.8329999636323963,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4274104228606179e-06,
+      "loss": 0.8831,
+      "step": 22905
+    },
+    {
+      "epoch": 0.8330363312361349,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4268040184304732e-06,
+      "loss": 0.9364,
+      "step": 22906
+    },
+    {
+      "epoch": 0.8330726988398734,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4261977329412546e-06,
+      "loss": 0.9282,
+      "step": 22907
+    },
+    {
+      "epoch": 0.833109066443612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4255915664013808e-06,
+      "loss": 0.9913,
+      "step": 22908
+    },
+    {
+      "epoch": 0.8331454340473506,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4249855188192584e-06,
+      "loss": 0.7565,
+      "step": 22909
+    },
+    {
+      "epoch": 0.8331818016510892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4243795902032953e-06,
+      "loss": 0.8952,
+      "step": 22910
+    },
+    {
+      "epoch": 0.8332181692548278,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4237737805618956e-06,
+      "loss": 0.8655,
+      "step": 22911
+    },
+    {
+      "epoch": 0.8332545368585664,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4231680899034718e-06,
+      "loss": 0.9848,
+      "step": 22912
+    },
+    {
+      "epoch": 0.833290904462305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4225625182364168e-06,
+      "loss": 0.8578,
+      "step": 22913
+    },
+    {
+      "epoch": 0.8333272720660436,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4219570655691395e-06,
+      "loss": 0.8928,
+      "step": 22914
+    },
+    {
+      "epoch": 0.8333636396697821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4213517319100367e-06,
+      "loss": 0.8708,
+      "step": 22915
+    },
+    {
+      "epoch": 0.8334000072735207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4207465172675083e-06,
+      "loss": 0.9865,
+      "step": 22916
+    },
+    {
+      "epoch": 0.8334363748772593,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4201414216499466e-06,
+      "loss": 0.9834,
+      "step": 22917
+    },
+    {
+      "epoch": 0.8334727424809979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4195364450657534e-06,
+      "loss": 0.9239,
+      "step": 22918
+    },
+    {
+      "epoch": 0.8335091100847365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4189315875233166e-06,
+      "loss": 1.0072,
+      "step": 22919
+    },
+    {
+      "epoch": 0.8335454776884751,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4183268490310265e-06,
+      "loss": 0.8895,
+      "step": 22920
+    },
+    {
+      "epoch": 0.8335818452922137,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4177222295972781e-06,
+      "loss": 0.8537,
+      "step": 22921
+    },
+    {
+      "epoch": 0.8336182128959523,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4171177292304571e-06,
+      "loss": 0.9961,
+      "step": 22922
+    },
+    {
+      "epoch": 0.8336545804996909,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4165133479389493e-06,
+      "loss": 0.8296,
+      "step": 22923
+    },
+    {
+      "epoch": 0.8336909481034295,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4159090857311386e-06,
+      "loss": 0.8912,
+      "step": 22924
+    },
+    {
+      "epoch": 0.8337273157071681,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4153049426154153e-06,
+      "loss": 0.8575,
+      "step": 22925
+    },
+    {
+      "epoch": 0.8337636833109067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4147009186001503e-06,
+      "loss": 1.0274,
+      "step": 22926
+    },
+    {
+      "epoch": 0.8338000509146453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4140970136937327e-06,
+      "loss": 0.9515,
+      "step": 22927
+    },
+    {
+      "epoch": 0.8338364185183839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4134932279045355e-06,
+      "loss": 0.8912,
+      "step": 22928
+    },
+    {
+      "epoch": 0.8338727861221225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4128895612409387e-06,
+      "loss": 1.0851,
+      "step": 22929
+    },
+    {
+      "epoch": 0.833909153725861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4122860137113126e-06,
+      "loss": 0.969,
+      "step": 22930
+    },
+    {
+      "epoch": 0.8339455213295995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.411682585324038e-06,
+      "loss": 0.8436,
+      "step": 22931
+    },
+    {
+      "epoch": 0.8339818889333381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.411079276087478e-06,
+      "loss": 0.9638,
+      "step": 22932
+    },
+    {
+      "epoch": 0.8340182565370767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4104760860100098e-06,
+      "loss": 0.8865,
+      "step": 22933
+    },
+    {
+      "epoch": 0.8340546241408153,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4098730150999985e-06,
+      "loss": 0.9072,
+      "step": 22934
+    },
+    {
+      "epoch": 0.8340909917445539,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4092700633658108e-06,
+      "loss": 0.9477,
+      "step": 22935
+    },
+    {
+      "epoch": 0.8341273593482925,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4086672308158112e-06,
+      "loss": 0.9405,
+      "step": 22936
+    },
+    {
+      "epoch": 0.8341637269520311,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4080645174583652e-06,
+      "loss": 1.1198,
+      "step": 22937
+    },
+    {
+      "epoch": 0.8342000945557697,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4074619233018371e-06,
+      "loss": 0.9112,
+      "step": 22938
+    },
+    {
+      "epoch": 0.8342364621595083,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4068594483545783e-06,
+      "loss": 0.9069,
+      "step": 22939
+    },
+    {
+      "epoch": 0.8342728297632469,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4062570926249553e-06,
+      "loss": 0.9253,
+      "step": 22940
+    },
+    {
+      "epoch": 0.8343091973669855,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4056548561213223e-06,
+      "loss": 0.8491,
+      "step": 22941
+    },
+    {
+      "epoch": 0.8343455649707241,
+      "grad_norm": 0.0,
+      "learning_rate": 1.405052738852034e-06,
+      "loss": 0.7965,
+      "step": 22942
+    },
+    {
+      "epoch": 0.8343819325744627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4044507408254426e-06,
+      "loss": 0.9048,
+      "step": 22943
+    },
+    {
+      "epoch": 0.8344183001782013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4038488620499057e-06,
+      "loss": 0.8002,
+      "step": 22944
+    },
+    {
+      "epoch": 0.8344546677819399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4032471025337645e-06,
+      "loss": 0.9642,
+      "step": 22945
+    },
+    {
+      "epoch": 0.8344910353856785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4026454622853759e-06,
+      "loss": 0.8754,
+      "step": 22946
+    },
+    {
+      "epoch": 0.8345274029894171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4020439413130826e-06,
+      "loss": 0.9039,
+      "step": 22947
+    },
+    {
+      "epoch": 0.8345637705931556,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4014425396252307e-06,
+      "loss": 0.9678,
+      "step": 22948
+    },
+    {
+      "epoch": 0.8346001381968942,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4008412572301622e-06,
+      "loss": 1.0571,
+      "step": 22949
+    },
+    {
+      "epoch": 0.8346365058006328,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4002400941362227e-06,
+      "loss": 0.8954,
+      "step": 22950
+    },
+    {
+      "epoch": 0.8346728734043714,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3996390503517509e-06,
+      "loss": 0.8031,
+      "step": 22951
+    },
+    {
+      "epoch": 0.83470924100811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3990381258850838e-06,
+      "loss": 0.8915,
+      "step": 22952
+    },
+    {
+      "epoch": 0.8347456086118485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.39843732074456e-06,
+      "loss": 0.9833,
+      "step": 22953
+    },
+    {
+      "epoch": 0.8347819762155871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3978366349385154e-06,
+      "loss": 0.954,
+      "step": 22954
+    },
+    {
+      "epoch": 0.8348183438193257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3972360684752784e-06,
+      "loss": 0.7532,
+      "step": 22955
+    },
+    {
+      "epoch": 0.8348547114230643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3966356213631893e-06,
+      "loss": 0.9255,
+      "step": 22956
+    },
+    {
+      "epoch": 0.8348910790268029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.396035293610576e-06,
+      "loss": 0.9336,
+      "step": 22957
+    },
+    {
+      "epoch": 0.8349274466305415,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3954350852257602e-06,
+      "loss": 0.8438,
+      "step": 22958
+    },
+    {
+      "epoch": 0.8349638142342801,
+      "grad_norm": 0.0,
+      "learning_rate": 1.394834996217077e-06,
+      "loss": 0.9872,
+      "step": 22959
+    },
+    {
+      "epoch": 0.8350001818380187,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3942350265928495e-06,
+      "loss": 0.9105,
+      "step": 22960
+    },
+    {
+      "epoch": 0.8350365494417573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3936351763614008e-06,
+      "loss": 0.9179,
+      "step": 22961
+    },
+    {
+      "epoch": 0.8350729170454959,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3930354455310513e-06,
+      "loss": 0.7854,
+      "step": 22962
+    },
+    {
+      "epoch": 0.8351092846492345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3924358341101274e-06,
+      "loss": 0.8777,
+      "step": 22963
+    },
+    {
+      "epoch": 0.8351456522529731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3918363421069403e-06,
+      "loss": 1.0153,
+      "step": 22964
+    },
+    {
+      "epoch": 0.8351820198567116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3912369695298111e-06,
+      "loss": 1.0437,
+      "step": 22965
+    },
+    {
+      "epoch": 0.8352183874604502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3906377163870567e-06,
+      "loss": 0.8094,
+      "step": 22966
+    },
+    {
+      "epoch": 0.8352547550641888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3900385826869867e-06,
+      "loss": 0.9574,
+      "step": 22967
+    },
+    {
+      "epoch": 0.8352911226679274,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3894395684379147e-06,
+      "loss": 0.9014,
+      "step": 22968
+    },
+    {
+      "epoch": 0.835327490271666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.388840673648154e-06,
+      "loss": 0.8138,
+      "step": 22969
+    },
+    {
+      "epoch": 0.8353638578754046,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3882418983260103e-06,
+      "loss": 0.9032,
+      "step": 22970
+    },
+    {
+      "epoch": 0.8354002254791432,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3876432424797925e-06,
+      "loss": 0.9917,
+      "step": 22971
+    },
+    {
+      "epoch": 0.8354365930828818,
+      "grad_norm": 0.0,
+      "learning_rate": 1.387044706117806e-06,
+      "loss": 0.8875,
+      "step": 22972
+    },
+    {
+      "epoch": 0.8354729606866204,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3864462892483533e-06,
+      "loss": 0.961,
+      "step": 22973
+    },
+    {
+      "epoch": 0.835509328290359,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3858479918797353e-06,
+      "loss": 0.8311,
+      "step": 22974
+    },
+    {
+      "epoch": 0.8355456958940976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3852498140202563e-06,
+      "loss": 0.8876,
+      "step": 22975
+    },
+    {
+      "epoch": 0.8355820634978361,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3846517556782157e-06,
+      "loss": 1.0148,
+      "step": 22976
+    },
+    {
+      "epoch": 0.8356184311015747,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3840538168619045e-06,
+      "loss": 1.0231,
+      "step": 22977
+    },
+    {
+      "epoch": 0.8356547987053133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3834559975796235e-06,
+      "loss": 0.8783,
+      "step": 22978
+    },
+    {
+      "epoch": 0.8356911663090519,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3828582978396665e-06,
+      "loss": 0.9196,
+      "step": 22979
+    },
+    {
+      "epoch": 0.8357275339127905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3822607176503233e-06,
+      "loss": 0.9009,
+      "step": 22980
+    },
+    {
+      "epoch": 0.8357639015165291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.381663257019884e-06,
+      "loss": 0.9354,
+      "step": 22981
+    },
+    {
+      "epoch": 0.8358002691202676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3810659159566408e-06,
+      "loss": 0.7992,
+      "step": 22982
+    },
+    {
+      "epoch": 0.8358366367240062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3804686944688805e-06,
+      "loss": 1.0334,
+      "step": 22983
+    },
+    {
+      "epoch": 0.8358730043277448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3798715925648865e-06,
+      "loss": 0.8332,
+      "step": 22984
+    },
+    {
+      "epoch": 0.8359093719314834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3792746102529442e-06,
+      "loss": 0.9787,
+      "step": 22985
+    },
+    {
+      "epoch": 0.835945739535222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3786777475413359e-06,
+      "loss": 0.8743,
+      "step": 22986
+    },
+    {
+      "epoch": 0.8359821071389606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3780810044383397e-06,
+      "loss": 0.9527,
+      "step": 22987
+    },
+    {
+      "epoch": 0.8360184747426992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3774843809522376e-06,
+      "loss": 0.9668,
+      "step": 22988
+    },
+    {
+      "epoch": 0.8360548423464378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3768878770913075e-06,
+      "loss": 1.0176,
+      "step": 22989
+    },
+    {
+      "epoch": 0.8360912099501764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3762914928638227e-06,
+      "loss": 0.9246,
+      "step": 22990
+    },
+    {
+      "epoch": 0.836127577553915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3756952282780556e-06,
+      "loss": 0.9471,
+      "step": 22991
+    },
+    {
+      "epoch": 0.8361639451576536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.375099083342285e-06,
+      "loss": 0.9012,
+      "step": 22992
+    },
+    {
+      "epoch": 0.8362003127613922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3745030580647745e-06,
+      "loss": 0.9068,
+      "step": 22993
+    },
+    {
+      "epoch": 0.8362366803651308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3739071524537973e-06,
+      "loss": 0.9482,
+      "step": 22994
+    },
+    {
+      "epoch": 0.8362730479688694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3733113665176222e-06,
+      "loss": 0.9446,
+      "step": 22995
+    },
+    {
+      "epoch": 0.836309415572608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3727157002645074e-06,
+      "loss": 0.8623,
+      "step": 22996
+    },
+    {
+      "epoch": 0.8363457831763466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.372120153702724e-06,
+      "loss": 0.8962,
+      "step": 22997
+    },
+    {
+      "epoch": 0.8363821507800852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3715247268405318e-06,
+      "loss": 0.9391,
+      "step": 22998
+    },
+    {
+      "epoch": 0.8364185183838236,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3709294196861922e-06,
+      "loss": 0.9744,
+      "step": 22999
+    },
+    {
+      "epoch": 0.8364548859875622,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3703342322479618e-06,
+      "loss": 0.7251,
+      "step": 23000
+    },
+    {
+      "epoch": 0.8364912535913008,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3697391645341019e-06,
+      "loss": 0.9517,
+      "step": 23001
+    },
+    {
+      "epoch": 0.8365276211950394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3691442165528667e-06,
+      "loss": 1.017,
+      "step": 23002
+    },
+    {
+      "epoch": 0.836563988798778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.368549388312509e-06,
+      "loss": 0.9916,
+      "step": 23003
+    },
+    {
+      "epoch": 0.8366003564025166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3679546798212807e-06,
+      "loss": 0.9095,
+      "step": 23004
+    },
+    {
+      "epoch": 0.8366367240062552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3673600910874386e-06,
+      "loss": 0.9751,
+      "step": 23005
+    },
+    {
+      "epoch": 0.8366730916099938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3667656221192226e-06,
+      "loss": 0.8304,
+      "step": 23006
+    },
+    {
+      "epoch": 0.8367094592137324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3661712729248866e-06,
+      "loss": 0.9139,
+      "step": 23007
+    },
+    {
+      "epoch": 0.836745826817471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3655770435126747e-06,
+      "loss": 1.0023,
+      "step": 23008
+    },
+    {
+      "epoch": 0.8367821944212096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3649829338908306e-06,
+      "loss": 0.9219,
+      "step": 23009
+    },
+    {
+      "epoch": 0.8368185620249482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.364388944067594e-06,
+      "loss": 0.8033,
+      "step": 23010
+    },
+    {
+      "epoch": 0.8368549296286868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.363795074051213e-06,
+      "loss": 0.9249,
+      "step": 23011
+    },
+    {
+      "epoch": 0.8368912972324254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3632013238499175e-06,
+      "loss": 0.7814,
+      "step": 23012
+    },
+    {
+      "epoch": 0.836927664836164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3626076934719534e-06,
+      "loss": 1.0354,
+      "step": 23013
+    },
+    {
+      "epoch": 0.8369640324399026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3620141829255506e-06,
+      "loss": 1.0538,
+      "step": 23014
+    },
+    {
+      "epoch": 0.8370004000436412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.361420792218947e-06,
+      "loss": 0.9339,
+      "step": 23015
+    },
+    {
+      "epoch": 0.8370367676473797,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3608275213603729e-06,
+      "loss": 0.9382,
+      "step": 23016
+    },
+    {
+      "epoch": 0.8370731352511183,
+      "grad_norm": 0.0,
+      "learning_rate": 1.360234370358059e-06,
+      "loss": 1.029,
+      "step": 23017
+    },
+    {
+      "epoch": 0.8371095028548569,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3596413392202358e-06,
+      "loss": 0.8478,
+      "step": 23018
+    },
+    {
+      "epoch": 0.8371458704585955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3590484279551275e-06,
+      "loss": 1.059,
+      "step": 23019
+    },
+    {
+      "epoch": 0.837182238062334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3584556365709644e-06,
+      "loss": 0.8672,
+      "step": 23020
+    },
+    {
+      "epoch": 0.8372186056660726,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3578629650759678e-06,
+      "loss": 0.9879,
+      "step": 23021
+    },
+    {
+      "epoch": 0.8372549732698112,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3572704134783621e-06,
+      "loss": 0.7958,
+      "step": 23022
+    },
+    {
+      "epoch": 0.8372913408735498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.356677981786364e-06,
+      "loss": 0.8867,
+      "step": 23023
+    },
+    {
+      "epoch": 0.8373277084772884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3560856700082003e-06,
+      "loss": 0.9437,
+      "step": 23024
+    },
+    {
+      "epoch": 0.837364076081027,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3554934781520801e-06,
+      "loss": 0.9369,
+      "step": 23025
+    },
+    {
+      "epoch": 0.8374004436847656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3549014062262245e-06,
+      "loss": 0.8559,
+      "step": 23026
+    },
+    {
+      "epoch": 0.8374368112885042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3543094542388469e-06,
+      "loss": 0.9097,
+      "step": 23027
+    },
+    {
+      "epoch": 0.8374731788922428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3537176221981575e-06,
+      "loss": 0.9029,
+      "step": 23028
+    },
+    {
+      "epoch": 0.8375095464959814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3531259101123674e-06,
+      "loss": 0.9141,
+      "step": 23029
+    },
+    {
+      "epoch": 0.83754591409972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3525343179896911e-06,
+      "loss": 0.9431,
+      "step": 23030
+    },
+    {
+      "epoch": 0.8375822817034586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3519428458383276e-06,
+      "loss": 0.9884,
+      "step": 23031
+    },
+    {
+      "epoch": 0.8376186493071971,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3513514936664896e-06,
+      "loss": 0.9312,
+      "step": 23032
+    },
+    {
+      "epoch": 0.8376550169109357,
+      "grad_norm": 0.0,
+      "learning_rate": 1.350760261482379e-06,
+      "loss": 0.9294,
+      "step": 23033
+    },
+    {
+      "epoch": 0.8376913845146743,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3501691492941983e-06,
+      "loss": 1.0121,
+      "step": 23034
+    },
+    {
+      "epoch": 0.8377277521184129,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3495781571101473e-06,
+      "loss": 0.8583,
+      "step": 23035
+    },
+    {
+      "epoch": 0.8377641197221515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3489872849384233e-06,
+      "loss": 0.9908,
+      "step": 23036
+    },
+    {
+      "epoch": 0.8378004873258901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3483965327872318e-06,
+      "loss": 0.9514,
+      "step": 23037
+    },
+    {
+      "epoch": 0.8378368549296287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3478059006647593e-06,
+      "loss": 0.9052,
+      "step": 23038
+    },
+    {
+      "epoch": 0.8378732225333673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3472153885792061e-06,
+      "loss": 0.9563,
+      "step": 23039
+    },
+    {
+      "epoch": 0.8379095901371059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3466249965387623e-06,
+      "loss": 0.9544,
+      "step": 23040
+    },
+    {
+      "epoch": 0.8379459577408445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3460347245516193e-06,
+      "loss": 0.996,
+      "step": 23041
+    },
+    {
+      "epoch": 0.8379823253445831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3454445726259645e-06,
+      "loss": 1.0261,
+      "step": 23042
+    },
+    {
+      "epoch": 0.8380186929483217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.344854540769991e-06,
+      "loss": 0.9747,
+      "step": 23043
+    },
+    {
+      "epoch": 0.8380550605520602,
+      "grad_norm": 0.0,
+      "learning_rate": 1.344264628991876e-06,
+      "loss": 0.9684,
+      "step": 23044
+    },
+    {
+      "epoch": 0.8380914281557988,
+      "grad_norm": 0.0,
+      "learning_rate": 1.343674837299811e-06,
+      "loss": 0.9391,
+      "step": 23045
+    },
+    {
+      "epoch": 0.8381277957595374,
+      "grad_norm": 0.0,
+      "learning_rate": 1.343085165701975e-06,
+      "loss": 0.9611,
+      "step": 23046
+    },
+    {
+      "epoch": 0.838164163363276,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3424956142065503e-06,
+      "loss": 1.0245,
+      "step": 23047
+    },
+    {
+      "epoch": 0.8382005309670146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3419061828217139e-06,
+      "loss": 0.8921,
+      "step": 23048
+    },
+    {
+      "epoch": 0.8382368985707531,
+      "grad_norm": 0.0,
+      "learning_rate": 1.341316871555647e-06,
+      "loss": 0.8488,
+      "step": 23049
+    },
+    {
+      "epoch": 0.8382732661744917,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3407276804165238e-06,
+      "loss": 0.9524,
+      "step": 23050
+    },
+    {
+      "epoch": 0.8383096337782303,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3401386094125168e-06,
+      "loss": 0.8004,
+      "step": 23051
+    },
+    {
+      "epoch": 0.8383460013819689,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3395496585518009e-06,
+      "loss": 0.7912,
+      "step": 23052
+    },
+    {
+      "epoch": 0.8383823689857075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.338960827842546e-06,
+      "loss": 0.8341,
+      "step": 23053
+    },
+    {
+      "epoch": 0.8384187365894461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3383721172929209e-06,
+      "loss": 0.8041,
+      "step": 23054
+    },
+    {
+      "epoch": 0.8384551041931847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3377835269110905e-06,
+      "loss": 0.9343,
+      "step": 23055
+    },
+    {
+      "epoch": 0.8384914717969233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3371950567052294e-06,
+      "loss": 0.941,
+      "step": 23056
+    },
+    {
+      "epoch": 0.8385278394006619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3366067066834908e-06,
+      "loss": 0.8495,
+      "step": 23057
+    },
+    {
+      "epoch": 0.8385642070044005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3360184768540442e-06,
+      "loss": 0.8647,
+      "step": 23058
+    },
+    {
+      "epoch": 0.8386005746081391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3354303672250491e-06,
+      "loss": 0.9707,
+      "step": 23059
+    },
+    {
+      "epoch": 0.8386369422118777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3348423778046648e-06,
+      "loss": 1.0163,
+      "step": 23060
+    },
+    {
+      "epoch": 0.8386733098156163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.334254508601045e-06,
+      "loss": 0.9175,
+      "step": 23061
+    },
+    {
+      "epoch": 0.8387096774193549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3336667596223541e-06,
+      "loss": 0.8451,
+      "step": 23062
+    },
+    {
+      "epoch": 0.8387460450230935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3330791308767366e-06,
+      "loss": 1.0254,
+      "step": 23063
+    },
+    {
+      "epoch": 0.8387824126268321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3324916223723504e-06,
+      "loss": 0.8349,
+      "step": 23064
+    },
+    {
+      "epoch": 0.8388187802305707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3319042341173471e-06,
+      "loss": 0.908,
+      "step": 23065
+    },
+    {
+      "epoch": 0.8388551478343091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3313169661198734e-06,
+      "loss": 1.023,
+      "step": 23066
+    },
+    {
+      "epoch": 0.8388915154380477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3307298183880745e-06,
+      "loss": 0.8907,
+      "step": 23067
+    },
+    {
+      "epoch": 0.8389278830417863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3301427909301023e-06,
+      "loss": 0.8373,
+      "step": 23068
+    },
+    {
+      "epoch": 0.8389642506455249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3295558837540979e-06,
+      "loss": 0.9457,
+      "step": 23069
+    },
+    {
+      "epoch": 0.8390006182492635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3289690968682035e-06,
+      "loss": 0.8739,
+      "step": 23070
+    },
+    {
+      "epoch": 0.8390369858530021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3283824302805602e-06,
+      "loss": 0.7495,
+      "step": 23071
+    },
+    {
+      "epoch": 0.8390733534567407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3277958839993077e-06,
+      "loss": 0.9086,
+      "step": 23072
+    },
+    {
+      "epoch": 0.8391097210604793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3272094580325834e-06,
+      "loss": 0.9248,
+      "step": 23073
+    },
+    {
+      "epoch": 0.8391460886642179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3266231523885199e-06,
+      "loss": 0.8901,
+      "step": 23074
+    },
+    {
+      "epoch": 0.8391824562679565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3260369670752593e-06,
+      "loss": 0.9885,
+      "step": 23075
+    },
+    {
+      "epoch": 0.8392188238716951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3254509021009242e-06,
+      "loss": 0.8736,
+      "step": 23076
+    },
+    {
+      "epoch": 0.8392551914754337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3248649574736528e-06,
+      "loss": 0.9496,
+      "step": 23077
+    },
+    {
+      "epoch": 0.8392915590791723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3242791332015714e-06,
+      "loss": 0.89,
+      "step": 23078
+    },
+    {
+      "epoch": 0.8393279266829109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3236934292928071e-06,
+      "loss": 0.9368,
+      "step": 23079
+    },
+    {
+      "epoch": 0.8393642942866495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3231078457554857e-06,
+      "loss": 0.9004,
+      "step": 23080
+    },
+    {
+      "epoch": 0.8394006618903881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3225223825977329e-06,
+      "loss": 0.8862,
+      "step": 23081
+    },
+    {
+      "epoch": 0.8394370294941267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.321937039827671e-06,
+      "loss": 0.8505,
+      "step": 23082
+    },
+    {
+      "epoch": 0.8394733970978652,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3213518174534201e-06,
+      "loss": 0.9921,
+      "step": 23083
+    },
+    {
+      "epoch": 0.8395097647016038,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3207667154830995e-06,
+      "loss": 1.0599,
+      "step": 23084
+    },
+    {
+      "epoch": 0.8395461323053424,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3201817339248257e-06,
+      "loss": 0.89,
+      "step": 23085
+    },
+    {
+      "epoch": 0.839582499909081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3195968727867137e-06,
+      "loss": 0.8644,
+      "step": 23086
+    },
+    {
+      "epoch": 0.8396188675128196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3190121320768811e-06,
+      "loss": 0.9513,
+      "step": 23087
+    },
+    {
+      "epoch": 0.8396552351165582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3184275118034394e-06,
+      "loss": 0.88,
+      "step": 23088
+    },
+    {
+      "epoch": 0.8396916027202967,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3178430119744978e-06,
+      "loss": 0.8591,
+      "step": 23089
+    },
+    {
+      "epoch": 0.8397279703240353,
+      "grad_norm": 0.0,
+      "learning_rate": 1.317258632598165e-06,
+      "loss": 0.8861,
+      "step": 23090
+    },
+    {
+      "epoch": 0.8397643379277739,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3166743736825516e-06,
+      "loss": 0.9778,
+      "step": 23091
+    },
+    {
+      "epoch": 0.8398007055315125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3160902352357596e-06,
+      "loss": 0.7955,
+      "step": 23092
+    },
+    {
+      "epoch": 0.8398370731352511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3155062172658927e-06,
+      "loss": 0.9291,
+      "step": 23093
+    },
+    {
+      "epoch": 0.8398734407389897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3149223197810578e-06,
+      "loss": 0.9546,
+      "step": 23094
+    },
+    {
+      "epoch": 0.8399098083427283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.314338542789354e-06,
+      "loss": 0.8533,
+      "step": 23095
+    },
+    {
+      "epoch": 0.8399461759464669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.313754886298878e-06,
+      "loss": 0.9061,
+      "step": 23096
+    },
+    {
+      "epoch": 0.8399825435502055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3131713503177301e-06,
+      "loss": 0.8751,
+      "step": 23097
+    },
+    {
+      "epoch": 0.8400189111539441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3125879348540038e-06,
+      "loss": 0.9664,
+      "step": 23098
+    },
+    {
+      "epoch": 0.8400552787576827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3120046399157926e-06,
+      "loss": 0.9855,
+      "step": 23099
+    },
+    {
+      "epoch": 0.8400916463614212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3114214655111923e-06,
+      "loss": 0.9159,
+      "step": 23100
+    },
+    {
+      "epoch": 0.8401280139651598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.310838411648292e-06,
+      "loss": 0.8982,
+      "step": 23101
+    },
+    {
+      "epoch": 0.8401643815688984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3102554783351794e-06,
+      "loss": 0.902,
+      "step": 23102
+    },
+    {
+      "epoch": 0.840200749172637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3096726655799418e-06,
+      "loss": 0.8985,
+      "step": 23103
+    },
+    {
+      "epoch": 0.8402371167763756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3090899733906703e-06,
+      "loss": 0.923,
+      "step": 23104
+    },
+    {
+      "epoch": 0.8402734843801142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3085074017754395e-06,
+      "loss": 0.8238,
+      "step": 23105
+    },
+    {
+      "epoch": 0.8403098519838528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3079249507423398e-06,
+      "loss": 0.8974,
+      "step": 23106
+    },
+    {
+      "epoch": 0.8403462195875914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3073426202994488e-06,
+      "loss": 0.9709,
+      "step": 23107
+    },
+    {
+      "epoch": 0.84038258719133,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3067604104548447e-06,
+      "loss": 0.8956,
+      "step": 23108
+    },
+    {
+      "epoch": 0.8404189547950686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3061783212166047e-06,
+      "loss": 0.8716,
+      "step": 23109
+    },
+    {
+      "epoch": 0.8404553223988072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3055963525928095e-06,
+      "loss": 0.9351,
+      "step": 23110
+    },
+    {
+      "epoch": 0.8404916900025458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3050145045915275e-06,
+      "loss": 0.8971,
+      "step": 23111
+    },
+    {
+      "epoch": 0.8405280576062844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.30443277722083e-06,
+      "loss": 0.9964,
+      "step": 23112
+    },
+    {
+      "epoch": 0.840564425210023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3038511704887935e-06,
+      "loss": 0.9647,
+      "step": 23113
+    },
+    {
+      "epoch": 0.8406007928137615,
+      "grad_norm": 0.0,
+      "learning_rate": 1.303269684403482e-06,
+      "loss": 0.8642,
+      "step": 23114
+    },
+    {
+      "epoch": 0.8406371604175001,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3026883189729656e-06,
+      "loss": 0.9938,
+      "step": 23115
+    },
+    {
+      "epoch": 0.8406735280212386,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3021070742053077e-06,
+      "loss": 0.9177,
+      "step": 23116
+    },
+    {
+      "epoch": 0.8407098956249772,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3015259501085763e-06,
+      "loss": 0.9885,
+      "step": 23117
+    },
+    {
+      "epoch": 0.8407462632287158,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3009449466908264e-06,
+      "loss": 0.9588,
+      "step": 23118
+    },
+    {
+      "epoch": 0.8407826308324544,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3003640639601267e-06,
+      "loss": 0.9281,
+      "step": 23119
+    },
+    {
+      "epoch": 0.840818998436193,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2997833019245309e-06,
+      "loss": 0.9801,
+      "step": 23120
+    },
+    {
+      "epoch": 0.8408553660399316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2992026605920982e-06,
+      "loss": 0.9442,
+      "step": 23121
+    },
+    {
+      "epoch": 0.8408917336436702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.298622139970882e-06,
+      "loss": 0.9093,
+      "step": 23122
+    },
+    {
+      "epoch": 0.8409281012474088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2980417400689427e-06,
+      "loss": 0.8481,
+      "step": 23123
+    },
+    {
+      "epoch": 0.8409644688511474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2974614608943226e-06,
+      "loss": 0.8403,
+      "step": 23124
+    },
+    {
+      "epoch": 0.841000836454886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2968813024550796e-06,
+      "loss": 0.9978,
+      "step": 23125
+    },
+    {
+      "epoch": 0.8410372040586246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.296301264759261e-06,
+      "loss": 1.0087,
+      "step": 23126
+    },
+    {
+      "epoch": 0.8410735716623632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2957213478149132e-06,
+      "loss": 0.826,
+      "step": 23127
+    },
+    {
+      "epoch": 0.8411099392661018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2951415516300792e-06,
+      "loss": 0.9409,
+      "step": 23128
+    },
+    {
+      "epoch": 0.8411463068698404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.294561876212811e-06,
+      "loss": 0.9481,
+      "step": 23129
+    },
+    {
+      "epoch": 0.841182674473579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2939823215711422e-06,
+      "loss": 0.8916,
+      "step": 23130
+    },
+    {
+      "epoch": 0.8412190420773176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2934028877131145e-06,
+      "loss": 0.8439,
+      "step": 23131
+    },
+    {
+      "epoch": 0.8412554096810562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.292823574646771e-06,
+      "loss": 0.8973,
+      "step": 23132
+    },
+    {
+      "epoch": 0.8412917772847947,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2922443823801466e-06,
+      "loss": 0.8811,
+      "step": 23133
+    },
+    {
+      "epoch": 0.8413281448885332,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2916653109212773e-06,
+      "loss": 0.9781,
+      "step": 23134
+    },
+    {
+      "epoch": 0.8413645124922718,
+      "grad_norm": 0.0,
+      "learning_rate": 1.291086360278193e-06,
+      "loss": 0.9652,
+      "step": 23135
+    },
+    {
+      "epoch": 0.8414008800960104,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2905075304589343e-06,
+      "loss": 0.8752,
+      "step": 23136
+    },
+    {
+      "epoch": 0.841437247699749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2899288214715212e-06,
+      "loss": 0.867,
+      "step": 23137
+    },
+    {
+      "epoch": 0.8414736153034876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2893502333239904e-06,
+      "loss": 0.8887,
+      "step": 23138
+    },
+    {
+      "epoch": 0.8415099829072262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2887717660243671e-06,
+      "loss": 0.9858,
+      "step": 23139
+    },
+    {
+      "epoch": 0.8415463505109648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2881934195806745e-06,
+      "loss": 0.8834,
+      "step": 23140
+    },
+    {
+      "epoch": 0.8415827181147034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2876151940009362e-06,
+      "loss": 0.9004,
+      "step": 23141
+    },
+    {
+      "epoch": 0.841619085718442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2870370892931804e-06,
+      "loss": 0.8997,
+      "step": 23142
+    },
+    {
+      "epoch": 0.8416554533221806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2864591054654184e-06,
+      "loss": 0.958,
+      "step": 23143
+    },
+    {
+      "epoch": 0.8416918209259192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2858812425256762e-06,
+      "loss": 1.0452,
+      "step": 23144
+    },
+    {
+      "epoch": 0.8417281885296578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.285303500481968e-06,
+      "loss": 0.9106,
+      "step": 23145
+    },
+    {
+      "epoch": 0.8417645561333964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.284725879342308e-06,
+      "loss": 1.0282,
+      "step": 23146
+    },
+    {
+      "epoch": 0.841800923737135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2841483791147102e-06,
+      "loss": 1.0447,
+      "step": 23147
+    },
+    {
+      "epoch": 0.8418372913408736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2835709998071888e-06,
+      "loss": 0.8153,
+      "step": 23148
+    },
+    {
+      "epoch": 0.8418736589446122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.282993741427755e-06,
+      "loss": 0.8437,
+      "step": 23149
+    },
+    {
+      "epoch": 0.8419100265483507,
+      "grad_norm": 0.0,
+      "learning_rate": 1.282416603984411e-06,
+      "loss": 0.9419,
+      "step": 23150
+    },
+    {
+      "epoch": 0.8419463941520893,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2818395874851708e-06,
+      "loss": 0.944,
+      "step": 23151
+    },
+    {
+      "epoch": 0.8419827617558279,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2812626919380356e-06,
+      "loss": 0.9858,
+      "step": 23152
+    },
+    {
+      "epoch": 0.8420191293595665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.280685917351011e-06,
+      "loss": 0.9724,
+      "step": 23153
+    },
+    {
+      "epoch": 0.8420554969633051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2801092637320966e-06,
+      "loss": 0.9533,
+      "step": 23154
+    },
+    {
+      "epoch": 0.8420918645670437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.279532731089298e-06,
+      "loss": 0.9428,
+      "step": 23155
+    },
+    {
+      "epoch": 0.8421282321707823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2789563194306053e-06,
+      "loss": 0.8257,
+      "step": 23156
+    },
+    {
+      "epoch": 0.8421645997745209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2783800287640225e-06,
+      "loss": 0.8642,
+      "step": 23157
+    },
+    {
+      "epoch": 0.8422009673782594,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2778038590975428e-06,
+      "loss": 0.9857,
+      "step": 23158
+    },
+    {
+      "epoch": 0.842237334981998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.277227810439159e-06,
+      "loss": 0.8204,
+      "step": 23159
+    },
+    {
+      "epoch": 0.8422737025857366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.276651882796862e-06,
+      "loss": 1.0568,
+      "step": 23160
+    },
+    {
+      "epoch": 0.8423100701894752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2760760761786462e-06,
+      "loss": 0.8064,
+      "step": 23161
+    },
+    {
+      "epoch": 0.8423464377932138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2755003905924945e-06,
+      "loss": 0.8807,
+      "step": 23162
+    },
+    {
+      "epoch": 0.8423828053969524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2749248260463976e-06,
+      "loss": 0.8633,
+      "step": 23163
+    },
+    {
+      "epoch": 0.842419173000691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.27434938254834e-06,
+      "loss": 0.8585,
+      "step": 23164
+    },
+    {
+      "epoch": 0.8424555406044296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2737740601063042e-06,
+      "loss": 1.0058,
+      "step": 23165
+    },
+    {
+      "epoch": 0.8424919082081682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2731988587282707e-06,
+      "loss": 0.9129,
+      "step": 23166
+    },
+    {
+      "epoch": 0.8425282758119067,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2726237784222228e-06,
+      "loss": 0.8583,
+      "step": 23167
+    },
+    {
+      "epoch": 0.8425646434156453,
+      "grad_norm": 0.0,
+      "learning_rate": 1.27204881919614e-06,
+      "loss": 0.8157,
+      "step": 23168
+    },
+    {
+      "epoch": 0.8426010110193839,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2714739810579923e-06,
+      "loss": 0.9983,
+      "step": 23169
+    },
+    {
+      "epoch": 0.8426373786231225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2708992640157602e-06,
+      "loss": 0.9403,
+      "step": 23170
+    },
+    {
+      "epoch": 0.8426737462268611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2703246680774174e-06,
+      "loss": 0.9615,
+      "step": 23171
+    },
+    {
+      "epoch": 0.8427101138305997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2697501932509326e-06,
+      "loss": 0.9719,
+      "step": 23172
+    },
+    {
+      "epoch": 0.8427464814343383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2691758395442754e-06,
+      "loss": 0.8986,
+      "step": 23173
+    },
+    {
+      "epoch": 0.8427828490380769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2686016069654216e-06,
+      "loss": 0.8464,
+      "step": 23174
+    },
+    {
+      "epoch": 0.8428192166418155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.268027495522327e-06,
+      "loss": 0.9479,
+      "step": 23175
+    },
+    {
+      "epoch": 0.8428555842455541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2674535052229642e-06,
+      "loss": 0.8877,
+      "step": 23176
+    },
+    {
+      "epoch": 0.8428919518492927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2668796360752933e-06,
+      "loss": 0.8466,
+      "step": 23177
+    },
+    {
+      "epoch": 0.8429283194530313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.266305888087278e-06,
+      "loss": 0.774,
+      "step": 23178
+    },
+    {
+      "epoch": 0.8429646870567699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2657322612668743e-06,
+      "loss": 0.9655,
+      "step": 23179
+    },
+    {
+      "epoch": 0.8430010546605085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2651587556220447e-06,
+      "loss": 0.9703,
+      "step": 23180
+    },
+    {
+      "epoch": 0.843037422264247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2645853711607447e-06,
+      "loss": 0.9285,
+      "step": 23181
+    },
+    {
+      "epoch": 0.8430737898679856,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2640121078909294e-06,
+      "loss": 1.0198,
+      "step": 23182
+    },
+    {
+      "epoch": 0.8431101574717242,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2634389658205503e-06,
+      "loss": 0.9187,
+      "step": 23183
+    },
+    {
+      "epoch": 0.8431465250754627,
+      "grad_norm": 0.0,
+      "learning_rate": 1.262865944957561e-06,
+      "loss": 0.8247,
+      "step": 23184
+    },
+    {
+      "epoch": 0.8431828926792013,
+      "grad_norm": 0.0,
+      "learning_rate": 1.262293045309908e-06,
+      "loss": 0.9065,
+      "step": 23185
+    },
+    {
+      "epoch": 0.8432192602829399,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2617202668855433e-06,
+      "loss": 0.9878,
+      "step": 23186
+    },
+    {
+      "epoch": 0.8432556278866785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2611476096924146e-06,
+      "loss": 1.0389,
+      "step": 23187
+    },
+    {
+      "epoch": 0.8432919954904171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2605750737384591e-06,
+      "loss": 0.8783,
+      "step": 23188
+    },
+    {
+      "epoch": 0.8433283630941557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2600026590316282e-06,
+      "loss": 0.9471,
+      "step": 23189
+    },
+    {
+      "epoch": 0.8433647306978943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.259430365579859e-06,
+      "loss": 0.7717,
+      "step": 23190
+    },
+    {
+      "epoch": 0.8434010983016329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2588581933910927e-06,
+      "loss": 0.8519,
+      "step": 23191
+    },
+    {
+      "epoch": 0.8434374659053715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.258286142473264e-06,
+      "loss": 0.9563,
+      "step": 23192
+    },
+    {
+      "epoch": 0.8434738335091101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2577142128343155e-06,
+      "loss": 0.8468,
+      "step": 23193
+    },
+    {
+      "epoch": 0.8435102011128487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2571424044821778e-06,
+      "loss": 0.9157,
+      "step": 23194
+    },
+    {
+      "epoch": 0.8435465687165873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2565707174247855e-06,
+      "loss": 0.8177,
+      "step": 23195
+    },
+    {
+      "epoch": 0.8435829363203259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2559991516700676e-06,
+      "loss": 0.9391,
+      "step": 23196
+    },
+    {
+      "epoch": 0.8436193039240645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.255427707225957e-06,
+      "loss": 0.8932,
+      "step": 23197
+    },
+    {
+      "epoch": 0.8436556715278031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.254856384100377e-06,
+      "loss": 0.8546,
+      "step": 23198
+    },
+    {
+      "epoch": 0.8436920391315417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2542851823012592e-06,
+      "loss": 0.9555,
+      "step": 23199
+    },
+    {
+      "epoch": 0.8437284067352803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2537141018365261e-06,
+      "loss": 0.7985,
+      "step": 23200
+    },
+    {
+      "epoch": 0.8437647743390188,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2531431427141006e-06,
+      "loss": 0.9198,
+      "step": 23201
+    },
+    {
+      "epoch": 0.8438011419427573,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2525723049419015e-06,
+      "loss": 0.9496,
+      "step": 23202
+    },
+    {
+      "epoch": 0.843837509546496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2520015885278547e-06,
+      "loss": 0.9626,
+      "step": 23203
+    },
+    {
+      "epoch": 0.8438738771502345,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2514309934798707e-06,
+      "loss": 0.8622,
+      "step": 23204
+    },
+    {
+      "epoch": 0.8439102447539731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.250860519805871e-06,
+      "loss": 0.9414,
+      "step": 23205
+    },
+    {
+      "epoch": 0.8439466123577117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.250290167513769e-06,
+      "loss": 0.975,
+      "step": 23206
+    },
+    {
+      "epoch": 0.8439829799614503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2497199366114743e-06,
+      "loss": 0.8654,
+      "step": 23207
+    },
+    {
+      "epoch": 0.8440193475651889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2491498271069013e-06,
+      "loss": 0.7538,
+      "step": 23208
+    },
+    {
+      "epoch": 0.8440557151689275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2485798390079596e-06,
+      "loss": 0.9517,
+      "step": 23209
+    },
+    {
+      "epoch": 0.8440920827726661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2480099723225547e-06,
+      "loss": 0.9236,
+      "step": 23210
+    },
+    {
+      "epoch": 0.8441284503764047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.247440227058593e-06,
+      "loss": 0.9086,
+      "step": 23211
+    },
+    {
+      "epoch": 0.8441648179801433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2468706032239808e-06,
+      "loss": 0.9033,
+      "step": 23212
+    },
+    {
+      "epoch": 0.8442011855838819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2463011008266214e-06,
+      "loss": 0.8753,
+      "step": 23213
+    },
+    {
+      "epoch": 0.8442375531876205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2457317198744135e-06,
+      "loss": 0.8364,
+      "step": 23214
+    },
+    {
+      "epoch": 0.8442739207913591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2451624603752554e-06,
+      "loss": 0.8749,
+      "step": 23215
+    },
+    {
+      "epoch": 0.8443102883950977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.244593322337051e-06,
+      "loss": 0.9358,
+      "step": 23216
+    },
+    {
+      "epoch": 0.8443466559988362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.244024305767687e-06,
+      "loss": 0.9117,
+      "step": 23217
+    },
+    {
+      "epoch": 0.8443830236025748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2434554106750652e-06,
+      "loss": 0.7408,
+      "step": 23218
+    },
+    {
+      "epoch": 0.8444193912063134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2428866370670767e-06,
+      "loss": 0.9096,
+      "step": 23219
+    },
+    {
+      "epoch": 0.844455758810052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.24231798495161e-06,
+      "loss": 0.9094,
+      "step": 23220
+    },
+    {
+      "epoch": 0.8444921264137906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2417494543365539e-06,
+      "loss": 0.9505,
+      "step": 23221
+    },
+    {
+      "epoch": 0.8445284940175292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2411810452298024e-06,
+      "loss": 0.9597,
+      "step": 23222
+    },
+    {
+      "epoch": 0.8445648616212678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2406127576392323e-06,
+      "loss": 0.9162,
+      "step": 23223
+    },
+    {
+      "epoch": 0.8446012292250064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2400445915727343e-06,
+      "loss": 0.8997,
+      "step": 23224
+    },
+    {
+      "epoch": 0.844637596828745,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2394765470381886e-06,
+      "loss": 0.9339,
+      "step": 23225
+    },
+    {
+      "epoch": 0.8446739644324835,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2389086240434766e-06,
+      "loss": 0.8017,
+      "step": 23226
+    },
+    {
+      "epoch": 0.8447103320362221,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2383408225964778e-06,
+      "loss": 0.8402,
+      "step": 23227
+    },
+    {
+      "epoch": 0.8447466996399607,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2377731427050677e-06,
+      "loss": 0.913,
+      "step": 23228
+    },
+    {
+      "epoch": 0.8447830672436993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2372055843771235e-06,
+      "loss": 1.0187,
+      "step": 23229
+    },
+    {
+      "epoch": 0.8448194348474379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.236638147620518e-06,
+      "loss": 0.8246,
+      "step": 23230
+    },
+    {
+      "epoch": 0.8448558024511765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.236070832443126e-06,
+      "loss": 0.9806,
+      "step": 23231
+    },
+    {
+      "epoch": 0.8448921700549151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2355036388528164e-06,
+      "loss": 1.0117,
+      "step": 23232
+    },
+    {
+      "epoch": 0.8449285376586537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2349365668574588e-06,
+      "loss": 0.874,
+      "step": 23233
+    },
+    {
+      "epoch": 0.8449649052623922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.234369616464919e-06,
+      "loss": 0.9348,
+      "step": 23234
+    },
+    {
+      "epoch": 0.8450012728661308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.233802787683067e-06,
+      "loss": 0.915,
+      "step": 23235
+    },
+    {
+      "epoch": 0.8450376404698694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2332360805197597e-06,
+      "loss": 0.9847,
+      "step": 23236
+    },
+    {
+      "epoch": 0.845074008073608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2326694949828665e-06,
+      "loss": 0.8806,
+      "step": 23237
+    },
+    {
+      "epoch": 0.8451103756773466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.232103031080244e-06,
+      "loss": 0.962,
+      "step": 23238
+    },
+    {
+      "epoch": 0.8451467432810852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2315366888197522e-06,
+      "loss": 0.8974,
+      "step": 23239
+    },
+    {
+      "epoch": 0.8451831108848238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2309704682092461e-06,
+      "loss": 0.8602,
+      "step": 23240
+    },
+    {
+      "epoch": 0.8452194784885624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.230404369256587e-06,
+      "loss": 0.8264,
+      "step": 23241
+    },
+    {
+      "epoch": 0.845255846092301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2298383919696199e-06,
+      "loss": 0.9146,
+      "step": 23242
+    },
+    {
+      "epoch": 0.8452922136960396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2292725363562052e-06,
+      "loss": 0.8143,
+      "step": 23243
+    },
+    {
+      "epoch": 0.8453285812997782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.22870680242419e-06,
+      "loss": 0.9173,
+      "step": 23244
+    },
+    {
+      "epoch": 0.8453649489035168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2281411901814222e-06,
+      "loss": 0.8771,
+      "step": 23245
+    },
+    {
+      "epoch": 0.8454013165072554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2275756996357503e-06,
+      "loss": 1.0057,
+      "step": 23246
+    },
+    {
+      "epoch": 0.845437684110994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2270103307950165e-06,
+      "loss": 0.854,
+      "step": 23247
+    },
+    {
+      "epoch": 0.8454740517147326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2264450836670727e-06,
+      "loss": 0.9687,
+      "step": 23248
+    },
+    {
+      "epoch": 0.8455104193184712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2258799582597502e-06,
+      "loss": 0.8907,
+      "step": 23249
+    },
+    {
+      "epoch": 0.8455467869222097,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2253149545808962e-06,
+      "loss": 0.8582,
+      "step": 23250
+    },
+    {
+      "epoch": 0.8455831545259482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2247500726383476e-06,
+      "loss": 0.8097,
+      "step": 23251
+    },
+    {
+      "epoch": 0.8456195221296868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2241853124399417e-06,
+      "loss": 0.8762,
+      "step": 23252
+    },
+    {
+      "epoch": 0.8456558897334254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2236206739935107e-06,
+      "loss": 0.9248,
+      "step": 23253
+    },
+    {
+      "epoch": 0.845692257337164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2230561573068956e-06,
+      "loss": 1.0095,
+      "step": 23254
+    },
+    {
+      "epoch": 0.8457286249409026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2224917623879196e-06,
+      "loss": 0.8936,
+      "step": 23255
+    },
+    {
+      "epoch": 0.8457649925446412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2219274892444177e-06,
+      "loss": 0.8541,
+      "step": 23256
+    },
+    {
+      "epoch": 0.8458013601483798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2213633378842183e-06,
+      "loss": 0.9115,
+      "step": 23257
+    },
+    {
+      "epoch": 0.8458377277521184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2207993083151471e-06,
+      "loss": 0.9139,
+      "step": 23258
+    },
+    {
+      "epoch": 0.845874095355857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2202354005450268e-06,
+      "loss": 0.9481,
+      "step": 23259
+    },
+    {
+      "epoch": 0.8459104629595956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2196716145816866e-06,
+      "loss": 0.9444,
+      "step": 23260
+    },
+    {
+      "epoch": 0.8459468305633342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2191079504329439e-06,
+      "loss": 0.7664,
+      "step": 23261
+    },
+    {
+      "epoch": 0.8459831981670728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2185444081066211e-06,
+      "loss": 0.9335,
+      "step": 23262
+    },
+    {
+      "epoch": 0.8460195657708114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2179809876105352e-06,
+      "loss": 0.8541,
+      "step": 23263
+    },
+    {
+      "epoch": 0.84605593337455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2174176889525025e-06,
+      "loss": 0.7331,
+      "step": 23264
+    },
+    {
+      "epoch": 0.8460923009782886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2168545121403386e-06,
+      "loss": 0.8984,
+      "step": 23265
+    },
+    {
+      "epoch": 0.8461286685820272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2162914571818552e-06,
+      "loss": 0.8176,
+      "step": 23266
+    },
+    {
+      "epoch": 0.8461650361857658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2157285240848682e-06,
+      "loss": 0.8912,
+      "step": 23267
+    },
+    {
+      "epoch": 0.8462014037895043,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2151657128571814e-06,
+      "loss": 0.8965,
+      "step": 23268
+    },
+    {
+      "epoch": 0.8462377713932429,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2146030235066085e-06,
+      "loss": 0.9996,
+      "step": 23269
+    },
+    {
+      "epoch": 0.8462741389969815,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2140404560409536e-06,
+      "loss": 1.0179,
+      "step": 23270
+    },
+    {
+      "epoch": 0.84631050660072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.213478010468022e-06,
+      "loss": 0.7444,
+      "step": 23271
+    },
+    {
+      "epoch": 0.8463468742044586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2129156867956149e-06,
+      "loss": 0.865,
+      "step": 23272
+    },
+    {
+      "epoch": 0.8463832418081972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2123534850315388e-06,
+      "loss": 0.9371,
+      "step": 23273
+    },
+    {
+      "epoch": 0.8464196094119358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2117914051835855e-06,
+      "loss": 0.8735,
+      "step": 23274
+    },
+    {
+      "epoch": 0.8464559770156744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2112294472595598e-06,
+      "loss": 0.9739,
+      "step": 23275
+    },
+    {
+      "epoch": 0.846492344619413,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2106676112672567e-06,
+      "loss": 0.8477,
+      "step": 23276
+    },
+    {
+      "epoch": 0.8465287122231516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2101058972144698e-06,
+      "loss": 0.8566,
+      "step": 23277
+    },
+    {
+      "epoch": 0.8465650798268902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2095443051089906e-06,
+      "loss": 0.8552,
+      "step": 23278
+    },
+    {
+      "epoch": 0.8466014474306288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.208982834958613e-06,
+      "loss": 0.8042,
+      "step": 23279
+    },
+    {
+      "epoch": 0.8466378150343674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2084214867711263e-06,
+      "loss": 0.8733,
+      "step": 23280
+    },
+    {
+      "epoch": 0.846674182638106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2078602605543189e-06,
+      "loss": 0.87,
+      "step": 23281
+    },
+    {
+      "epoch": 0.8467105502418446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2072991563159742e-06,
+      "loss": 0.9534,
+      "step": 23282
+    },
+    {
+      "epoch": 0.8467469178455832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2067381740638794e-06,
+      "loss": 0.8586,
+      "step": 23283
+    },
+    {
+      "epoch": 0.8467832854493218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2061773138058176e-06,
+      "loss": 0.9749,
+      "step": 23284
+    },
+    {
+      "epoch": 0.8468196530530603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2056165755495652e-06,
+      "loss": 1.023,
+      "step": 23285
+    },
+    {
+      "epoch": 0.8468560206567989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.20505595930291e-06,
+      "loss": 0.8604,
+      "step": 23286
+    },
+    {
+      "epoch": 0.8468923882605375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2044954650736197e-06,
+      "loss": 0.8107,
+      "step": 23287
+    },
+    {
+      "epoch": 0.8469287558642761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2039350928694781e-06,
+      "loss": 0.9335,
+      "step": 23288
+    },
+    {
+      "epoch": 0.8469651234680147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2033748426982584e-06,
+      "loss": 0.8064,
+      "step": 23289
+    },
+    {
+      "epoch": 0.8470014910717533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2028147145677306e-06,
+      "loss": 0.9197,
+      "step": 23290
+    },
+    {
+      "epoch": 0.8470378586754919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2022547084856662e-06,
+      "loss": 1.0743,
+      "step": 23291
+    },
+    {
+      "epoch": 0.8470742262792305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.201694824459836e-06,
+      "loss": 0.8704,
+      "step": 23292
+    },
+    {
+      "epoch": 0.847110593882969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2011350624980079e-06,
+      "loss": 0.8918,
+      "step": 23293
+    },
+    {
+      "epoch": 0.8471469614867077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2005754226079469e-06,
+      "loss": 0.9235,
+      "step": 23294
+    },
+    {
+      "epoch": 0.8471833290904462,
+      "grad_norm": 0.0,
+      "learning_rate": 1.200015904797417e-06,
+      "loss": 1.0154,
+      "step": 23295
+    },
+    {
+      "epoch": 0.8472196966941848,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1994565090741806e-06,
+      "loss": 0.9936,
+      "step": 23296
+    },
+    {
+      "epoch": 0.8472560642979234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.198897235445997e-06,
+      "loss": 0.9176,
+      "step": 23297
+    },
+    {
+      "epoch": 0.847292431901662,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1983380839206283e-06,
+      "loss": 0.9378,
+      "step": 23298
+    },
+    {
+      "epoch": 0.8473287995054006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.197779054505832e-06,
+      "loss": 0.8171,
+      "step": 23299
+    },
+    {
+      "epoch": 0.8473651671091392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1972201472093625e-06,
+      "loss": 0.8371,
+      "step": 23300
+    },
+    {
+      "epoch": 0.8474015347128777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1966613620389734e-06,
+      "loss": 0.8761,
+      "step": 23301
+    },
+    {
+      "epoch": 0.8474379023166163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1961026990024172e-06,
+      "loss": 0.8792,
+      "step": 23302
+    },
+    {
+      "epoch": 0.8474742699203549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1955441581074456e-06,
+      "loss": 0.8723,
+      "step": 23303
+    },
+    {
+      "epoch": 0.8475106375240935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1949857393618047e-06,
+      "loss": 1.0346,
+      "step": 23304
+    },
+    {
+      "epoch": 0.8475470051278321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1944274427732483e-06,
+      "loss": 0.8306,
+      "step": 23305
+    },
+    {
+      "epoch": 0.8475833727315707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1938692683495133e-06,
+      "loss": 0.8119,
+      "step": 23306
+    },
+    {
+      "epoch": 0.8476197403353093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1933112160983506e-06,
+      "loss": 0.8848,
+      "step": 23307
+    },
+    {
+      "epoch": 0.8476561079390479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1927532860274993e-06,
+      "loss": 0.9077,
+      "step": 23308
+    },
+    {
+      "epoch": 0.8476924755427865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1921954781446998e-06,
+      "loss": 0.8869,
+      "step": 23309
+    },
+    {
+      "epoch": 0.8477288431465251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1916377924576905e-06,
+      "loss": 0.8822,
+      "step": 23310
+    },
+    {
+      "epoch": 0.8477652107502637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1910802289742107e-06,
+      "loss": 0.9872,
+      "step": 23311
+    },
+    {
+      "epoch": 0.8478015783540023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1905227877019943e-06,
+      "loss": 0.9825,
+      "step": 23312
+    },
+    {
+      "epoch": 0.8478379459577409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1899654686487761e-06,
+      "loss": 0.9457,
+      "step": 23313
+    },
+    {
+      "epoch": 0.8478743135614795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1894082718222844e-06,
+      "loss": 1.0257,
+      "step": 23314
+    },
+    {
+      "epoch": 0.8479106811652181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1888511972302562e-06,
+      "loss": 0.8759,
+      "step": 23315
+    },
+    {
+      "epoch": 0.8479470487689567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1882942448804124e-06,
+      "loss": 0.8945,
+      "step": 23316
+    },
+    {
+      "epoch": 0.8479834163726953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1877374147804854e-06,
+      "loss": 0.912,
+      "step": 23317
+    },
+    {
+      "epoch": 0.8480197839764337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1871807069382e-06,
+      "loss": 0.9402,
+      "step": 23318
+    },
+    {
+      "epoch": 0.8480561515801723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1866241213612783e-06,
+      "loss": 0.8641,
+      "step": 23319
+    },
+    {
+      "epoch": 0.8480925191839109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1860676580574415e-06,
+      "loss": 1.0117,
+      "step": 23320
+    },
+    {
+      "epoch": 0.8481288867876495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1855113170344113e-06,
+      "loss": 0.8628,
+      "step": 23321
+    },
+    {
+      "epoch": 0.8481652543913881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1849550982999048e-06,
+      "loss": 0.9311,
+      "step": 23322
+    },
+    {
+      "epoch": 0.8482016219951267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.184399001861638e-06,
+      "loss": 0.914,
+      "step": 23323
+    },
+    {
+      "epoch": 0.8482379895988653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1838430277273293e-06,
+      "loss": 0.994,
+      "step": 23324
+    },
+    {
+      "epoch": 0.8482743572026039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1832871759046893e-06,
+      "loss": 0.8771,
+      "step": 23325
+    },
+    {
+      "epoch": 0.8483107248063425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1827314464014316e-06,
+      "loss": 0.8691,
+      "step": 23326
+    },
+    {
+      "epoch": 0.8483470924100811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1821758392252646e-06,
+      "loss": 0.9208,
+      "step": 23327
+    },
+    {
+      "epoch": 0.8483834600138197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1816203543838967e-06,
+      "loss": 0.9962,
+      "step": 23328
+    },
+    {
+      "epoch": 0.8484198276175583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1810649918850325e-06,
+      "loss": 0.8703,
+      "step": 23329
+    },
+    {
+      "epoch": 0.8484561952212969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1805097517363817e-06,
+      "loss": 0.9171,
+      "step": 23330
+    },
+    {
+      "epoch": 0.8484925628250355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1799546339456447e-06,
+      "loss": 0.8864,
+      "step": 23331
+    },
+    {
+      "epoch": 0.8485289304287741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1793996385205242e-06,
+      "loss": 0.791,
+      "step": 23332
+    },
+    {
+      "epoch": 0.8485652980325127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1788447654687151e-06,
+      "loss": 1.0448,
+      "step": 23333
+    },
+    {
+      "epoch": 0.8486016656362513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1782900147979247e-06,
+      "loss": 0.8593,
+      "step": 23334
+    },
+    {
+      "epoch": 0.8486380332399898,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1777353865158393e-06,
+      "loss": 0.9742,
+      "step": 23335
+    },
+    {
+      "epoch": 0.8486744008437284,
+      "grad_norm": 0.0,
+      "learning_rate": 1.17718088063016e-06,
+      "loss": 0.8859,
+      "step": 23336
+    },
+    {
+      "epoch": 0.848710768447467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1766264971485787e-06,
+      "loss": 0.8508,
+      "step": 23337
+    },
+    {
+      "epoch": 0.8487471360512056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1760722360787856e-06,
+      "loss": 0.781,
+      "step": 23338
+    },
+    {
+      "epoch": 0.8487835036549441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1755180974284686e-06,
+      "loss": 1.076,
+      "step": 23339
+    },
+    {
+      "epoch": 0.8488198712586827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.174964081205322e-06,
+      "loss": 0.8356,
+      "step": 23340
+    },
+    {
+      "epoch": 0.8488562388624213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1744101874170256e-06,
+      "loss": 0.8427,
+      "step": 23341
+    },
+    {
+      "epoch": 0.8488926064661599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.173856416071264e-06,
+      "loss": 0.864,
+      "step": 23342
+    },
+    {
+      "epoch": 0.8489289740698985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1733027671757236e-06,
+      "loss": 0.9266,
+      "step": 23343
+    },
+    {
+      "epoch": 0.8489653416736371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1727492407380847e-06,
+      "loss": 0.873,
+      "step": 23344
+    },
+    {
+      "epoch": 0.8490017092773757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1721958367660247e-06,
+      "loss": 0.9003,
+      "step": 23345
+    },
+    {
+      "epoch": 0.8490380768811143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1716425552672205e-06,
+      "loss": 0.9879,
+      "step": 23346
+    },
+    {
+      "epoch": 0.8490744444848529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1710893962493552e-06,
+      "loss": 1.0,
+      "step": 23347
+    },
+    {
+      "epoch": 0.8491108120885915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1705363597200935e-06,
+      "loss": 0.8759,
+      "step": 23348
+    },
+    {
+      "epoch": 0.8491471796923301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.169983445687114e-06,
+      "loss": 0.8377,
+      "step": 23349
+    },
+    {
+      "epoch": 0.8491835472960687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1694306541580868e-06,
+      "loss": 1.062,
+      "step": 23350
+    },
+    {
+      "epoch": 0.8492199148998073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1688779851406795e-06,
+      "loss": 0.9197,
+      "step": 23351
+    },
+    {
+      "epoch": 0.8492562825035458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.168325438642559e-06,
+      "loss": 0.8611,
+      "step": 23352
+    },
+    {
+      "epoch": 0.8492926501072844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1677730146713962e-06,
+      "loss": 0.928,
+      "step": 23353
+    },
+    {
+      "epoch": 0.849329017711023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1672207132348478e-06,
+      "loss": 0.9487,
+      "step": 23354
+    },
+    {
+      "epoch": 0.8493653853147616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1666685343405816e-06,
+      "loss": 0.9038,
+      "step": 23355
+    },
+    {
+      "epoch": 0.8494017529185002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1661164779962576e-06,
+      "loss": 0.9454,
+      "step": 23356
+    },
+    {
+      "epoch": 0.8494381205222388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.165564544209532e-06,
+      "loss": 0.8093,
+      "step": 23357
+    },
+    {
+      "epoch": 0.8494744881259774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1650127329880634e-06,
+      "loss": 0.877,
+      "step": 23358
+    },
+    {
+      "epoch": 0.849510855729716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1644610443395098e-06,
+      "loss": 0.8301,
+      "step": 23359
+    },
+    {
+      "epoch": 0.8495472233334546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1639094782715243e-06,
+      "loss": 0.8386,
+      "step": 23360
+    },
+    {
+      "epoch": 0.8495835909371932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1633580347917551e-06,
+      "loss": 0.998,
+      "step": 23361
+    },
+    {
+      "epoch": 0.8496199585409318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.162806713907857e-06,
+      "loss": 0.8785,
+      "step": 23362
+    },
+    {
+      "epoch": 0.8496563261446703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1622555156274773e-06,
+      "loss": 0.9868,
+      "step": 23363
+    },
+    {
+      "epoch": 0.849692693748409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1617044399582633e-06,
+      "loss": 0.9711,
+      "step": 23364
+    },
+    {
+      "epoch": 0.8497290613521475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1611534869078588e-06,
+      "loss": 0.862,
+      "step": 23365
+    },
+    {
+      "epoch": 0.8497654289558861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1606026564839123e-06,
+      "loss": 0.8833,
+      "step": 23366
+    },
+    {
+      "epoch": 0.8498017965596247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1600519486940587e-06,
+      "loss": 0.8441,
+      "step": 23367
+    },
+    {
+      "epoch": 0.8498381641633633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1595013635459452e-06,
+      "loss": 0.9027,
+      "step": 23368
+    },
+    {
+      "epoch": 0.8498745317671018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1589509010472055e-06,
+      "loss": 0.9136,
+      "step": 23369
+    },
+    {
+      "epoch": 0.8499108993708404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1584005612054794e-06,
+      "loss": 0.8143,
+      "step": 23370
+    },
+    {
+      "epoch": 0.849947266974579,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1578503440283983e-06,
+      "loss": 0.8187,
+      "step": 23371
+    },
+    {
+      "epoch": 0.8499836345783176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.157300249523603e-06,
+      "loss": 0.9414,
+      "step": 23372
+    },
+    {
+      "epoch": 0.8500200021820562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.156750277698715e-06,
+      "loss": 0.9438,
+      "step": 23373
+    },
+    {
+      "epoch": 0.8500563697857948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1562004285613737e-06,
+      "loss": 0.9626,
+      "step": 23374
+    },
+    {
+      "epoch": 0.8500927373895334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.155650702119203e-06,
+      "loss": 0.9911,
+      "step": 23375
+    },
+    {
+      "epoch": 0.850129104993272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1551010983798305e-06,
+      "loss": 0.9348,
+      "step": 23376
+    },
+    {
+      "epoch": 0.8501654725970106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1545516173508786e-06,
+      "loss": 0.7415,
+      "step": 23377
+    },
+    {
+      "epoch": 0.8502018402007492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1540022590399759e-06,
+      "loss": 0.9476,
+      "step": 23378
+    },
+    {
+      "epoch": 0.8502382078044878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1534530234547415e-06,
+      "loss": 1.0252,
+      "step": 23379
+    },
+    {
+      "epoch": 0.8502745754082264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1529039106027916e-06,
+      "loss": 0.9723,
+      "step": 23380
+    },
+    {
+      "epoch": 0.850310943011965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1523549204917506e-06,
+      "loss": 0.9076,
+      "step": 23381
+    },
+    {
+      "epoch": 0.8503473106157036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1518060531292307e-06,
+      "loss": 0.9206,
+      "step": 23382
+    },
+    {
+      "epoch": 0.8503836782194422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1512573085228474e-06,
+      "loss": 0.8896,
+      "step": 23383
+    },
+    {
+      "epoch": 0.8504200458231808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.150708686680213e-06,
+      "loss": 0.9361,
+      "step": 23384
+    },
+    {
+      "epoch": 0.8504564134269194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1501601876089442e-06,
+      "loss": 0.9131,
+      "step": 23385
+    },
+    {
+      "epoch": 0.8504927810306578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.149611811316642e-06,
+      "loss": 0.9689,
+      "step": 23386
+    },
+    {
+      "epoch": 0.8505291486343964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1490635578109222e-06,
+      "loss": 0.8696,
+      "step": 23387
+    },
+    {
+      "epoch": 0.850565516238135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1485154270993858e-06,
+      "loss": 0.9662,
+      "step": 23388
+    },
+    {
+      "epoch": 0.8506018838418736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1479674191896407e-06,
+      "loss": 0.8059,
+      "step": 23389
+    },
+    {
+      "epoch": 0.8506382514456122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1474195340892857e-06,
+      "loss": 0.8813,
+      "step": 23390
+    },
+    {
+      "epoch": 0.8506746190493508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1468717718059264e-06,
+      "loss": 0.9406,
+      "step": 23391
+    },
+    {
+      "epoch": 0.8507109866530894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1463241323471596e-06,
+      "loss": 0.8276,
+      "step": 23392
+    },
+    {
+      "epoch": 0.850747354256828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1457766157205852e-06,
+      "loss": 0.9408,
+      "step": 23393
+    },
+    {
+      "epoch": 0.8507837218605666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1452292219337967e-06,
+      "loss": 0.8955,
+      "step": 23394
+    },
+    {
+      "epoch": 0.8508200894643052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1446819509943906e-06,
+      "loss": 0.9701,
+      "step": 23395
+    },
+    {
+      "epoch": 0.8508564570680438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1441348029099552e-06,
+      "loss": 0.9852,
+      "step": 23396
+    },
+    {
+      "epoch": 0.8508928246717824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1435877776880865e-06,
+      "loss": 0.865,
+      "step": 23397
+    },
+    {
+      "epoch": 0.850929192275521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1430408753363753e-06,
+      "loss": 0.9041,
+      "step": 23398
+    },
+    {
+      "epoch": 0.8509655598792596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1424940958623998e-06,
+      "loss": 0.9247,
+      "step": 23399
+    },
+    {
+      "epoch": 0.8510019274829982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.141947439273755e-06,
+      "loss": 0.8443,
+      "step": 23400
+    },
+    {
+      "epoch": 0.8510382950867368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1414009055780206e-06,
+      "loss": 0.8216,
+      "step": 23401
+    },
+    {
+      "epoch": 0.8510746626904753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1408544947827805e-06,
+      "loss": 0.9595,
+      "step": 23402
+    },
+    {
+      "epoch": 0.8511110302942139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1403082068956118e-06,
+      "loss": 0.8212,
+      "step": 23403
+    },
+    {
+      "epoch": 0.8511473978979525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1397620419240995e-06,
+      "loss": 0.9016,
+      "step": 23404
+    },
+    {
+      "epoch": 0.8511837655016911,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1392159998758178e-06,
+      "loss": 0.9811,
+      "step": 23405
+    },
+    {
+      "epoch": 0.8512201331054297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1386700807583417e-06,
+      "loss": 0.8468,
+      "step": 23406
+    },
+    {
+      "epoch": 0.8512565007091683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1381242845792462e-06,
+      "loss": 0.8654,
+      "step": 23407
+    },
+    {
+      "epoch": 0.8512928683129068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1375786113461029e-06,
+      "loss": 0.9523,
+      "step": 23408
+    },
+    {
+      "epoch": 0.8513292359166454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1370330610664792e-06,
+      "loss": 0.8141,
+      "step": 23409
+    },
+    {
+      "epoch": 0.851365603520384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1364876337479502e-06,
+      "loss": 0.9371,
+      "step": 23410
+    },
+    {
+      "epoch": 0.8514019711241226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1359423293980787e-06,
+      "loss": 0.9628,
+      "step": 23411
+    },
+    {
+      "epoch": 0.8514383387278612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1353971480244318e-06,
+      "loss": 0.9099,
+      "step": 23412
+    },
+    {
+      "epoch": 0.8514747063315998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1348520896345694e-06,
+      "loss": 0.8782,
+      "step": 23413
+    },
+    {
+      "epoch": 0.8515110739353384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1343071542360607e-06,
+      "loss": 0.9936,
+      "step": 23414
+    },
+    {
+      "epoch": 0.851547441539077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1337623418364575e-06,
+      "loss": 0.9542,
+      "step": 23415
+    },
+    {
+      "epoch": 0.8515838091428156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1332176524433236e-06,
+      "loss": 0.9019,
+      "step": 23416
+    },
+    {
+      "epoch": 0.8516201767465542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1326730860642178e-06,
+      "loss": 0.9985,
+      "step": 23417
+    },
+    {
+      "epoch": 0.8516565443502928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1321286427066869e-06,
+      "loss": 0.884,
+      "step": 23418
+    },
+    {
+      "epoch": 0.8516929119540313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1315843223782908e-06,
+      "loss": 1.0017,
+      "step": 23419
+    },
+    {
+      "epoch": 0.8517292795577699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.13104012508658e-06,
+      "loss": 0.8519,
+      "step": 23420
+    },
+    {
+      "epoch": 0.8517656471615085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1304960508391037e-06,
+      "loss": 0.9532,
+      "step": 23421
+    },
+    {
+      "epoch": 0.8518020147652471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.12995209964341e-06,
+      "loss": 1.0318,
+      "step": 23422
+    },
+    {
+      "epoch": 0.8518383823689857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1294082715070464e-06,
+      "loss": 0.8944,
+      "step": 23423
+    },
+    {
+      "epoch": 0.8518747499727243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1288645664375574e-06,
+      "loss": 0.9078,
+      "step": 23424
+    },
+    {
+      "epoch": 0.8519111175764629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.128320984442487e-06,
+      "loss": 1.0009,
+      "step": 23425
+    },
+    {
+      "epoch": 0.8519474851802015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1277775255293733e-06,
+      "loss": 0.9116,
+      "step": 23426
+    },
+    {
+      "epoch": 0.8519838527839401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1272341897057627e-06,
+      "loss": 0.9788,
+      "step": 23427
+    },
+    {
+      "epoch": 0.8520202203876787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1266909769791845e-06,
+      "loss": 0.8554,
+      "step": 23428
+    },
+    {
+      "epoch": 0.8520565879914173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1261478873571807e-06,
+      "loss": 0.9005,
+      "step": 23429
+    },
+    {
+      "epoch": 0.8520929555951559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1256049208472853e-06,
+      "loss": 0.9004,
+      "step": 23430
+    },
+    {
+      "epoch": 0.8521293231988945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1250620774570308e-06,
+      "loss": 0.9177,
+      "step": 23431
+    },
+    {
+      "epoch": 0.852165690802633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1245193571939462e-06,
+      "loss": 0.9291,
+      "step": 23432
+    },
+    {
+      "epoch": 0.8522020584063716,
+      "grad_norm": 0.0,
+      "learning_rate": 1.123976760065566e-06,
+      "loss": 0.931,
+      "step": 23433
+    },
+    {
+      "epoch": 0.8522384260101102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1234342860794112e-06,
+      "loss": 0.9598,
+      "step": 23434
+    },
+    {
+      "epoch": 0.8522747936138488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1228919352430135e-06,
+      "loss": 1.0601,
+      "step": 23435
+    },
+    {
+      "epoch": 0.8523111612175873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1223497075638957e-06,
+      "loss": 0.9928,
+      "step": 23436
+    },
+    {
+      "epoch": 0.8523475288213259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1218076030495795e-06,
+      "loss": 1.0114,
+      "step": 23437
+    },
+    {
+      "epoch": 0.8523838964250645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1212656217075868e-06,
+      "loss": 0.8745,
+      "step": 23438
+    },
+    {
+      "epoch": 0.8524202640288031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1207237635454348e-06,
+      "loss": 0.9389,
+      "step": 23439
+    },
+    {
+      "epoch": 0.8524566316325417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1201820285706443e-06,
+      "loss": 0.9131,
+      "step": 23440
+    },
+    {
+      "epoch": 0.8524929992362803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.119640416790726e-06,
+      "loss": 0.9097,
+      "step": 23441
+    },
+    {
+      "epoch": 0.8525293668400189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1190989282132002e-06,
+      "loss": 0.8845,
+      "step": 23442
+    },
+    {
+      "epoch": 0.8525657344437575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1185575628455769e-06,
+      "loss": 0.9136,
+      "step": 23443
+    },
+    {
+      "epoch": 0.8526021020474961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1180163206953654e-06,
+      "loss": 0.8884,
+      "step": 23444
+    },
+    {
+      "epoch": 0.8526384696512347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1174752017700741e-06,
+      "loss": 0.8014,
+      "step": 23445
+    },
+    {
+      "epoch": 0.8526748372549733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1169342060772147e-06,
+      "loss": 0.8982,
+      "step": 23446
+    },
+    {
+      "epoch": 0.8527112048587119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.116393333624287e-06,
+      "loss": 0.952,
+      "step": 23447
+    },
+    {
+      "epoch": 0.8527475724624505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1158525844187984e-06,
+      "loss": 0.9472,
+      "step": 23448
+    },
+    {
+      "epoch": 0.8527839400661891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1153119584682514e-06,
+      "loss": 0.8008,
+      "step": 23449
+    },
+    {
+      "epoch": 0.8528203076699277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1147714557801449e-06,
+      "loss": 0.9317,
+      "step": 23450
+    },
+    {
+      "epoch": 0.8528566752736663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1142310763619756e-06,
+      "loss": 0.8424,
+      "step": 23451
+    },
+    {
+      "epoch": 0.8528930428774049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.113690820221247e-06,
+      "loss": 0.7699,
+      "step": 23452
+    },
+    {
+      "epoch": 0.8529294104811433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1131506873654464e-06,
+      "loss": 0.964,
+      "step": 23453
+    },
+    {
+      "epoch": 0.8529657780848819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1126106778020729e-06,
+      "loss": 0.9704,
+      "step": 23454
+    },
+    {
+      "epoch": 0.8530021456886205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1120707915386176e-06,
+      "loss": 0.88,
+      "step": 23455
+    },
+    {
+      "epoch": 0.8530385132923591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1115310285825687e-06,
+      "loss": 0.8726,
+      "step": 23456
+    },
+    {
+      "epoch": 0.8530748808960977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1109913889414158e-06,
+      "loss": 1.0085,
+      "step": 23457
+    },
+    {
+      "epoch": 0.8531112484998363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1104518726226443e-06,
+      "loss": 1.0115,
+      "step": 23458
+    },
+    {
+      "epoch": 0.8531476161035749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1099124796337435e-06,
+      "loss": 0.8988,
+      "step": 23459
+    },
+    {
+      "epoch": 0.8531839837073135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1093732099821909e-06,
+      "loss": 0.9663,
+      "step": 23460
+    },
+    {
+      "epoch": 0.8532203513110521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1088340636754714e-06,
+      "loss": 1.0536,
+      "step": 23461
+    },
+    {
+      "epoch": 0.8532567189147907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1082950407210646e-06,
+      "loss": 0.8187,
+      "step": 23462
+    },
+    {
+      "epoch": 0.8532930865185293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1077561411264491e-06,
+      "loss": 0.8675,
+      "step": 23463
+    },
+    {
+      "epoch": 0.8533294541222679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.107217364899099e-06,
+      "loss": 0.9341,
+      "step": 23464
+    },
+    {
+      "epoch": 0.8533658217260065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1066787120464938e-06,
+      "loss": 0.669,
+      "step": 23465
+    },
+    {
+      "epoch": 0.8534021893297451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1061401825760999e-06,
+      "loss": 1.1241,
+      "step": 23466
+    },
+    {
+      "epoch": 0.8534385569334837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1056017764953941e-06,
+      "loss": 0.8743,
+      "step": 23467
+    },
+    {
+      "epoch": 0.8534749245372223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1050634938118443e-06,
+      "loss": 0.8614,
+      "step": 23468
+    },
+    {
+      "epoch": 0.8535112921409609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1045253345329188e-06,
+      "loss": 0.9205,
+      "step": 23469
+    },
+    {
+      "epoch": 0.8535476597446994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1039872986660805e-06,
+      "loss": 0.9157,
+      "step": 23470
+    },
+    {
+      "epoch": 0.853584027348438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1034493862188012e-06,
+      "loss": 0.9456,
+      "step": 23471
+    },
+    {
+      "epoch": 0.8536203949521766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1029115971985338e-06,
+      "loss": 0.8772,
+      "step": 23472
+    },
+    {
+      "epoch": 0.8536567625559152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1023739316127479e-06,
+      "loss": 0.9761,
+      "step": 23473
+    },
+    {
+      "epoch": 0.8536931301596538,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1018363894689e-06,
+      "loss": 0.9607,
+      "step": 23474
+    },
+    {
+      "epoch": 0.8537294977633924,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1012989707744458e-06,
+      "loss": 1.0268,
+      "step": 23475
+    },
+    {
+      "epoch": 0.853765865367131,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1007616755368445e-06,
+      "loss": 0.9676,
+      "step": 23476
+    },
+    {
+      "epoch": 0.8538022329708695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.100224503763545e-06,
+      "loss": 0.9435,
+      "step": 23477
+    },
+    {
+      "epoch": 0.8538386005746081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0996874554620086e-06,
+      "loss": 0.9335,
+      "step": 23478
+    },
+    {
+      "epoch": 0.8538749681783467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0991505306396766e-06,
+      "loss": 0.8689,
+      "step": 23479
+    },
+    {
+      "epoch": 0.8539113357820853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0986137293040034e-06,
+      "loss": 0.8326,
+      "step": 23480
+    },
+    {
+      "epoch": 0.8539477033858239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0980770514624361e-06,
+      "loss": 0.9809,
+      "step": 23481
+    },
+    {
+      "epoch": 0.8539840709895625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0975404971224201e-06,
+      "loss": 1.0443,
+      "step": 23482
+    },
+    {
+      "epoch": 0.8540204385933011,
+      "grad_norm": 0.0,
+      "learning_rate": 1.097004066291396e-06,
+      "loss": 0.9501,
+      "step": 23483
+    },
+    {
+      "epoch": 0.8540568061970397,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0964677589768125e-06,
+      "loss": 0.9986,
+      "step": 23484
+    },
+    {
+      "epoch": 0.8540931738007783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0959315751861033e-06,
+      "loss": 0.8981,
+      "step": 23485
+    },
+    {
+      "epoch": 0.8541295414045168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0953955149267114e-06,
+      "loss": 0.8596,
+      "step": 23486
+    },
+    {
+      "epoch": 0.8541659090082554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0948595782060734e-06,
+      "loss": 0.9719,
+      "step": 23487
+    },
+    {
+      "epoch": 0.854202276611994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0943237650316229e-06,
+      "loss": 0.8861,
+      "step": 23488
+    },
+    {
+      "epoch": 0.8542386442157326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0937880754107943e-06,
+      "loss": 0.9623,
+      "step": 23489
+    },
+    {
+      "epoch": 0.8542750118194712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0932525093510204e-06,
+      "loss": 0.8526,
+      "step": 23490
+    },
+    {
+      "epoch": 0.8543113794232098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.092717066859732e-06,
+      "loss": 0.9841,
+      "step": 23491
+    },
+    {
+      "epoch": 0.8543477470269484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0921817479443565e-06,
+      "loss": 0.821,
+      "step": 23492
+    },
+    {
+      "epoch": 0.854384114630687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0916465526123199e-06,
+      "loss": 0.9042,
+      "step": 23493
+    },
+    {
+      "epoch": 0.8544204822344256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0911114808710488e-06,
+      "loss": 0.869,
+      "step": 23494
+    },
+    {
+      "epoch": 0.8544568498381642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0905765327279649e-06,
+      "loss": 0.8875,
+      "step": 23495
+    },
+    {
+      "epoch": 0.8544932174419028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0900417081904901e-06,
+      "loss": 0.9409,
+      "step": 23496
+    },
+    {
+      "epoch": 0.8545295850456414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0895070072660473e-06,
+      "loss": 0.9233,
+      "step": 23497
+    },
+    {
+      "epoch": 0.85456595264938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0889724299620498e-06,
+      "loss": 0.8402,
+      "step": 23498
+    },
+    {
+      "epoch": 0.8546023202531186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0884379762859187e-06,
+      "loss": 0.9154,
+      "step": 23499
+    },
+    {
+      "epoch": 0.8546386878568571,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0879036462450664e-06,
+      "loss": 0.8774,
+      "step": 23500
+    },
+    {
+      "epoch": 0.8546750554605957,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0873694398469071e-06,
+      "loss": 0.9498,
+      "step": 23501
+    },
+    {
+      "epoch": 0.8547114230643343,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0868353570988487e-06,
+      "loss": 0.8919,
+      "step": 23502
+    },
+    {
+      "epoch": 0.8547477906680728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0863013980083059e-06,
+      "loss": 0.9764,
+      "step": 23503
+    },
+    {
+      "epoch": 0.8547841582718114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0857675625826836e-06,
+      "loss": 0.8964,
+      "step": 23504
+    },
+    {
+      "epoch": 0.85482052587555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0852338508293892e-06,
+      "loss": 0.8192,
+      "step": 23505
+    },
+    {
+      "epoch": 0.8548568934792886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0847002627558268e-06,
+      "loss": 0.9377,
+      "step": 23506
+    },
+    {
+      "epoch": 0.8548932610830272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0841667983693993e-06,
+      "loss": 0.9715,
+      "step": 23507
+    },
+    {
+      "epoch": 0.8549296286867658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0836334576775055e-06,
+      "loss": 0.8194,
+      "step": 23508
+    },
+    {
+      "epoch": 0.8549659962905044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.083100240687549e-06,
+      "loss": 0.9787,
+      "step": 23509
+    },
+    {
+      "epoch": 0.855002363894243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0825671474069243e-06,
+      "loss": 0.8222,
+      "step": 23510
+    },
+    {
+      "epoch": 0.8550387314979816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0820341778430298e-06,
+      "loss": 0.8157,
+      "step": 23511
+    },
+    {
+      "epoch": 0.8550750991017202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0815013320032563e-06,
+      "loss": 0.9088,
+      "step": 23512
+    },
+    {
+      "epoch": 0.8551114667054588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.080968609895e-06,
+      "loss": 0.9264,
+      "step": 23513
+    },
+    {
+      "epoch": 0.8551478343091974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0804360115256485e-06,
+      "loss": 0.7892,
+      "step": 23514
+    },
+    {
+      "epoch": 0.855184201912936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.07990353690259e-06,
+      "loss": 0.9414,
+      "step": 23515
+    },
+    {
+      "epoch": 0.8552205695166746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.079371186033219e-06,
+      "loss": 0.8769,
+      "step": 23516
+    },
+    {
+      "epoch": 0.8552569371204132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0788389589249115e-06,
+      "loss": 0.9522,
+      "step": 23517
+    },
+    {
+      "epoch": 0.8552933047241518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0783068555850585e-06,
+      "loss": 0.9419,
+      "step": 23518
+    },
+    {
+      "epoch": 0.8553296723278904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0777748760210382e-06,
+      "loss": 0.8642,
+      "step": 23519
+    },
+    {
+      "epoch": 0.8553660399316289,
+      "grad_norm": 0.0,
+      "learning_rate": 1.077243020240234e-06,
+      "loss": 0.9673,
+      "step": 23520
+    },
+    {
+      "epoch": 0.8554024075353674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0767112882500207e-06,
+      "loss": 0.8491,
+      "step": 23521
+    },
+    {
+      "epoch": 0.855438775139106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0761796800577794e-06,
+      "loss": 1.0355,
+      "step": 23522
+    },
+    {
+      "epoch": 0.8554751427428446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0756481956708841e-06,
+      "loss": 0.9704,
+      "step": 23523
+    },
+    {
+      "epoch": 0.8555115103465832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.075116835096708e-06,
+      "loss": 0.9205,
+      "step": 23524
+    },
+    {
+      "epoch": 0.8555478779503218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0745855983426211e-06,
+      "loss": 0.8338,
+      "step": 23525
+    },
+    {
+      "epoch": 0.8555842455540604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0740544854159995e-06,
+      "loss": 0.8238,
+      "step": 23526
+    },
+    {
+      "epoch": 0.855620613157799,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0735234963242036e-06,
+      "loss": 0.9918,
+      "step": 23527
+    },
+    {
+      "epoch": 0.8556569807615376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0729926310746064e-06,
+      "loss": 0.9564,
+      "step": 23528
+    },
+    {
+      "epoch": 0.8556933483652762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.072461889674571e-06,
+      "loss": 1.0001,
+      "step": 23529
+    },
+    {
+      "epoch": 0.8557297159690148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0719312721314601e-06,
+      "loss": 0.8233,
+      "step": 23530
+    },
+    {
+      "epoch": 0.8557660835727534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0714007784526348e-06,
+      "loss": 0.8907,
+      "step": 23531
+    },
+    {
+      "epoch": 0.855802451176492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.070870408645457e-06,
+      "loss": 0.8356,
+      "step": 23532
+    },
+    {
+      "epoch": 0.8558388187802306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0703401627172827e-06,
+      "loss": 0.9585,
+      "step": 23533
+    },
+    {
+      "epoch": 0.8558751863839692,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0698100406754663e-06,
+      "loss": 0.881,
+      "step": 23534
+    },
+    {
+      "epoch": 0.8559115539877078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0692800425273685e-06,
+      "loss": 0.9758,
+      "step": 23535
+    },
+    {
+      "epoch": 0.8559479215914464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0687501682803392e-06,
+      "loss": 0.9516,
+      "step": 23536
+    },
+    {
+      "epoch": 0.8559842891951849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0682204179417287e-06,
+      "loss": 0.8571,
+      "step": 23537
+    },
+    {
+      "epoch": 0.8560206567989235,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0676907915188872e-06,
+      "loss": 0.9292,
+      "step": 23538
+    },
+    {
+      "epoch": 0.8560570244026621,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0671612890191641e-06,
+      "loss": 0.9299,
+      "step": 23539
+    },
+    {
+      "epoch": 0.8560933920064007,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0666319104499011e-06,
+      "loss": 0.8924,
+      "step": 23540
+    },
+    {
+      "epoch": 0.8561297596101393,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0661026558184473e-06,
+      "loss": 1.0704,
+      "step": 23541
+    },
+    {
+      "epoch": 0.8561661272138779,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0655735251321441e-06,
+      "loss": 0.846,
+      "step": 23542
+    },
+    {
+      "epoch": 0.8562024948176165,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0650445183983315e-06,
+      "loss": 0.9095,
+      "step": 23543
+    },
+    {
+      "epoch": 0.856238862421355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.064515635624348e-06,
+      "loss": 0.8914,
+      "step": 23544
+    },
+    {
+      "epoch": 0.8562752300250936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0639868768175354e-06,
+      "loss": 0.8707,
+      "step": 23545
+    },
+    {
+      "epoch": 0.8563115976288322,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0634582419852224e-06,
+      "loss": 0.834,
+      "step": 23546
+    },
+    {
+      "epoch": 0.8563479652325708,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0629297311347498e-06,
+      "loss": 0.8429,
+      "step": 23547
+    },
+    {
+      "epoch": 0.8563843328363094,
+      "grad_norm": 0.0,
+      "learning_rate": 1.062401344273446e-06,
+      "loss": 1.0112,
+      "step": 23548
+    },
+    {
+      "epoch": 0.856420700440048,
+      "grad_norm": 0.0,
+      "learning_rate": 1.061873081408643e-06,
+      "loss": 0.8869,
+      "step": 23549
+    },
+    {
+      "epoch": 0.8564570680437866,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0613449425476696e-06,
+      "loss": 0.8426,
+      "step": 23550
+    },
+    {
+      "epoch": 0.8564934356475252,
+      "grad_norm": 0.0,
+      "learning_rate": 1.060816927697853e-06,
+      "loss": 0.7889,
+      "step": 23551
+    },
+    {
+      "epoch": 0.8565298032512638,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0602890368665186e-06,
+      "loss": 0.9121,
+      "step": 23552
+    },
+    {
+      "epoch": 0.8565661708550024,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0597612700609882e-06,
+      "loss": 0.9341,
+      "step": 23553
+    },
+    {
+      "epoch": 0.8566025384587409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0592336272885872e-06,
+      "loss": 0.9304,
+      "step": 23554
+    },
+    {
+      "epoch": 0.8566389060624795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0587061085566352e-06,
+      "loss": 0.8955,
+      "step": 23555
+    },
+    {
+      "epoch": 0.8566752736662181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0581787138724487e-06,
+      "loss": 0.8693,
+      "step": 23556
+    },
+    {
+      "epoch": 0.8567116412699567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0576514432433439e-06,
+      "loss": 0.908,
+      "step": 23557
+    },
+    {
+      "epoch": 0.8567480088736953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0571242966766426e-06,
+      "loss": 0.8853,
+      "step": 23558
+    },
+    {
+      "epoch": 0.8567843764774339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0565972741796494e-06,
+      "loss": 0.9469,
+      "step": 23559
+    },
+    {
+      "epoch": 0.8568207440811725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0560703757596813e-06,
+      "loss": 0.7131,
+      "step": 23560
+    },
+    {
+      "epoch": 0.8568571116849111,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0555436014240471e-06,
+      "loss": 0.8736,
+      "step": 23561
+    },
+    {
+      "epoch": 0.8568934792886497,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0550169511800545e-06,
+      "loss": 0.8935,
+      "step": 23562
+    },
+    {
+      "epoch": 0.8569298468923883,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0544904250350097e-06,
+      "loss": 0.956,
+      "step": 23563
+    },
+    {
+      "epoch": 0.8569662144961269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0539640229962212e-06,
+      "loss": 0.7826,
+      "step": 23564
+    },
+    {
+      "epoch": 0.8570025820998655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0534377450709853e-06,
+      "loss": 1.0165,
+      "step": 23565
+    },
+    {
+      "epoch": 0.8570389497036041,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0529115912666088e-06,
+      "loss": 0.9619,
+      "step": 23566
+    },
+    {
+      "epoch": 0.8570753173073427,
+      "grad_norm": 0.0,
+      "learning_rate": 1.052385561590391e-06,
+      "loss": 1.0417,
+      "step": 23567
+    },
+    {
+      "epoch": 0.8571116849110813,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0518596560496275e-06,
+      "loss": 0.9247,
+      "step": 23568
+    },
+    {
+      "epoch": 0.8571480525148198,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0513338746516134e-06,
+      "loss": 0.8778,
+      "step": 23569
+    },
+    {
+      "epoch": 0.8571844201185584,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0508082174036483e-06,
+      "loss": 0.9818,
+      "step": 23570
+    },
+    {
+      "epoch": 0.8572207877222969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0502826843130231e-06,
+      "loss": 0.9355,
+      "step": 23571
+    },
+    {
+      "epoch": 0.8572571553260355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0497572753870233e-06,
+      "loss": 0.8753,
+      "step": 23572
+    },
+    {
+      "epoch": 0.8572935229297741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0492319906329452e-06,
+      "loss": 0.9748,
+      "step": 23573
+    },
+    {
+      "epoch": 0.8573298905335127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0487068300580739e-06,
+      "loss": 0.88,
+      "step": 23574
+    },
+    {
+      "epoch": 0.8573662581372513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0481817936696948e-06,
+      "loss": 0.84,
+      "step": 23575
+    },
+    {
+      "epoch": 0.8574026257409899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0476568814750899e-06,
+      "loss": 0.9048,
+      "step": 23576
+    },
+    {
+      "epoch": 0.8574389933447285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0471320934815488e-06,
+      "loss": 0.9894,
+      "step": 23577
+    },
+    {
+      "epoch": 0.8574753609484671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0466074296963435e-06,
+      "loss": 0.9211,
+      "step": 23578
+    },
+    {
+      "epoch": 0.8575117285522057,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0460828901267583e-06,
+      "loss": 0.7981,
+      "step": 23579
+    },
+    {
+      "epoch": 0.8575480961559443,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0455584747800696e-06,
+      "loss": 0.8475,
+      "step": 23580
+    },
+    {
+      "epoch": 0.8575844637596829,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0450341836635524e-06,
+      "loss": 0.9502,
+      "step": 23581
+    },
+    {
+      "epoch": 0.8576208313634215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.044510016784478e-06,
+      "loss": 0.9375,
+      "step": 23582
+    },
+    {
+      "epoch": 0.8576571989671601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0439859741501245e-06,
+      "loss": 0.8448,
+      "step": 23583
+    },
+    {
+      "epoch": 0.8576935665708987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0434620557677555e-06,
+      "loss": 0.8578,
+      "step": 23584
+    },
+    {
+      "epoch": 0.8577299341746373,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0429382616446448e-06,
+      "loss": 0.9554,
+      "step": 23585
+    },
+    {
+      "epoch": 0.8577663017783759,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0424145917880569e-06,
+      "loss": 0.741,
+      "step": 23586
+    },
+    {
+      "epoch": 0.8578026693821144,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0418910462052577e-06,
+      "loss": 0.9839,
+      "step": 23587
+    },
+    {
+      "epoch": 0.857839036985853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0413676249035086e-06,
+      "loss": 0.9211,
+      "step": 23588
+    },
+    {
+      "epoch": 0.8578754045895916,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0408443278900747e-06,
+      "loss": 0.8983,
+      "step": 23589
+    },
+    {
+      "epoch": 0.8579117721933301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.040321155172217e-06,
+      "loss": 0.9627,
+      "step": 23590
+    },
+    {
+      "epoch": 0.8579481397970687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0397981067571861e-06,
+      "loss": 0.9058,
+      "step": 23591
+    },
+    {
+      "epoch": 0.8579845074008073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0392751826522462e-06,
+      "loss": 0.8913,
+      "step": 23592
+    },
+    {
+      "epoch": 0.8580208750045459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0387523828646495e-06,
+      "loss": 0.958,
+      "step": 23593
+    },
+    {
+      "epoch": 0.8580572426082845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.038229707401649e-06,
+      "loss": 0.8052,
+      "step": 23594
+    },
+    {
+      "epoch": 0.8580936102120231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0377071562704944e-06,
+      "loss": 0.8556,
+      "step": 23595
+    },
+    {
+      "epoch": 0.8581299778157617,
+      "grad_norm": 0.0,
+      "learning_rate": 1.037184729478441e-06,
+      "loss": 1.0745,
+      "step": 23596
+    },
+    {
+      "epoch": 0.8581663454195003,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0366624270327286e-06,
+      "loss": 0.8665,
+      "step": 23597
+    },
+    {
+      "epoch": 0.8582027130232389,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0361402489406102e-06,
+      "loss": 0.8971,
+      "step": 23598
+    },
+    {
+      "epoch": 0.8582390806269775,
+      "grad_norm": 0.0,
+      "learning_rate": 1.035618195209328e-06,
+      "loss": 0.9743,
+      "step": 23599
+    },
+    {
+      "epoch": 0.8582754482307161,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0350962658461238e-06,
+      "loss": 0.9386,
+      "step": 23600
+    },
+    {
+      "epoch": 0.8583118158344547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0345744608582376e-06,
+      "loss": 0.8585,
+      "step": 23601
+    },
+    {
+      "epoch": 0.8583481834381933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0340527802529132e-06,
+      "loss": 0.9542,
+      "step": 23602
+    },
+    {
+      "epoch": 0.8583845510419319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0335312240373841e-06,
+      "loss": 0.8859,
+      "step": 23603
+    },
+    {
+      "epoch": 0.8584209186456704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0330097922188886e-06,
+      "loss": 0.8859,
+      "step": 23604
+    },
+    {
+      "epoch": 0.858457286249409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0324884848046601e-06,
+      "loss": 0.9005,
+      "step": 23605
+    },
+    {
+      "epoch": 0.8584936538531476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0319673018019294e-06,
+      "loss": 0.9547,
+      "step": 23606
+    },
+    {
+      "epoch": 0.8585300214568862,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0314462432179273e-06,
+      "loss": 0.8816,
+      "step": 23607
+    },
+    {
+      "epoch": 0.8585663890606248,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0309253090598858e-06,
+      "loss": 0.9078,
+      "step": 23608
+    },
+    {
+      "epoch": 0.8586027566643634,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0304044993350314e-06,
+      "loss": 0.9158,
+      "step": 23609
+    },
+    {
+      "epoch": 0.858639124268102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.029883814050585e-06,
+      "loss": 0.8731,
+      "step": 23610
+    },
+    {
+      "epoch": 0.8586754918718406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0293632532137755e-06,
+      "loss": 0.8902,
+      "step": 23611
+    },
+    {
+      "epoch": 0.8587118594755792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0288428168318244e-06,
+      "loss": 0.8872,
+      "step": 23612
+    },
+    {
+      "epoch": 0.8587482270793177,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0283225049119494e-06,
+      "loss": 0.8875,
+      "step": 23613
+    },
+    {
+      "epoch": 0.8587845946830563,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0278023174613682e-06,
+      "loss": 0.8565,
+      "step": 23614
+    },
+    {
+      "epoch": 0.8588209622867949,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0272822544873051e-06,
+      "loss": 0.9877,
+      "step": 23615
+    },
+    {
+      "epoch": 0.8588573298905335,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0267623159969652e-06,
+      "loss": 0.8148,
+      "step": 23616
+    },
+    {
+      "epoch": 0.8588936974942721,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0262425019975697e-06,
+      "loss": 0.8977,
+      "step": 23617
+    },
+    {
+      "epoch": 0.8589300650980107,
+      "grad_norm": 0.0,
+      "learning_rate": 1.025722812496327e-06,
+      "loss": 1.0456,
+      "step": 23618
+    },
+    {
+      "epoch": 0.8589664327017493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.025203247500447e-06,
+      "loss": 0.9251,
+      "step": 23619
+    },
+    {
+      "epoch": 0.8590028003054879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0246838070171373e-06,
+      "loss": 0.9241,
+      "step": 23620
+    },
+    {
+      "epoch": 0.8590391679092264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0241644910536063e-06,
+      "loss": 0.8898,
+      "step": 23621
+    },
+    {
+      "epoch": 0.859075535512965,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0236452996170588e-06,
+      "loss": 0.9674,
+      "step": 23622
+    },
+    {
+      "epoch": 0.8591119031167036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0231262327146973e-06,
+      "loss": 1.0322,
+      "step": 23623
+    },
+    {
+      "epoch": 0.8591482707204422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0226072903537198e-06,
+      "loss": 0.7417,
+      "step": 23624
+    },
+    {
+      "epoch": 0.8591846383241808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0220884725413337e-06,
+      "loss": 0.7958,
+      "step": 23625
+    },
+    {
+      "epoch": 0.8592210059279194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0215697792847278e-06,
+      "loss": 0.8714,
+      "step": 23626
+    },
+    {
+      "epoch": 0.859257373531658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0210512105911042e-06,
+      "loss": 1.0082,
+      "step": 23627
+    },
+    {
+      "epoch": 0.8592937411353966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0205327664676578e-06,
+      "loss": 0.8763,
+      "step": 23628
+    },
+    {
+      "epoch": 0.8593301087391352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0200144469215745e-06,
+      "loss": 0.9199,
+      "step": 23629
+    },
+    {
+      "epoch": 0.8593664763428738,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0194962519600525e-06,
+      "loss": 0.8466,
+      "step": 23630
+    },
+    {
+      "epoch": 0.8594028439466124,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0189781815902789e-06,
+      "loss": 0.9878,
+      "step": 23631
+    },
+    {
+      "epoch": 0.859439211550351,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0184602358194395e-06,
+      "loss": 0.926,
+      "step": 23632
+    },
+    {
+      "epoch": 0.8594755791540896,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0179424146547202e-06,
+      "loss": 0.8585,
+      "step": 23633
+    },
+    {
+      "epoch": 0.8595119467578282,
+      "grad_norm": 0.0,
+      "learning_rate": 1.017424718103307e-06,
+      "loss": 0.899,
+      "step": 23634
+    },
+    {
+      "epoch": 0.8595483143615668,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0169071461723812e-06,
+      "loss": 0.9191,
+      "step": 23635
+    },
+    {
+      "epoch": 0.8595846819653054,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0163896988691247e-06,
+      "loss": 1.0303,
+      "step": 23636
+    },
+    {
+      "epoch": 0.859621049569044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0158723762007138e-06,
+      "loss": 0.8789,
+      "step": 23637
+    },
+    {
+      "epoch": 0.8596574171727824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0153551781743275e-06,
+      "loss": 0.9453,
+      "step": 23638
+    },
+    {
+      "epoch": 0.859693784776521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0148381047971378e-06,
+      "loss": 0.9087,
+      "step": 23639
+    },
+    {
+      "epoch": 0.8597301523802596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0143211560763245e-06,
+      "loss": 0.9176,
+      "step": 23640
+    },
+    {
+      "epoch": 0.8597665199839982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0138043320190548e-06,
+      "loss": 0.9545,
+      "step": 23641
+    },
+    {
+      "epoch": 0.8598028875877368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0132876326325014e-06,
+      "loss": 0.998,
+      "step": 23642
+    },
+    {
+      "epoch": 0.8598392551914754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0127710579238281e-06,
+      "loss": 0.9245,
+      "step": 23643
+    },
+    {
+      "epoch": 0.859875622795214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0122546079002105e-06,
+      "loss": 0.9293,
+      "step": 23644
+    },
+    {
+      "epoch": 0.8599119903989526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0117382825688037e-06,
+      "loss": 0.9042,
+      "step": 23645
+    },
+    {
+      "epoch": 0.8599483580026912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0112220819367779e-06,
+      "loss": 0.9615,
+      "step": 23646
+    },
+    {
+      "epoch": 0.8599847256064298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0107060060112917e-06,
+      "loss": 0.8324,
+      "step": 23647
+    },
+    {
+      "epoch": 0.8600210932101684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0101900547995057e-06,
+      "loss": 0.9193,
+      "step": 23648
+    },
+    {
+      "epoch": 0.860057460813907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.009674228308578e-06,
+      "loss": 0.9001,
+      "step": 23649
+    },
+    {
+      "epoch": 0.8600938284176456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0091585265456649e-06,
+      "loss": 0.8999,
+      "step": 23650
+    },
+    {
+      "epoch": 0.8601301960213842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0086429495179217e-06,
+      "loss": 1.1138,
+      "step": 23651
+    },
+    {
+      "epoch": 0.8601665636251228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0081274972324972e-06,
+      "loss": 0.998,
+      "step": 23652
+    },
+    {
+      "epoch": 0.8602029312288614,
+      "grad_norm": 0.0,
+      "learning_rate": 1.007612169696548e-06,
+      "loss": 0.8381,
+      "step": 23653
+    },
+    {
+      "epoch": 0.8602392988326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0070969669172226e-06,
+      "loss": 0.8876,
+      "step": 23654
+    },
+    {
+      "epoch": 0.8602756664363385,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0065818889016665e-06,
+      "loss": 0.9647,
+      "step": 23655
+    },
+    {
+      "epoch": 0.8603120340400771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.006066935657024e-06,
+      "loss": 0.9503,
+      "step": 23656
+    },
+    {
+      "epoch": 0.8603484016438157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0055521071904473e-06,
+      "loss": 0.932,
+      "step": 23657
+    },
+    {
+      "epoch": 0.8603847692475542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0050374035090693e-06,
+      "loss": 0.919,
+      "step": 23658
+    },
+    {
+      "epoch": 0.8604211368512928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0045228246200367e-06,
+      "loss": 0.8083,
+      "step": 23659
+    },
+    {
+      "epoch": 0.8604575044550314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0040083705304882e-06,
+      "loss": 0.9364,
+      "step": 23660
+    },
+    {
+      "epoch": 0.86049387205877,
+      "grad_norm": 0.0,
+      "learning_rate": 1.003494041247559e-06,
+      "loss": 0.9932,
+      "step": 23661
+    },
+    {
+      "epoch": 0.8605302396625086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0029798367783838e-06,
+      "loss": 0.9769,
+      "step": 23662
+    },
+    {
+      "epoch": 0.8605666072662472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0024657571301023e-06,
+      "loss": 0.9383,
+      "step": 23663
+    },
+    {
+      "epoch": 0.8606029748699858,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0019518023098385e-06,
+      "loss": 0.9734,
+      "step": 23664
+    },
+    {
+      "epoch": 0.8606393424737244,
+      "grad_norm": 0.0,
+      "learning_rate": 1.001437972324728e-06,
+      "loss": 0.786,
+      "step": 23665
+    },
+    {
+      "epoch": 0.860675710077463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0009242671818998e-06,
+      "loss": 0.8679,
+      "step": 23666
+    },
+    {
+      "epoch": 0.8607120776812016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.000410686888479e-06,
+      "loss": 0.8509,
+      "step": 23667
+    },
+    {
+      "epoch": 0.8607484452849402,
+      "grad_norm": 0.0,
+      "learning_rate": 9.99897231451591e-07,
+      "loss": 1.0437,
+      "step": 23668
+    },
+    {
+      "epoch": 0.8607848128886788,
+      "grad_norm": 0.0,
+      "learning_rate": 9.993839008783568e-07,
+      "loss": 0.9445,
+      "step": 23669
+    },
+    {
+      "epoch": 0.8608211804924174,
+      "grad_norm": 0.0,
+      "learning_rate": 9.988706951759052e-07,
+      "loss": 0.8738,
+      "step": 23670
+    },
+    {
+      "epoch": 0.8608575480961559,
+      "grad_norm": 0.0,
+      "learning_rate": 9.983576143513484e-07,
+      "loss": 0.9277,
+      "step": 23671
+    },
+    {
+      "epoch": 0.8608939156998945,
+      "grad_norm": 0.0,
+      "learning_rate": 9.978446584118096e-07,
+      "loss": 1.0023,
+      "step": 23672
+    },
+    {
+      "epoch": 0.8609302833036331,
+      "grad_norm": 0.0,
+      "learning_rate": 9.973318273644028e-07,
+      "loss": 0.8439,
+      "step": 23673
+    },
+    {
+      "epoch": 0.8609666509073717,
+      "grad_norm": 0.0,
+      "learning_rate": 9.968191212162447e-07,
+      "loss": 0.9239,
+      "step": 23674
+    },
+    {
+      "epoch": 0.8610030185111103,
+      "grad_norm": 0.0,
+      "learning_rate": 9.963065399744464e-07,
+      "loss": 0.8515,
+      "step": 23675
+    },
+    {
+      "epoch": 0.8610393861148489,
+      "grad_norm": 0.0,
+      "learning_rate": 9.957940836461234e-07,
+      "loss": 0.7969,
+      "step": 23676
+    },
+    {
+      "epoch": 0.8610757537185875,
+      "grad_norm": 0.0,
+      "learning_rate": 9.952817522383772e-07,
+      "loss": 0.9554,
+      "step": 23677
+    },
+    {
+      "epoch": 0.8611121213223261,
+      "grad_norm": 0.0,
+      "learning_rate": 9.94769545758324e-07,
+      "loss": 0.7707,
+      "step": 23678
+    },
+    {
+      "epoch": 0.8611484889260647,
+      "grad_norm": 0.0,
+      "learning_rate": 9.942574642130665e-07,
+      "loss": 0.9056,
+      "step": 23679
+    },
+    {
+      "epoch": 0.8611848565298033,
+      "grad_norm": 0.0,
+      "learning_rate": 9.93745507609709e-07,
+      "loss": 0.8695,
+      "step": 23680
+    },
+    {
+      "epoch": 0.8612212241335419,
+      "grad_norm": 0.0,
+      "learning_rate": 9.932336759553519e-07,
+      "loss": 0.8596,
+      "step": 23681
+    },
+    {
+      "epoch": 0.8612575917372804,
+      "grad_norm": 0.0,
+      "learning_rate": 9.927219692571033e-07,
+      "loss": 0.889,
+      "step": 23682
+    },
+    {
+      "epoch": 0.861293959341019,
+      "grad_norm": 0.0,
+      "learning_rate": 9.922103875220534e-07,
+      "loss": 0.8596,
+      "step": 23683
+    },
+    {
+      "epoch": 0.8613303269447576,
+      "grad_norm": 0.0,
+      "learning_rate": 9.916989307573056e-07,
+      "loss": 0.9292,
+      "step": 23684
+    },
+    {
+      "epoch": 0.8613666945484962,
+      "grad_norm": 0.0,
+      "learning_rate": 9.911875989699548e-07,
+      "loss": 0.994,
+      "step": 23685
+    },
+    {
+      "epoch": 0.8614030621522348,
+      "grad_norm": 0.0,
+      "learning_rate": 9.906763921670936e-07,
+      "loss": 0.8113,
+      "step": 23686
+    },
+    {
+      "epoch": 0.8614394297559734,
+      "grad_norm": 0.0,
+      "learning_rate": 9.90165310355815e-07,
+      "loss": 0.9506,
+      "step": 23687
+    },
+    {
+      "epoch": 0.8614757973597119,
+      "grad_norm": 0.0,
+      "learning_rate": 9.896543535432079e-07,
+      "loss": 0.8568,
+      "step": 23688
+    },
+    {
+      "epoch": 0.8615121649634505,
+      "grad_norm": 0.0,
+      "learning_rate": 9.891435217363666e-07,
+      "loss": 0.9117,
+      "step": 23689
+    },
+    {
+      "epoch": 0.8615485325671891,
+      "grad_norm": 0.0,
+      "learning_rate": 9.886328149423697e-07,
+      "loss": 0.9098,
+      "step": 23690
+    },
+    {
+      "epoch": 0.8615849001709277,
+      "grad_norm": 0.0,
+      "learning_rate": 9.881222331683095e-07,
+      "loss": 0.9849,
+      "step": 23691
+    },
+    {
+      "epoch": 0.8616212677746663,
+      "grad_norm": 0.0,
+      "learning_rate": 9.876117764212668e-07,
+      "loss": 0.994,
+      "step": 23692
+    },
+    {
+      "epoch": 0.8616576353784049,
+      "grad_norm": 0.0,
+      "learning_rate": 9.871014447083239e-07,
+      "loss": 0.8351,
+      "step": 23693
+    },
+    {
+      "epoch": 0.8616940029821435,
+      "grad_norm": 0.0,
+      "learning_rate": 9.865912380365594e-07,
+      "loss": 0.9118,
+      "step": 23694
+    },
+    {
+      "epoch": 0.8617303705858821,
+      "grad_norm": 0.0,
+      "learning_rate": 9.860811564130556e-07,
+      "loss": 0.9394,
+      "step": 23695
+    },
+    {
+      "epoch": 0.8617667381896207,
+      "grad_norm": 0.0,
+      "learning_rate": 9.855711998448846e-07,
+      "loss": 0.9141,
+      "step": 23696
+    },
+    {
+      "epoch": 0.8618031057933593,
+      "grad_norm": 0.0,
+      "learning_rate": 9.850613683391242e-07,
+      "loss": 0.9265,
+      "step": 23697
+    },
+    {
+      "epoch": 0.8618394733970979,
+      "grad_norm": 0.0,
+      "learning_rate": 9.845516619028473e-07,
+      "loss": 0.9383,
+      "step": 23698
+    },
+    {
+      "epoch": 0.8618758410008365,
+      "grad_norm": 0.0,
+      "learning_rate": 9.840420805431239e-07,
+      "loss": 0.8516,
+      "step": 23699
+    },
+    {
+      "epoch": 0.8619122086045751,
+      "grad_norm": 0.0,
+      "learning_rate": 9.83532624267024e-07,
+      "loss": 0.8933,
+      "step": 23700
+    },
+    {
+      "epoch": 0.8619485762083137,
+      "grad_norm": 0.0,
+      "learning_rate": 9.830232930816174e-07,
+      "loss": 0.8977,
+      "step": 23701
+    },
+    {
+      "epoch": 0.8619849438120523,
+      "grad_norm": 0.0,
+      "learning_rate": 9.825140869939686e-07,
+      "loss": 0.9515,
+      "step": 23702
+    },
+    {
+      "epoch": 0.8620213114157909,
+      "grad_norm": 0.0,
+      "learning_rate": 9.82005006011144e-07,
+      "loss": 0.929,
+      "step": 23703
+    },
+    {
+      "epoch": 0.8620576790195295,
+      "grad_norm": 0.0,
+      "learning_rate": 9.814960501402048e-07,
+      "loss": 0.9158,
+      "step": 23704
+    },
+    {
+      "epoch": 0.8620940466232679,
+      "grad_norm": 0.0,
+      "learning_rate": 9.80987219388212e-07,
+      "loss": 0.7991,
+      "step": 23705
+    },
+    {
+      "epoch": 0.8621304142270065,
+      "grad_norm": 0.0,
+      "learning_rate": 9.804785137622253e-07,
+      "loss": 0.9326,
+      "step": 23706
+    },
+    {
+      "epoch": 0.8621667818307451,
+      "grad_norm": 0.0,
+      "learning_rate": 9.799699332693002e-07,
+      "loss": 0.9352,
+      "step": 23707
+    },
+    {
+      "epoch": 0.8622031494344837,
+      "grad_norm": 0.0,
+      "learning_rate": 9.79461477916498e-07,
+      "loss": 0.9215,
+      "step": 23708
+    },
+    {
+      "epoch": 0.8622395170382223,
+      "grad_norm": 0.0,
+      "learning_rate": 9.78953147710866e-07,
+      "loss": 0.9301,
+      "step": 23709
+    },
+    {
+      "epoch": 0.8622758846419609,
+      "grad_norm": 0.0,
+      "learning_rate": 9.784449426594622e-07,
+      "loss": 0.9569,
+      "step": 23710
+    },
+    {
+      "epoch": 0.8623122522456995,
+      "grad_norm": 0.0,
+      "learning_rate": 9.779368627693343e-07,
+      "loss": 0.9891,
+      "step": 23711
+    },
+    {
+      "epoch": 0.8623486198494381,
+      "grad_norm": 0.0,
+      "learning_rate": 9.774289080475319e-07,
+      "loss": 0.865,
+      "step": 23712
+    },
+    {
+      "epoch": 0.8623849874531767,
+      "grad_norm": 0.0,
+      "learning_rate": 9.769210785010995e-07,
+      "loss": 0.7632,
+      "step": 23713
+    },
+    {
+      "epoch": 0.8624213550569153,
+      "grad_norm": 0.0,
+      "learning_rate": 9.764133741370886e-07,
+      "loss": 0.9008,
+      "step": 23714
+    },
+    {
+      "epoch": 0.8624577226606539,
+      "grad_norm": 0.0,
+      "learning_rate": 9.759057949625383e-07,
+      "loss": 1.005,
+      "step": 23715
+    },
+    {
+      "epoch": 0.8624940902643925,
+      "grad_norm": 0.0,
+      "learning_rate": 9.753983409844913e-07,
+      "loss": 0.9454,
+      "step": 23716
+    },
+    {
+      "epoch": 0.8625304578681311,
+      "grad_norm": 0.0,
+      "learning_rate": 9.748910122099886e-07,
+      "loss": 0.8177,
+      "step": 23717
+    },
+    {
+      "epoch": 0.8625668254718697,
+      "grad_norm": 0.0,
+      "learning_rate": 9.743838086460677e-07,
+      "loss": 0.8682,
+      "step": 23718
+    },
+    {
+      "epoch": 0.8626031930756083,
+      "grad_norm": 0.0,
+      "learning_rate": 9.738767302997643e-07,
+      "loss": 0.9896,
+      "step": 23719
+    },
+    {
+      "epoch": 0.8626395606793469,
+      "grad_norm": 0.0,
+      "learning_rate": 9.73369777178117e-07,
+      "loss": 0.9684,
+      "step": 23720
+    },
+    {
+      "epoch": 0.8626759282830855,
+      "grad_norm": 0.0,
+      "learning_rate": 9.728629492881557e-07,
+      "loss": 0.9533,
+      "step": 23721
+    },
+    {
+      "epoch": 0.862712295886824,
+      "grad_norm": 0.0,
+      "learning_rate": 9.72356246636914e-07,
+      "loss": 0.88,
+      "step": 23722
+    },
+    {
+      "epoch": 0.8627486634905626,
+      "grad_norm": 0.0,
+      "learning_rate": 9.718496692314217e-07,
+      "loss": 0.885,
+      "step": 23723
+    },
+    {
+      "epoch": 0.8627850310943012,
+      "grad_norm": 0.0,
+      "learning_rate": 9.713432170787051e-07,
+      "loss": 0.8324,
+      "step": 23724
+    },
+    {
+      "epoch": 0.8628213986980398,
+      "grad_norm": 0.0,
+      "learning_rate": 9.708368901857912e-07,
+      "loss": 0.9279,
+      "step": 23725
+    },
+    {
+      "epoch": 0.8628577663017784,
+      "grad_norm": 0.0,
+      "learning_rate": 9.70330688559703e-07,
+      "loss": 0.8996,
+      "step": 23726
+    },
+    {
+      "epoch": 0.862894133905517,
+      "grad_norm": 0.0,
+      "learning_rate": 9.698246122074684e-07,
+      "loss": 0.8624,
+      "step": 23727
+    },
+    {
+      "epoch": 0.8629305015092555,
+      "grad_norm": 0.0,
+      "learning_rate": 9.693186611361017e-07,
+      "loss": 1.0234,
+      "step": 23728
+    },
+    {
+      "epoch": 0.8629668691129941,
+      "grad_norm": 0.0,
+      "learning_rate": 9.688128353526283e-07,
+      "loss": 0.9335,
+      "step": 23729
+    },
+    {
+      "epoch": 0.8630032367167327,
+      "grad_norm": 0.0,
+      "learning_rate": 9.683071348640627e-07,
+      "loss": 0.9038,
+      "step": 23730
+    },
+    {
+      "epoch": 0.8630396043204713,
+      "grad_norm": 0.0,
+      "learning_rate": 9.678015596774215e-07,
+      "loss": 0.8896,
+      "step": 23731
+    },
+    {
+      "epoch": 0.8630759719242099,
+      "grad_norm": 0.0,
+      "learning_rate": 9.672961097997158e-07,
+      "loss": 0.9364,
+      "step": 23732
+    },
+    {
+      "epoch": 0.8631123395279485,
+      "grad_norm": 0.0,
+      "learning_rate": 9.667907852379643e-07,
+      "loss": 0.8911,
+      "step": 23733
+    },
+    {
+      "epoch": 0.8631487071316871,
+      "grad_norm": 0.0,
+      "learning_rate": 9.662855859991726e-07,
+      "loss": 0.9436,
+      "step": 23734
+    },
+    {
+      "epoch": 0.8631850747354257,
+      "grad_norm": 0.0,
+      "learning_rate": 9.65780512090353e-07,
+      "loss": 0.8934,
+      "step": 23735
+    },
+    {
+      "epoch": 0.8632214423391643,
+      "grad_norm": 0.0,
+      "learning_rate": 9.652755635185073e-07,
+      "loss": 0.8545,
+      "step": 23736
+    },
+    {
+      "epoch": 0.8632578099429029,
+      "grad_norm": 0.0,
+      "learning_rate": 9.647707402906492e-07,
+      "loss": 0.9222,
+      "step": 23737
+    },
+    {
+      "epoch": 0.8632941775466415,
+      "grad_norm": 0.0,
+      "learning_rate": 9.64266042413774e-07,
+      "loss": 1.0301,
+      "step": 23738
+    },
+    {
+      "epoch": 0.86333054515038,
+      "grad_norm": 0.0,
+      "learning_rate": 9.637614698948893e-07,
+      "loss": 0.8173,
+      "step": 23739
+    },
+    {
+      "epoch": 0.8633669127541186,
+      "grad_norm": 0.0,
+      "learning_rate": 9.632570227409932e-07,
+      "loss": 0.9503,
+      "step": 23740
+    },
+    {
+      "epoch": 0.8634032803578572,
+      "grad_norm": 0.0,
+      "learning_rate": 9.627527009590843e-07,
+      "loss": 0.9017,
+      "step": 23741
+    },
+    {
+      "epoch": 0.8634396479615958,
+      "grad_norm": 0.0,
+      "learning_rate": 9.622485045561602e-07,
+      "loss": 0.854,
+      "step": 23742
+    },
+    {
+      "epoch": 0.8634760155653344,
+      "grad_norm": 0.0,
+      "learning_rate": 9.617444335392146e-07,
+      "loss": 0.9156,
+      "step": 23743
+    },
+    {
+      "epoch": 0.863512383169073,
+      "grad_norm": 0.0,
+      "learning_rate": 9.612404879152416e-07,
+      "loss": 1.0349,
+      "step": 23744
+    },
+    {
+      "epoch": 0.8635487507728116,
+      "grad_norm": 0.0,
+      "learning_rate": 9.60736667691231e-07,
+      "loss": 1.0032,
+      "step": 23745
+    },
+    {
+      "epoch": 0.8635851183765502,
+      "grad_norm": 0.0,
+      "learning_rate": 9.602329728741756e-07,
+      "loss": 0.8506,
+      "step": 23746
+    },
+    {
+      "epoch": 0.8636214859802888,
+      "grad_norm": 0.0,
+      "learning_rate": 9.597294034710614e-07,
+      "loss": 0.9014,
+      "step": 23747
+    },
+    {
+      "epoch": 0.8636578535840274,
+      "grad_norm": 0.0,
+      "learning_rate": 9.592259594888764e-07,
+      "loss": 0.9142,
+      "step": 23748
+    },
+    {
+      "epoch": 0.863694221187766,
+      "grad_norm": 0.0,
+      "learning_rate": 9.587226409346028e-07,
+      "loss": 0.8515,
+      "step": 23749
+    },
+    {
+      "epoch": 0.8637305887915045,
+      "grad_norm": 0.0,
+      "learning_rate": 9.58219447815225e-07,
+      "loss": 0.8536,
+      "step": 23750
+    },
+    {
+      "epoch": 0.8637669563952431,
+      "grad_norm": 0.0,
+      "learning_rate": 9.577163801377232e-07,
+      "loss": 0.9017,
+      "step": 23751
+    },
+    {
+      "epoch": 0.8638033239989817,
+      "grad_norm": 0.0,
+      "learning_rate": 9.57213437909078e-07,
+      "loss": 0.9177,
+      "step": 23752
+    },
+    {
+      "epoch": 0.8638396916027203,
+      "grad_norm": 0.0,
+      "learning_rate": 9.567106211362664e-07,
+      "loss": 1.0155,
+      "step": 23753
+    },
+    {
+      "epoch": 0.8638760592064589,
+      "grad_norm": 0.0,
+      "learning_rate": 9.562079298262638e-07,
+      "loss": 0.9893,
+      "step": 23754
+    },
+    {
+      "epoch": 0.8639124268101975,
+      "grad_norm": 0.0,
+      "learning_rate": 9.557053639860436e-07,
+      "loss": 0.8929,
+      "step": 23755
+    },
+    {
+      "epoch": 0.863948794413936,
+      "grad_norm": 0.0,
+      "learning_rate": 9.552029236225824e-07,
+      "loss": 0.9064,
+      "step": 23756
+    },
+    {
+      "epoch": 0.8639851620176746,
+      "grad_norm": 0.0,
+      "learning_rate": 9.547006087428445e-07,
+      "loss": 1.0233,
+      "step": 23757
+    },
+    {
+      "epoch": 0.8640215296214132,
+      "grad_norm": 0.0,
+      "learning_rate": 9.541984193538033e-07,
+      "loss": 1.0427,
+      "step": 23758
+    },
+    {
+      "epoch": 0.8640578972251518,
+      "grad_norm": 0.0,
+      "learning_rate": 9.536963554624257e-07,
+      "loss": 0.8737,
+      "step": 23759
+    },
+    {
+      "epoch": 0.8640942648288904,
+      "grad_norm": 0.0,
+      "learning_rate": 9.531944170756757e-07,
+      "loss": 0.8299,
+      "step": 23760
+    },
+    {
+      "epoch": 0.864130632432629,
+      "grad_norm": 0.0,
+      "learning_rate": 9.526926042005169e-07,
+      "loss": 0.9658,
+      "step": 23761
+    },
+    {
+      "epoch": 0.8641670000363676,
+      "grad_norm": 0.0,
+      "learning_rate": 9.521909168439125e-07,
+      "loss": 0.9662,
+      "step": 23762
+    },
+    {
+      "epoch": 0.8642033676401062,
+      "grad_norm": 0.0,
+      "learning_rate": 9.516893550128215e-07,
+      "loss": 0.9173,
+      "step": 23763
+    },
+    {
+      "epoch": 0.8642397352438448,
+      "grad_norm": 0.0,
+      "learning_rate": 9.511879187142015e-07,
+      "loss": 0.934,
+      "step": 23764
+    },
+    {
+      "epoch": 0.8642761028475834,
+      "grad_norm": 0.0,
+      "learning_rate": 9.506866079550126e-07,
+      "loss": 0.8896,
+      "step": 23765
+    },
+    {
+      "epoch": 0.864312470451322,
+      "grad_norm": 0.0,
+      "learning_rate": 9.501854227422069e-07,
+      "loss": 0.931,
+      "step": 23766
+    },
+    {
+      "epoch": 0.8643488380550606,
+      "grad_norm": 0.0,
+      "learning_rate": 9.496843630827379e-07,
+      "loss": 0.9211,
+      "step": 23767
+    },
+    {
+      "epoch": 0.8643852056587992,
+      "grad_norm": 0.0,
+      "learning_rate": 9.491834289835566e-07,
+      "loss": 0.9992,
+      "step": 23768
+    },
+    {
+      "epoch": 0.8644215732625378,
+      "grad_norm": 0.0,
+      "learning_rate": 9.486826204516175e-07,
+      "loss": 0.7959,
+      "step": 23769
+    },
+    {
+      "epoch": 0.8644579408662764,
+      "grad_norm": 0.0,
+      "learning_rate": 9.481819374938617e-07,
+      "loss": 0.9421,
+      "step": 23770
+    },
+    {
+      "epoch": 0.864494308470015,
+      "grad_norm": 0.0,
+      "learning_rate": 9.476813801172402e-07,
+      "loss": 0.9302,
+      "step": 23771
+    },
+    {
+      "epoch": 0.8645306760737534,
+      "grad_norm": 0.0,
+      "learning_rate": 9.471809483286953e-07,
+      "loss": 0.9301,
+      "step": 23772
+    },
+    {
+      "epoch": 0.864567043677492,
+      "grad_norm": 0.0,
+      "learning_rate": 9.466806421351715e-07,
+      "loss": 0.8512,
+      "step": 23773
+    },
+    {
+      "epoch": 0.8646034112812306,
+      "grad_norm": 0.0,
+      "learning_rate": 9.461804615436066e-07,
+      "loss": 0.9464,
+      "step": 23774
+    },
+    {
+      "epoch": 0.8646397788849692,
+      "grad_norm": 0.0,
+      "learning_rate": 9.456804065609448e-07,
+      "loss": 0.9779,
+      "step": 23775
+    },
+    {
+      "epoch": 0.8646761464887078,
+      "grad_norm": 0.0,
+      "learning_rate": 9.451804771941187e-07,
+      "loss": 0.8287,
+      "step": 23776
+    },
+    {
+      "epoch": 0.8647125140924464,
+      "grad_norm": 0.0,
+      "learning_rate": 9.44680673450068e-07,
+      "loss": 0.9128,
+      "step": 23777
+    },
+    {
+      "epoch": 0.864748881696185,
+      "grad_norm": 0.0,
+      "learning_rate": 9.441809953357239e-07,
+      "loss": 0.8734,
+      "step": 23778
+    },
+    {
+      "epoch": 0.8647852492999236,
+      "grad_norm": 0.0,
+      "learning_rate": 9.43681442858021e-07,
+      "loss": 0.9403,
+      "step": 23779
+    },
+    {
+      "epoch": 0.8648216169036622,
+      "grad_norm": 0.0,
+      "learning_rate": 9.431820160238891e-07,
+      "loss": 0.8173,
+      "step": 23780
+    },
+    {
+      "epoch": 0.8648579845074008,
+      "grad_norm": 0.0,
+      "learning_rate": 9.426827148402562e-07,
+      "loss": 0.922,
+      "step": 23781
+    },
+    {
+      "epoch": 0.8648943521111394,
+      "grad_norm": 0.0,
+      "learning_rate": 9.421835393140499e-07,
+      "loss": 1.0031,
+      "step": 23782
+    },
+    {
+      "epoch": 0.864930719714878,
+      "grad_norm": 0.0,
+      "learning_rate": 9.416844894521937e-07,
+      "loss": 0.9265,
+      "step": 23783
+    },
+    {
+      "epoch": 0.8649670873186166,
+      "grad_norm": 0.0,
+      "learning_rate": 9.411855652616153e-07,
+      "loss": 0.8979,
+      "step": 23784
+    },
+    {
+      "epoch": 0.8650034549223552,
+      "grad_norm": 0.0,
+      "learning_rate": 9.406867667492347e-07,
+      "loss": 0.8666,
+      "step": 23785
+    },
+    {
+      "epoch": 0.8650398225260938,
+      "grad_norm": 0.0,
+      "learning_rate": 9.401880939219721e-07,
+      "loss": 0.8463,
+      "step": 23786
+    },
+    {
+      "epoch": 0.8650761901298324,
+      "grad_norm": 0.0,
+      "learning_rate": 9.396895467867428e-07,
+      "loss": 0.8944,
+      "step": 23787
+    },
+    {
+      "epoch": 0.865112557733571,
+      "grad_norm": 0.0,
+      "learning_rate": 9.391911253504693e-07,
+      "loss": 1.0185,
+      "step": 23788
+    },
+    {
+      "epoch": 0.8651489253373095,
+      "grad_norm": 0.0,
+      "learning_rate": 9.386928296200603e-07,
+      "loss": 0.899,
+      "step": 23789
+    },
+    {
+      "epoch": 0.8651852929410481,
+      "grad_norm": 0.0,
+      "learning_rate": 9.381946596024349e-07,
+      "loss": 0.9747,
+      "step": 23790
+    },
+    {
+      "epoch": 0.8652216605447867,
+      "grad_norm": 0.0,
+      "learning_rate": 9.376966153044997e-07,
+      "loss": 0.8587,
+      "step": 23791
+    },
+    {
+      "epoch": 0.8652580281485253,
+      "grad_norm": 0.0,
+      "learning_rate": 9.37198696733167e-07,
+      "loss": 0.9032,
+      "step": 23792
+    },
+    {
+      "epoch": 0.8652943957522639,
+      "grad_norm": 0.0,
+      "learning_rate": 9.367009038953423e-07,
+      "loss": 0.9643,
+      "step": 23793
+    },
+    {
+      "epoch": 0.8653307633560025,
+      "grad_norm": 0.0,
+      "learning_rate": 9.362032367979379e-07,
+      "loss": 1.0238,
+      "step": 23794
+    },
+    {
+      "epoch": 0.865367130959741,
+      "grad_norm": 0.0,
+      "learning_rate": 9.357056954478482e-07,
+      "loss": 0.9342,
+      "step": 23795
+    },
+    {
+      "epoch": 0.8654034985634796,
+      "grad_norm": 0.0,
+      "learning_rate": 9.352082798519846e-07,
+      "loss": 0.8693,
+      "step": 23796
+    },
+    {
+      "epoch": 0.8654398661672182,
+      "grad_norm": 0.0,
+      "learning_rate": 9.347109900172446e-07,
+      "loss": 0.7859,
+      "step": 23797
+    },
+    {
+      "epoch": 0.8654762337709568,
+      "grad_norm": 0.0,
+      "learning_rate": 9.342138259505273e-07,
+      "loss": 0.8703,
+      "step": 23798
+    },
+    {
+      "epoch": 0.8655126013746954,
+      "grad_norm": 0.0,
+      "learning_rate": 9.337167876587305e-07,
+      "loss": 0.9139,
+      "step": 23799
+    },
+    {
+      "epoch": 0.865548968978434,
+      "grad_norm": 0.0,
+      "learning_rate": 9.332198751487486e-07,
+      "loss": 0.8705,
+      "step": 23800
+    },
+    {
+      "epoch": 0.8655853365821726,
+      "grad_norm": 0.0,
+      "learning_rate": 9.327230884274807e-07,
+      "loss": 0.8586,
+      "step": 23801
+    },
+    {
+      "epoch": 0.8656217041859112,
+      "grad_norm": 0.0,
+      "learning_rate": 9.322264275018111e-07,
+      "loss": 0.8133,
+      "step": 23802
+    },
+    {
+      "epoch": 0.8656580717896498,
+      "grad_norm": 0.0,
+      "learning_rate": 9.317298923786355e-07,
+      "loss": 0.7906,
+      "step": 23803
+    },
+    {
+      "epoch": 0.8656944393933884,
+      "grad_norm": 0.0,
+      "learning_rate": 9.312334830648417e-07,
+      "loss": 0.9225,
+      "step": 23804
+    },
+    {
+      "epoch": 0.865730806997127,
+      "grad_norm": 0.0,
+      "learning_rate": 9.307371995673176e-07,
+      "loss": 0.9453,
+      "step": 23805
+    },
+    {
+      "epoch": 0.8657671746008655,
+      "grad_norm": 0.0,
+      "learning_rate": 9.302410418929442e-07,
+      "loss": 0.9631,
+      "step": 23806
+    },
+    {
+      "epoch": 0.8658035422046041,
+      "grad_norm": 0.0,
+      "learning_rate": 9.297450100486127e-07,
+      "loss": 0.9756,
+      "step": 23807
+    },
+    {
+      "epoch": 0.8658399098083427,
+      "grad_norm": 0.0,
+      "learning_rate": 9.292491040411955e-07,
+      "loss": 0.9227,
+      "step": 23808
+    },
+    {
+      "epoch": 0.8658762774120813,
+      "grad_norm": 0.0,
+      "learning_rate": 9.287533238775792e-07,
+      "loss": 0.8039,
+      "step": 23809
+    },
+    {
+      "epoch": 0.8659126450158199,
+      "grad_norm": 0.0,
+      "learning_rate": 9.282576695646406e-07,
+      "loss": 1.0001,
+      "step": 23810
+    },
+    {
+      "epoch": 0.8659490126195585,
+      "grad_norm": 0.0,
+      "learning_rate": 9.277621411092552e-07,
+      "loss": 1.0433,
+      "step": 23811
+    },
+    {
+      "epoch": 0.8659853802232971,
+      "grad_norm": 0.0,
+      "learning_rate": 9.272667385182976e-07,
+      "loss": 0.9621,
+      "step": 23812
+    },
+    {
+      "epoch": 0.8660217478270357,
+      "grad_norm": 0.0,
+      "learning_rate": 9.267714617986423e-07,
+      "loss": 0.9373,
+      "step": 23813
+    },
+    {
+      "epoch": 0.8660581154307743,
+      "grad_norm": 0.0,
+      "learning_rate": 9.262763109571593e-07,
+      "loss": 0.9133,
+      "step": 23814
+    },
+    {
+      "epoch": 0.8660944830345129,
+      "grad_norm": 0.0,
+      "learning_rate": 9.257812860007198e-07,
+      "loss": 0.8293,
+      "step": 23815
+    },
+    {
+      "epoch": 0.8661308506382515,
+      "grad_norm": 0.0,
+      "learning_rate": 9.252863869361894e-07,
+      "loss": 0.9294,
+      "step": 23816
+    },
+    {
+      "epoch": 0.8661672182419901,
+      "grad_norm": 0.0,
+      "learning_rate": 9.247916137704349e-07,
+      "loss": 0.8885,
+      "step": 23817
+    },
+    {
+      "epoch": 0.8662035858457287,
+      "grad_norm": 0.0,
+      "learning_rate": 9.242969665103197e-07,
+      "loss": 0.8679,
+      "step": 23818
+    },
+    {
+      "epoch": 0.8662399534494672,
+      "grad_norm": 0.0,
+      "learning_rate": 9.238024451627092e-07,
+      "loss": 0.9259,
+      "step": 23819
+    },
+    {
+      "epoch": 0.8662763210532058,
+      "grad_norm": 0.0,
+      "learning_rate": 9.233080497344637e-07,
+      "loss": 0.8179,
+      "step": 23820
+    },
+    {
+      "epoch": 0.8663126886569444,
+      "grad_norm": 0.0,
+      "learning_rate": 9.228137802324366e-07,
+      "loss": 1.0232,
+      "step": 23821
+    },
+    {
+      "epoch": 0.866349056260683,
+      "grad_norm": 0.0,
+      "learning_rate": 9.223196366634924e-07,
+      "loss": 0.881,
+      "step": 23822
+    },
+    {
+      "epoch": 0.8663854238644215,
+      "grad_norm": 0.0,
+      "learning_rate": 9.218256190344832e-07,
+      "loss": 0.7968,
+      "step": 23823
+    },
+    {
+      "epoch": 0.8664217914681601,
+      "grad_norm": 0.0,
+      "learning_rate": 9.213317273522637e-07,
+      "loss": 0.9769,
+      "step": 23824
+    },
+    {
+      "epoch": 0.8664581590718987,
+      "grad_norm": 0.0,
+      "learning_rate": 9.208379616236828e-07,
+      "loss": 0.9177,
+      "step": 23825
+    },
+    {
+      "epoch": 0.8664945266756373,
+      "grad_norm": 0.0,
+      "learning_rate": 9.203443218555985e-07,
+      "loss": 0.8549,
+      "step": 23826
+    },
+    {
+      "epoch": 0.8665308942793759,
+      "grad_norm": 0.0,
+      "learning_rate": 9.198508080548497e-07,
+      "loss": 0.85,
+      "step": 23827
+    },
+    {
+      "epoch": 0.8665672618831145,
+      "grad_norm": 0.0,
+      "learning_rate": 9.193574202282895e-07,
+      "loss": 0.9659,
+      "step": 23828
+    },
+    {
+      "epoch": 0.8666036294868531,
+      "grad_norm": 0.0,
+      "learning_rate": 9.188641583827629e-07,
+      "loss": 0.8936,
+      "step": 23829
+    },
+    {
+      "epoch": 0.8666399970905917,
+      "grad_norm": 0.0,
+      "learning_rate": 9.183710225251097e-07,
+      "loss": 0.9357,
+      "step": 23830
+    },
+    {
+      "epoch": 0.8666763646943303,
+      "grad_norm": 0.0,
+      "learning_rate": 9.178780126621734e-07,
+      "loss": 0.9612,
+      "step": 23831
+    },
+    {
+      "epoch": 0.8667127322980689,
+      "grad_norm": 0.0,
+      "learning_rate": 9.17385128800794e-07,
+      "loss": 0.9034,
+      "step": 23832
+    },
+    {
+      "epoch": 0.8667490999018075,
+      "grad_norm": 0.0,
+      "learning_rate": 9.168923709478106e-07,
+      "loss": 0.8603,
+      "step": 23833
+    },
+    {
+      "epoch": 0.8667854675055461,
+      "grad_norm": 0.0,
+      "learning_rate": 9.163997391100577e-07,
+      "loss": 0.8589,
+      "step": 23834
+    },
+    {
+      "epoch": 0.8668218351092847,
+      "grad_norm": 0.0,
+      "learning_rate": 9.159072332943686e-07,
+      "loss": 0.9548,
+      "step": 23835
+    },
+    {
+      "epoch": 0.8668582027130233,
+      "grad_norm": 0.0,
+      "learning_rate": 9.154148535075824e-07,
+      "loss": 0.8433,
+      "step": 23836
+    },
+    {
+      "epoch": 0.8668945703167619,
+      "grad_norm": 0.0,
+      "learning_rate": 9.149225997565214e-07,
+      "loss": 1.0009,
+      "step": 23837
+    },
+    {
+      "epoch": 0.8669309379205005,
+      "grad_norm": 0.0,
+      "learning_rate": 9.144304720480212e-07,
+      "loss": 0.903,
+      "step": 23838
+    },
+    {
+      "epoch": 0.8669673055242391,
+      "grad_norm": 0.0,
+      "learning_rate": 9.139384703889098e-07,
+      "loss": 0.9097,
+      "step": 23839
+    },
+    {
+      "epoch": 0.8670036731279775,
+      "grad_norm": 0.0,
+      "learning_rate": 9.134465947860061e-07,
+      "loss": 0.7366,
+      "step": 23840
+    },
+    {
+      "epoch": 0.8670400407317161,
+      "grad_norm": 0.0,
+      "learning_rate": 9.129548452461412e-07,
+      "loss": 0.8857,
+      "step": 23841
+    },
+    {
+      "epoch": 0.8670764083354547,
+      "grad_norm": 0.0,
+      "learning_rate": 9.124632217761354e-07,
+      "loss": 0.9403,
+      "step": 23842
+    },
+    {
+      "epoch": 0.8671127759391933,
+      "grad_norm": 0.0,
+      "learning_rate": 9.119717243828075e-07,
+      "loss": 1.0001,
+      "step": 23843
+    },
+    {
+      "epoch": 0.8671491435429319,
+      "grad_norm": 0.0,
+      "learning_rate": 9.114803530729777e-07,
+      "loss": 0.9161,
+      "step": 23844
+    },
+    {
+      "epoch": 0.8671855111466705,
+      "grad_norm": 0.0,
+      "learning_rate": 9.109891078534639e-07,
+      "loss": 0.8991,
+      "step": 23845
+    },
+    {
+      "epoch": 0.8672218787504091,
+      "grad_norm": 0.0,
+      "learning_rate": 9.104979887310805e-07,
+      "loss": 0.7728,
+      "step": 23846
+    },
+    {
+      "epoch": 0.8672582463541477,
+      "grad_norm": 0.0,
+      "learning_rate": 9.100069957126412e-07,
+      "loss": 0.9706,
+      "step": 23847
+    },
+    {
+      "epoch": 0.8672946139578863,
+      "grad_norm": 0.0,
+      "learning_rate": 9.095161288049581e-07,
+      "loss": 0.9273,
+      "step": 23848
+    },
+    {
+      "epoch": 0.8673309815616249,
+      "grad_norm": 0.0,
+      "learning_rate": 9.090253880148403e-07,
+      "loss": 0.9409,
+      "step": 23849
+    },
+    {
+      "epoch": 0.8673673491653635,
+      "grad_norm": 0.0,
+      "learning_rate": 9.085347733490956e-07,
+      "loss": 0.9392,
+      "step": 23850
+    },
+    {
+      "epoch": 0.8674037167691021,
+      "grad_norm": 0.0,
+      "learning_rate": 9.080442848145344e-07,
+      "loss": 1.0091,
+      "step": 23851
+    },
+    {
+      "epoch": 0.8674400843728407,
+      "grad_norm": 0.0,
+      "learning_rate": 9.075539224179574e-07,
+      "loss": 0.8685,
+      "step": 23852
+    },
+    {
+      "epoch": 0.8674764519765793,
+      "grad_norm": 0.0,
+      "learning_rate": 9.070636861661708e-07,
+      "loss": 0.8761,
+      "step": 23853
+    },
+    {
+      "epoch": 0.8675128195803179,
+      "grad_norm": 0.0,
+      "learning_rate": 9.065735760659711e-07,
+      "loss": 0.8676,
+      "step": 23854
+    },
+    {
+      "epoch": 0.8675491871840565,
+      "grad_norm": 0.0,
+      "learning_rate": 9.060835921241651e-07,
+      "loss": 0.8657,
+      "step": 23855
+    },
+    {
+      "epoch": 0.867585554787795,
+      "grad_norm": 0.0,
+      "learning_rate": 9.055937343475441e-07,
+      "loss": 0.9933,
+      "step": 23856
+    },
+    {
+      "epoch": 0.8676219223915336,
+      "grad_norm": 0.0,
+      "learning_rate": 9.051040027429081e-07,
+      "loss": 0.8016,
+      "step": 23857
+    },
+    {
+      "epoch": 0.8676582899952722,
+      "grad_norm": 0.0,
+      "learning_rate": 9.046143973170496e-07,
+      "loss": 0.8727,
+      "step": 23858
+    },
+    {
+      "epoch": 0.8676946575990108,
+      "grad_norm": 0.0,
+      "learning_rate": 9.04124918076762e-07,
+      "loss": 0.8446,
+      "step": 23859
+    },
+    {
+      "epoch": 0.8677310252027494,
+      "grad_norm": 0.0,
+      "learning_rate": 9.036355650288353e-07,
+      "loss": 0.9765,
+      "step": 23860
+    },
+    {
+      "epoch": 0.867767392806488,
+      "grad_norm": 0.0,
+      "learning_rate": 9.031463381800598e-07,
+      "loss": 0.8454,
+      "step": 23861
+    },
+    {
+      "epoch": 0.8678037604102266,
+      "grad_norm": 0.0,
+      "learning_rate": 9.026572375372222e-07,
+      "loss": 0.9345,
+      "step": 23862
+    },
+    {
+      "epoch": 0.8678401280139652,
+      "grad_norm": 0.0,
+      "learning_rate": 9.02168263107105e-07,
+      "loss": 1.0313,
+      "step": 23863
+    },
+    {
+      "epoch": 0.8678764956177037,
+      "grad_norm": 0.0,
+      "learning_rate": 9.01679414896498e-07,
+      "loss": 0.9095,
+      "step": 23864
+    },
+    {
+      "epoch": 0.8679128632214423,
+      "grad_norm": 0.0,
+      "learning_rate": 9.011906929121794e-07,
+      "loss": 0.8973,
+      "step": 23865
+    },
+    {
+      "epoch": 0.8679492308251809,
+      "grad_norm": 0.0,
+      "learning_rate": 9.007020971609315e-07,
+      "loss": 0.917,
+      "step": 23866
+    },
+    {
+      "epoch": 0.8679855984289195,
+      "grad_norm": 0.0,
+      "learning_rate": 9.002136276495288e-07,
+      "loss": 0.8387,
+      "step": 23867
+    },
+    {
+      "epoch": 0.8680219660326581,
+      "grad_norm": 0.0,
+      "learning_rate": 8.997252843847559e-07,
+      "loss": 0.9134,
+      "step": 23868
+    },
+    {
+      "epoch": 0.8680583336363967,
+      "grad_norm": 0.0,
+      "learning_rate": 8.992370673733786e-07,
+      "loss": 0.8841,
+      "step": 23869
+    },
+    {
+      "epoch": 0.8680947012401353,
+      "grad_norm": 0.0,
+      "learning_rate": 8.987489766221769e-07,
+      "loss": 0.8805,
+      "step": 23870
+    },
+    {
+      "epoch": 0.8681310688438739,
+      "grad_norm": 0.0,
+      "learning_rate": 8.982610121379198e-07,
+      "loss": 0.9801,
+      "step": 23871
+    },
+    {
+      "epoch": 0.8681674364476125,
+      "grad_norm": 0.0,
+      "learning_rate": 8.977731739273776e-07,
+      "loss": 0.875,
+      "step": 23872
+    },
+    {
+      "epoch": 0.868203804051351,
+      "grad_norm": 0.0,
+      "learning_rate": 8.97285461997317e-07,
+      "loss": 0.8523,
+      "step": 23873
+    },
+    {
+      "epoch": 0.8682401716550896,
+      "grad_norm": 0.0,
+      "learning_rate": 8.967978763545093e-07,
+      "loss": 1.0724,
+      "step": 23874
+    },
+    {
+      "epoch": 0.8682765392588282,
+      "grad_norm": 0.0,
+      "learning_rate": 8.963104170057102e-07,
+      "loss": 0.9057,
+      "step": 23875
+    },
+    {
+      "epoch": 0.8683129068625668,
+      "grad_norm": 0.0,
+      "learning_rate": 8.958230839576909e-07,
+      "loss": 0.9873,
+      "step": 23876
+    },
+    {
+      "epoch": 0.8683492744663054,
+      "grad_norm": 0.0,
+      "learning_rate": 8.953358772172083e-07,
+      "loss": 1.0618,
+      "step": 23877
+    },
+    {
+      "epoch": 0.868385642070044,
+      "grad_norm": 0.0,
+      "learning_rate": 8.948487967910235e-07,
+      "loss": 0.8692,
+      "step": 23878
+    },
+    {
+      "epoch": 0.8684220096737826,
+      "grad_norm": 0.0,
+      "learning_rate": 8.943618426858925e-07,
+      "loss": 0.9281,
+      "step": 23879
+    },
+    {
+      "epoch": 0.8684583772775212,
+      "grad_norm": 0.0,
+      "learning_rate": 8.938750149085695e-07,
+      "loss": 0.9452,
+      "step": 23880
+    },
+    {
+      "epoch": 0.8684947448812598,
+      "grad_norm": 0.0,
+      "learning_rate": 8.93388313465815e-07,
+      "loss": 0.9123,
+      "step": 23881
+    },
+    {
+      "epoch": 0.8685311124849984,
+      "grad_norm": 0.0,
+      "learning_rate": 8.929017383643735e-07,
+      "loss": 0.8974,
+      "step": 23882
+    },
+    {
+      "epoch": 0.868567480088737,
+      "grad_norm": 0.0,
+      "learning_rate": 8.924152896110005e-07,
+      "loss": 0.817,
+      "step": 23883
+    },
+    {
+      "epoch": 0.8686038476924756,
+      "grad_norm": 0.0,
+      "learning_rate": 8.919289672124432e-07,
+      "loss": 0.9738,
+      "step": 23884
+    },
+    {
+      "epoch": 0.8686402152962142,
+      "grad_norm": 0.0,
+      "learning_rate": 8.914427711754492e-07,
+      "loss": 0.808,
+      "step": 23885
+    },
+    {
+      "epoch": 0.8686765828999528,
+      "grad_norm": 0.0,
+      "learning_rate": 8.909567015067622e-07,
+      "loss": 0.9315,
+      "step": 23886
+    },
+    {
+      "epoch": 0.8687129505036914,
+      "grad_norm": 0.0,
+      "learning_rate": 8.9047075821313e-07,
+      "loss": 0.9478,
+      "step": 23887
+    },
+    {
+      "epoch": 0.86874931810743,
+      "grad_norm": 0.0,
+      "learning_rate": 8.899849413012874e-07,
+      "loss": 0.9006,
+      "step": 23888
+    },
+    {
+      "epoch": 0.8687856857111685,
+      "grad_norm": 0.0,
+      "learning_rate": 8.894992507779798e-07,
+      "loss": 0.9509,
+      "step": 23889
+    },
+    {
+      "epoch": 0.868822053314907,
+      "grad_norm": 0.0,
+      "learning_rate": 8.890136866499443e-07,
+      "loss": 0.9101,
+      "step": 23890
+    },
+    {
+      "epoch": 0.8688584209186456,
+      "grad_norm": 0.0,
+      "learning_rate": 8.885282489239177e-07,
+      "loss": 0.8315,
+      "step": 23891
+    },
+    {
+      "epoch": 0.8688947885223842,
+      "grad_norm": 0.0,
+      "learning_rate": 8.880429376066324e-07,
+      "loss": 0.8932,
+      "step": 23892
+    },
+    {
+      "epoch": 0.8689311561261228,
+      "grad_norm": 0.0,
+      "learning_rate": 8.875577527048262e-07,
+      "loss": 0.9125,
+      "step": 23893
+    },
+    {
+      "epoch": 0.8689675237298614,
+      "grad_norm": 0.0,
+      "learning_rate": 8.870726942252239e-07,
+      "loss": 0.9507,
+      "step": 23894
+    },
+    {
+      "epoch": 0.8690038913336,
+      "grad_norm": 0.0,
+      "learning_rate": 8.8658776217456e-07,
+      "loss": 0.8078,
+      "step": 23895
+    },
+    {
+      "epoch": 0.8690402589373386,
+      "grad_norm": 0.0,
+      "learning_rate": 8.861029565595602e-07,
+      "loss": 0.9026,
+      "step": 23896
+    },
+    {
+      "epoch": 0.8690766265410772,
+      "grad_norm": 0.0,
+      "learning_rate": 8.856182773869526e-07,
+      "loss": 0.9156,
+      "step": 23897
+    },
+    {
+      "epoch": 0.8691129941448158,
+      "grad_norm": 0.0,
+      "learning_rate": 8.851337246634584e-07,
+      "loss": 0.8551,
+      "step": 23898
+    },
+    {
+      "epoch": 0.8691493617485544,
+      "grad_norm": 0.0,
+      "learning_rate": 8.846492983957999e-07,
+      "loss": 0.8991,
+      "step": 23899
+    },
+    {
+      "epoch": 0.869185729352293,
+      "grad_norm": 0.0,
+      "learning_rate": 8.84164998590703e-07,
+      "loss": 0.7407,
+      "step": 23900
+    },
+    {
+      "epoch": 0.8692220969560316,
+      "grad_norm": 0.0,
+      "learning_rate": 8.836808252548801e-07,
+      "loss": 1.0112,
+      "step": 23901
+    },
+    {
+      "epoch": 0.8692584645597702,
+      "grad_norm": 0.0,
+      "learning_rate": 8.831967783950524e-07,
+      "loss": 0.9345,
+      "step": 23902
+    },
+    {
+      "epoch": 0.8692948321635088,
+      "grad_norm": 0.0,
+      "learning_rate": 8.827128580179345e-07,
+      "loss": 0.9121,
+      "step": 23903
+    },
+    {
+      "epoch": 0.8693311997672474,
+      "grad_norm": 0.0,
+      "learning_rate": 8.8222906413024e-07,
+      "loss": 0.83,
+      "step": 23904
+    },
+    {
+      "epoch": 0.869367567370986,
+      "grad_norm": 0.0,
+      "learning_rate": 8.817453967386791e-07,
+      "loss": 0.7761,
+      "step": 23905
+    },
+    {
+      "epoch": 0.8694039349747246,
+      "grad_norm": 0.0,
+      "learning_rate": 8.812618558499674e-07,
+      "loss": 0.9094,
+      "step": 23906
+    },
+    {
+      "epoch": 0.869440302578463,
+      "grad_norm": 0.0,
+      "learning_rate": 8.807784414708054e-07,
+      "loss": 0.8931,
+      "step": 23907
+    },
+    {
+      "epoch": 0.8694766701822016,
+      "grad_norm": 0.0,
+      "learning_rate": 8.802951536079063e-07,
+      "loss": 0.9639,
+      "step": 23908
+    },
+    {
+      "epoch": 0.8695130377859402,
+      "grad_norm": 0.0,
+      "learning_rate": 8.798119922679737e-07,
+      "loss": 0.961,
+      "step": 23909
+    },
+    {
+      "epoch": 0.8695494053896788,
+      "grad_norm": 0.0,
+      "learning_rate": 8.793289574577091e-07,
+      "loss": 0.9202,
+      "step": 23910
+    },
+    {
+      "epoch": 0.8695857729934174,
+      "grad_norm": 0.0,
+      "learning_rate": 8.788460491838124e-07,
+      "loss": 0.9704,
+      "step": 23911
+    },
+    {
+      "epoch": 0.869622140597156,
+      "grad_norm": 0.0,
+      "learning_rate": 8.783632674529874e-07,
+      "loss": 0.848,
+      "step": 23912
+    },
+    {
+      "epoch": 0.8696585082008946,
+      "grad_norm": 0.0,
+      "learning_rate": 8.778806122719308e-07,
+      "loss": 0.8736,
+      "step": 23913
+    },
+    {
+      "epoch": 0.8696948758046332,
+      "grad_norm": 0.0,
+      "learning_rate": 8.773980836473373e-07,
+      "loss": 1.039,
+      "step": 23914
+    },
+    {
+      "epoch": 0.8697312434083718,
+      "grad_norm": 0.0,
+      "learning_rate": 8.769156815859026e-07,
+      "loss": 0.8602,
+      "step": 23915
+    },
+    {
+      "epoch": 0.8697676110121104,
+      "grad_norm": 0.0,
+      "learning_rate": 8.764334060943192e-07,
+      "loss": 0.9202,
+      "step": 23916
+    },
+    {
+      "epoch": 0.869803978615849,
+      "grad_norm": 0.0,
+      "learning_rate": 8.759512571792772e-07,
+      "loss": 0.8989,
+      "step": 23917
+    },
+    {
+      "epoch": 0.8698403462195876,
+      "grad_norm": 0.0,
+      "learning_rate": 8.754692348474647e-07,
+      "loss": 0.8354,
+      "step": 23918
+    },
+    {
+      "epoch": 0.8698767138233262,
+      "grad_norm": 0.0,
+      "learning_rate": 8.749873391055741e-07,
+      "loss": 0.9794,
+      "step": 23919
+    },
+    {
+      "epoch": 0.8699130814270648,
+      "grad_norm": 0.0,
+      "learning_rate": 8.745055699602834e-07,
+      "loss": 0.873,
+      "step": 23920
+    },
+    {
+      "epoch": 0.8699494490308034,
+      "grad_norm": 0.0,
+      "learning_rate": 8.740239274182826e-07,
+      "loss": 0.9109,
+      "step": 23921
+    },
+    {
+      "epoch": 0.869985816634542,
+      "grad_norm": 0.0,
+      "learning_rate": 8.735424114862523e-07,
+      "loss": 0.8602,
+      "step": 23922
+    },
+    {
+      "epoch": 0.8700221842382806,
+      "grad_norm": 0.0,
+      "learning_rate": 8.730610221708713e-07,
+      "loss": 0.9906,
+      "step": 23923
+    },
+    {
+      "epoch": 0.8700585518420191,
+      "grad_norm": 0.0,
+      "learning_rate": 8.725797594788176e-07,
+      "loss": 0.9261,
+      "step": 23924
+    },
+    {
+      "epoch": 0.8700949194457577,
+      "grad_norm": 0.0,
+      "learning_rate": 8.720986234167716e-07,
+      "loss": 0.9845,
+      "step": 23925
+    },
+    {
+      "epoch": 0.8701312870494963,
+      "grad_norm": 0.0,
+      "learning_rate": 8.716176139914067e-07,
+      "loss": 0.9637,
+      "step": 23926
+    },
+    {
+      "epoch": 0.8701676546532349,
+      "grad_norm": 0.0,
+      "learning_rate": 8.711367312093955e-07,
+      "loss": 0.9526,
+      "step": 23927
+    },
+    {
+      "epoch": 0.8702040222569735,
+      "grad_norm": 0.0,
+      "learning_rate": 8.706559750774102e-07,
+      "loss": 0.8337,
+      "step": 23928
+    },
+    {
+      "epoch": 0.8702403898607121,
+      "grad_norm": 0.0,
+      "learning_rate": 8.7017534560212e-07,
+      "loss": 0.9861,
+      "step": 23929
+    },
+    {
+      "epoch": 0.8702767574644507,
+      "grad_norm": 0.0,
+      "learning_rate": 8.696948427901931e-07,
+      "loss": 0.9868,
+      "step": 23930
+    },
+    {
+      "epoch": 0.8703131250681893,
+      "grad_norm": 0.0,
+      "learning_rate": 8.69214466648296e-07,
+      "loss": 0.7684,
+      "step": 23931
+    },
+    {
+      "epoch": 0.8703494926719278,
+      "grad_norm": 0.0,
+      "learning_rate": 8.687342171830937e-07,
+      "loss": 0.9054,
+      "step": 23932
+    },
+    {
+      "epoch": 0.8703858602756664,
+      "grad_norm": 0.0,
+      "learning_rate": 8.682540944012496e-07,
+      "loss": 0.9044,
+      "step": 23933
+    },
+    {
+      "epoch": 0.870422227879405,
+      "grad_norm": 0.0,
+      "learning_rate": 8.67774098309423e-07,
+      "loss": 0.9461,
+      "step": 23934
+    },
+    {
+      "epoch": 0.8704585954831436,
+      "grad_norm": 0.0,
+      "learning_rate": 8.672942289142727e-07,
+      "loss": 0.9164,
+      "step": 23935
+    },
+    {
+      "epoch": 0.8704949630868822,
+      "grad_norm": 0.0,
+      "learning_rate": 8.668144862224593e-07,
+      "loss": 1.0045,
+      "step": 23936
+    },
+    {
+      "epoch": 0.8705313306906208,
+      "grad_norm": 0.0,
+      "learning_rate": 8.663348702406337e-07,
+      "loss": 0.9979,
+      "step": 23937
+    },
+    {
+      "epoch": 0.8705676982943594,
+      "grad_norm": 0.0,
+      "learning_rate": 8.658553809754567e-07,
+      "loss": 0.9744,
+      "step": 23938
+    },
+    {
+      "epoch": 0.870604065898098,
+      "grad_norm": 0.0,
+      "learning_rate": 8.653760184335724e-07,
+      "loss": 0.8992,
+      "step": 23939
+    },
+    {
+      "epoch": 0.8706404335018366,
+      "grad_norm": 0.0,
+      "learning_rate": 8.648967826216382e-07,
+      "loss": 1.105,
+      "step": 23940
+    },
+    {
+      "epoch": 0.8706768011055751,
+      "grad_norm": 0.0,
+      "learning_rate": 8.644176735462995e-07,
+      "loss": 0.8547,
+      "step": 23941
+    },
+    {
+      "epoch": 0.8707131687093137,
+      "grad_norm": 0.0,
+      "learning_rate": 8.639386912142045e-07,
+      "loss": 0.9326,
+      "step": 23942
+    },
+    {
+      "epoch": 0.8707495363130523,
+      "grad_norm": 0.0,
+      "learning_rate": 8.634598356319956e-07,
+      "loss": 0.8306,
+      "step": 23943
+    },
+    {
+      "epoch": 0.8707859039167909,
+      "grad_norm": 0.0,
+      "learning_rate": 8.62981106806321e-07,
+      "loss": 0.9359,
+      "step": 23944
+    },
+    {
+      "epoch": 0.8708222715205295,
+      "grad_norm": 0.0,
+      "learning_rate": 8.625025047438185e-07,
+      "loss": 1.0108,
+      "step": 23945
+    },
+    {
+      "epoch": 0.8708586391242681,
+      "grad_norm": 0.0,
+      "learning_rate": 8.620240294511295e-07,
+      "loss": 0.9481,
+      "step": 23946
+    },
+    {
+      "epoch": 0.8708950067280067,
+      "grad_norm": 0.0,
+      "learning_rate": 8.615456809348932e-07,
+      "loss": 0.9364,
+      "step": 23947
+    },
+    {
+      "epoch": 0.8709313743317453,
+      "grad_norm": 0.0,
+      "learning_rate": 8.610674592017443e-07,
+      "loss": 0.7252,
+      "step": 23948
+    },
+    {
+      "epoch": 0.8709677419354839,
+      "grad_norm": 0.0,
+      "learning_rate": 8.605893642583152e-07,
+      "loss": 0.9615,
+      "step": 23949
+    },
+    {
+      "epoch": 0.8710041095392225,
+      "grad_norm": 0.0,
+      "learning_rate": 8.60111396111244e-07,
+      "loss": 1.0389,
+      "step": 23950
+    },
+    {
+      "epoch": 0.8710404771429611,
+      "grad_norm": 0.0,
+      "learning_rate": 8.596335547671597e-07,
+      "loss": 0.9565,
+      "step": 23951
+    },
+    {
+      "epoch": 0.8710768447466997,
+      "grad_norm": 0.0,
+      "learning_rate": 8.591558402326905e-07,
+      "loss": 0.9199,
+      "step": 23952
+    },
+    {
+      "epoch": 0.8711132123504383,
+      "grad_norm": 0.0,
+      "learning_rate": 8.586782525144655e-07,
+      "loss": 0.9455,
+      "step": 23953
+    },
+    {
+      "epoch": 0.8711495799541769,
+      "grad_norm": 0.0,
+      "learning_rate": 8.582007916191104e-07,
+      "loss": 0.9747,
+      "step": 23954
+    },
+    {
+      "epoch": 0.8711859475579155,
+      "grad_norm": 0.0,
+      "learning_rate": 8.577234575532479e-07,
+      "loss": 0.9927,
+      "step": 23955
+    },
+    {
+      "epoch": 0.871222315161654,
+      "grad_norm": 0.0,
+      "learning_rate": 8.572462503235002e-07,
+      "loss": 0.9663,
+      "step": 23956
+    },
+    {
+      "epoch": 0.8712586827653925,
+      "grad_norm": 0.0,
+      "learning_rate": 8.5676916993649e-07,
+      "loss": 0.9945,
+      "step": 23957
+    },
+    {
+      "epoch": 0.8712950503691311,
+      "grad_norm": 0.0,
+      "learning_rate": 8.562922163988352e-07,
+      "loss": 0.9065,
+      "step": 23958
+    },
+    {
+      "epoch": 0.8713314179728697,
+      "grad_norm": 0.0,
+      "learning_rate": 8.558153897171529e-07,
+      "loss": 0.9357,
+      "step": 23959
+    },
+    {
+      "epoch": 0.8713677855766083,
+      "grad_norm": 0.0,
+      "learning_rate": 8.553386898980576e-07,
+      "loss": 0.8294,
+      "step": 23960
+    },
+    {
+      "epoch": 0.8714041531803469,
+      "grad_norm": 0.0,
+      "learning_rate": 8.548621169481641e-07,
+      "loss": 0.8095,
+      "step": 23961
+    },
+    {
+      "epoch": 0.8714405207840855,
+      "grad_norm": 0.0,
+      "learning_rate": 8.543856708740816e-07,
+      "loss": 0.9283,
+      "step": 23962
+    },
+    {
+      "epoch": 0.8714768883878241,
+      "grad_norm": 0.0,
+      "learning_rate": 8.539093516824238e-07,
+      "loss": 0.899,
+      "step": 23963
+    },
+    {
+      "epoch": 0.8715132559915627,
+      "grad_norm": 0.0,
+      "learning_rate": 8.534331593797962e-07,
+      "loss": 0.9447,
+      "step": 23964
+    },
+    {
+      "epoch": 0.8715496235953013,
+      "grad_norm": 0.0,
+      "learning_rate": 8.529570939728071e-07,
+      "loss": 0.7931,
+      "step": 23965
+    },
+    {
+      "epoch": 0.8715859911990399,
+      "grad_norm": 0.0,
+      "learning_rate": 8.52481155468059e-07,
+      "loss": 0.8903,
+      "step": 23966
+    },
+    {
+      "epoch": 0.8716223588027785,
+      "grad_norm": 0.0,
+      "learning_rate": 8.520053438721588e-07,
+      "loss": 0.8667,
+      "step": 23967
+    },
+    {
+      "epoch": 0.8716587264065171,
+      "grad_norm": 0.0,
+      "learning_rate": 8.515296591917021e-07,
+      "loss": 0.8373,
+      "step": 23968
+    },
+    {
+      "epoch": 0.8716950940102557,
+      "grad_norm": 0.0,
+      "learning_rate": 8.51054101433293e-07,
+      "loss": 1.0598,
+      "step": 23969
+    },
+    {
+      "epoch": 0.8717314616139943,
+      "grad_norm": 0.0,
+      "learning_rate": 8.505786706035279e-07,
+      "loss": 0.9978,
+      "step": 23970
+    },
+    {
+      "epoch": 0.8717678292177329,
+      "grad_norm": 0.0,
+      "learning_rate": 8.50103366709002e-07,
+      "loss": 0.9522,
+      "step": 23971
+    },
+    {
+      "epoch": 0.8718041968214715,
+      "grad_norm": 0.0,
+      "learning_rate": 8.496281897563096e-07,
+      "loss": 1.0273,
+      "step": 23972
+    },
+    {
+      "epoch": 0.8718405644252101,
+      "grad_norm": 0.0,
+      "learning_rate": 8.491531397520447e-07,
+      "loss": 0.9242,
+      "step": 23973
+    },
+    {
+      "epoch": 0.8718769320289486,
+      "grad_norm": 0.0,
+      "learning_rate": 8.486782167027952e-07,
+      "loss": 0.8672,
+      "step": 23974
+    },
+    {
+      "epoch": 0.8719132996326872,
+      "grad_norm": 0.0,
+      "learning_rate": 8.482034206151502e-07,
+      "loss": 0.9621,
+      "step": 23975
+    },
+    {
+      "epoch": 0.8719496672364258,
+      "grad_norm": 0.0,
+      "learning_rate": 8.477287514957e-07,
+      "loss": 0.966,
+      "step": 23976
+    },
+    {
+      "epoch": 0.8719860348401643,
+      "grad_norm": 0.0,
+      "learning_rate": 8.472542093510283e-07,
+      "loss": 0.8094,
+      "step": 23977
+    },
+    {
+      "epoch": 0.8720224024439029,
+      "grad_norm": 0.0,
+      "learning_rate": 8.467797941877176e-07,
+      "loss": 0.9195,
+      "step": 23978
+    },
+    {
+      "epoch": 0.8720587700476415,
+      "grad_norm": 0.0,
+      "learning_rate": 8.463055060123493e-07,
+      "loss": 0.903,
+      "step": 23979
+    },
+    {
+      "epoch": 0.8720951376513801,
+      "grad_norm": 0.0,
+      "learning_rate": 8.458313448315081e-07,
+      "loss": 0.9276,
+      "step": 23980
+    },
+    {
+      "epoch": 0.8721315052551187,
+      "grad_norm": 0.0,
+      "learning_rate": 8.453573106517654e-07,
+      "loss": 0.9002,
+      "step": 23981
+    },
+    {
+      "epoch": 0.8721678728588573,
+      "grad_norm": 0.0,
+      "learning_rate": 8.448834034797027e-07,
+      "loss": 0.8912,
+      "step": 23982
+    },
+    {
+      "epoch": 0.8722042404625959,
+      "grad_norm": 0.0,
+      "learning_rate": 8.444096233218935e-07,
+      "loss": 1.0462,
+      "step": 23983
+    },
+    {
+      "epoch": 0.8722406080663345,
+      "grad_norm": 0.0,
+      "learning_rate": 8.439359701849114e-07,
+      "loss": 0.9586,
+      "step": 23984
+    },
+    {
+      "epoch": 0.8722769756700731,
+      "grad_norm": 0.0,
+      "learning_rate": 8.434624440753247e-07,
+      "loss": 0.8756,
+      "step": 23985
+    },
+    {
+      "epoch": 0.8723133432738117,
+      "grad_norm": 0.0,
+      "learning_rate": 8.42989044999708e-07,
+      "loss": 0.8055,
+      "step": 23986
+    },
+    {
+      "epoch": 0.8723497108775503,
+      "grad_norm": 0.0,
+      "learning_rate": 8.425157729646238e-07,
+      "loss": 0.8559,
+      "step": 23987
+    },
+    {
+      "epoch": 0.8723860784812889,
+      "grad_norm": 0.0,
+      "learning_rate": 8.420426279766414e-07,
+      "loss": 0.8718,
+      "step": 23988
+    },
+    {
+      "epoch": 0.8724224460850275,
+      "grad_norm": 0.0,
+      "learning_rate": 8.415696100423243e-07,
+      "loss": 0.922,
+      "step": 23989
+    },
+    {
+      "epoch": 0.8724588136887661,
+      "grad_norm": 0.0,
+      "learning_rate": 8.410967191682351e-07,
+      "loss": 0.9637,
+      "step": 23990
+    },
+    {
+      "epoch": 0.8724951812925046,
+      "grad_norm": 0.0,
+      "learning_rate": 8.406239553609341e-07,
+      "loss": 0.9066,
+      "step": 23991
+    },
+    {
+      "epoch": 0.8725315488962432,
+      "grad_norm": 0.0,
+      "learning_rate": 8.401513186269794e-07,
+      "loss": 0.9168,
+      "step": 23992
+    },
+    {
+      "epoch": 0.8725679164999818,
+      "grad_norm": 0.0,
+      "learning_rate": 8.396788089729302e-07,
+      "loss": 0.8415,
+      "step": 23993
+    },
+    {
+      "epoch": 0.8726042841037204,
+      "grad_norm": 0.0,
+      "learning_rate": 8.39206426405339e-07,
+      "loss": 0.977,
+      "step": 23994
+    },
+    {
+      "epoch": 0.872640651707459,
+      "grad_norm": 0.0,
+      "learning_rate": 8.387341709307617e-07,
+      "loss": 0.9155,
+      "step": 23995
+    },
+    {
+      "epoch": 0.8726770193111976,
+      "grad_norm": 0.0,
+      "learning_rate": 8.382620425557508e-07,
+      "loss": 0.8156,
+      "step": 23996
+    },
+    {
+      "epoch": 0.8727133869149362,
+      "grad_norm": 0.0,
+      "learning_rate": 8.377900412868545e-07,
+      "loss": 0.8831,
+      "step": 23997
+    },
+    {
+      "epoch": 0.8727497545186748,
+      "grad_norm": 0.0,
+      "learning_rate": 8.373181671306208e-07,
+      "loss": 0.8634,
+      "step": 23998
+    },
+    {
+      "epoch": 0.8727861221224134,
+      "grad_norm": 0.0,
+      "learning_rate": 8.368464200935999e-07,
+      "loss": 0.8818,
+      "step": 23999
+    },
+    {
+      "epoch": 0.872822489726152,
+      "grad_norm": 0.0,
+      "learning_rate": 8.363748001823313e-07,
+      "loss": 1.015,
+      "step": 24000
+    },
+    {
+      "epoch": 0.8728588573298905,
+      "grad_norm": 0.0,
+      "learning_rate": 8.359033074033629e-07,
+      "loss": 0.8721,
+      "step": 24001
+    },
+    {
+      "epoch": 0.8728952249336291,
+      "grad_norm": 0.0,
+      "learning_rate": 8.35431941763234e-07,
+      "loss": 0.8972,
+      "step": 24002
+    },
+    {
+      "epoch": 0.8729315925373677,
+      "grad_norm": 0.0,
+      "learning_rate": 8.349607032684836e-07,
+      "loss": 0.9353,
+      "step": 24003
+    },
+    {
+      "epoch": 0.8729679601411063,
+      "grad_norm": 0.0,
+      "learning_rate": 8.344895919256479e-07,
+      "loss": 0.9265,
+      "step": 24004
+    },
+    {
+      "epoch": 0.8730043277448449,
+      "grad_norm": 0.0,
+      "learning_rate": 8.340186077412693e-07,
+      "loss": 0.864,
+      "step": 24005
+    },
+    {
+      "epoch": 0.8730406953485835,
+      "grad_norm": 0.0,
+      "learning_rate": 8.335477507218737e-07,
+      "loss": 1.0604,
+      "step": 24006
+    },
+    {
+      "epoch": 0.8730770629523221,
+      "grad_norm": 0.0,
+      "learning_rate": 8.330770208739991e-07,
+      "loss": 0.949,
+      "step": 24007
+    },
+    {
+      "epoch": 0.8731134305560606,
+      "grad_norm": 0.0,
+      "learning_rate": 8.326064182041748e-07,
+      "loss": 0.9299,
+      "step": 24008
+    },
+    {
+      "epoch": 0.8731497981597992,
+      "grad_norm": 0.0,
+      "learning_rate": 8.321359427189302e-07,
+      "loss": 0.8828,
+      "step": 24009
+    },
+    {
+      "epoch": 0.8731861657635378,
+      "grad_norm": 0.0,
+      "learning_rate": 8.316655944247909e-07,
+      "loss": 1.0281,
+      "step": 24010
+    },
+    {
+      "epoch": 0.8732225333672764,
+      "grad_norm": 0.0,
+      "learning_rate": 8.311953733282818e-07,
+      "loss": 0.8549,
+      "step": 24011
+    },
+    {
+      "epoch": 0.873258900971015,
+      "grad_norm": 0.0,
+      "learning_rate": 8.307252794359321e-07,
+      "loss": 0.9547,
+      "step": 24012
+    },
+    {
+      "epoch": 0.8732952685747536,
+      "grad_norm": 0.0,
+      "learning_rate": 8.302553127542567e-07,
+      "loss": 0.9113,
+      "step": 24013
+    },
+    {
+      "epoch": 0.8733316361784922,
+      "grad_norm": 0.0,
+      "learning_rate": 8.297854732897792e-07,
+      "loss": 0.8051,
+      "step": 24014
+    },
+    {
+      "epoch": 0.8733680037822308,
+      "grad_norm": 0.0,
+      "learning_rate": 8.293157610490176e-07,
+      "loss": 0.8349,
+      "step": 24015
+    },
+    {
+      "epoch": 0.8734043713859694,
+      "grad_norm": 0.0,
+      "learning_rate": 8.288461760384892e-07,
+      "loss": 1.0807,
+      "step": 24016
+    },
+    {
+      "epoch": 0.873440738989708,
+      "grad_norm": 0.0,
+      "learning_rate": 8.283767182647051e-07,
+      "loss": 0.9413,
+      "step": 24017
+    },
+    {
+      "epoch": 0.8734771065934466,
+      "grad_norm": 0.0,
+      "learning_rate": 8.279073877341859e-07,
+      "loss": 0.8665,
+      "step": 24018
+    },
+    {
+      "epoch": 0.8735134741971852,
+      "grad_norm": 0.0,
+      "learning_rate": 8.27438184453433e-07,
+      "loss": 0.8507,
+      "step": 24019
+    },
+    {
+      "epoch": 0.8735498418009238,
+      "grad_norm": 0.0,
+      "learning_rate": 8.269691084289644e-07,
+      "loss": 0.9172,
+      "step": 24020
+    },
+    {
+      "epoch": 0.8735862094046624,
+      "grad_norm": 0.0,
+      "learning_rate": 8.26500159667284e-07,
+      "loss": 0.8322,
+      "step": 24021
+    },
+    {
+      "epoch": 0.873622577008401,
+      "grad_norm": 0.0,
+      "learning_rate": 8.260313381748985e-07,
+      "loss": 0.8891,
+      "step": 24022
+    },
+    {
+      "epoch": 0.8736589446121396,
+      "grad_norm": 0.0,
+      "learning_rate": 8.255626439583098e-07,
+      "loss": 1.0327,
+      "step": 24023
+    },
+    {
+      "epoch": 0.8736953122158782,
+      "grad_norm": 0.0,
+      "learning_rate": 8.250940770240246e-07,
+      "loss": 0.8817,
+      "step": 24024
+    },
+    {
+      "epoch": 0.8737316798196166,
+      "grad_norm": 0.0,
+      "learning_rate": 8.246256373785421e-07,
+      "loss": 0.8827,
+      "step": 24025
+    },
+    {
+      "epoch": 0.8737680474233552,
+      "grad_norm": 0.0,
+      "learning_rate": 8.241573250283608e-07,
+      "loss": 0.8435,
+      "step": 24026
+    },
+    {
+      "epoch": 0.8738044150270938,
+      "grad_norm": 0.0,
+      "learning_rate": 8.236891399799773e-07,
+      "loss": 0.8004,
+      "step": 24027
+    },
+    {
+      "epoch": 0.8738407826308324,
+      "grad_norm": 0.0,
+      "learning_rate": 8.232210822398878e-07,
+      "loss": 0.8858,
+      "step": 24028
+    },
+    {
+      "epoch": 0.873877150234571,
+      "grad_norm": 0.0,
+      "learning_rate": 8.227531518145848e-07,
+      "loss": 0.9372,
+      "step": 24029
+    },
+    {
+      "epoch": 0.8739135178383096,
+      "grad_norm": 0.0,
+      "learning_rate": 8.222853487105598e-07,
+      "loss": 0.7113,
+      "step": 24030
+    },
+    {
+      "epoch": 0.8739498854420482,
+      "grad_norm": 0.0,
+      "learning_rate": 8.218176729343075e-07,
+      "loss": 0.8622,
+      "step": 24031
+    },
+    {
+      "epoch": 0.8739862530457868,
+      "grad_norm": 0.0,
+      "learning_rate": 8.213501244923094e-07,
+      "loss": 0.9327,
+      "step": 24032
+    },
+    {
+      "epoch": 0.8740226206495254,
+      "grad_norm": 0.0,
+      "learning_rate": 8.208827033910583e-07,
+      "loss": 0.9359,
+      "step": 24033
+    },
+    {
+      "epoch": 0.874058988253264,
+      "grad_norm": 0.0,
+      "learning_rate": 8.204154096370354e-07,
+      "loss": 0.9039,
+      "step": 24034
+    },
+    {
+      "epoch": 0.8740953558570026,
+      "grad_norm": 0.0,
+      "learning_rate": 8.199482432367245e-07,
+      "loss": 0.9024,
+      "step": 24035
+    },
+    {
+      "epoch": 0.8741317234607412,
+      "grad_norm": 0.0,
+      "learning_rate": 8.19481204196605e-07,
+      "loss": 0.9154,
+      "step": 24036
+    },
+    {
+      "epoch": 0.8741680910644798,
+      "grad_norm": 0.0,
+      "learning_rate": 8.190142925231614e-07,
+      "loss": 0.8639,
+      "step": 24037
+    },
+    {
+      "epoch": 0.8742044586682184,
+      "grad_norm": 0.0,
+      "learning_rate": 8.185475082228655e-07,
+      "loss": 1.0257,
+      "step": 24038
+    },
+    {
+      "epoch": 0.874240826271957,
+      "grad_norm": 0.0,
+      "learning_rate": 8.180808513021987e-07,
+      "loss": 0.828,
+      "step": 24039
+    },
+    {
+      "epoch": 0.8742771938756956,
+      "grad_norm": 0.0,
+      "learning_rate": 8.176143217676313e-07,
+      "loss": 0.8638,
+      "step": 24040
+    },
+    {
+      "epoch": 0.8743135614794341,
+      "grad_norm": 0.0,
+      "learning_rate": 8.171479196256382e-07,
+      "loss": 0.9631,
+      "step": 24041
+    },
+    {
+      "epoch": 0.8743499290831727,
+      "grad_norm": 0.0,
+      "learning_rate": 8.166816448826864e-07,
+      "loss": 1.0403,
+      "step": 24042
+    },
+    {
+      "epoch": 0.8743862966869113,
+      "grad_norm": 0.0,
+      "learning_rate": 8.162154975452486e-07,
+      "loss": 0.8865,
+      "step": 24043
+    },
+    {
+      "epoch": 0.8744226642906499,
+      "grad_norm": 0.0,
+      "learning_rate": 8.157494776197916e-07,
+      "loss": 0.9236,
+      "step": 24044
+    },
+    {
+      "epoch": 0.8744590318943884,
+      "grad_norm": 0.0,
+      "learning_rate": 8.152835851127805e-07,
+      "loss": 0.8612,
+      "step": 24045
+    },
+    {
+      "epoch": 0.874495399498127,
+      "grad_norm": 0.0,
+      "learning_rate": 8.148178200306756e-07,
+      "loss": 0.9021,
+      "step": 24046
+    },
+    {
+      "epoch": 0.8745317671018656,
+      "grad_norm": 0.0,
+      "learning_rate": 8.14352182379945e-07,
+      "loss": 0.9125,
+      "step": 24047
+    },
+    {
+      "epoch": 0.8745681347056042,
+      "grad_norm": 0.0,
+      "learning_rate": 8.138866721670413e-07,
+      "loss": 0.8755,
+      "step": 24048
+    },
+    {
+      "epoch": 0.8746045023093428,
+      "grad_norm": 0.0,
+      "learning_rate": 8.134212893984284e-07,
+      "loss": 0.8901,
+      "step": 24049
+    },
+    {
+      "epoch": 0.8746408699130814,
+      "grad_norm": 0.0,
+      "learning_rate": 8.129560340805632e-07,
+      "loss": 0.964,
+      "step": 24050
+    },
+    {
+      "epoch": 0.87467723751682,
+      "grad_norm": 0.0,
+      "learning_rate": 8.124909062198949e-07,
+      "loss": 0.8524,
+      "step": 24051
+    },
+    {
+      "epoch": 0.8747136051205586,
+      "grad_norm": 0.0,
+      "learning_rate": 8.120259058228807e-07,
+      "loss": 1.0125,
+      "step": 24052
+    },
+    {
+      "epoch": 0.8747499727242972,
+      "grad_norm": 0.0,
+      "learning_rate": 8.115610328959722e-07,
+      "loss": 0.946,
+      "step": 24053
+    },
+    {
+      "epoch": 0.8747863403280358,
+      "grad_norm": 0.0,
+      "learning_rate": 8.110962874456174e-07,
+      "loss": 0.8849,
+      "step": 24054
+    },
+    {
+      "epoch": 0.8748227079317744,
+      "grad_norm": 0.0,
+      "learning_rate": 8.106316694782623e-07,
+      "loss": 0.8978,
+      "step": 24055
+    },
+    {
+      "epoch": 0.874859075535513,
+      "grad_norm": 0.0,
+      "learning_rate": 8.101671790003562e-07,
+      "loss": 0.9066,
+      "step": 24056
+    },
+    {
+      "epoch": 0.8748954431392516,
+      "grad_norm": 0.0,
+      "learning_rate": 8.097028160183429e-07,
+      "loss": 0.7818,
+      "step": 24057
+    },
+    {
+      "epoch": 0.8749318107429901,
+      "grad_norm": 0.0,
+      "learning_rate": 8.092385805386627e-07,
+      "loss": 0.8883,
+      "step": 24058
+    },
+    {
+      "epoch": 0.8749681783467287,
+      "grad_norm": 0.0,
+      "learning_rate": 8.087744725677571e-07,
+      "loss": 0.9308,
+      "step": 24059
+    },
+    {
+      "epoch": 0.8750045459504673,
+      "grad_norm": 0.0,
+      "learning_rate": 8.083104921120654e-07,
+      "loss": 1.0651,
+      "step": 24060
+    },
+    {
+      "epoch": 0.8750409135542059,
+      "grad_norm": 0.0,
+      "learning_rate": 8.078466391780226e-07,
+      "loss": 0.927,
+      "step": 24061
+    },
+    {
+      "epoch": 0.8750772811579445,
+      "grad_norm": 0.0,
+      "learning_rate": 8.073829137720679e-07,
+      "loss": 0.9481,
+      "step": 24062
+    },
+    {
+      "epoch": 0.8751136487616831,
+      "grad_norm": 0.0,
+      "learning_rate": 8.069193159006328e-07,
+      "loss": 0.906,
+      "step": 24063
+    },
+    {
+      "epoch": 0.8751500163654217,
+      "grad_norm": 0.0,
+      "learning_rate": 8.064558455701477e-07,
+      "loss": 1.0379,
+      "step": 24064
+    },
+    {
+      "epoch": 0.8751863839691603,
+      "grad_norm": 0.0,
+      "learning_rate": 8.05992502787043e-07,
+      "loss": 0.981,
+      "step": 24065
+    },
+    {
+      "epoch": 0.8752227515728989,
+      "grad_norm": 0.0,
+      "learning_rate": 8.055292875577514e-07,
+      "loss": 0.9603,
+      "step": 24066
+    },
+    {
+      "epoch": 0.8752591191766375,
+      "grad_norm": 0.0,
+      "learning_rate": 8.050661998886911e-07,
+      "loss": 0.8121,
+      "step": 24067
+    },
+    {
+      "epoch": 0.875295486780376,
+      "grad_norm": 0.0,
+      "learning_rate": 8.046032397862946e-07,
+      "loss": 0.9937,
+      "step": 24068
+    },
+    {
+      "epoch": 0.8753318543841146,
+      "grad_norm": 0.0,
+      "learning_rate": 8.041404072569803e-07,
+      "loss": 0.9101,
+      "step": 24069
+    },
+    {
+      "epoch": 0.8753682219878532,
+      "grad_norm": 0.0,
+      "learning_rate": 8.036777023071707e-07,
+      "loss": 0.8347,
+      "step": 24070
+    },
+    {
+      "epoch": 0.8754045895915918,
+      "grad_norm": 0.0,
+      "learning_rate": 8.032151249432863e-07,
+      "loss": 0.9746,
+      "step": 24071
+    },
+    {
+      "epoch": 0.8754409571953304,
+      "grad_norm": 0.0,
+      "learning_rate": 8.027526751717418e-07,
+      "loss": 0.9216,
+      "step": 24072
+    },
+    {
+      "epoch": 0.875477324799069,
+      "grad_norm": 0.0,
+      "learning_rate": 8.022903529989567e-07,
+      "loss": 0.8809,
+      "step": 24073
+    },
+    {
+      "epoch": 0.8755136924028076,
+      "grad_norm": 0.0,
+      "learning_rate": 8.018281584313403e-07,
+      "loss": 0.9194,
+      "step": 24074
+    },
+    {
+      "epoch": 0.8755500600065461,
+      "grad_norm": 0.0,
+      "learning_rate": 8.013660914753085e-07,
+      "loss": 0.9111,
+      "step": 24075
+    },
+    {
+      "epoch": 0.8755864276102847,
+      "grad_norm": 0.0,
+      "learning_rate": 8.009041521372729e-07,
+      "loss": 0.9532,
+      "step": 24076
+    },
+    {
+      "epoch": 0.8756227952140233,
+      "grad_norm": 0.0,
+      "learning_rate": 8.004423404236395e-07,
+      "loss": 0.8086,
+      "step": 24077
+    },
+    {
+      "epoch": 0.8756591628177619,
+      "grad_norm": 0.0,
+      "learning_rate": 7.999806563408141e-07,
+      "loss": 0.9759,
+      "step": 24078
+    },
+    {
+      "epoch": 0.8756955304215005,
+      "grad_norm": 0.0,
+      "learning_rate": 7.995190998952074e-07,
+      "loss": 1.086,
+      "step": 24079
+    },
+    {
+      "epoch": 0.8757318980252391,
+      "grad_norm": 0.0,
+      "learning_rate": 7.990576710932163e-07,
+      "loss": 0.9126,
+      "step": 24080
+    },
+    {
+      "epoch": 0.8757682656289777,
+      "grad_norm": 0.0,
+      "learning_rate": 7.98596369941248e-07,
+      "loss": 0.9356,
+      "step": 24081
+    },
+    {
+      "epoch": 0.8758046332327163,
+      "grad_norm": 0.0,
+      "learning_rate": 7.981351964456985e-07,
+      "loss": 0.8131,
+      "step": 24082
+    },
+    {
+      "epoch": 0.8758410008364549,
+      "grad_norm": 0.0,
+      "learning_rate": 7.976741506129681e-07,
+      "loss": 0.8938,
+      "step": 24083
+    },
+    {
+      "epoch": 0.8758773684401935,
+      "grad_norm": 0.0,
+      "learning_rate": 7.972132324494508e-07,
+      "loss": 0.7641,
+      "step": 24084
+    },
+    {
+      "epoch": 0.8759137360439321,
+      "grad_norm": 0.0,
+      "learning_rate": 7.967524419615469e-07,
+      "loss": 0.9238,
+      "step": 24085
+    },
+    {
+      "epoch": 0.8759501036476707,
+      "grad_norm": 0.0,
+      "learning_rate": 7.962917791556413e-07,
+      "loss": 0.7857,
+      "step": 24086
+    },
+    {
+      "epoch": 0.8759864712514093,
+      "grad_norm": 0.0,
+      "learning_rate": 7.958312440381299e-07,
+      "loss": 1.079,
+      "step": 24087
+    },
+    {
+      "epoch": 0.8760228388551479,
+      "grad_norm": 0.0,
+      "learning_rate": 7.953708366154023e-07,
+      "loss": 0.9418,
+      "step": 24088
+    },
+    {
+      "epoch": 0.8760592064588865,
+      "grad_norm": 0.0,
+      "learning_rate": 7.949105568938442e-07,
+      "loss": 0.8698,
+      "step": 24089
+    },
+    {
+      "epoch": 0.8760955740626251,
+      "grad_norm": 0.0,
+      "learning_rate": 7.944504048798407e-07,
+      "loss": 0.9909,
+      "step": 24090
+    },
+    {
+      "epoch": 0.8761319416663637,
+      "grad_norm": 0.0,
+      "learning_rate": 7.939903805797777e-07,
+      "loss": 0.8357,
+      "step": 24091
+    },
+    {
+      "epoch": 0.8761683092701021,
+      "grad_norm": 0.0,
+      "learning_rate": 7.935304840000369e-07,
+      "loss": 0.9962,
+      "step": 24092
+    },
+    {
+      "epoch": 0.8762046768738407,
+      "grad_norm": 0.0,
+      "learning_rate": 7.930707151469963e-07,
+      "loss": 0.9321,
+      "step": 24093
+    },
+    {
+      "epoch": 0.8762410444775793,
+      "grad_norm": 0.0,
+      "learning_rate": 7.926110740270388e-07,
+      "loss": 0.9032,
+      "step": 24094
+    },
+    {
+      "epoch": 0.8762774120813179,
+      "grad_norm": 0.0,
+      "learning_rate": 7.92151560646538e-07,
+      "loss": 0.8723,
+      "step": 24095
+    },
+    {
+      "epoch": 0.8763137796850565,
+      "grad_norm": 0.0,
+      "learning_rate": 7.916921750118711e-07,
+      "loss": 0.879,
+      "step": 24096
+    },
+    {
+      "epoch": 0.8763501472887951,
+      "grad_norm": 0.0,
+      "learning_rate": 7.912329171294087e-07,
+      "loss": 0.908,
+      "step": 24097
+    },
+    {
+      "epoch": 0.8763865148925337,
+      "grad_norm": 0.0,
+      "learning_rate": 7.907737870055266e-07,
+      "loss": 0.9976,
+      "step": 24098
+    },
+    {
+      "epoch": 0.8764228824962723,
+      "grad_norm": 0.0,
+      "learning_rate": 7.903147846465887e-07,
+      "loss": 0.9156,
+      "step": 24099
+    },
+    {
+      "epoch": 0.8764592501000109,
+      "grad_norm": 0.0,
+      "learning_rate": 7.898559100589686e-07,
+      "loss": 0.8508,
+      "step": 24100
+    },
+    {
+      "epoch": 0.8764956177037495,
+      "grad_norm": 0.0,
+      "learning_rate": 7.893971632490304e-07,
+      "loss": 1.0067,
+      "step": 24101
+    },
+    {
+      "epoch": 0.8765319853074881,
+      "grad_norm": 0.0,
+      "learning_rate": 7.889385442231378e-07,
+      "loss": 0.8853,
+      "step": 24102
+    },
+    {
+      "epoch": 0.8765683529112267,
+      "grad_norm": 0.0,
+      "learning_rate": 7.884800529876524e-07,
+      "loss": 0.8589,
+      "step": 24103
+    },
+    {
+      "epoch": 0.8766047205149653,
+      "grad_norm": 0.0,
+      "learning_rate": 7.880216895489412e-07,
+      "loss": 0.9332,
+      "step": 24104
+    },
+    {
+      "epoch": 0.8766410881187039,
+      "grad_norm": 0.0,
+      "learning_rate": 7.875634539133548e-07,
+      "loss": 0.8638,
+      "step": 24105
+    },
+    {
+      "epoch": 0.8766774557224425,
+      "grad_norm": 0.0,
+      "learning_rate": 7.871053460872558e-07,
+      "loss": 0.9688,
+      "step": 24106
+    },
+    {
+      "epoch": 0.8767138233261811,
+      "grad_norm": 0.0,
+      "learning_rate": 7.866473660770002e-07,
+      "loss": 0.8187,
+      "step": 24107
+    },
+    {
+      "epoch": 0.8767501909299197,
+      "grad_norm": 0.0,
+      "learning_rate": 7.861895138889398e-07,
+      "loss": 1.0073,
+      "step": 24108
+    },
+    {
+      "epoch": 0.8767865585336582,
+      "grad_norm": 0.0,
+      "learning_rate": 7.85731789529427e-07,
+      "loss": 0.8593,
+      "step": 24109
+    },
+    {
+      "epoch": 0.8768229261373968,
+      "grad_norm": 0.0,
+      "learning_rate": 7.852741930048102e-07,
+      "loss": 0.9142,
+      "step": 24110
+    },
+    {
+      "epoch": 0.8768592937411354,
+      "grad_norm": 0.0,
+      "learning_rate": 7.848167243214433e-07,
+      "loss": 0.9791,
+      "step": 24111
+    },
+    {
+      "epoch": 0.876895661344874,
+      "grad_norm": 0.0,
+      "learning_rate": 7.843593834856677e-07,
+      "loss": 1.0461,
+      "step": 24112
+    },
+    {
+      "epoch": 0.8769320289486126,
+      "grad_norm": 0.0,
+      "learning_rate": 7.839021705038308e-07,
+      "loss": 0.8051,
+      "step": 24113
+    },
+    {
+      "epoch": 0.8769683965523511,
+      "grad_norm": 0.0,
+      "learning_rate": 7.834450853822762e-07,
+      "loss": 0.9481,
+      "step": 24114
+    },
+    {
+      "epoch": 0.8770047641560897,
+      "grad_norm": 0.0,
+      "learning_rate": 7.829881281273432e-07,
+      "loss": 0.9441,
+      "step": 24115
+    },
+    {
+      "epoch": 0.8770411317598283,
+      "grad_norm": 0.0,
+      "learning_rate": 7.825312987453715e-07,
+      "loss": 0.988,
+      "step": 24116
+    },
+    {
+      "epoch": 0.8770774993635669,
+      "grad_norm": 0.0,
+      "learning_rate": 7.820745972427047e-07,
+      "loss": 0.8939,
+      "step": 24117
+    },
+    {
+      "epoch": 0.8771138669673055,
+      "grad_norm": 0.0,
+      "learning_rate": 7.816180236256699e-07,
+      "loss": 0.9461,
+      "step": 24118
+    },
+    {
+      "epoch": 0.8771502345710441,
+      "grad_norm": 0.0,
+      "learning_rate": 7.811615779006077e-07,
+      "loss": 0.9904,
+      "step": 24119
+    },
+    {
+      "epoch": 0.8771866021747827,
+      "grad_norm": 0.0,
+      "learning_rate": 7.807052600738485e-07,
+      "loss": 0.8876,
+      "step": 24120
+    },
+    {
+      "epoch": 0.8772229697785213,
+      "grad_norm": 0.0,
+      "learning_rate": 7.80249070151724e-07,
+      "loss": 0.9097,
+      "step": 24121
+    },
+    {
+      "epoch": 0.8772593373822599,
+      "grad_norm": 0.0,
+      "learning_rate": 7.797930081405602e-07,
+      "loss": 0.8691,
+      "step": 24122
+    },
+    {
+      "epoch": 0.8772957049859985,
+      "grad_norm": 0.0,
+      "learning_rate": 7.793370740466888e-07,
+      "loss": 0.8531,
+      "step": 24123
+    },
+    {
+      "epoch": 0.8773320725897371,
+      "grad_norm": 0.0,
+      "learning_rate": 7.788812678764324e-07,
+      "loss": 0.8718,
+      "step": 24124
+    },
+    {
+      "epoch": 0.8773684401934757,
+      "grad_norm": 0.0,
+      "learning_rate": 7.784255896361159e-07,
+      "loss": 0.9051,
+      "step": 24125
+    },
+    {
+      "epoch": 0.8774048077972142,
+      "grad_norm": 0.0,
+      "learning_rate": 7.77970039332061e-07,
+      "loss": 0.7986,
+      "step": 24126
+    },
+    {
+      "epoch": 0.8774411754009528,
+      "grad_norm": 0.0,
+      "learning_rate": 7.775146169705871e-07,
+      "loss": 0.8869,
+      "step": 24127
+    },
+    {
+      "epoch": 0.8774775430046914,
+      "grad_norm": 0.0,
+      "learning_rate": 7.770593225580126e-07,
+      "loss": 0.8195,
+      "step": 24128
+    },
+    {
+      "epoch": 0.87751391060843,
+      "grad_norm": 0.0,
+      "learning_rate": 7.766041561006532e-07,
+      "loss": 0.8255,
+      "step": 24129
+    },
+    {
+      "epoch": 0.8775502782121686,
+      "grad_norm": 0.0,
+      "learning_rate": 7.761491176048274e-07,
+      "loss": 0.9332,
+      "step": 24130
+    },
+    {
+      "epoch": 0.8775866458159072,
+      "grad_norm": 0.0,
+      "learning_rate": 7.756942070768437e-07,
+      "loss": 1.0671,
+      "step": 24131
+    },
+    {
+      "epoch": 0.8776230134196458,
+      "grad_norm": 0.0,
+      "learning_rate": 7.752394245230155e-07,
+      "loss": 0.9239,
+      "step": 24132
+    },
+    {
+      "epoch": 0.8776593810233844,
+      "grad_norm": 0.0,
+      "learning_rate": 7.747847699496525e-07,
+      "loss": 0.8778,
+      "step": 24133
+    },
+    {
+      "epoch": 0.877695748627123,
+      "grad_norm": 0.0,
+      "learning_rate": 7.743302433630628e-07,
+      "loss": 1.0253,
+      "step": 24134
+    },
+    {
+      "epoch": 0.8777321162308616,
+      "grad_norm": 0.0,
+      "learning_rate": 7.738758447695494e-07,
+      "loss": 0.9237,
+      "step": 24135
+    },
+    {
+      "epoch": 0.8777684838346002,
+      "grad_norm": 0.0,
+      "learning_rate": 7.734215741754214e-07,
+      "loss": 0.9384,
+      "step": 24136
+    },
+    {
+      "epoch": 0.8778048514383388,
+      "grad_norm": 0.0,
+      "learning_rate": 7.729674315869751e-07,
+      "loss": 0.6643,
+      "step": 24137
+    },
+    {
+      "epoch": 0.8778412190420773,
+      "grad_norm": 0.0,
+      "learning_rate": 7.725134170105153e-07,
+      "loss": 0.9735,
+      "step": 24138
+    },
+    {
+      "epoch": 0.8778775866458159,
+      "grad_norm": 0.0,
+      "learning_rate": 7.720595304523404e-07,
+      "loss": 0.9887,
+      "step": 24139
+    },
+    {
+      "epoch": 0.8779139542495545,
+      "grad_norm": 0.0,
+      "learning_rate": 7.716057719187464e-07,
+      "loss": 0.9409,
+      "step": 24140
+    },
+    {
+      "epoch": 0.8779503218532931,
+      "grad_norm": 0.0,
+      "learning_rate": 7.711521414160283e-07,
+      "loss": 0.8179,
+      "step": 24141
+    },
+    {
+      "epoch": 0.8779866894570316,
+      "grad_norm": 0.0,
+      "learning_rate": 7.706986389504811e-07,
+      "loss": 0.8458,
+      "step": 24142
+    },
+    {
+      "epoch": 0.8780230570607702,
+      "grad_norm": 0.0,
+      "learning_rate": 7.702452645283953e-07,
+      "loss": 0.9421,
+      "step": 24143
+    },
+    {
+      "epoch": 0.8780594246645088,
+      "grad_norm": 0.0,
+      "learning_rate": 7.697920181560615e-07,
+      "loss": 0.8713,
+      "step": 24144
+    },
+    {
+      "epoch": 0.8780957922682474,
+      "grad_norm": 0.0,
+      "learning_rate": 7.693388998397677e-07,
+      "loss": 0.8598,
+      "step": 24145
+    },
+    {
+      "epoch": 0.878132159871986,
+      "grad_norm": 0.0,
+      "learning_rate": 7.688859095858003e-07,
+      "loss": 0.9252,
+      "step": 24146
+    },
+    {
+      "epoch": 0.8781685274757246,
+      "grad_norm": 0.0,
+      "learning_rate": 7.684330474004431e-07,
+      "loss": 0.859,
+      "step": 24147
+    },
+    {
+      "epoch": 0.8782048950794632,
+      "grad_norm": 0.0,
+      "learning_rate": 7.679803132899776e-07,
+      "loss": 0.978,
+      "step": 24148
+    },
+    {
+      "epoch": 0.8782412626832018,
+      "grad_norm": 0.0,
+      "learning_rate": 7.675277072606902e-07,
+      "loss": 0.7689,
+      "step": 24149
+    },
+    {
+      "epoch": 0.8782776302869404,
+      "grad_norm": 0.0,
+      "learning_rate": 7.670752293188544e-07,
+      "loss": 0.9402,
+      "step": 24150
+    },
+    {
+      "epoch": 0.878313997890679,
+      "grad_norm": 0.0,
+      "learning_rate": 7.666228794707508e-07,
+      "loss": 0.9632,
+      "step": 24151
+    },
+    {
+      "epoch": 0.8783503654944176,
+      "grad_norm": 0.0,
+      "learning_rate": 7.661706577226535e-07,
+      "loss": 1.0121,
+      "step": 24152
+    },
+    {
+      "epoch": 0.8783867330981562,
+      "grad_norm": 0.0,
+      "learning_rate": 7.657185640808384e-07,
+      "loss": 1.0702,
+      "step": 24153
+    },
+    {
+      "epoch": 0.8784231007018948,
+      "grad_norm": 0.0,
+      "learning_rate": 7.65266598551575e-07,
+      "loss": 0.8619,
+      "step": 24154
+    },
+    {
+      "epoch": 0.8784594683056334,
+      "grad_norm": 0.0,
+      "learning_rate": 7.648147611411361e-07,
+      "loss": 0.8005,
+      "step": 24155
+    },
+    {
+      "epoch": 0.878495835909372,
+      "grad_norm": 0.0,
+      "learning_rate": 7.643630518557898e-07,
+      "loss": 0.9512,
+      "step": 24156
+    },
+    {
+      "epoch": 0.8785322035131106,
+      "grad_norm": 0.0,
+      "learning_rate": 7.639114707018025e-07,
+      "loss": 0.8892,
+      "step": 24157
+    },
+    {
+      "epoch": 0.8785685711168492,
+      "grad_norm": 0.0,
+      "learning_rate": 7.63460017685439e-07,
+      "loss": 1.053,
+      "step": 24158
+    },
+    {
+      "epoch": 0.8786049387205876,
+      "grad_norm": 0.0,
+      "learning_rate": 7.630086928129632e-07,
+      "loss": 0.8839,
+      "step": 24159
+    },
+    {
+      "epoch": 0.8786413063243262,
+      "grad_norm": 0.0,
+      "learning_rate": 7.625574960906335e-07,
+      "loss": 0.9572,
+      "step": 24160
+    },
+    {
+      "epoch": 0.8786776739280648,
+      "grad_norm": 0.0,
+      "learning_rate": 7.621064275247148e-07,
+      "loss": 0.8182,
+      "step": 24161
+    },
+    {
+      "epoch": 0.8787140415318034,
+      "grad_norm": 0.0,
+      "learning_rate": 7.616554871214621e-07,
+      "loss": 1.1035,
+      "step": 24162
+    },
+    {
+      "epoch": 0.878750409135542,
+      "grad_norm": 0.0,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.8788,
+      "step": 24163
+    },
+    {
+      "epoch": 0.8787867767392806,
+      "grad_norm": 0.0,
+      "learning_rate": 7.607539908279804e-07,
+      "loss": 0.97,
+      "step": 24164
+    },
+    {
+      "epoch": 0.8788231443430192,
+      "grad_norm": 0.0,
+      "learning_rate": 7.60303434950257e-07,
+      "loss": 0.8871,
+      "step": 24165
+    },
+    {
+      "epoch": 0.8788595119467578,
+      "grad_norm": 0.0,
+      "learning_rate": 7.59853007260215e-07,
+      "loss": 0.859,
+      "step": 24166
+    },
+    {
+      "epoch": 0.8788958795504964,
+      "grad_norm": 0.0,
+      "learning_rate": 7.594027077640997e-07,
+      "loss": 0.9727,
+      "step": 24167
+    },
+    {
+      "epoch": 0.878932247154235,
+      "grad_norm": 0.0,
+      "learning_rate": 7.589525364681627e-07,
+      "loss": 0.9052,
+      "step": 24168
+    },
+    {
+      "epoch": 0.8789686147579736,
+      "grad_norm": 0.0,
+      "learning_rate": 7.585024933786489e-07,
+      "loss": 0.9502,
+      "step": 24169
+    },
+    {
+      "epoch": 0.8790049823617122,
+      "grad_norm": 0.0,
+      "learning_rate": 7.580525785018012e-07,
+      "loss": 0.8476,
+      "step": 24170
+    },
+    {
+      "epoch": 0.8790413499654508,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5760279184386e-07,
+      "loss": 0.9817,
+      "step": 24171
+    },
+    {
+      "epoch": 0.8790777175691894,
+      "grad_norm": 0.0,
+      "learning_rate": 7.571531334110682e-07,
+      "loss": 0.9333,
+      "step": 24172
+    },
+    {
+      "epoch": 0.879114085172928,
+      "grad_norm": 0.0,
+      "learning_rate": 7.567036032096598e-07,
+      "loss": 0.9297,
+      "step": 24173
+    },
+    {
+      "epoch": 0.8791504527766666,
+      "grad_norm": 0.0,
+      "learning_rate": 7.562542012458773e-07,
+      "loss": 0.9267,
+      "step": 24174
+    },
+    {
+      "epoch": 0.8791868203804052,
+      "grad_norm": 0.0,
+      "learning_rate": 7.558049275259516e-07,
+      "loss": 0.8651,
+      "step": 24175
+    },
+    {
+      "epoch": 0.8792231879841437,
+      "grad_norm": 0.0,
+      "learning_rate": 7.553557820561175e-07,
+      "loss": 0.9918,
+      "step": 24176
+    },
+    {
+      "epoch": 0.8792595555878823,
+      "grad_norm": 0.0,
+      "learning_rate": 7.549067648426034e-07,
+      "loss": 0.8513,
+      "step": 24177
+    },
+    {
+      "epoch": 0.8792959231916209,
+      "grad_norm": 0.0,
+      "learning_rate": 7.544578758916443e-07,
+      "loss": 0.9525,
+      "step": 24178
+    },
+    {
+      "epoch": 0.8793322907953595,
+      "grad_norm": 0.0,
+      "learning_rate": 7.540091152094609e-07,
+      "loss": 0.9126,
+      "step": 24179
+    },
+    {
+      "epoch": 0.8793686583990981,
+      "grad_norm": 0.0,
+      "learning_rate": 7.535604828022847e-07,
+      "loss": 0.8552,
+      "step": 24180
+    },
+    {
+      "epoch": 0.8794050260028367,
+      "grad_norm": 0.0,
+      "learning_rate": 7.531119786763375e-07,
+      "loss": 0.7783,
+      "step": 24181
+    },
+    {
+      "epoch": 0.8794413936065753,
+      "grad_norm": 0.0,
+      "learning_rate": 7.526636028378431e-07,
+      "loss": 0.9708,
+      "step": 24182
+    },
+    {
+      "epoch": 0.8794777612103138,
+      "grad_norm": 0.0,
+      "learning_rate": 7.5221535529302e-07,
+      "loss": 0.7708,
+      "step": 24183
+    },
+    {
+      "epoch": 0.8795141288140524,
+      "grad_norm": 0.0,
+      "learning_rate": 7.517672360480876e-07,
+      "loss": 0.9288,
+      "step": 24184
+    },
+    {
+      "epoch": 0.879550496417791,
+      "grad_norm": 0.0,
+      "learning_rate": 7.513192451092633e-07,
+      "loss": 0.9265,
+      "step": 24185
+    },
+    {
+      "epoch": 0.8795868640215296,
+      "grad_norm": 0.0,
+      "learning_rate": 7.508713824827618e-07,
+      "loss": 0.8582,
+      "step": 24186
+    },
+    {
+      "epoch": 0.8796232316252682,
+      "grad_norm": 0.0,
+      "learning_rate": 7.504236481747973e-07,
+      "loss": 0.8827,
+      "step": 24187
+    },
+    {
+      "epoch": 0.8796595992290068,
+      "grad_norm": 0.0,
+      "learning_rate": 7.499760421915825e-07,
+      "loss": 0.8668,
+      "step": 24188
+    },
+    {
+      "epoch": 0.8796959668327454,
+      "grad_norm": 0.0,
+      "learning_rate": 7.495285645393247e-07,
+      "loss": 0.8984,
+      "step": 24189
+    },
+    {
+      "epoch": 0.879732334436484,
+      "grad_norm": 0.0,
+      "learning_rate": 7.490812152242332e-07,
+      "loss": 0.9796,
+      "step": 24190
+    },
+    {
+      "epoch": 0.8797687020402226,
+      "grad_norm": 0.0,
+      "learning_rate": 7.486339942525167e-07,
+      "loss": 0.9579,
+      "step": 24191
+    },
+    {
+      "epoch": 0.8798050696439612,
+      "grad_norm": 0.0,
+      "learning_rate": 7.481869016303744e-07,
+      "loss": 0.8995,
+      "step": 24192
+    },
+    {
+      "epoch": 0.8798414372476997,
+      "grad_norm": 0.0,
+      "learning_rate": 7.477399373640148e-07,
+      "loss": 1.0229,
+      "step": 24193
+    },
+    {
+      "epoch": 0.8798778048514383,
+      "grad_norm": 0.0,
+      "learning_rate": 7.472931014596352e-07,
+      "loss": 0.8461,
+      "step": 24194
+    },
+    {
+      "epoch": 0.8799141724551769,
+      "grad_norm": 0.0,
+      "learning_rate": 7.468463939234361e-07,
+      "loss": 0.9036,
+      "step": 24195
+    },
+    {
+      "epoch": 0.8799505400589155,
+      "grad_norm": 0.0,
+      "learning_rate": 7.463998147616136e-07,
+      "loss": 0.8861,
+      "step": 24196
+    },
+    {
+      "epoch": 0.8799869076626541,
+      "grad_norm": 0.0,
+      "learning_rate": 7.459533639803673e-07,
+      "loss": 1.0206,
+      "step": 24197
+    },
+    {
+      "epoch": 0.8800232752663927,
+      "grad_norm": 0.0,
+      "learning_rate": 7.455070415858856e-07,
+      "loss": 0.9309,
+      "step": 24198
+    },
+    {
+      "epoch": 0.8800596428701313,
+      "grad_norm": 0.0,
+      "learning_rate": 7.450608475843646e-07,
+      "loss": 0.9879,
+      "step": 24199
+    },
+    {
+      "epoch": 0.8800960104738699,
+      "grad_norm": 0.0,
+      "learning_rate": 7.446147819819938e-07,
+      "loss": 1.0055,
+      "step": 24200
+    },
+    {
+      "epoch": 0.8801323780776085,
+      "grad_norm": 0.0,
+      "learning_rate": 7.441688447849604e-07,
+      "loss": 0.9268,
+      "step": 24201
+    },
+    {
+      "epoch": 0.8801687456813471,
+      "grad_norm": 0.0,
+      "learning_rate": 7.437230359994529e-07,
+      "loss": 0.889,
+      "step": 24202
+    },
+    {
+      "epoch": 0.8802051132850857,
+      "grad_norm": 0.0,
+      "learning_rate": 7.432773556316541e-07,
+      "loss": 0.9573,
+      "step": 24203
+    },
+    {
+      "epoch": 0.8802414808888243,
+      "grad_norm": 0.0,
+      "learning_rate": 7.428318036877491e-07,
+      "loss": 0.9154,
+      "step": 24204
+    },
+    {
+      "epoch": 0.8802778484925629,
+      "grad_norm": 0.0,
+      "learning_rate": 7.423863801739173e-07,
+      "loss": 0.994,
+      "step": 24205
+    },
+    {
+      "epoch": 0.8803142160963014,
+      "grad_norm": 0.0,
+      "learning_rate": 7.419410850963415e-07,
+      "loss": 0.8616,
+      "step": 24206
+    },
+    {
+      "epoch": 0.88035058370004,
+      "grad_norm": 0.0,
+      "learning_rate": 7.41495918461197e-07,
+      "loss": 0.9066,
+      "step": 24207
+    },
+    {
+      "epoch": 0.8803869513037786,
+      "grad_norm": 0.0,
+      "learning_rate": 7.410508802746608e-07,
+      "loss": 0.91,
+      "step": 24208
+    },
+    {
+      "epoch": 0.8804233189075172,
+      "grad_norm": 0.0,
+      "learning_rate": 7.406059705429047e-07,
+      "loss": 0.9059,
+      "step": 24209
+    },
+    {
+      "epoch": 0.8804596865112557,
+      "grad_norm": 0.0,
+      "learning_rate": 7.401611892721073e-07,
+      "loss": 0.8714,
+      "step": 24210
+    },
+    {
+      "epoch": 0.8804960541149943,
+      "grad_norm": 0.0,
+      "learning_rate": 7.397165364684322e-07,
+      "loss": 0.9883,
+      "step": 24211
+    },
+    {
+      "epoch": 0.8805324217187329,
+      "grad_norm": 0.0,
+      "learning_rate": 7.392720121380526e-07,
+      "loss": 0.8442,
+      "step": 24212
+    },
+    {
+      "epoch": 0.8805687893224715,
+      "grad_norm": 0.0,
+      "learning_rate": 7.388276162871344e-07,
+      "loss": 0.8372,
+      "step": 24213
+    },
+    {
+      "epoch": 0.8806051569262101,
+      "grad_norm": 0.0,
+      "learning_rate": 7.383833489218439e-07,
+      "loss": 0.918,
+      "step": 24214
+    },
+    {
+      "epoch": 0.8806415245299487,
+      "grad_norm": 0.0,
+      "learning_rate": 7.379392100483418e-07,
+      "loss": 0.9893,
+      "step": 24215
+    },
+    {
+      "epoch": 0.8806778921336873,
+      "grad_norm": 0.0,
+      "learning_rate": 7.374951996727963e-07,
+      "loss": 0.9944,
+      "step": 24216
+    },
+    {
+      "epoch": 0.8807142597374259,
+      "grad_norm": 0.0,
+      "learning_rate": 7.370513178013583e-07,
+      "loss": 0.9531,
+      "step": 24217
+    },
+    {
+      "epoch": 0.8807506273411645,
+      "grad_norm": 0.0,
+      "learning_rate": 7.366075644401938e-07,
+      "loss": 0.9924,
+      "step": 24218
+    },
+    {
+      "epoch": 0.8807869949449031,
+      "grad_norm": 0.0,
+      "learning_rate": 7.361639395954556e-07,
+      "loss": 0.8291,
+      "step": 24219
+    },
+    {
+      "epoch": 0.8808233625486417,
+      "grad_norm": 0.0,
+      "learning_rate": 7.357204432732989e-07,
+      "loss": 0.8805,
+      "step": 24220
+    },
+    {
+      "epoch": 0.8808597301523803,
+      "grad_norm": 0.0,
+      "learning_rate": 7.352770754798777e-07,
+      "loss": 0.9311,
+      "step": 24221
+    },
+    {
+      "epoch": 0.8808960977561189,
+      "grad_norm": 0.0,
+      "learning_rate": 7.348338362213403e-07,
+      "loss": 0.9433,
+      "step": 24222
+    },
+    {
+      "epoch": 0.8809324653598575,
+      "grad_norm": 0.0,
+      "learning_rate": 7.343907255038408e-07,
+      "loss": 1.1605,
+      "step": 24223
+    },
+    {
+      "epoch": 0.8809688329635961,
+      "grad_norm": 0.0,
+      "learning_rate": 7.339477433335196e-07,
+      "loss": 1.0455,
+      "step": 24224
+    },
+    {
+      "epoch": 0.8810052005673347,
+      "grad_norm": 0.0,
+      "learning_rate": 7.335048897165297e-07,
+      "loss": 1.015,
+      "step": 24225
+    },
+    {
+      "epoch": 0.8810415681710732,
+      "grad_norm": 0.0,
+      "learning_rate": 7.33062164659012e-07,
+      "loss": 0.9003,
+      "step": 24226
+    },
+    {
+      "epoch": 0.8810779357748117,
+      "grad_norm": 0.0,
+      "learning_rate": 7.326195681671078e-07,
+      "loss": 0.9779,
+      "step": 24227
+    },
+    {
+      "epoch": 0.8811143033785503,
+      "grad_norm": 0.0,
+      "learning_rate": 7.321771002469569e-07,
+      "loss": 0.9233,
+      "step": 24228
+    },
+    {
+      "epoch": 0.8811506709822889,
+      "grad_norm": 0.0,
+      "learning_rate": 7.317347609047032e-07,
+      "loss": 0.9383,
+      "step": 24229
+    },
+    {
+      "epoch": 0.8811870385860275,
+      "grad_norm": 0.0,
+      "learning_rate": 7.312925501464763e-07,
+      "loss": 1.0144,
+      "step": 24230
+    },
+    {
+      "epoch": 0.8812234061897661,
+      "grad_norm": 0.0,
+      "learning_rate": 7.308504679784157e-07,
+      "loss": 0.9642,
+      "step": 24231
+    },
+    {
+      "epoch": 0.8812597737935047,
+      "grad_norm": 0.0,
+      "learning_rate": 7.304085144066542e-07,
+      "loss": 0.9398,
+      "step": 24232
+    },
+    {
+      "epoch": 0.8812961413972433,
+      "grad_norm": 0.0,
+      "learning_rate": 7.299666894373236e-07,
+      "loss": 1.0045,
+      "step": 24233
+    },
+    {
+      "epoch": 0.8813325090009819,
+      "grad_norm": 0.0,
+      "learning_rate": 7.295249930765503e-07,
+      "loss": 0.8351,
+      "step": 24234
+    },
+    {
+      "epoch": 0.8813688766047205,
+      "grad_norm": 0.0,
+      "learning_rate": 7.290834253304657e-07,
+      "loss": 1.0408,
+      "step": 24235
+    },
+    {
+      "epoch": 0.8814052442084591,
+      "grad_norm": 0.0,
+      "learning_rate": 7.286419862051953e-07,
+      "loss": 0.885,
+      "step": 24236
+    },
+    {
+      "epoch": 0.8814416118121977,
+      "grad_norm": 0.0,
+      "learning_rate": 7.282006757068627e-07,
+      "loss": 0.8353,
+      "step": 24237
+    },
+    {
+      "epoch": 0.8814779794159363,
+      "grad_norm": 0.0,
+      "learning_rate": 7.27759493841591e-07,
+      "loss": 0.9277,
+      "step": 24238
+    },
+    {
+      "epoch": 0.8815143470196749,
+      "grad_norm": 0.0,
+      "learning_rate": 7.273184406155009e-07,
+      "loss": 0.8585,
+      "step": 24239
+    },
+    {
+      "epoch": 0.8815507146234135,
+      "grad_norm": 0.0,
+      "learning_rate": 7.268775160347119e-07,
+      "loss": 0.9219,
+      "step": 24240
+    },
+    {
+      "epoch": 0.8815870822271521,
+      "grad_norm": 0.0,
+      "learning_rate": 7.264367201053379e-07,
+      "loss": 0.9142,
+      "step": 24241
+    },
+    {
+      "epoch": 0.8816234498308907,
+      "grad_norm": 0.0,
+      "learning_rate": 7.259960528335008e-07,
+      "loss": 0.7793,
+      "step": 24242
+    },
+    {
+      "epoch": 0.8816598174346292,
+      "grad_norm": 0.0,
+      "learning_rate": 7.255555142253068e-07,
+      "loss": 0.8505,
+      "step": 24243
+    },
+    {
+      "epoch": 0.8816961850383678,
+      "grad_norm": 0.0,
+      "learning_rate": 7.251151042868731e-07,
+      "loss": 0.9626,
+      "step": 24244
+    },
+    {
+      "epoch": 0.8817325526421064,
+      "grad_norm": 0.0,
+      "learning_rate": 7.246748230243073e-07,
+      "loss": 0.8856,
+      "step": 24245
+    },
+    {
+      "epoch": 0.881768920245845,
+      "grad_norm": 0.0,
+      "learning_rate": 7.242346704437198e-07,
+      "loss": 0.9593,
+      "step": 24246
+    },
+    {
+      "epoch": 0.8818052878495836,
+      "grad_norm": 0.0,
+      "learning_rate": 7.237946465512124e-07,
+      "loss": 1.0173,
+      "step": 24247
+    },
+    {
+      "epoch": 0.8818416554533222,
+      "grad_norm": 0.0,
+      "learning_rate": 7.233547513528971e-07,
+      "loss": 0.8495,
+      "step": 24248
+    },
+    {
+      "epoch": 0.8818780230570608,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2291498485487e-07,
+      "loss": 0.8404,
+      "step": 24249
+    },
+    {
+      "epoch": 0.8819143906607994,
+      "grad_norm": 0.0,
+      "learning_rate": 7.224753470632362e-07,
+      "loss": 0.9937,
+      "step": 24250
+    },
+    {
+      "epoch": 0.881950758264538,
+      "grad_norm": 0.0,
+      "learning_rate": 7.220358379840931e-07,
+      "loss": 0.9399,
+      "step": 24251
+    },
+    {
+      "epoch": 0.8819871258682765,
+      "grad_norm": 0.0,
+      "learning_rate": 7.215964576235401e-07,
+      "loss": 0.9047,
+      "step": 24252
+    },
+    {
+      "epoch": 0.8820234934720151,
+      "grad_norm": 0.0,
+      "learning_rate": 7.211572059876703e-07,
+      "loss": 0.8951,
+      "step": 24253
+    },
+    {
+      "epoch": 0.8820598610757537,
+      "grad_norm": 0.0,
+      "learning_rate": 7.2071808308258e-07,
+      "loss": 0.8557,
+      "step": 24254
+    },
+    {
+      "epoch": 0.8820962286794923,
+      "grad_norm": 0.0,
+      "learning_rate": 7.202790889143619e-07,
+      "loss": 0.9938,
+      "step": 24255
+    },
+    {
+      "epoch": 0.8821325962832309,
+      "grad_norm": 0.0,
+      "learning_rate": 7.198402234891044e-07,
+      "loss": 0.8968,
+      "step": 24256
+    },
+    {
+      "epoch": 0.8821689638869695,
+      "grad_norm": 0.0,
+      "learning_rate": 7.194014868128973e-07,
+      "loss": 0.9911,
+      "step": 24257
+    },
+    {
+      "epoch": 0.8822053314907081,
+      "grad_norm": 0.0,
+      "learning_rate": 7.189628788918268e-07,
+      "loss": 1.0306,
+      "step": 24258
+    },
+    {
+      "epoch": 0.8822416990944467,
+      "grad_norm": 0.0,
+      "learning_rate": 7.185243997319791e-07,
+      "loss": 0.9058,
+      "step": 24259
+    },
+    {
+      "epoch": 0.8822780666981852,
+      "grad_norm": 0.0,
+      "learning_rate": 7.180860493394337e-07,
+      "loss": 0.7747,
+      "step": 24260
+    },
+    {
+      "epoch": 0.8823144343019238,
+      "grad_norm": 0.0,
+      "learning_rate": 7.176478277202792e-07,
+      "loss": 0.994,
+      "step": 24261
+    },
+    {
+      "epoch": 0.8823508019056624,
+      "grad_norm": 0.0,
+      "learning_rate": 7.172097348805874e-07,
+      "loss": 1.0139,
+      "step": 24262
+    },
+    {
+      "epoch": 0.882387169509401,
+      "grad_norm": 0.0,
+      "learning_rate": 7.167717708264422e-07,
+      "loss": 0.9625,
+      "step": 24263
+    },
+    {
+      "epoch": 0.8824235371131396,
+      "grad_norm": 0.0,
+      "learning_rate": 7.163339355639176e-07,
+      "loss": 0.8364,
+      "step": 24264
+    },
+    {
+      "epoch": 0.8824599047168782,
+      "grad_norm": 0.0,
+      "learning_rate": 7.158962290990878e-07,
+      "loss": 0.8798,
+      "step": 24265
+    },
+    {
+      "epoch": 0.8824962723206168,
+      "grad_norm": 0.0,
+      "learning_rate": 7.154586514380235e-07,
+      "loss": 0.8652,
+      "step": 24266
+    },
+    {
+      "epoch": 0.8825326399243554,
+      "grad_norm": 0.0,
+      "learning_rate": 7.150212025867987e-07,
+      "loss": 0.8579,
+      "step": 24267
+    },
+    {
+      "epoch": 0.882569007528094,
+      "grad_norm": 0.0,
+      "learning_rate": 7.145838825514817e-07,
+      "loss": 0.9148,
+      "step": 24268
+    },
+    {
+      "epoch": 0.8826053751318326,
+      "grad_norm": 0.0,
+      "learning_rate": 7.141466913381378e-07,
+      "loss": 0.9781,
+      "step": 24269
+    },
+    {
+      "epoch": 0.8826417427355712,
+      "grad_norm": 0.0,
+      "learning_rate": 7.137096289528345e-07,
+      "loss": 0.9061,
+      "step": 24270
+    },
+    {
+      "epoch": 0.8826781103393098,
+      "grad_norm": 0.0,
+      "learning_rate": 7.132726954016356e-07,
+      "loss": 0.9428,
+      "step": 24271
+    },
+    {
+      "epoch": 0.8827144779430484,
+      "grad_norm": 0.0,
+      "learning_rate": 7.128358906905986e-07,
+      "loss": 0.9086,
+      "step": 24272
+    },
+    {
+      "epoch": 0.882750845546787,
+      "grad_norm": 0.0,
+      "learning_rate": 7.123992148257896e-07,
+      "loss": 0.8737,
+      "step": 24273
+    },
+    {
+      "epoch": 0.8827872131505256,
+      "grad_norm": 0.0,
+      "learning_rate": 7.119626678132629e-07,
+      "loss": 0.8768,
+      "step": 24274
+    },
+    {
+      "epoch": 0.8828235807542641,
+      "grad_norm": 0.0,
+      "learning_rate": 7.11526249659078e-07,
+      "loss": 0.9057,
+      "step": 24275
+    },
+    {
+      "epoch": 0.8828599483580027,
+      "grad_norm": 0.0,
+      "learning_rate": 7.110899603692845e-07,
+      "loss": 0.9535,
+      "step": 24276
+    },
+    {
+      "epoch": 0.8828963159617412,
+      "grad_norm": 0.0,
+      "learning_rate": 7.106537999499429e-07,
+      "loss": 0.9024,
+      "step": 24277
+    },
+    {
+      "epoch": 0.8829326835654798,
+      "grad_norm": 0.0,
+      "learning_rate": 7.102177684070977e-07,
+      "loss": 0.9181,
+      "step": 24278
+    },
+    {
+      "epoch": 0.8829690511692184,
+      "grad_norm": 0.0,
+      "learning_rate": 7.097818657467992e-07,
+      "loss": 0.7845,
+      "step": 24279
+    },
+    {
+      "epoch": 0.883005418772957,
+      "grad_norm": 0.0,
+      "learning_rate": 7.093460919750994e-07,
+      "loss": 0.9096,
+      "step": 24280
+    },
+    {
+      "epoch": 0.8830417863766956,
+      "grad_norm": 0.0,
+      "learning_rate": 7.089104470980368e-07,
+      "loss": 0.9256,
+      "step": 24281
+    },
+    {
+      "epoch": 0.8830781539804342,
+      "grad_norm": 0.0,
+      "learning_rate": 7.084749311216621e-07,
+      "loss": 0.9197,
+      "step": 24282
+    },
+    {
+      "epoch": 0.8831145215841728,
+      "grad_norm": 0.0,
+      "learning_rate": 7.080395440520138e-07,
+      "loss": 0.9469,
+      "step": 24283
+    },
+    {
+      "epoch": 0.8831508891879114,
+      "grad_norm": 0.0,
+      "learning_rate": 7.076042858951349e-07,
+      "loss": 0.9285,
+      "step": 24284
+    },
+    {
+      "epoch": 0.88318725679165,
+      "grad_norm": 0.0,
+      "learning_rate": 7.071691566570593e-07,
+      "loss": 1.005,
+      "step": 24285
+    },
+    {
+      "epoch": 0.8832236243953886,
+      "grad_norm": 0.0,
+      "learning_rate": 7.067341563438291e-07,
+      "loss": 0.9107,
+      "step": 24286
+    },
+    {
+      "epoch": 0.8832599919991272,
+      "grad_norm": 0.0,
+      "learning_rate": 7.062992849614769e-07,
+      "loss": 0.9862,
+      "step": 24287
+    },
+    {
+      "epoch": 0.8832963596028658,
+      "grad_norm": 0.0,
+      "learning_rate": 7.05864542516036e-07,
+      "loss": 0.8694,
+      "step": 24288
+    },
+    {
+      "epoch": 0.8833327272066044,
+      "grad_norm": 0.0,
+      "learning_rate": 7.054299290135358e-07,
+      "loss": 0.9731,
+      "step": 24289
+    },
+    {
+      "epoch": 0.883369094810343,
+      "grad_norm": 0.0,
+      "learning_rate": 7.049954444600115e-07,
+      "loss": 0.9193,
+      "step": 24290
+    },
+    {
+      "epoch": 0.8834054624140816,
+      "grad_norm": 0.0,
+      "learning_rate": 7.045610888614851e-07,
+      "loss": 0.998,
+      "step": 24291
+    },
+    {
+      "epoch": 0.8834418300178202,
+      "grad_norm": 0.0,
+      "learning_rate": 7.041268622239861e-07,
+      "loss": 0.9493,
+      "step": 24292
+    },
+    {
+      "epoch": 0.8834781976215588,
+      "grad_norm": 0.0,
+      "learning_rate": 7.036927645535385e-07,
+      "loss": 0.9634,
+      "step": 24293
+    },
+    {
+      "epoch": 0.8835145652252973,
+      "grad_norm": 0.0,
+      "learning_rate": 7.032587958561632e-07,
+      "loss": 0.9321,
+      "step": 24294
+    },
+    {
+      "epoch": 0.8835509328290359,
+      "grad_norm": 0.0,
+      "learning_rate": 7.028249561378797e-07,
+      "loss": 0.9006,
+      "step": 24295
+    },
+    {
+      "epoch": 0.8835873004327744,
+      "grad_norm": 0.0,
+      "learning_rate": 7.023912454047132e-07,
+      "loss": 1.01,
+      "step": 24296
+    },
+    {
+      "epoch": 0.883623668036513,
+      "grad_norm": 0.0,
+      "learning_rate": 7.019576636626734e-07,
+      "loss": 0.9531,
+      "step": 24297
+    },
+    {
+      "epoch": 0.8836600356402516,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0152421091778e-07,
+      "loss": 0.9223,
+      "step": 24298
+    },
+    {
+      "epoch": 0.8836964032439902,
+      "grad_norm": 0.0,
+      "learning_rate": 7.010908871760458e-07,
+      "loss": 0.754,
+      "step": 24299
+    },
+    {
+      "epoch": 0.8837327708477288,
+      "grad_norm": 0.0,
+      "learning_rate": 7.006576924434815e-07,
+      "loss": 0.9368,
+      "step": 24300
+    },
+    {
+      "epoch": 0.8837691384514674,
+      "grad_norm": 0.0,
+      "learning_rate": 7.002246267260982e-07,
+      "loss": 0.9412,
+      "step": 24301
+    },
+    {
+      "epoch": 0.883805506055206,
+      "grad_norm": 0.0,
+      "learning_rate": 6.997916900299029e-07,
+      "loss": 0.9574,
+      "step": 24302
+    },
+    {
+      "epoch": 0.8838418736589446,
+      "grad_norm": 0.0,
+      "learning_rate": 6.993588823609032e-07,
+      "loss": 0.8658,
+      "step": 24303
+    },
+    {
+      "epoch": 0.8838782412626832,
+      "grad_norm": 0.0,
+      "learning_rate": 6.98926203725101e-07,
+      "loss": 0.8766,
+      "step": 24304
+    },
+    {
+      "epoch": 0.8839146088664218,
+      "grad_norm": 0.0,
+      "learning_rate": 6.984936541285037e-07,
+      "loss": 0.9224,
+      "step": 24305
+    },
+    {
+      "epoch": 0.8839509764701604,
+      "grad_norm": 0.0,
+      "learning_rate": 6.980612335771087e-07,
+      "loss": 0.8731,
+      "step": 24306
+    },
+    {
+      "epoch": 0.883987344073899,
+      "grad_norm": 0.0,
+      "learning_rate": 6.976289420769167e-07,
+      "loss": 0.9819,
+      "step": 24307
+    },
+    {
+      "epoch": 0.8840237116776376,
+      "grad_norm": 0.0,
+      "learning_rate": 6.97196779633924e-07,
+      "loss": 0.8898,
+      "step": 24308
+    },
+    {
+      "epoch": 0.8840600792813762,
+      "grad_norm": 0.0,
+      "learning_rate": 6.967647462541293e-07,
+      "loss": 1.0146,
+      "step": 24309
+    },
+    {
+      "epoch": 0.8840964468851148,
+      "grad_norm": 0.0,
+      "learning_rate": 6.963328419435211e-07,
+      "loss": 0.8391,
+      "step": 24310
+    },
+    {
+      "epoch": 0.8841328144888533,
+      "grad_norm": 0.0,
+      "learning_rate": 6.959010667080968e-07,
+      "loss": 0.9068,
+      "step": 24311
+    },
+    {
+      "epoch": 0.8841691820925919,
+      "grad_norm": 0.0,
+      "learning_rate": 6.954694205538437e-07,
+      "loss": 1.0031,
+      "step": 24312
+    },
+    {
+      "epoch": 0.8842055496963305,
+      "grad_norm": 0.0,
+      "learning_rate": 6.950379034867516e-07,
+      "loss": 0.8237,
+      "step": 24313
+    },
+    {
+      "epoch": 0.8842419173000691,
+      "grad_norm": 0.0,
+      "learning_rate": 6.946065155128046e-07,
+      "loss": 0.9011,
+      "step": 24314
+    },
+    {
+      "epoch": 0.8842782849038077,
+      "grad_norm": 0.0,
+      "learning_rate": 6.941752566379934e-07,
+      "loss": 0.9552,
+      "step": 24315
+    },
+    {
+      "epoch": 0.8843146525075463,
+      "grad_norm": 0.0,
+      "learning_rate": 6.937441268682932e-07,
+      "loss": 0.8785,
+      "step": 24316
+    },
+    {
+      "epoch": 0.8843510201112849,
+      "grad_norm": 0.0,
+      "learning_rate": 6.933131262096914e-07,
+      "loss": 0.9647,
+      "step": 24317
+    },
+    {
+      "epoch": 0.8843873877150235,
+      "grad_norm": 0.0,
+      "learning_rate": 6.928822546681657e-07,
+      "loss": 0.9563,
+      "step": 24318
+    },
+    {
+      "epoch": 0.884423755318762,
+      "grad_norm": 0.0,
+      "learning_rate": 6.924515122496934e-07,
+      "loss": 1.0811,
+      "step": 24319
+    },
+    {
+      "epoch": 0.8844601229225006,
+      "grad_norm": 0.0,
+      "learning_rate": 6.920208989602505e-07,
+      "loss": 0.8786,
+      "step": 24320
+    },
+    {
+      "epoch": 0.8844964905262392,
+      "grad_norm": 0.0,
+      "learning_rate": 6.915904148058095e-07,
+      "loss": 0.8619,
+      "step": 24321
+    },
+    {
+      "epoch": 0.8845328581299778,
+      "grad_norm": 0.0,
+      "learning_rate": 6.911600597923485e-07,
+      "loss": 0.8015,
+      "step": 24322
+    },
+    {
+      "epoch": 0.8845692257337164,
+      "grad_norm": 0.0,
+      "learning_rate": 6.907298339258306e-07,
+      "loss": 0.8889,
+      "step": 24323
+    },
+    {
+      "epoch": 0.884605593337455,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9029973721223e-07,
+      "loss": 0.9148,
+      "step": 24324
+    },
+    {
+      "epoch": 0.8846419609411936,
+      "grad_norm": 0.0,
+      "learning_rate": 6.898697696575108e-07,
+      "loss": 0.9157,
+      "step": 24325
+    },
+    {
+      "epoch": 0.8846783285449322,
+      "grad_norm": 0.0,
+      "learning_rate": 6.894399312676403e-07,
+      "loss": 0.8807,
+      "step": 24326
+    },
+    {
+      "epoch": 0.8847146961486707,
+      "grad_norm": 0.0,
+      "learning_rate": 6.890102220485784e-07,
+      "loss": 0.84,
+      "step": 24327
+    },
+    {
+      "epoch": 0.8847510637524093,
+      "grad_norm": 0.0,
+      "learning_rate": 6.885806420062935e-07,
+      "loss": 0.9477,
+      "step": 24328
+    },
+    {
+      "epoch": 0.8847874313561479,
+      "grad_norm": 0.0,
+      "learning_rate": 6.881511911467365e-07,
+      "loss": 0.9055,
+      "step": 24329
+    },
+    {
+      "epoch": 0.8848237989598865,
+      "grad_norm": 0.0,
+      "learning_rate": 6.877218694758725e-07,
+      "loss": 0.8731,
+      "step": 24330
+    },
+    {
+      "epoch": 0.8848601665636251,
+      "grad_norm": 0.0,
+      "learning_rate": 6.872926769996546e-07,
+      "loss": 0.8728,
+      "step": 24331
+    },
+    {
+      "epoch": 0.8848965341673637,
+      "grad_norm": 0.0,
+      "learning_rate": 6.868636137240381e-07,
+      "loss": 0.905,
+      "step": 24332
+    },
+    {
+      "epoch": 0.8849329017711023,
+      "grad_norm": 0.0,
+      "learning_rate": 6.864346796549736e-07,
+      "loss": 0.9281,
+      "step": 24333
+    },
+    {
+      "epoch": 0.8849692693748409,
+      "grad_norm": 0.0,
+      "learning_rate": 6.860058747984155e-07,
+      "loss": 0.8443,
+      "step": 24334
+    },
+    {
+      "epoch": 0.8850056369785795,
+      "grad_norm": 0.0,
+      "learning_rate": 6.855771991603099e-07,
+      "loss": 0.9651,
+      "step": 24335
+    },
+    {
+      "epoch": 0.8850420045823181,
+      "grad_norm": 0.0,
+      "learning_rate": 6.851486527466067e-07,
+      "loss": 0.8785,
+      "step": 24336
+    },
+    {
+      "epoch": 0.8850783721860567,
+      "grad_norm": 0.0,
+      "learning_rate": 6.847202355632498e-07,
+      "loss": 0.9106,
+      "step": 24337
+    },
+    {
+      "epoch": 0.8851147397897953,
+      "grad_norm": 0.0,
+      "learning_rate": 6.842919476161824e-07,
+      "loss": 0.8821,
+      "step": 24338
+    },
+    {
+      "epoch": 0.8851511073935339,
+      "grad_norm": 0.0,
+      "learning_rate": 6.838637889113475e-07,
+      "loss": 0.8559,
+      "step": 24339
+    },
+    {
+      "epoch": 0.8851874749972725,
+      "grad_norm": 0.0,
+      "learning_rate": 6.834357594546825e-07,
+      "loss": 0.923,
+      "step": 24340
+    },
+    {
+      "epoch": 0.8852238426010111,
+      "grad_norm": 0.0,
+      "learning_rate": 6.830078592521306e-07,
+      "loss": 0.9221,
+      "step": 24341
+    },
+    {
+      "epoch": 0.8852602102047497,
+      "grad_norm": 0.0,
+      "learning_rate": 6.825800883096234e-07,
+      "loss": 0.7333,
+      "step": 24342
+    },
+    {
+      "epoch": 0.8852965778084882,
+      "grad_norm": 0.0,
+      "learning_rate": 6.821524466330987e-07,
+      "loss": 0.8895,
+      "step": 24343
+    },
+    {
+      "epoch": 0.8853329454122267,
+      "grad_norm": 0.0,
+      "learning_rate": 6.817249342284893e-07,
+      "loss": 0.9985,
+      "step": 24344
+    },
+    {
+      "epoch": 0.8853693130159653,
+      "grad_norm": 0.0,
+      "learning_rate": 6.81297551101725e-07,
+      "loss": 0.9572,
+      "step": 24345
+    },
+    {
+      "epoch": 0.8854056806197039,
+      "grad_norm": 0.0,
+      "learning_rate": 6.808702972587344e-07,
+      "loss": 0.8262,
+      "step": 24346
+    },
+    {
+      "epoch": 0.8854420482234425,
+      "grad_norm": 0.0,
+      "learning_rate": 6.804431727054495e-07,
+      "loss": 0.8992,
+      "step": 24347
+    },
+    {
+      "epoch": 0.8854784158271811,
+      "grad_norm": 0.0,
+      "learning_rate": 6.800161774477887e-07,
+      "loss": 0.8901,
+      "step": 24348
+    },
+    {
+      "epoch": 0.8855147834309197,
+      "grad_norm": 0.0,
+      "learning_rate": 6.795893114916829e-07,
+      "loss": 0.8669,
+      "step": 24349
+    },
+    {
+      "epoch": 0.8855511510346583,
+      "grad_norm": 0.0,
+      "learning_rate": 6.791625748430508e-07,
+      "loss": 0.8079,
+      "step": 24350
+    },
+    {
+      "epoch": 0.8855875186383969,
+      "grad_norm": 0.0,
+      "learning_rate": 6.787359675078131e-07,
+      "loss": 0.8107,
+      "step": 24351
+    },
+    {
+      "epoch": 0.8856238862421355,
+      "grad_norm": 0.0,
+      "learning_rate": 6.783094894918885e-07,
+      "loss": 0.9476,
+      "step": 24352
+    },
+    {
+      "epoch": 0.8856602538458741,
+      "grad_norm": 0.0,
+      "learning_rate": 6.778831408011943e-07,
+      "loss": 0.9158,
+      "step": 24353
+    },
+    {
+      "epoch": 0.8856966214496127,
+      "grad_norm": 0.0,
+      "learning_rate": 6.77456921441646e-07,
+      "loss": 0.9543,
+      "step": 24354
+    },
+    {
+      "epoch": 0.8857329890533513,
+      "grad_norm": 0.0,
+      "learning_rate": 6.770308314191554e-07,
+      "loss": 0.8906,
+      "step": 24355
+    },
+    {
+      "epoch": 0.8857693566570899,
+      "grad_norm": 0.0,
+      "learning_rate": 6.766048707396345e-07,
+      "loss": 0.8873,
+      "step": 24356
+    },
+    {
+      "epoch": 0.8858057242608285,
+      "grad_norm": 0.0,
+      "learning_rate": 6.761790394089929e-07,
+      "loss": 0.8664,
+      "step": 24357
+    },
+    {
+      "epoch": 0.8858420918645671,
+      "grad_norm": 0.0,
+      "learning_rate": 6.757533374331382e-07,
+      "loss": 0.886,
+      "step": 24358
+    },
+    {
+      "epoch": 0.8858784594683057,
+      "grad_norm": 0.0,
+      "learning_rate": 6.753277648179746e-07,
+      "loss": 0.8883,
+      "step": 24359
+    },
+    {
+      "epoch": 0.8859148270720443,
+      "grad_norm": 0.0,
+      "learning_rate": 6.749023215694129e-07,
+      "loss": 0.9412,
+      "step": 24360
+    },
+    {
+      "epoch": 0.8859511946757828,
+      "grad_norm": 0.0,
+      "learning_rate": 6.74477007693346e-07,
+      "loss": 0.9651,
+      "step": 24361
+    },
+    {
+      "epoch": 0.8859875622795214,
+      "grad_norm": 0.0,
+      "learning_rate": 6.740518231956827e-07,
+      "loss": 0.8407,
+      "step": 24362
+    },
+    {
+      "epoch": 0.88602392988326,
+      "grad_norm": 0.0,
+      "learning_rate": 6.736267680823172e-07,
+      "loss": 0.8673,
+      "step": 24363
+    },
+    {
+      "epoch": 0.8860602974869985,
+      "grad_norm": 0.0,
+      "learning_rate": 6.73201842359149e-07,
+      "loss": 0.9328,
+      "step": 24364
+    },
+    {
+      "epoch": 0.8860966650907371,
+      "grad_norm": 0.0,
+      "learning_rate": 6.727770460320693e-07,
+      "loss": 0.8769,
+      "step": 24365
+    },
+    {
+      "epoch": 0.8861330326944757,
+      "grad_norm": 0.0,
+      "learning_rate": 6.723523791069775e-07,
+      "loss": 0.7754,
+      "step": 24366
+    },
+    {
+      "epoch": 0.8861694002982143,
+      "grad_norm": 0.0,
+      "learning_rate": 6.719278415897613e-07,
+      "loss": 0.9102,
+      "step": 24367
+    },
+    {
+      "epoch": 0.8862057679019529,
+      "grad_norm": 0.0,
+      "learning_rate": 6.715034334863124e-07,
+      "loss": 1.0518,
+      "step": 24368
+    },
+    {
+      "epoch": 0.8862421355056915,
+      "grad_norm": 0.0,
+      "learning_rate": 6.710791548025164e-07,
+      "loss": 0.9054,
+      "step": 24369
+    },
+    {
+      "epoch": 0.8862785031094301,
+      "grad_norm": 0.0,
+      "learning_rate": 6.706550055442629e-07,
+      "loss": 0.9459,
+      "step": 24370
+    },
+    {
+      "epoch": 0.8863148707131687,
+      "grad_norm": 0.0,
+      "learning_rate": 6.702309857174316e-07,
+      "loss": 0.9189,
+      "step": 24371
+    },
+    {
+      "epoch": 0.8863512383169073,
+      "grad_norm": 0.0,
+      "learning_rate": 6.698070953279101e-07,
+      "loss": 0.9934,
+      "step": 24372
+    },
+    {
+      "epoch": 0.8863876059206459,
+      "grad_norm": 0.0,
+      "learning_rate": 6.693833343815781e-07,
+      "loss": 0.9103,
+      "step": 24373
+    },
+    {
+      "epoch": 0.8864239735243845,
+      "grad_norm": 0.0,
+      "learning_rate": 6.689597028843131e-07,
+      "loss": 0.909,
+      "step": 24374
+    },
+    {
+      "epoch": 0.8864603411281231,
+      "grad_norm": 0.0,
+      "learning_rate": 6.685362008419938e-07,
+      "loss": 0.8735,
+      "step": 24375
+    },
+    {
+      "epoch": 0.8864967087318617,
+      "grad_norm": 0.0,
+      "learning_rate": 6.681128282604942e-07,
+      "loss": 0.8979,
+      "step": 24376
+    },
+    {
+      "epoch": 0.8865330763356003,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6768958514569e-07,
+      "loss": 0.9169,
+      "step": 24377
+    },
+    {
+      "epoch": 0.8865694439393388,
+      "grad_norm": 0.0,
+      "learning_rate": 6.672664715034494e-07,
+      "loss": 1.1085,
+      "step": 24378
+    },
+    {
+      "epoch": 0.8866058115430774,
+      "grad_norm": 0.0,
+      "learning_rate": 6.668434873396478e-07,
+      "loss": 0.9559,
+      "step": 24379
+    },
+    {
+      "epoch": 0.886642179146816,
+      "grad_norm": 0.0,
+      "learning_rate": 6.664206326601497e-07,
+      "loss": 0.8199,
+      "step": 24380
+    },
+    {
+      "epoch": 0.8866785467505546,
+      "grad_norm": 0.0,
+      "learning_rate": 6.659979074708234e-07,
+      "loss": 0.9785,
+      "step": 24381
+    },
+    {
+      "epoch": 0.8867149143542932,
+      "grad_norm": 0.0,
+      "learning_rate": 6.655753117775332e-07,
+      "loss": 0.9355,
+      "step": 24382
+    },
+    {
+      "epoch": 0.8867512819580318,
+      "grad_norm": 0.0,
+      "learning_rate": 6.651528455861411e-07,
+      "loss": 0.9333,
+      "step": 24383
+    },
+    {
+      "epoch": 0.8867876495617704,
+      "grad_norm": 0.0,
+      "learning_rate": 6.647305089025069e-07,
+      "loss": 0.9198,
+      "step": 24384
+    },
+    {
+      "epoch": 0.886824017165509,
+      "grad_norm": 0.0,
+      "learning_rate": 6.643083017324936e-07,
+      "loss": 0.9377,
+      "step": 24385
+    },
+    {
+      "epoch": 0.8868603847692476,
+      "grad_norm": 0.0,
+      "learning_rate": 6.638862240819576e-07,
+      "loss": 1.0712,
+      "step": 24386
+    },
+    {
+      "epoch": 0.8868967523729862,
+      "grad_norm": 0.0,
+      "learning_rate": 6.634642759567544e-07,
+      "loss": 0.8794,
+      "step": 24387
+    },
+    {
+      "epoch": 0.8869331199767247,
+      "grad_norm": 0.0,
+      "learning_rate": 6.630424573627348e-07,
+      "loss": 1.0533,
+      "step": 24388
+    },
+    {
+      "epoch": 0.8869694875804633,
+      "grad_norm": 0.0,
+      "learning_rate": 6.626207683057573e-07,
+      "loss": 0.9961,
+      "step": 24389
+    },
+    {
+      "epoch": 0.8870058551842019,
+      "grad_norm": 0.0,
+      "learning_rate": 6.621992087916662e-07,
+      "loss": 0.9617,
+      "step": 24390
+    },
+    {
+      "epoch": 0.8870422227879405,
+      "grad_norm": 0.0,
+      "learning_rate": 6.617777788263124e-07,
+      "loss": 0.8652,
+      "step": 24391
+    },
+    {
+      "epoch": 0.8870785903916791,
+      "grad_norm": 0.0,
+      "learning_rate": 6.613564784155446e-07,
+      "loss": 0.9213,
+      "step": 24392
+    },
+    {
+      "epoch": 0.8871149579954177,
+      "grad_norm": 0.0,
+      "learning_rate": 6.609353075652047e-07,
+      "loss": 0.9253,
+      "step": 24393
+    },
+    {
+      "epoch": 0.8871513255991563,
+      "grad_norm": 0.0,
+      "learning_rate": 6.605142662811371e-07,
+      "loss": 0.9405,
+      "step": 24394
+    },
+    {
+      "epoch": 0.8871876932028948,
+      "grad_norm": 0.0,
+      "learning_rate": 6.600933545691834e-07,
+      "loss": 0.8696,
+      "step": 24395
+    },
+    {
+      "epoch": 0.8872240608066334,
+      "grad_norm": 0.0,
+      "learning_rate": 6.596725724351827e-07,
+      "loss": 1.0326,
+      "step": 24396
+    },
+    {
+      "epoch": 0.887260428410372,
+      "grad_norm": 0.0,
+      "learning_rate": 6.592519198849712e-07,
+      "loss": 0.9244,
+      "step": 24397
+    },
+    {
+      "epoch": 0.8872967960141106,
+      "grad_norm": 0.0,
+      "learning_rate": 6.588313969243887e-07,
+      "loss": 0.8371,
+      "step": 24398
+    },
+    {
+      "epoch": 0.8873331636178492,
+      "grad_norm": 0.0,
+      "learning_rate": 6.584110035592661e-07,
+      "loss": 0.8797,
+      "step": 24399
+    },
+    {
+      "epoch": 0.8873695312215878,
+      "grad_norm": 0.0,
+      "learning_rate": 6.579907397954377e-07,
+      "loss": 0.9279,
+      "step": 24400
+    },
+    {
+      "epoch": 0.8874058988253264,
+      "grad_norm": 0.0,
+      "learning_rate": 6.575706056387332e-07,
+      "loss": 0.9891,
+      "step": 24401
+    },
+    {
+      "epoch": 0.887442266429065,
+      "grad_norm": 0.0,
+      "learning_rate": 6.571506010949802e-07,
+      "loss": 0.8297,
+      "step": 24402
+    },
+    {
+      "epoch": 0.8874786340328036,
+      "grad_norm": 0.0,
+      "learning_rate": 6.567307261700063e-07,
+      "loss": 0.8441,
+      "step": 24403
+    },
+    {
+      "epoch": 0.8875150016365422,
+      "grad_norm": 0.0,
+      "learning_rate": 6.563109808696389e-07,
+      "loss": 0.9887,
+      "step": 24404
+    },
+    {
+      "epoch": 0.8875513692402808,
+      "grad_norm": 0.0,
+      "learning_rate": 6.55891365199699e-07,
+      "loss": 0.9151,
+      "step": 24405
+    },
+    {
+      "epoch": 0.8875877368440194,
+      "grad_norm": 0.0,
+      "learning_rate": 6.554718791660097e-07,
+      "loss": 0.8353,
+      "step": 24406
+    },
+    {
+      "epoch": 0.887624104447758,
+      "grad_norm": 0.0,
+      "learning_rate": 6.550525227743876e-07,
+      "loss": 0.94,
+      "step": 24407
+    },
+    {
+      "epoch": 0.8876604720514966,
+      "grad_norm": 0.0,
+      "learning_rate": 6.546332960306567e-07,
+      "loss": 0.9209,
+      "step": 24408
+    },
+    {
+      "epoch": 0.8876968396552352,
+      "grad_norm": 0.0,
+      "learning_rate": 6.542141989406259e-07,
+      "loss": 0.879,
+      "step": 24409
+    },
+    {
+      "epoch": 0.8877332072589738,
+      "grad_norm": 0.0,
+      "learning_rate": 6.537952315101148e-07,
+      "loss": 0.9183,
+      "step": 24410
+    },
+    {
+      "epoch": 0.8877695748627122,
+      "grad_norm": 0.0,
+      "learning_rate": 6.533763937449344e-07,
+      "loss": 0.9284,
+      "step": 24411
+    },
+    {
+      "epoch": 0.8878059424664508,
+      "grad_norm": 0.0,
+      "learning_rate": 6.529576856508957e-07,
+      "loss": 0.9263,
+      "step": 24412
+    },
+    {
+      "epoch": 0.8878423100701894,
+      "grad_norm": 0.0,
+      "learning_rate": 6.525391072338072e-07,
+      "loss": 0.8738,
+      "step": 24413
+    },
+    {
+      "epoch": 0.887878677673928,
+      "grad_norm": 0.0,
+      "learning_rate": 6.521206584994766e-07,
+      "loss": 0.9398,
+      "step": 24414
+    },
+    {
+      "epoch": 0.8879150452776666,
+      "grad_norm": 0.0,
+      "learning_rate": 6.517023394537081e-07,
+      "loss": 0.902,
+      "step": 24415
+    },
+    {
+      "epoch": 0.8879514128814052,
+      "grad_norm": 0.0,
+      "learning_rate": 6.512841501023049e-07,
+      "loss": 0.973,
+      "step": 24416
+    },
+    {
+      "epoch": 0.8879877804851438,
+      "grad_norm": 0.0,
+      "learning_rate": 6.508660904510722e-07,
+      "loss": 0.8115,
+      "step": 24417
+    },
+    {
+      "epoch": 0.8880241480888824,
+      "grad_norm": 0.0,
+      "learning_rate": 6.504481605058066e-07,
+      "loss": 0.9464,
+      "step": 24418
+    },
+    {
+      "epoch": 0.888060515692621,
+      "grad_norm": 0.0,
+      "learning_rate": 6.500303602723091e-07,
+      "loss": 0.9396,
+      "step": 24419
+    },
+    {
+      "epoch": 0.8880968832963596,
+      "grad_norm": 0.0,
+      "learning_rate": 6.496126897563715e-07,
+      "loss": 1.0163,
+      "step": 24420
+    },
+    {
+      "epoch": 0.8881332509000982,
+      "grad_norm": 0.0,
+      "learning_rate": 6.491951489637949e-07,
+      "loss": 0.9452,
+      "step": 24421
+    },
+    {
+      "epoch": 0.8881696185038368,
+      "grad_norm": 0.0,
+      "learning_rate": 6.487777379003657e-07,
+      "loss": 0.8089,
+      "step": 24422
+    },
+    {
+      "epoch": 0.8882059861075754,
+      "grad_norm": 0.0,
+      "learning_rate": 6.483604565718793e-07,
+      "loss": 0.928,
+      "step": 24423
+    },
+    {
+      "epoch": 0.888242353711314,
+      "grad_norm": 0.0,
+      "learning_rate": 6.479433049841233e-07,
+      "loss": 0.8911,
+      "step": 24424
+    },
+    {
+      "epoch": 0.8882787213150526,
+      "grad_norm": 0.0,
+      "learning_rate": 6.475262831428852e-07,
+      "loss": 0.8905,
+      "step": 24425
+    },
+    {
+      "epoch": 0.8883150889187912,
+      "grad_norm": 0.0,
+      "learning_rate": 6.471093910539484e-07,
+      "loss": 0.9419,
+      "step": 24426
+    },
+    {
+      "epoch": 0.8883514565225298,
+      "grad_norm": 0.0,
+      "learning_rate": 6.466926287231023e-07,
+      "loss": 0.9261,
+      "step": 24427
+    },
+    {
+      "epoch": 0.8883878241262683,
+      "grad_norm": 0.0,
+      "learning_rate": 6.462759961561227e-07,
+      "loss": 0.9984,
+      "step": 24428
+    },
+    {
+      "epoch": 0.8884241917300069,
+      "grad_norm": 0.0,
+      "learning_rate": 6.458594933587936e-07,
+      "loss": 0.9428,
+      "step": 24429
+    },
+    {
+      "epoch": 0.8884605593337455,
+      "grad_norm": 0.0,
+      "learning_rate": 6.454431203368916e-07,
+      "loss": 0.8162,
+      "step": 24430
+    },
+    {
+      "epoch": 0.888496926937484,
+      "grad_norm": 0.0,
+      "learning_rate": 6.450268770961943e-07,
+      "loss": 0.8773,
+      "step": 24431
+    },
+    {
+      "epoch": 0.8885332945412227,
+      "grad_norm": 0.0,
+      "learning_rate": 6.446107636424769e-07,
+      "loss": 0.9314,
+      "step": 24432
+    },
+    {
+      "epoch": 0.8885696621449612,
+      "grad_norm": 0.0,
+      "learning_rate": 6.441947799815096e-07,
+      "loss": 0.8257,
+      "step": 24433
+    },
+    {
+      "epoch": 0.8886060297486998,
+      "grad_norm": 0.0,
+      "learning_rate": 6.437789261190686e-07,
+      "loss": 0.9791,
+      "step": 24434
+    },
+    {
+      "epoch": 0.8886423973524384,
+      "grad_norm": 0.0,
+      "learning_rate": 6.433632020609159e-07,
+      "loss": 0.8537,
+      "step": 24435
+    },
+    {
+      "epoch": 0.888678764956177,
+      "grad_norm": 0.0,
+      "learning_rate": 6.429476078128271e-07,
+      "loss": 0.9213,
+      "step": 24436
+    },
+    {
+      "epoch": 0.8887151325599156,
+      "grad_norm": 0.0,
+      "learning_rate": 6.425321433805631e-07,
+      "loss": 0.8171,
+      "step": 24437
+    },
+    {
+      "epoch": 0.8887515001636542,
+      "grad_norm": 0.0,
+      "learning_rate": 6.421168087698893e-07,
+      "loss": 0.9628,
+      "step": 24438
+    },
+    {
+      "epoch": 0.8887878677673928,
+      "grad_norm": 0.0,
+      "learning_rate": 6.417016039865664e-07,
+      "loss": 0.875,
+      "step": 24439
+    },
+    {
+      "epoch": 0.8888242353711314,
+      "grad_norm": 0.0,
+      "learning_rate": 6.41286529036359e-07,
+      "loss": 0.865,
+      "step": 24440
+    },
+    {
+      "epoch": 0.88886060297487,
+      "grad_norm": 0.0,
+      "learning_rate": 6.408715839250191e-07,
+      "loss": 0.937,
+      "step": 24441
+    },
+    {
+      "epoch": 0.8888969705786086,
+      "grad_norm": 0.0,
+      "learning_rate": 6.404567686583096e-07,
+      "loss": 0.7882,
+      "step": 24442
+    },
+    {
+      "epoch": 0.8889333381823472,
+      "grad_norm": 0.0,
+      "learning_rate": 6.400420832419818e-07,
+      "loss": 0.8403,
+      "step": 24443
+    },
+    {
+      "epoch": 0.8889697057860858,
+      "grad_norm": 0.0,
+      "learning_rate": 6.396275276817909e-07,
+      "loss": 0.86,
+      "step": 24444
+    },
+    {
+      "epoch": 0.8890060733898243,
+      "grad_norm": 0.0,
+      "learning_rate": 6.392131019834858e-07,
+      "loss": 0.8418,
+      "step": 24445
+    },
+    {
+      "epoch": 0.8890424409935629,
+      "grad_norm": 0.0,
+      "learning_rate": 6.387988061528216e-07,
+      "loss": 0.8486,
+      "step": 24446
+    },
+    {
+      "epoch": 0.8890788085973015,
+      "grad_norm": 0.0,
+      "learning_rate": 6.383846401955373e-07,
+      "loss": 0.8169,
+      "step": 24447
+    },
+    {
+      "epoch": 0.8891151762010401,
+      "grad_norm": 0.0,
+      "learning_rate": 6.379706041173861e-07,
+      "loss": 0.969,
+      "step": 24448
+    },
+    {
+      "epoch": 0.8891515438047787,
+      "grad_norm": 0.0,
+      "learning_rate": 6.375566979241099e-07,
+      "loss": 0.9263,
+      "step": 24449
+    },
+    {
+      "epoch": 0.8891879114085173,
+      "grad_norm": 0.0,
+      "learning_rate": 6.371429216214509e-07,
+      "loss": 1.0477,
+      "step": 24450
+    },
+    {
+      "epoch": 0.8892242790122559,
+      "grad_norm": 0.0,
+      "learning_rate": 6.367292752151499e-07,
+      "loss": 0.8286,
+      "step": 24451
+    },
+    {
+      "epoch": 0.8892606466159945,
+      "grad_norm": 0.0,
+      "learning_rate": 6.363157587109436e-07,
+      "loss": 0.8528,
+      "step": 24452
+    },
+    {
+      "epoch": 0.8892970142197331,
+      "grad_norm": 0.0,
+      "learning_rate": 6.359023721145741e-07,
+      "loss": 0.885,
+      "step": 24453
+    },
+    {
+      "epoch": 0.8893333818234717,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3548911543177e-07,
+      "loss": 1.026,
+      "step": 24454
+    },
+    {
+      "epoch": 0.8893697494272103,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3507598866827e-07,
+      "loss": 1.008,
+      "step": 24455
+    },
+    {
+      "epoch": 0.8894061170309489,
+      "grad_norm": 0.0,
+      "learning_rate": 6.346629918298031e-07,
+      "loss": 0.9191,
+      "step": 24456
+    },
+    {
+      "epoch": 0.8894424846346874,
+      "grad_norm": 0.0,
+      "learning_rate": 6.342501249221e-07,
+      "loss": 0.7808,
+      "step": 24457
+    },
+    {
+      "epoch": 0.889478852238426,
+      "grad_norm": 0.0,
+      "learning_rate": 6.338373879508852e-07,
+      "loss": 0.9409,
+      "step": 24458
+    },
+    {
+      "epoch": 0.8895152198421646,
+      "grad_norm": 0.0,
+      "learning_rate": 6.334247809218929e-07,
+      "loss": 0.8874,
+      "step": 24459
+    },
+    {
+      "epoch": 0.8895515874459032,
+      "grad_norm": 0.0,
+      "learning_rate": 6.330123038408376e-07,
+      "loss": 0.9271,
+      "step": 24460
+    },
+    {
+      "epoch": 0.8895879550496418,
+      "grad_norm": 0.0,
+      "learning_rate": 6.325999567134489e-07,
+      "loss": 0.9552,
+      "step": 24461
+    },
+    {
+      "epoch": 0.8896243226533803,
+      "grad_norm": 0.0,
+      "learning_rate": 6.321877395454446e-07,
+      "loss": 0.8675,
+      "step": 24462
+    },
+    {
+      "epoch": 0.8896606902571189,
+      "grad_norm": 0.0,
+      "learning_rate": 6.317756523425445e-07,
+      "loss": 0.9653,
+      "step": 24463
+    },
+    {
+      "epoch": 0.8896970578608575,
+      "grad_norm": 0.0,
+      "learning_rate": 6.31363695110464e-07,
+      "loss": 0.8622,
+      "step": 24464
+    },
+    {
+      "epoch": 0.8897334254645961,
+      "grad_norm": 0.0,
+      "learning_rate": 6.30951867854922e-07,
+      "loss": 0.9467,
+      "step": 24465
+    },
+    {
+      "epoch": 0.8897697930683347,
+      "grad_norm": 0.0,
+      "learning_rate": 6.305401705816294e-07,
+      "loss": 0.9584,
+      "step": 24466
+    },
+    {
+      "epoch": 0.8898061606720733,
+      "grad_norm": 0.0,
+      "learning_rate": 6.301286032962983e-07,
+      "loss": 1.0323,
+      "step": 24467
+    },
+    {
+      "epoch": 0.8898425282758119,
+      "grad_norm": 0.0,
+      "learning_rate": 6.297171660046387e-07,
+      "loss": 0.8417,
+      "step": 24468
+    },
+    {
+      "epoch": 0.8898788958795505,
+      "grad_norm": 0.0,
+      "learning_rate": 6.293058587123591e-07,
+      "loss": 0.8543,
+      "step": 24469
+    },
+    {
+      "epoch": 0.8899152634832891,
+      "grad_norm": 0.0,
+      "learning_rate": 6.288946814251651e-07,
+      "loss": 0.9579,
+      "step": 24470
+    },
+    {
+      "epoch": 0.8899516310870277,
+      "grad_norm": 0.0,
+      "learning_rate": 6.284836341487599e-07,
+      "loss": 0.9185,
+      "step": 24471
+    },
+    {
+      "epoch": 0.8899879986907663,
+      "grad_norm": 0.0,
+      "learning_rate": 6.280727168888512e-07,
+      "loss": 0.8288,
+      "step": 24472
+    },
+    {
+      "epoch": 0.8900243662945049,
+      "grad_norm": 0.0,
+      "learning_rate": 6.276619296511333e-07,
+      "loss": 0.9915,
+      "step": 24473
+    },
+    {
+      "epoch": 0.8900607338982435,
+      "grad_norm": 0.0,
+      "learning_rate": 6.272512724413105e-07,
+      "loss": 0.9479,
+      "step": 24474
+    },
+    {
+      "epoch": 0.8900971015019821,
+      "grad_norm": 0.0,
+      "learning_rate": 6.268407452650771e-07,
+      "loss": 0.9781,
+      "step": 24475
+    },
+    {
+      "epoch": 0.8901334691057207,
+      "grad_norm": 0.0,
+      "learning_rate": 6.264303481281297e-07,
+      "loss": 0.9904,
+      "step": 24476
+    },
+    {
+      "epoch": 0.8901698367094593,
+      "grad_norm": 0.0,
+      "learning_rate": 6.260200810361617e-07,
+      "loss": 0.9592,
+      "step": 24477
+    },
+    {
+      "epoch": 0.8902062043131979,
+      "grad_norm": 0.0,
+      "learning_rate": 6.256099439948648e-07,
+      "loss": 0.8385,
+      "step": 24478
+    },
+    {
+      "epoch": 0.8902425719169363,
+      "grad_norm": 0.0,
+      "learning_rate": 6.251999370099305e-07,
+      "loss": 0.8743,
+      "step": 24479
+    },
+    {
+      "epoch": 0.8902789395206749,
+      "grad_norm": 0.0,
+      "learning_rate": 6.247900600870461e-07,
+      "loss": 0.8942,
+      "step": 24480
+    },
+    {
+      "epoch": 0.8903153071244135,
+      "grad_norm": 0.0,
+      "learning_rate": 6.243803132318982e-07,
+      "loss": 1.061,
+      "step": 24481
+    },
+    {
+      "epoch": 0.8903516747281521,
+      "grad_norm": 0.0,
+      "learning_rate": 6.239706964501701e-07,
+      "loss": 0.9594,
+      "step": 24482
+    },
+    {
+      "epoch": 0.8903880423318907,
+      "grad_norm": 0.0,
+      "learning_rate": 6.235612097475452e-07,
+      "loss": 0.9092,
+      "step": 24483
+    },
+    {
+      "epoch": 0.8904244099356293,
+      "grad_norm": 0.0,
+      "learning_rate": 6.231518531297054e-07,
+      "loss": 0.8793,
+      "step": 24484
+    },
+    {
+      "epoch": 0.8904607775393679,
+      "grad_norm": 0.0,
+      "learning_rate": 6.227426266023307e-07,
+      "loss": 0.9295,
+      "step": 24485
+    },
+    {
+      "epoch": 0.8904971451431065,
+      "grad_norm": 0.0,
+      "learning_rate": 6.223335301710975e-07,
+      "loss": 0.8725,
+      "step": 24486
+    },
+    {
+      "epoch": 0.8905335127468451,
+      "grad_norm": 0.0,
+      "learning_rate": 6.219245638416794e-07,
+      "loss": 0.8641,
+      "step": 24487
+    },
+    {
+      "epoch": 0.8905698803505837,
+      "grad_norm": 0.0,
+      "learning_rate": 6.215157276197559e-07,
+      "loss": 0.88,
+      "step": 24488
+    },
+    {
+      "epoch": 0.8906062479543223,
+      "grad_norm": 0.0,
+      "learning_rate": 6.211070215109927e-07,
+      "loss": 0.8673,
+      "step": 24489
+    },
+    {
+      "epoch": 0.8906426155580609,
+      "grad_norm": 0.0,
+      "learning_rate": 6.206984455210619e-07,
+      "loss": 0.8858,
+      "step": 24490
+    },
+    {
+      "epoch": 0.8906789831617995,
+      "grad_norm": 0.0,
+      "learning_rate": 6.202899996556355e-07,
+      "loss": 0.8213,
+      "step": 24491
+    },
+    {
+      "epoch": 0.8907153507655381,
+      "grad_norm": 0.0,
+      "learning_rate": 6.198816839203747e-07,
+      "loss": 0.88,
+      "step": 24492
+    },
+    {
+      "epoch": 0.8907517183692767,
+      "grad_norm": 0.0,
+      "learning_rate": 6.194734983209483e-07,
+      "loss": 0.9135,
+      "step": 24493
+    },
+    {
+      "epoch": 0.8907880859730153,
+      "grad_norm": 0.0,
+      "learning_rate": 6.190654428630171e-07,
+      "loss": 0.7718,
+      "step": 24494
+    },
+    {
+      "epoch": 0.8908244535767539,
+      "grad_norm": 0.0,
+      "learning_rate": 6.186575175522436e-07,
+      "loss": 0.8319,
+      "step": 24495
+    },
+    {
+      "epoch": 0.8908608211804924,
+      "grad_norm": 0.0,
+      "learning_rate": 6.182497223942852e-07,
+      "loss": 0.8687,
+      "step": 24496
+    },
+    {
+      "epoch": 0.890897188784231,
+      "grad_norm": 0.0,
+      "learning_rate": 6.178420573948029e-07,
+      "loss": 0.9554,
+      "step": 24497
+    },
+    {
+      "epoch": 0.8909335563879696,
+      "grad_norm": 0.0,
+      "learning_rate": 6.174345225594502e-07,
+      "loss": 0.9478,
+      "step": 24498
+    },
+    {
+      "epoch": 0.8909699239917082,
+      "grad_norm": 0.0,
+      "learning_rate": 6.170271178938814e-07,
+      "loss": 0.9671,
+      "step": 24499
+    },
+    {
+      "epoch": 0.8910062915954468,
+      "grad_norm": 0.0,
+      "learning_rate": 6.166198434037462e-07,
+      "loss": 0.9293,
+      "step": 24500
+    },
+    {
+      "epoch": 0.8910426591991853,
+      "grad_norm": 0.0,
+      "learning_rate": 6.162126990947015e-07,
+      "loss": 0.8117,
+      "step": 24501
+    },
+    {
+      "epoch": 0.891079026802924,
+      "grad_norm": 0.0,
+      "learning_rate": 6.158056849723882e-07,
+      "loss": 0.8754,
+      "step": 24502
+    },
+    {
+      "epoch": 0.8911153944066625,
+      "grad_norm": 0.0,
+      "learning_rate": 6.153988010424583e-07,
+      "loss": 0.9672,
+      "step": 24503
+    },
+    {
+      "epoch": 0.8911517620104011,
+      "grad_norm": 0.0,
+      "learning_rate": 6.149920473105553e-07,
+      "loss": 0.9688,
+      "step": 24504
+    },
+    {
+      "epoch": 0.8911881296141397,
+      "grad_norm": 0.0,
+      "learning_rate": 6.145854237823212e-07,
+      "loss": 0.864,
+      "step": 24505
+    },
+    {
+      "epoch": 0.8912244972178783,
+      "grad_norm": 0.0,
+      "learning_rate": 6.141789304633971e-07,
+      "loss": 0.9475,
+      "step": 24506
+    },
+    {
+      "epoch": 0.8912608648216169,
+      "grad_norm": 0.0,
+      "learning_rate": 6.137725673594275e-07,
+      "loss": 0.8907,
+      "step": 24507
+    },
+    {
+      "epoch": 0.8912972324253555,
+      "grad_norm": 0.0,
+      "learning_rate": 6.133663344760444e-07,
+      "loss": 0.8384,
+      "step": 24508
+    },
+    {
+      "epoch": 0.8913336000290941,
+      "grad_norm": 0.0,
+      "learning_rate": 6.129602318188832e-07,
+      "loss": 0.8377,
+      "step": 24509
+    },
+    {
+      "epoch": 0.8913699676328327,
+      "grad_norm": 0.0,
+      "learning_rate": 6.125542593935819e-07,
+      "loss": 0.9597,
+      "step": 24510
+    },
+    {
+      "epoch": 0.8914063352365713,
+      "grad_norm": 0.0,
+      "learning_rate": 6.121484172057712e-07,
+      "loss": 0.9859,
+      "step": 24511
+    },
+    {
+      "epoch": 0.8914427028403098,
+      "grad_norm": 0.0,
+      "learning_rate": 6.117427052610825e-07,
+      "loss": 0.9828,
+      "step": 24512
+    },
+    {
+      "epoch": 0.8914790704440484,
+      "grad_norm": 0.0,
+      "learning_rate": 6.113371235651433e-07,
+      "loss": 0.9062,
+      "step": 24513
+    },
+    {
+      "epoch": 0.891515438047787,
+      "grad_norm": 0.0,
+      "learning_rate": 6.109316721235803e-07,
+      "loss": 0.934,
+      "step": 24514
+    },
+    {
+      "epoch": 0.8915518056515256,
+      "grad_norm": 0.0,
+      "learning_rate": 6.105263509420178e-07,
+      "loss": 0.8667,
+      "step": 24515
+    },
+    {
+      "epoch": 0.8915881732552642,
+      "grad_norm": 0.0,
+      "learning_rate": 6.101211600260815e-07,
+      "loss": 0.9781,
+      "step": 24516
+    },
+    {
+      "epoch": 0.8916245408590028,
+      "grad_norm": 0.0,
+      "learning_rate": 6.097160993813911e-07,
+      "loss": 0.8571,
+      "step": 24517
+    },
+    {
+      "epoch": 0.8916609084627414,
+      "grad_norm": 0.0,
+      "learning_rate": 6.093111690135678e-07,
+      "loss": 1.0069,
+      "step": 24518
+    },
+    {
+      "epoch": 0.89169727606648,
+      "grad_norm": 0.0,
+      "learning_rate": 6.08906368928226e-07,
+      "loss": 0.8344,
+      "step": 24519
+    },
+    {
+      "epoch": 0.8917336436702186,
+      "grad_norm": 0.0,
+      "learning_rate": 6.085016991309878e-07,
+      "loss": 0.8483,
+      "step": 24520
+    },
+    {
+      "epoch": 0.8917700112739572,
+      "grad_norm": 0.0,
+      "learning_rate": 6.08097159627461e-07,
+      "loss": 0.9104,
+      "step": 24521
+    },
+    {
+      "epoch": 0.8918063788776958,
+      "grad_norm": 0.0,
+      "learning_rate": 6.076927504232611e-07,
+      "loss": 1.003,
+      "step": 24522
+    },
+    {
+      "epoch": 0.8918427464814344,
+      "grad_norm": 0.0,
+      "learning_rate": 6.07288471523999e-07,
+      "loss": 0.9421,
+      "step": 24523
+    },
+    {
+      "epoch": 0.891879114085173,
+      "grad_norm": 0.0,
+      "learning_rate": 6.068843229352827e-07,
+      "loss": 0.8964,
+      "step": 24524
+    },
+    {
+      "epoch": 0.8919154816889115,
+      "grad_norm": 0.0,
+      "learning_rate": 6.064803046627177e-07,
+      "loss": 0.9792,
+      "step": 24525
+    },
+    {
+      "epoch": 0.8919518492926501,
+      "grad_norm": 0.0,
+      "learning_rate": 6.060764167119149e-07,
+      "loss": 1.0288,
+      "step": 24526
+    },
+    {
+      "epoch": 0.8919882168963887,
+      "grad_norm": 0.0,
+      "learning_rate": 6.056726590884698e-07,
+      "loss": 0.8727,
+      "step": 24527
+    },
+    {
+      "epoch": 0.8920245845001273,
+      "grad_norm": 0.0,
+      "learning_rate": 6.052690317979893e-07,
+      "loss": 1.0692,
+      "step": 24528
+    },
+    {
+      "epoch": 0.8920609521038658,
+      "grad_norm": 0.0,
+      "learning_rate": 6.048655348460708e-07,
+      "loss": 1.0434,
+      "step": 24529
+    },
+    {
+      "epoch": 0.8920973197076044,
+      "grad_norm": 0.0,
+      "learning_rate": 6.044621682383133e-07,
+      "loss": 0.8923,
+      "step": 24530
+    },
+    {
+      "epoch": 0.892133687311343,
+      "grad_norm": 0.0,
+      "learning_rate": 6.040589319803136e-07,
+      "loss": 1.011,
+      "step": 24531
+    },
+    {
+      "epoch": 0.8921700549150816,
+      "grad_norm": 0.0,
+      "learning_rate": 6.036558260776637e-07,
+      "loss": 0.9025,
+      "step": 24532
+    },
+    {
+      "epoch": 0.8922064225188202,
+      "grad_norm": 0.0,
+      "learning_rate": 6.032528505359592e-07,
+      "loss": 0.9088,
+      "step": 24533
+    },
+    {
+      "epoch": 0.8922427901225588,
+      "grad_norm": 0.0,
+      "learning_rate": 6.028500053607877e-07,
+      "loss": 0.9429,
+      "step": 24534
+    },
+    {
+      "epoch": 0.8922791577262974,
+      "grad_norm": 0.0,
+      "learning_rate": 6.024472905577406e-07,
+      "loss": 0.9484,
+      "step": 24535
+    },
+    {
+      "epoch": 0.892315525330036,
+      "grad_norm": 0.0,
+      "learning_rate": 6.020447061324031e-07,
+      "loss": 0.9574,
+      "step": 24536
+    },
+    {
+      "epoch": 0.8923518929337746,
+      "grad_norm": 0.0,
+      "learning_rate": 6.01642252090362e-07,
+      "loss": 0.9756,
+      "step": 24537
+    },
+    {
+      "epoch": 0.8923882605375132,
+      "grad_norm": 0.0,
+      "learning_rate": 6.012399284371983e-07,
+      "loss": 0.8935,
+      "step": 24538
+    },
+    {
+      "epoch": 0.8924246281412518,
+      "grad_norm": 0.0,
+      "learning_rate": 6.008377351784989e-07,
+      "loss": 0.8903,
+      "step": 24539
+    },
+    {
+      "epoch": 0.8924609957449904,
+      "grad_norm": 0.0,
+      "learning_rate": 6.004356723198368e-07,
+      "loss": 0.9568,
+      "step": 24540
+    },
+    {
+      "epoch": 0.892497363348729,
+      "grad_norm": 0.0,
+      "learning_rate": 6.000337398667955e-07,
+      "loss": 0.9523,
+      "step": 24541
+    },
+    {
+      "epoch": 0.8925337309524676,
+      "grad_norm": 0.0,
+      "learning_rate": 5.996319378249493e-07,
+      "loss": 0.9003,
+      "step": 24542
+    },
+    {
+      "epoch": 0.8925700985562062,
+      "grad_norm": 0.0,
+      "learning_rate": 5.992302661998728e-07,
+      "loss": 0.7697,
+      "step": 24543
+    },
+    {
+      "epoch": 0.8926064661599448,
+      "grad_norm": 0.0,
+      "learning_rate": 5.988287249971359e-07,
+      "loss": 0.8741,
+      "step": 24544
+    },
+    {
+      "epoch": 0.8926428337636834,
+      "grad_norm": 0.0,
+      "learning_rate": 5.98427314222314e-07,
+      "loss": 0.9854,
+      "step": 24545
+    },
+    {
+      "epoch": 0.8926792013674218,
+      "grad_norm": 0.0,
+      "learning_rate": 5.98026033880974e-07,
+      "loss": 0.8637,
+      "step": 24546
+    },
+    {
+      "epoch": 0.8927155689711604,
+      "grad_norm": 0.0,
+      "learning_rate": 5.976248839786835e-07,
+      "loss": 0.8941,
+      "step": 24547
+    },
+    {
+      "epoch": 0.892751936574899,
+      "grad_norm": 0.0,
+      "learning_rate": 5.972238645210071e-07,
+      "loss": 1.0076,
+      "step": 24548
+    },
+    {
+      "epoch": 0.8927883041786376,
+      "grad_norm": 0.0,
+      "learning_rate": 5.968229755135091e-07,
+      "loss": 0.8804,
+      "step": 24549
+    },
+    {
+      "epoch": 0.8928246717823762,
+      "grad_norm": 0.0,
+      "learning_rate": 5.964222169617506e-07,
+      "loss": 0.9761,
+      "step": 24550
+    },
+    {
+      "epoch": 0.8928610393861148,
+      "grad_norm": 0.0,
+      "learning_rate": 5.960215888712905e-07,
+      "loss": 0.9373,
+      "step": 24551
+    },
+    {
+      "epoch": 0.8928974069898534,
+      "grad_norm": 0.0,
+      "learning_rate": 5.956210912476912e-07,
+      "loss": 0.9301,
+      "step": 24552
+    },
+    {
+      "epoch": 0.892933774593592,
+      "grad_norm": 0.0,
+      "learning_rate": 5.952207240965036e-07,
+      "loss": 0.8649,
+      "step": 24553
+    },
+    {
+      "epoch": 0.8929701421973306,
+      "grad_norm": 0.0,
+      "learning_rate": 5.948204874232854e-07,
+      "loss": 0.9507,
+      "step": 24554
+    },
+    {
+      "epoch": 0.8930065098010692,
+      "grad_norm": 0.0,
+      "learning_rate": 5.944203812335891e-07,
+      "loss": 0.958,
+      "step": 24555
+    },
+    {
+      "epoch": 0.8930428774048078,
+      "grad_norm": 0.0,
+      "learning_rate": 5.940204055329657e-07,
+      "loss": 0.9506,
+      "step": 24556
+    },
+    {
+      "epoch": 0.8930792450085464,
+      "grad_norm": 0.0,
+      "learning_rate": 5.936205603269629e-07,
+      "loss": 0.8398,
+      "step": 24557
+    },
+    {
+      "epoch": 0.893115612612285,
+      "grad_norm": 0.0,
+      "learning_rate": 5.932208456211308e-07,
+      "loss": 0.9426,
+      "step": 24558
+    },
+    {
+      "epoch": 0.8931519802160236,
+      "grad_norm": 0.0,
+      "learning_rate": 5.928212614210105e-07,
+      "loss": 0.9022,
+      "step": 24559
+    },
+    {
+      "epoch": 0.8931883478197622,
+      "grad_norm": 0.0,
+      "learning_rate": 5.924218077321497e-07,
+      "loss": 0.9588,
+      "step": 24560
+    },
+    {
+      "epoch": 0.8932247154235008,
+      "grad_norm": 0.0,
+      "learning_rate": 5.920224845600886e-07,
+      "loss": 0.9596,
+      "step": 24561
+    },
+    {
+      "epoch": 0.8932610830272394,
+      "grad_norm": 0.0,
+      "learning_rate": 5.916232919103671e-07,
+      "loss": 0.8107,
+      "step": 24562
+    },
+    {
+      "epoch": 0.8932974506309779,
+      "grad_norm": 0.0,
+      "learning_rate": 5.912242297885218e-07,
+      "loss": 0.9091,
+      "step": 24563
+    },
+    {
+      "epoch": 0.8933338182347165,
+      "grad_norm": 0.0,
+      "learning_rate": 5.90825298200094e-07,
+      "loss": 0.9219,
+      "step": 24564
+    },
+    {
+      "epoch": 0.8933701858384551,
+      "grad_norm": 0.0,
+      "learning_rate": 5.904264971506135e-07,
+      "loss": 0.9411,
+      "step": 24565
+    },
+    {
+      "epoch": 0.8934065534421937,
+      "grad_norm": 0.0,
+      "learning_rate": 5.900278266456161e-07,
+      "loss": 0.8497,
+      "step": 24566
+    },
+    {
+      "epoch": 0.8934429210459323,
+      "grad_norm": 0.0,
+      "learning_rate": 5.896292866906317e-07,
+      "loss": 0.8841,
+      "step": 24567
+    },
+    {
+      "epoch": 0.8934792886496709,
+      "grad_norm": 0.0,
+      "learning_rate": 5.892308772911881e-07,
+      "loss": 0.9628,
+      "step": 24568
+    },
+    {
+      "epoch": 0.8935156562534095,
+      "grad_norm": 0.0,
+      "learning_rate": 5.888325984528154e-07,
+      "loss": 0.8216,
+      "step": 24569
+    },
+    {
+      "epoch": 0.893552023857148,
+      "grad_norm": 0.0,
+      "learning_rate": 5.884344501810357e-07,
+      "loss": 0.9954,
+      "step": 24570
+    },
+    {
+      "epoch": 0.8935883914608866,
+      "grad_norm": 0.0,
+      "learning_rate": 5.880364324813781e-07,
+      "loss": 0.9968,
+      "step": 24571
+    },
+    {
+      "epoch": 0.8936247590646252,
+      "grad_norm": 0.0,
+      "learning_rate": 5.87638545359357e-07,
+      "loss": 0.9488,
+      "step": 24572
+    },
+    {
+      "epoch": 0.8936611266683638,
+      "grad_norm": 0.0,
+      "learning_rate": 5.872407888204989e-07,
+      "loss": 0.7698,
+      "step": 24573
+    },
+    {
+      "epoch": 0.8936974942721024,
+      "grad_norm": 0.0,
+      "learning_rate": 5.868431628703197e-07,
+      "loss": 0.8059,
+      "step": 24574
+    },
+    {
+      "epoch": 0.893733861875841,
+      "grad_norm": 0.0,
+      "learning_rate": 5.864456675143371e-07,
+      "loss": 0.8779,
+      "step": 24575
+    },
+    {
+      "epoch": 0.8937702294795796,
+      "grad_norm": 0.0,
+      "learning_rate": 5.860483027580622e-07,
+      "loss": 0.895,
+      "step": 24576
+    },
+    {
+      "epoch": 0.8938065970833182,
+      "grad_norm": 0.0,
+      "learning_rate": 5.856510686070127e-07,
+      "loss": 0.9761,
+      "step": 24577
+    },
+    {
+      "epoch": 0.8938429646870568,
+      "grad_norm": 0.0,
+      "learning_rate": 5.852539650666967e-07,
+      "loss": 0.9056,
+      "step": 24578
+    },
+    {
+      "epoch": 0.8938793322907954,
+      "grad_norm": 0.0,
+      "learning_rate": 5.848569921426239e-07,
+      "loss": 1.0079,
+      "step": 24579
+    },
+    {
+      "epoch": 0.8939156998945339,
+      "grad_norm": 0.0,
+      "learning_rate": 5.844601498403025e-07,
+      "loss": 0.7664,
+      "step": 24580
+    },
+    {
+      "epoch": 0.8939520674982725,
+      "grad_norm": 0.0,
+      "learning_rate": 5.840634381652377e-07,
+      "loss": 0.8511,
+      "step": 24581
+    },
+    {
+      "epoch": 0.8939884351020111,
+      "grad_norm": 0.0,
+      "learning_rate": 5.836668571229321e-07,
+      "loss": 0.8737,
+      "step": 24582
+    },
+    {
+      "epoch": 0.8940248027057497,
+      "grad_norm": 0.0,
+      "learning_rate": 5.832704067188899e-07,
+      "loss": 0.9174,
+      "step": 24583
+    },
+    {
+      "epoch": 0.8940611703094883,
+      "grad_norm": 0.0,
+      "learning_rate": 5.828740869586102e-07,
+      "loss": 0.9297,
+      "step": 24584
+    },
+    {
+      "epoch": 0.8940975379132269,
+      "grad_norm": 0.0,
+      "learning_rate": 5.824778978475909e-07,
+      "loss": 0.9099,
+      "step": 24585
+    },
+    {
+      "epoch": 0.8941339055169655,
+      "grad_norm": 0.0,
+      "learning_rate": 5.820818393913296e-07,
+      "loss": 0.9519,
+      "step": 24586
+    },
+    {
+      "epoch": 0.8941702731207041,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8168591159532e-07,
+      "loss": 0.86,
+      "step": 24587
+    },
+    {
+      "epoch": 0.8942066407244427,
+      "grad_norm": 0.0,
+      "learning_rate": 5.812901144650562e-07,
+      "loss": 0.9905,
+      "step": 24588
+    },
+    {
+      "epoch": 0.8942430083281813,
+      "grad_norm": 0.0,
+      "learning_rate": 5.808944480060275e-07,
+      "loss": 0.989,
+      "step": 24589
+    },
+    {
+      "epoch": 0.8942793759319199,
+      "grad_norm": 0.0,
+      "learning_rate": 5.80498912223727e-07,
+      "loss": 0.9575,
+      "step": 24590
+    },
+    {
+      "epoch": 0.8943157435356585,
+      "grad_norm": 0.0,
+      "learning_rate": 5.801035071236361e-07,
+      "loss": 1.0404,
+      "step": 24591
+    },
+    {
+      "epoch": 0.894352111139397,
+      "grad_norm": 0.0,
+      "learning_rate": 5.797082327112458e-07,
+      "loss": 0.8851,
+      "step": 24592
+    },
+    {
+      "epoch": 0.8943884787431357,
+      "grad_norm": 0.0,
+      "learning_rate": 5.793130889920373e-07,
+      "loss": 0.9339,
+      "step": 24593
+    },
+    {
+      "epoch": 0.8944248463468742,
+      "grad_norm": 0.0,
+      "learning_rate": 5.789180759714941e-07,
+      "loss": 0.8358,
+      "step": 24594
+    },
+    {
+      "epoch": 0.8944612139506128,
+      "grad_norm": 0.0,
+      "learning_rate": 5.78523193655095e-07,
+      "loss": 0.9754,
+      "step": 24595
+    },
+    {
+      "epoch": 0.8944975815543513,
+      "grad_norm": 0.0,
+      "learning_rate": 5.781284420483202e-07,
+      "loss": 0.9284,
+      "step": 24596
+    },
+    {
+      "epoch": 0.8945339491580899,
+      "grad_norm": 0.0,
+      "learning_rate": 5.777338211566453e-07,
+      "loss": 0.9946,
+      "step": 24597
+    },
+    {
+      "epoch": 0.8945703167618285,
+      "grad_norm": 0.0,
+      "learning_rate": 5.773393309855457e-07,
+      "loss": 0.8573,
+      "step": 24598
+    },
+    {
+      "epoch": 0.8946066843655671,
+      "grad_norm": 0.0,
+      "learning_rate": 5.769449715404917e-07,
+      "loss": 0.9979,
+      "step": 24599
+    },
+    {
+      "epoch": 0.8946430519693057,
+      "grad_norm": 0.0,
+      "learning_rate": 5.765507428269601e-07,
+      "loss": 0.9349,
+      "step": 24600
+    },
+    {
+      "epoch": 0.8946794195730443,
+      "grad_norm": 0.0,
+      "learning_rate": 5.76156644850413e-07,
+      "loss": 0.7806,
+      "step": 24601
+    },
+    {
+      "epoch": 0.8947157871767829,
+      "grad_norm": 0.0,
+      "learning_rate": 5.757626776163239e-07,
+      "loss": 0.8518,
+      "step": 24602
+    },
+    {
+      "epoch": 0.8947521547805215,
+      "grad_norm": 0.0,
+      "learning_rate": 5.753688411301573e-07,
+      "loss": 1.0088,
+      "step": 24603
+    },
+    {
+      "epoch": 0.8947885223842601,
+      "grad_norm": 0.0,
+      "learning_rate": 5.749751353973743e-07,
+      "loss": 0.9154,
+      "step": 24604
+    },
+    {
+      "epoch": 0.8948248899879987,
+      "grad_norm": 0.0,
+      "learning_rate": 5.745815604234406e-07,
+      "loss": 0.935,
+      "step": 24605
+    },
+    {
+      "epoch": 0.8948612575917373,
+      "grad_norm": 0.0,
+      "learning_rate": 5.741881162138141e-07,
+      "loss": 0.8974,
+      "step": 24606
+    },
+    {
+      "epoch": 0.8948976251954759,
+      "grad_norm": 0.0,
+      "learning_rate": 5.737948027739537e-07,
+      "loss": 0.861,
+      "step": 24607
+    },
+    {
+      "epoch": 0.8949339927992145,
+      "grad_norm": 0.0,
+      "learning_rate": 5.734016201093151e-07,
+      "loss": 0.9357,
+      "step": 24608
+    },
+    {
+      "epoch": 0.8949703604029531,
+      "grad_norm": 0.0,
+      "learning_rate": 5.730085682253561e-07,
+      "loss": 0.9431,
+      "step": 24609
+    },
+    {
+      "epoch": 0.8950067280066917,
+      "grad_norm": 0.0,
+      "learning_rate": 5.72615647127529e-07,
+      "loss": 0.886,
+      "step": 24610
+    },
+    {
+      "epoch": 0.8950430956104303,
+      "grad_norm": 0.0,
+      "learning_rate": 5.722228568212828e-07,
+      "loss": 0.9028,
+      "step": 24611
+    },
+    {
+      "epoch": 0.8950794632141689,
+      "grad_norm": 0.0,
+      "learning_rate": 5.718301973120688e-07,
+      "loss": 0.9169,
+      "step": 24612
+    },
+    {
+      "epoch": 0.8951158308179074,
+      "grad_norm": 0.0,
+      "learning_rate": 5.714376686053336e-07,
+      "loss": 0.9883,
+      "step": 24613
+    },
+    {
+      "epoch": 0.895152198421646,
+      "grad_norm": 0.0,
+      "learning_rate": 5.710452707065217e-07,
+      "loss": 1.0637,
+      "step": 24614
+    },
+    {
+      "epoch": 0.8951885660253845,
+      "grad_norm": 0.0,
+      "learning_rate": 5.706530036210811e-07,
+      "loss": 0.9251,
+      "step": 24615
+    },
+    {
+      "epoch": 0.8952249336291231,
+      "grad_norm": 0.0,
+      "learning_rate": 5.702608673544508e-07,
+      "loss": 1.0087,
+      "step": 24616
+    },
+    {
+      "epoch": 0.8952613012328617,
+      "grad_norm": 0.0,
+      "learning_rate": 5.698688619120718e-07,
+      "loss": 0.8248,
+      "step": 24617
+    },
+    {
+      "epoch": 0.8952976688366003,
+      "grad_norm": 0.0,
+      "learning_rate": 5.694769872993822e-07,
+      "loss": 0.8965,
+      "step": 24618
+    },
+    {
+      "epoch": 0.8953340364403389,
+      "grad_norm": 0.0,
+      "learning_rate": 5.690852435218208e-07,
+      "loss": 0.8164,
+      "step": 24619
+    },
+    {
+      "epoch": 0.8953704040440775,
+      "grad_norm": 0.0,
+      "learning_rate": 5.68693630584819e-07,
+      "loss": 0.8896,
+      "step": 24620
+    },
+    {
+      "epoch": 0.8954067716478161,
+      "grad_norm": 0.0,
+      "learning_rate": 5.683021484938134e-07,
+      "loss": 1.0524,
+      "step": 24621
+    },
+    {
+      "epoch": 0.8954431392515547,
+      "grad_norm": 0.0,
+      "learning_rate": 5.67910797254233e-07,
+      "loss": 0.7792,
+      "step": 24622
+    },
+    {
+      "epoch": 0.8954795068552933,
+      "grad_norm": 0.0,
+      "learning_rate": 5.67519576871508e-07,
+      "loss": 0.8044,
+      "step": 24623
+    },
+    {
+      "epoch": 0.8955158744590319,
+      "grad_norm": 0.0,
+      "learning_rate": 5.671284873510663e-07,
+      "loss": 0.8638,
+      "step": 24624
+    },
+    {
+      "epoch": 0.8955522420627705,
+      "grad_norm": 0.0,
+      "learning_rate": 5.667375286983334e-07,
+      "loss": 1.0316,
+      "step": 24625
+    },
+    {
+      "epoch": 0.8955886096665091,
+      "grad_norm": 0.0,
+      "learning_rate": 5.663467009187329e-07,
+      "loss": 0.9037,
+      "step": 24626
+    },
+    {
+      "epoch": 0.8956249772702477,
+      "grad_norm": 0.0,
+      "learning_rate": 5.659560040176859e-07,
+      "loss": 0.8661,
+      "step": 24627
+    },
+    {
+      "epoch": 0.8956613448739863,
+      "grad_norm": 0.0,
+      "learning_rate": 5.655654380006159e-07,
+      "loss": 0.8576,
+      "step": 24628
+    },
+    {
+      "epoch": 0.8956977124777249,
+      "grad_norm": 0.0,
+      "learning_rate": 5.651750028729397e-07,
+      "loss": 0.9257,
+      "step": 24629
+    },
+    {
+      "epoch": 0.8957340800814634,
+      "grad_norm": 0.0,
+      "learning_rate": 5.647846986400741e-07,
+      "loss": 0.9764,
+      "step": 24630
+    },
+    {
+      "epoch": 0.895770447685202,
+      "grad_norm": 0.0,
+      "learning_rate": 5.643945253074323e-07,
+      "loss": 0.9416,
+      "step": 24631
+    },
+    {
+      "epoch": 0.8958068152889406,
+      "grad_norm": 0.0,
+      "learning_rate": 5.640044828804314e-07,
+      "loss": 0.7931,
+      "step": 24632
+    },
+    {
+      "epoch": 0.8958431828926792,
+      "grad_norm": 0.0,
+      "learning_rate": 5.636145713644781e-07,
+      "loss": 0.8646,
+      "step": 24633
+    },
+    {
+      "epoch": 0.8958795504964178,
+      "grad_norm": 0.0,
+      "learning_rate": 5.632247907649857e-07,
+      "loss": 0.9705,
+      "step": 24634
+    },
+    {
+      "epoch": 0.8959159181001564,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6283514108736e-07,
+      "loss": 0.9377,
+      "step": 24635
+    },
+    {
+      "epoch": 0.895952285703895,
+      "grad_norm": 0.0,
+      "learning_rate": 5.624456223370078e-07,
+      "loss": 0.8799,
+      "step": 24636
+    },
+    {
+      "epoch": 0.8959886533076336,
+      "grad_norm": 0.0,
+      "learning_rate": 5.620562345193303e-07,
+      "loss": 1.0103,
+      "step": 24637
+    },
+    {
+      "epoch": 0.8960250209113721,
+      "grad_norm": 0.0,
+      "learning_rate": 5.616669776397354e-07,
+      "loss": 0.9027,
+      "step": 24638
+    },
+    {
+      "epoch": 0.8960613885151107,
+      "grad_norm": 0.0,
+      "learning_rate": 5.612778517036166e-07,
+      "loss": 0.9342,
+      "step": 24639
+    },
+    {
+      "epoch": 0.8960977561188493,
+      "grad_norm": 0.0,
+      "learning_rate": 5.608888567163773e-07,
+      "loss": 1.0129,
+      "step": 24640
+    },
+    {
+      "epoch": 0.8961341237225879,
+      "grad_norm": 0.0,
+      "learning_rate": 5.604999926834121e-07,
+      "loss": 0.9215,
+      "step": 24641
+    },
+    {
+      "epoch": 0.8961704913263265,
+      "grad_norm": 0.0,
+      "learning_rate": 5.601112596101177e-07,
+      "loss": 0.8193,
+      "step": 24642
+    },
+    {
+      "epoch": 0.8962068589300651,
+      "grad_norm": 0.0,
+      "learning_rate": 5.597226575018855e-07,
+      "loss": 0.9212,
+      "step": 24643
+    },
+    {
+      "epoch": 0.8962432265338037,
+      "grad_norm": 0.0,
+      "learning_rate": 5.593341863641067e-07,
+      "loss": 0.8602,
+      "step": 24644
+    },
+    {
+      "epoch": 0.8962795941375423,
+      "grad_norm": 0.0,
+      "learning_rate": 5.589458462021736e-07,
+      "loss": 1.069,
+      "step": 24645
+    },
+    {
+      "epoch": 0.8963159617412809,
+      "grad_norm": 0.0,
+      "learning_rate": 5.585576370214696e-07,
+      "loss": 0.9505,
+      "step": 24646
+    },
+    {
+      "epoch": 0.8963523293450194,
+      "grad_norm": 0.0,
+      "learning_rate": 5.581695588273827e-07,
+      "loss": 0.9297,
+      "step": 24647
+    },
+    {
+      "epoch": 0.896388696948758,
+      "grad_norm": 0.0,
+      "learning_rate": 5.577816116252988e-07,
+      "loss": 0.9112,
+      "step": 24648
+    },
+    {
+      "epoch": 0.8964250645524966,
+      "grad_norm": 0.0,
+      "learning_rate": 5.573937954205977e-07,
+      "loss": 0.9851,
+      "step": 24649
+    },
+    {
+      "epoch": 0.8964614321562352,
+      "grad_norm": 0.0,
+      "learning_rate": 5.570061102186585e-07,
+      "loss": 0.8699,
+      "step": 24650
+    },
+    {
+      "epoch": 0.8964977997599738,
+      "grad_norm": 0.0,
+      "learning_rate": 5.566185560248649e-07,
+      "loss": 0.9634,
+      "step": 24651
+    },
+    {
+      "epoch": 0.8965341673637124,
+      "grad_norm": 0.0,
+      "learning_rate": 5.562311328445869e-07,
+      "loss": 0.8518,
+      "step": 24652
+    },
+    {
+      "epoch": 0.896570534967451,
+      "grad_norm": 0.0,
+      "learning_rate": 5.558438406832056e-07,
+      "loss": 1.0281,
+      "step": 24653
+    },
+    {
+      "epoch": 0.8966069025711896,
+      "grad_norm": 0.0,
+      "learning_rate": 5.554566795460903e-07,
+      "loss": 0.9126,
+      "step": 24654
+    },
+    {
+      "epoch": 0.8966432701749282,
+      "grad_norm": 0.0,
+      "learning_rate": 5.550696494386143e-07,
+      "loss": 0.7799,
+      "step": 24655
+    },
+    {
+      "epoch": 0.8966796377786668,
+      "grad_norm": 0.0,
+      "learning_rate": 5.546827503661445e-07,
+      "loss": 0.9303,
+      "step": 24656
+    },
+    {
+      "epoch": 0.8967160053824054,
+      "grad_norm": 0.0,
+      "learning_rate": 5.542959823340532e-07,
+      "loss": 1.026,
+      "step": 24657
+    },
+    {
+      "epoch": 0.896752372986144,
+      "grad_norm": 0.0,
+      "learning_rate": 5.539093453477007e-07,
+      "loss": 0.8514,
+      "step": 24658
+    },
+    {
+      "epoch": 0.8967887405898826,
+      "grad_norm": 0.0,
+      "learning_rate": 5.535228394124558e-07,
+      "loss": 0.8469,
+      "step": 24659
+    },
+    {
+      "epoch": 0.8968251081936212,
+      "grad_norm": 0.0,
+      "learning_rate": 5.531364645336778e-07,
+      "loss": 0.9236,
+      "step": 24660
+    },
+    {
+      "epoch": 0.8968614757973598,
+      "grad_norm": 0.0,
+      "learning_rate": 5.52750220716729e-07,
+      "loss": 0.9376,
+      "step": 24661
+    },
+    {
+      "epoch": 0.8968978434010983,
+      "grad_norm": 0.0,
+      "learning_rate": 5.523641079669662e-07,
+      "loss": 0.8512,
+      "step": 24662
+    },
+    {
+      "epoch": 0.896934211004837,
+      "grad_norm": 0.0,
+      "learning_rate": 5.51978126289745e-07,
+      "loss": 0.9368,
+      "step": 24663
+    },
+    {
+      "epoch": 0.8969705786085754,
+      "grad_norm": 0.0,
+      "learning_rate": 5.515922756904268e-07,
+      "loss": 0.8773,
+      "step": 24664
+    },
+    {
+      "epoch": 0.897006946212314,
+      "grad_norm": 0.0,
+      "learning_rate": 5.512065561743563e-07,
+      "loss": 0.8919,
+      "step": 24665
+    },
+    {
+      "epoch": 0.8970433138160526,
+      "grad_norm": 0.0,
+      "learning_rate": 5.508209677468901e-07,
+      "loss": 0.7369,
+      "step": 24666
+    },
+    {
+      "epoch": 0.8970796814197912,
+      "grad_norm": 0.0,
+      "learning_rate": 5.504355104133762e-07,
+      "loss": 0.83,
+      "step": 24667
+    },
+    {
+      "epoch": 0.8971160490235298,
+      "grad_norm": 0.0,
+      "learning_rate": 5.500501841791628e-07,
+      "loss": 0.9963,
+      "step": 24668
+    },
+    {
+      "epoch": 0.8971524166272684,
+      "grad_norm": 0.0,
+      "learning_rate": 5.496649890495942e-07,
+      "loss": 0.761,
+      "step": 24669
+    },
+    {
+      "epoch": 0.897188784231007,
+      "grad_norm": 0.0,
+      "learning_rate": 5.492799250300185e-07,
+      "loss": 0.8827,
+      "step": 24670
+    },
+    {
+      "epoch": 0.8972251518347456,
+      "grad_norm": 0.0,
+      "learning_rate": 5.488949921257714e-07,
+      "loss": 0.9096,
+      "step": 24671
+    },
+    {
+      "epoch": 0.8972615194384842,
+      "grad_norm": 0.0,
+      "learning_rate": 5.485101903421986e-07,
+      "loss": 0.8994,
+      "step": 24672
+    },
+    {
+      "epoch": 0.8972978870422228,
+      "grad_norm": 0.0,
+      "learning_rate": 5.48125519684638e-07,
+      "loss": 1.0218,
+      "step": 24673
+    },
+    {
+      "epoch": 0.8973342546459614,
+      "grad_norm": 0.0,
+      "learning_rate": 5.477409801584243e-07,
+      "loss": 0.8338,
+      "step": 24674
+    },
+    {
+      "epoch": 0.8973706222497,
+      "grad_norm": 0.0,
+      "learning_rate": 5.473565717688922e-07,
+      "loss": 0.8656,
+      "step": 24675
+    },
+    {
+      "epoch": 0.8974069898534386,
+      "grad_norm": 0.0,
+      "learning_rate": 5.469722945213774e-07,
+      "loss": 0.8701,
+      "step": 24676
+    },
+    {
+      "epoch": 0.8974433574571772,
+      "grad_norm": 0.0,
+      "learning_rate": 5.46588148421211e-07,
+      "loss": 1.0312,
+      "step": 24677
+    },
+    {
+      "epoch": 0.8974797250609158,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4620413347372e-07,
+      "loss": 0.9502,
+      "step": 24678
+    },
+    {
+      "epoch": 0.8975160926646544,
+      "grad_norm": 0.0,
+      "learning_rate": 5.458202496842346e-07,
+      "loss": 0.9445,
+      "step": 24679
+    },
+    {
+      "epoch": 0.897552460268393,
+      "grad_norm": 0.0,
+      "learning_rate": 5.454364970580783e-07,
+      "loss": 0.8616,
+      "step": 24680
+    },
+    {
+      "epoch": 0.8975888278721315,
+      "grad_norm": 0.0,
+      "learning_rate": 5.450528756005779e-07,
+      "loss": 0.8584,
+      "step": 24681
+    },
+    {
+      "epoch": 0.89762519547587,
+      "grad_norm": 0.0,
+      "learning_rate": 5.446693853170515e-07,
+      "loss": 0.8356,
+      "step": 24682
+    },
+    {
+      "epoch": 0.8976615630796086,
+      "grad_norm": 0.0,
+      "learning_rate": 5.442860262128247e-07,
+      "loss": 1.054,
+      "step": 24683
+    },
+    {
+      "epoch": 0.8976979306833472,
+      "grad_norm": 0.0,
+      "learning_rate": 5.43902798293211e-07,
+      "loss": 0.9418,
+      "step": 24684
+    },
+    {
+      "epoch": 0.8977342982870858,
+      "grad_norm": 0.0,
+      "learning_rate": 5.435197015635308e-07,
+      "loss": 0.9139,
+      "step": 24685
+    },
+    {
+      "epoch": 0.8977706658908244,
+      "grad_norm": 0.0,
+      "learning_rate": 5.431367360290984e-07,
+      "loss": 0.9332,
+      "step": 24686
+    },
+    {
+      "epoch": 0.897807033494563,
+      "grad_norm": 0.0,
+      "learning_rate": 5.427539016952255e-07,
+      "loss": 0.9232,
+      "step": 24687
+    },
+    {
+      "epoch": 0.8978434010983016,
+      "grad_norm": 0.0,
+      "learning_rate": 5.423711985672231e-07,
+      "loss": 0.8745,
+      "step": 24688
+    },
+    {
+      "epoch": 0.8978797687020402,
+      "grad_norm": 0.0,
+      "learning_rate": 5.419886266504026e-07,
+      "loss": 0.8677,
+      "step": 24689
+    },
+    {
+      "epoch": 0.8979161363057788,
+      "grad_norm": 0.0,
+      "learning_rate": 5.416061859500709e-07,
+      "loss": 0.9308,
+      "step": 24690
+    },
+    {
+      "epoch": 0.8979525039095174,
+      "grad_norm": 0.0,
+      "learning_rate": 5.412238764715338e-07,
+      "loss": 1.0378,
+      "step": 24691
+    },
+    {
+      "epoch": 0.897988871513256,
+      "grad_norm": 0.0,
+      "learning_rate": 5.408416982200959e-07,
+      "loss": 0.909,
+      "step": 24692
+    },
+    {
+      "epoch": 0.8980252391169946,
+      "grad_norm": 0.0,
+      "learning_rate": 5.404596512010574e-07,
+      "loss": 0.9491,
+      "step": 24693
+    },
+    {
+      "epoch": 0.8980616067207332,
+      "grad_norm": 0.0,
+      "learning_rate": 5.400777354197195e-07,
+      "loss": 0.9149,
+      "step": 24694
+    },
+    {
+      "epoch": 0.8980979743244718,
+      "grad_norm": 0.0,
+      "learning_rate": 5.396959508813815e-07,
+      "loss": 0.8524,
+      "step": 24695
+    },
+    {
+      "epoch": 0.8981343419282104,
+      "grad_norm": 0.0,
+      "learning_rate": 5.393142975913401e-07,
+      "loss": 0.9585,
+      "step": 24696
+    },
+    {
+      "epoch": 0.8981707095319489,
+      "grad_norm": 0.0,
+      "learning_rate": 5.389327755548901e-07,
+      "loss": 0.9043,
+      "step": 24697
+    },
+    {
+      "epoch": 0.8982070771356875,
+      "grad_norm": 0.0,
+      "learning_rate": 5.385513847773238e-07,
+      "loss": 0.789,
+      "step": 24698
+    },
+    {
+      "epoch": 0.8982434447394261,
+      "grad_norm": 0.0,
+      "learning_rate": 5.381701252639349e-07,
+      "loss": 0.914,
+      "step": 24699
+    },
+    {
+      "epoch": 0.8982798123431647,
+      "grad_norm": 0.0,
+      "learning_rate": 5.37788997020009e-07,
+      "loss": 0.8893,
+      "step": 24700
+    },
+    {
+      "epoch": 0.8983161799469033,
+      "grad_norm": 0.0,
+      "learning_rate": 5.374080000508353e-07,
+      "loss": 0.9464,
+      "step": 24701
+    },
+    {
+      "epoch": 0.8983525475506419,
+      "grad_norm": 0.0,
+      "learning_rate": 5.370271343617018e-07,
+      "loss": 0.8581,
+      "step": 24702
+    },
+    {
+      "epoch": 0.8983889151543805,
+      "grad_norm": 0.0,
+      "learning_rate": 5.366463999578885e-07,
+      "loss": 0.9027,
+      "step": 24703
+    },
+    {
+      "epoch": 0.8984252827581191,
+      "grad_norm": 0.0,
+      "learning_rate": 5.362657968446805e-07,
+      "loss": 0.8769,
+      "step": 24704
+    },
+    {
+      "epoch": 0.8984616503618577,
+      "grad_norm": 0.0,
+      "learning_rate": 5.358853250273577e-07,
+      "loss": 0.9738,
+      "step": 24705
+    },
+    {
+      "epoch": 0.8984980179655963,
+      "grad_norm": 0.0,
+      "learning_rate": 5.355049845111981e-07,
+      "loss": 0.9438,
+      "step": 24706
+    },
+    {
+      "epoch": 0.8985343855693348,
+      "grad_norm": 0.0,
+      "learning_rate": 5.351247753014777e-07,
+      "loss": 0.8657,
+      "step": 24707
+    },
+    {
+      "epoch": 0.8985707531730734,
+      "grad_norm": 0.0,
+      "learning_rate": 5.347446974034731e-07,
+      "loss": 0.8663,
+      "step": 24708
+    },
+    {
+      "epoch": 0.898607120776812,
+      "grad_norm": 0.0,
+      "learning_rate": 5.343647508224559e-07,
+      "loss": 1.0501,
+      "step": 24709
+    },
+    {
+      "epoch": 0.8986434883805506,
+      "grad_norm": 0.0,
+      "learning_rate": 5.339849355636984e-07,
+      "loss": 1.031,
+      "step": 24710
+    },
+    {
+      "epoch": 0.8986798559842892,
+      "grad_norm": 0.0,
+      "learning_rate": 5.336052516324675e-07,
+      "loss": 0.8771,
+      "step": 24711
+    },
+    {
+      "epoch": 0.8987162235880278,
+      "grad_norm": 0.0,
+      "learning_rate": 5.332256990340357e-07,
+      "loss": 0.8691,
+      "step": 24712
+    },
+    {
+      "epoch": 0.8987525911917664,
+      "grad_norm": 0.0,
+      "learning_rate": 5.328462777736631e-07,
+      "loss": 0.9111,
+      "step": 24713
+    },
+    {
+      "epoch": 0.8987889587955049,
+      "grad_norm": 0.0,
+      "learning_rate": 5.324669878566169e-07,
+      "loss": 0.9789,
+      "step": 24714
+    },
+    {
+      "epoch": 0.8988253263992435,
+      "grad_norm": 0.0,
+      "learning_rate": 5.320878292881593e-07,
+      "loss": 1.062,
+      "step": 24715
+    },
+    {
+      "epoch": 0.8988616940029821,
+      "grad_norm": 0.0,
+      "learning_rate": 5.317088020735494e-07,
+      "loss": 0.9724,
+      "step": 24716
+    },
+    {
+      "epoch": 0.8988980616067207,
+      "grad_norm": 0.0,
+      "learning_rate": 5.313299062180455e-07,
+      "loss": 0.8834,
+      "step": 24717
+    },
+    {
+      "epoch": 0.8989344292104593,
+      "grad_norm": 0.0,
+      "learning_rate": 5.309511417269076e-07,
+      "loss": 0.9137,
+      "step": 24718
+    },
+    {
+      "epoch": 0.8989707968141979,
+      "grad_norm": 0.0,
+      "learning_rate": 5.305725086053859e-07,
+      "loss": 0.9718,
+      "step": 24719
+    },
+    {
+      "epoch": 0.8990071644179365,
+      "grad_norm": 0.0,
+      "learning_rate": 5.301940068587342e-07,
+      "loss": 0.7549,
+      "step": 24720
+    },
+    {
+      "epoch": 0.8990435320216751,
+      "grad_norm": 0.0,
+      "learning_rate": 5.298156364922058e-07,
+      "loss": 0.8574,
+      "step": 24721
+    },
+    {
+      "epoch": 0.8990798996254137,
+      "grad_norm": 0.0,
+      "learning_rate": 5.29437397511049e-07,
+      "loss": 0.8335,
+      "step": 24722
+    },
+    {
+      "epoch": 0.8991162672291523,
+      "grad_norm": 0.0,
+      "learning_rate": 5.290592899205115e-07,
+      "loss": 0.9124,
+      "step": 24723
+    },
+    {
+      "epoch": 0.8991526348328909,
+      "grad_norm": 0.0,
+      "learning_rate": 5.286813137258384e-07,
+      "loss": 0.9264,
+      "step": 24724
+    },
+    {
+      "epoch": 0.8991890024366295,
+      "grad_norm": 0.0,
+      "learning_rate": 5.283034689322752e-07,
+      "loss": 0.8739,
+      "step": 24725
+    },
+    {
+      "epoch": 0.8992253700403681,
+      "grad_norm": 0.0,
+      "learning_rate": 5.279257555450601e-07,
+      "loss": 0.8799,
+      "step": 24726
+    },
+    {
+      "epoch": 0.8992617376441067,
+      "grad_norm": 0.0,
+      "learning_rate": 5.275481735694377e-07,
+      "loss": 0.9515,
+      "step": 24727
+    },
+    {
+      "epoch": 0.8992981052478453,
+      "grad_norm": 0.0,
+      "learning_rate": 5.271707230106459e-07,
+      "loss": 0.975,
+      "step": 24728
+    },
+    {
+      "epoch": 0.8993344728515839,
+      "grad_norm": 0.0,
+      "learning_rate": 5.267934038739186e-07,
+      "loss": 0.9219,
+      "step": 24729
+    },
+    {
+      "epoch": 0.8993708404553225,
+      "grad_norm": 0.0,
+      "learning_rate": 5.264162161644915e-07,
+      "loss": 0.9339,
+      "step": 24730
+    },
+    {
+      "epoch": 0.8994072080590609,
+      "grad_norm": 0.0,
+      "learning_rate": 5.260391598876014e-07,
+      "loss": 0.9021,
+      "step": 24731
+    },
+    {
+      "epoch": 0.8994435756627995,
+      "grad_norm": 0.0,
+      "learning_rate": 5.256622350484719e-07,
+      "loss": 0.8822,
+      "step": 24732
+    },
+    {
+      "epoch": 0.8994799432665381,
+      "grad_norm": 0.0,
+      "learning_rate": 5.252854416523389e-07,
+      "loss": 0.9665,
+      "step": 24733
+    },
+    {
+      "epoch": 0.8995163108702767,
+      "grad_norm": 0.0,
+      "learning_rate": 5.249087797044272e-07,
+      "loss": 0.9655,
+      "step": 24734
+    },
+    {
+      "epoch": 0.8995526784740153,
+      "grad_norm": 0.0,
+      "learning_rate": 5.245322492099636e-07,
+      "loss": 0.8431,
+      "step": 24735
+    },
+    {
+      "epoch": 0.8995890460777539,
+      "grad_norm": 0.0,
+      "learning_rate": 5.241558501741695e-07,
+      "loss": 0.9188,
+      "step": 24736
+    },
+    {
+      "epoch": 0.8996254136814925,
+      "grad_norm": 0.0,
+      "learning_rate": 5.237795826022707e-07,
+      "loss": 0.9476,
+      "step": 24737
+    },
+    {
+      "epoch": 0.8996617812852311,
+      "grad_norm": 0.0,
+      "learning_rate": 5.234034464994842e-07,
+      "loss": 0.9647,
+      "step": 24738
+    },
+    {
+      "epoch": 0.8996981488889697,
+      "grad_norm": 0.0,
+      "learning_rate": 5.230274418710279e-07,
+      "loss": 0.8274,
+      "step": 24739
+    },
+    {
+      "epoch": 0.8997345164927083,
+      "grad_norm": 0.0,
+      "learning_rate": 5.226515687221212e-07,
+      "loss": 0.9971,
+      "step": 24740
+    },
+    {
+      "epoch": 0.8997708840964469,
+      "grad_norm": 0.0,
+      "learning_rate": 5.222758270579764e-07,
+      "loss": 1.0081,
+      "step": 24741
+    },
+    {
+      "epoch": 0.8998072517001855,
+      "grad_norm": 0.0,
+      "learning_rate": 5.219002168838083e-07,
+      "loss": 0.8329,
+      "step": 24742
+    },
+    {
+      "epoch": 0.8998436193039241,
+      "grad_norm": 0.0,
+      "learning_rate": 5.215247382048261e-07,
+      "loss": 0.9089,
+      "step": 24743
+    },
+    {
+      "epoch": 0.8998799869076627,
+      "grad_norm": 0.0,
+      "learning_rate": 5.211493910262421e-07,
+      "loss": 0.8803,
+      "step": 24744
+    },
+    {
+      "epoch": 0.8999163545114013,
+      "grad_norm": 0.0,
+      "learning_rate": 5.207741753532591e-07,
+      "loss": 0.9708,
+      "step": 24745
+    },
+    {
+      "epoch": 0.8999527221151399,
+      "grad_norm": 0.0,
+      "learning_rate": 5.203990911910861e-07,
+      "loss": 0.9251,
+      "step": 24746
+    },
+    {
+      "epoch": 0.8999890897188785,
+      "grad_norm": 0.0,
+      "learning_rate": 5.200241385449256e-07,
+      "loss": 0.96,
+      "step": 24747
+    },
+    {
+      "epoch": 0.900025457322617,
+      "grad_norm": 0.0,
+      "learning_rate": 5.196493174199813e-07,
+      "loss": 0.9033,
+      "step": 24748
+    },
+    {
+      "epoch": 0.9000618249263556,
+      "grad_norm": 0.0,
+      "learning_rate": 5.192746278214489e-07,
+      "loss": 0.8978,
+      "step": 24749
+    },
+    {
+      "epoch": 0.9000981925300942,
+      "grad_norm": 0.0,
+      "learning_rate": 5.189000697545332e-07,
+      "loss": 0.9409,
+      "step": 24750
+    },
+    {
+      "epoch": 0.9001345601338328,
+      "grad_norm": 0.0,
+      "learning_rate": 5.185256432244246e-07,
+      "loss": 0.8741,
+      "step": 24751
+    },
+    {
+      "epoch": 0.9001709277375713,
+      "grad_norm": 0.0,
+      "learning_rate": 5.18151348236321e-07,
+      "loss": 0.8806,
+      "step": 24752
+    },
+    {
+      "epoch": 0.9002072953413099,
+      "grad_norm": 0.0,
+      "learning_rate": 5.177771847954149e-07,
+      "loss": 0.9489,
+      "step": 24753
+    },
+    {
+      "epoch": 0.9002436629450485,
+      "grad_norm": 0.0,
+      "learning_rate": 5.174031529068968e-07,
+      "loss": 0.951,
+      "step": 24754
+    },
+    {
+      "epoch": 0.9002800305487871,
+      "grad_norm": 0.0,
+      "learning_rate": 5.170292525759546e-07,
+      "loss": 0.9051,
+      "step": 24755
+    },
+    {
+      "epoch": 0.9003163981525257,
+      "grad_norm": 0.0,
+      "learning_rate": 5.166554838077786e-07,
+      "loss": 0.8834,
+      "step": 24756
+    },
+    {
+      "epoch": 0.9003527657562643,
+      "grad_norm": 0.0,
+      "learning_rate": 5.162818466075525e-07,
+      "loss": 0.8615,
+      "step": 24757
+    },
+    {
+      "epoch": 0.9003891333600029,
+      "grad_norm": 0.0,
+      "learning_rate": 5.159083409804588e-07,
+      "loss": 0.8904,
+      "step": 24758
+    },
+    {
+      "epoch": 0.9004255009637415,
+      "grad_norm": 0.0,
+      "learning_rate": 5.155349669316812e-07,
+      "loss": 0.8443,
+      "step": 24759
+    },
+    {
+      "epoch": 0.9004618685674801,
+      "grad_norm": 0.0,
+      "learning_rate": 5.151617244663998e-07,
+      "loss": 0.944,
+      "step": 24760
+    },
+    {
+      "epoch": 0.9004982361712187,
+      "grad_norm": 0.0,
+      "learning_rate": 5.147886135897929e-07,
+      "loss": 1.0034,
+      "step": 24761
+    },
+    {
+      "epoch": 0.9005346037749573,
+      "grad_norm": 0.0,
+      "learning_rate": 5.144156343070339e-07,
+      "loss": 0.8731,
+      "step": 24762
+    },
+    {
+      "epoch": 0.9005709713786959,
+      "grad_norm": 0.0,
+      "learning_rate": 5.140427866233033e-07,
+      "loss": 0.9627,
+      "step": 24763
+    },
+    {
+      "epoch": 0.9006073389824345,
+      "grad_norm": 0.0,
+      "learning_rate": 5.136700705437658e-07,
+      "loss": 0.8822,
+      "step": 24764
+    },
+    {
+      "epoch": 0.900643706586173,
+      "grad_norm": 0.0,
+      "learning_rate": 5.132974860735995e-07,
+      "loss": 0.9014,
+      "step": 24765
+    },
+    {
+      "epoch": 0.9006800741899116,
+      "grad_norm": 0.0,
+      "learning_rate": 5.129250332179702e-07,
+      "loss": 0.8927,
+      "step": 24766
+    },
+    {
+      "epoch": 0.9007164417936502,
+      "grad_norm": 0.0,
+      "learning_rate": 5.125527119820461e-07,
+      "loss": 0.9547,
+      "step": 24767
+    },
+    {
+      "epoch": 0.9007528093973888,
+      "grad_norm": 0.0,
+      "learning_rate": 5.121805223709908e-07,
+      "loss": 0.9262,
+      "step": 24768
+    },
+    {
+      "epoch": 0.9007891770011274,
+      "grad_norm": 0.0,
+      "learning_rate": 5.118084643899723e-07,
+      "loss": 0.9583,
+      "step": 24769
+    },
+    {
+      "epoch": 0.900825544604866,
+      "grad_norm": 0.0,
+      "learning_rate": 5.114365380441455e-07,
+      "loss": 0.9027,
+      "step": 24770
+    },
+    {
+      "epoch": 0.9008619122086046,
+      "grad_norm": 0.0,
+      "learning_rate": 5.110647433386761e-07,
+      "loss": 0.9385,
+      "step": 24771
+    },
+    {
+      "epoch": 0.9008982798123432,
+      "grad_norm": 0.0,
+      "learning_rate": 5.106930802787213e-07,
+      "loss": 0.9293,
+      "step": 24772
+    },
+    {
+      "epoch": 0.9009346474160818,
+      "grad_norm": 0.0,
+      "learning_rate": 5.103215488694357e-07,
+      "loss": 0.846,
+      "step": 24773
+    },
+    {
+      "epoch": 0.9009710150198204,
+      "grad_norm": 0.0,
+      "learning_rate": 5.099501491159731e-07,
+      "loss": 0.9698,
+      "step": 24774
+    },
+    {
+      "epoch": 0.901007382623559,
+      "grad_norm": 0.0,
+      "learning_rate": 5.095788810234881e-07,
+      "loss": 0.8981,
+      "step": 24775
+    },
+    {
+      "epoch": 0.9010437502272975,
+      "grad_norm": 0.0,
+      "learning_rate": 5.092077445971321e-07,
+      "loss": 0.88,
+      "step": 24776
+    },
+    {
+      "epoch": 0.9010801178310361,
+      "grad_norm": 0.0,
+      "learning_rate": 5.088367398420524e-07,
+      "loss": 0.9207,
+      "step": 24777
+    },
+    {
+      "epoch": 0.9011164854347747,
+      "grad_norm": 0.0,
+      "learning_rate": 5.084658667633968e-07,
+      "loss": 0.7822,
+      "step": 24778
+    },
+    {
+      "epoch": 0.9011528530385133,
+      "grad_norm": 0.0,
+      "learning_rate": 5.080951253663113e-07,
+      "loss": 0.9847,
+      "step": 24779
+    },
+    {
+      "epoch": 0.9011892206422519,
+      "grad_norm": 0.0,
+      "learning_rate": 5.077245156559385e-07,
+      "loss": 0.8162,
+      "step": 24780
+    },
+    {
+      "epoch": 0.9012255882459904,
+      "grad_norm": 0.0,
+      "learning_rate": 5.073540376374187e-07,
+      "loss": 0.8608,
+      "step": 24781
+    },
+    {
+      "epoch": 0.901261955849729,
+      "grad_norm": 0.0,
+      "learning_rate": 5.069836913158954e-07,
+      "loss": 1.0488,
+      "step": 24782
+    },
+    {
+      "epoch": 0.9012983234534676,
+      "grad_norm": 0.0,
+      "learning_rate": 5.066134766965025e-07,
+      "loss": 0.8767,
+      "step": 24783
+    },
+    {
+      "epoch": 0.9013346910572062,
+      "grad_norm": 0.0,
+      "learning_rate": 5.062433937843792e-07,
+      "loss": 0.7858,
+      "step": 24784
+    },
+    {
+      "epoch": 0.9013710586609448,
+      "grad_norm": 0.0,
+      "learning_rate": 5.05873442584659e-07,
+      "loss": 0.9684,
+      "step": 24785
+    },
+    {
+      "epoch": 0.9014074262646834,
+      "grad_norm": 0.0,
+      "learning_rate": 5.055036231024746e-07,
+      "loss": 0.9429,
+      "step": 24786
+    },
+    {
+      "epoch": 0.901443793868422,
+      "grad_norm": 0.0,
+      "learning_rate": 5.051339353429552e-07,
+      "loss": 0.9545,
+      "step": 24787
+    },
+    {
+      "epoch": 0.9014801614721606,
+      "grad_norm": 0.0,
+      "learning_rate": 5.047643793112323e-07,
+      "loss": 0.9993,
+      "step": 24788
+    },
+    {
+      "epoch": 0.9015165290758992,
+      "grad_norm": 0.0,
+      "learning_rate": 5.043949550124316e-07,
+      "loss": 0.8972,
+      "step": 24789
+    },
+    {
+      "epoch": 0.9015528966796378,
+      "grad_norm": 0.0,
+      "learning_rate": 5.040256624516792e-07,
+      "loss": 0.958,
+      "step": 24790
+    },
+    {
+      "epoch": 0.9015892642833764,
+      "grad_norm": 0.0,
+      "learning_rate": 5.036565016340966e-07,
+      "loss": 0.9704,
+      "step": 24791
+    },
+    {
+      "epoch": 0.901625631887115,
+      "grad_norm": 0.0,
+      "learning_rate": 5.032874725648074e-07,
+      "loss": 0.8975,
+      "step": 24792
+    },
+    {
+      "epoch": 0.9016619994908536,
+      "grad_norm": 0.0,
+      "learning_rate": 5.029185752489285e-07,
+      "loss": 0.8888,
+      "step": 24793
+    },
+    {
+      "epoch": 0.9016983670945922,
+      "grad_norm": 0.0,
+      "learning_rate": 5.025498096915804e-07,
+      "loss": 0.9228,
+      "step": 24794
+    },
+    {
+      "epoch": 0.9017347346983308,
+      "grad_norm": 0.0,
+      "learning_rate": 5.021811758978801e-07,
+      "loss": 0.9658,
+      "step": 24795
+    },
+    {
+      "epoch": 0.9017711023020694,
+      "grad_norm": 0.0,
+      "learning_rate": 5.018126738729389e-07,
+      "loss": 0.856,
+      "step": 24796
+    },
+    {
+      "epoch": 0.901807469905808,
+      "grad_norm": 0.0,
+      "learning_rate": 5.014443036218719e-07,
+      "loss": 0.8628,
+      "step": 24797
+    },
+    {
+      "epoch": 0.9018438375095464,
+      "grad_norm": 0.0,
+      "learning_rate": 5.01076065149787e-07,
+      "loss": 0.8864,
+      "step": 24798
+    },
+    {
+      "epoch": 0.901880205113285,
+      "grad_norm": 0.0,
+      "learning_rate": 5.007079584617958e-07,
+      "loss": 0.8472,
+      "step": 24799
+    },
+    {
+      "epoch": 0.9019165727170236,
+      "grad_norm": 0.0,
+      "learning_rate": 5.00339983563003e-07,
+      "loss": 0.9731,
+      "step": 24800
+    },
+    {
+      "epoch": 0.9019529403207622,
+      "grad_norm": 0.0,
+      "learning_rate": 4.999721404585156e-07,
+      "loss": 0.9478,
+      "step": 24801
+    },
+    {
+      "epoch": 0.9019893079245008,
+      "grad_norm": 0.0,
+      "learning_rate": 4.996044291534352e-07,
+      "loss": 0.885,
+      "step": 24802
+    },
+    {
+      "epoch": 0.9020256755282394,
+      "grad_norm": 0.0,
+      "learning_rate": 4.992368496528644e-07,
+      "loss": 0.9123,
+      "step": 24803
+    },
+    {
+      "epoch": 0.902062043131978,
+      "grad_norm": 0.0,
+      "learning_rate": 4.988694019619022e-07,
+      "loss": 0.9008,
+      "step": 24804
+    },
+    {
+      "epoch": 0.9020984107357166,
+      "grad_norm": 0.0,
+      "learning_rate": 4.985020860856471e-07,
+      "loss": 0.9215,
+      "step": 24805
+    },
+    {
+      "epoch": 0.9021347783394552,
+      "grad_norm": 0.0,
+      "learning_rate": 4.981349020291936e-07,
+      "loss": 0.9257,
+      "step": 24806
+    },
+    {
+      "epoch": 0.9021711459431938,
+      "grad_norm": 0.0,
+      "learning_rate": 4.977678497976379e-07,
+      "loss": 0.8692,
+      "step": 24807
+    },
+    {
+      "epoch": 0.9022075135469324,
+      "grad_norm": 0.0,
+      "learning_rate": 4.974009293960713e-07,
+      "loss": 0.8996,
+      "step": 24808
+    },
+    {
+      "epoch": 0.902243881150671,
+      "grad_norm": 0.0,
+      "learning_rate": 4.970341408295842e-07,
+      "loss": 0.7757,
+      "step": 24809
+    },
+    {
+      "epoch": 0.9022802487544096,
+      "grad_norm": 0.0,
+      "learning_rate": 4.966674841032648e-07,
+      "loss": 0.9057,
+      "step": 24810
+    },
+    {
+      "epoch": 0.9023166163581482,
+      "grad_norm": 0.0,
+      "learning_rate": 4.963009592222034e-07,
+      "loss": 1.0,
+      "step": 24811
+    },
+    {
+      "epoch": 0.9023529839618868,
+      "grad_norm": 0.0,
+      "learning_rate": 4.959345661914783e-07,
+      "loss": 0.951,
+      "step": 24812
+    },
+    {
+      "epoch": 0.9023893515656254,
+      "grad_norm": 0.0,
+      "learning_rate": 4.955683050161786e-07,
+      "loss": 0.9548,
+      "step": 24813
+    },
+    {
+      "epoch": 0.902425719169364,
+      "grad_norm": 0.0,
+      "learning_rate": 4.952021757013848e-07,
+      "loss": 1.1115,
+      "step": 24814
+    },
+    {
+      "epoch": 0.9024620867731025,
+      "grad_norm": 0.0,
+      "learning_rate": 4.948361782521737e-07,
+      "loss": 1.0707,
+      "step": 24815
+    },
+    {
+      "epoch": 0.9024984543768411,
+      "grad_norm": 0.0,
+      "learning_rate": 4.94470312673625e-07,
+      "loss": 0.7832,
+      "step": 24816
+    },
+    {
+      "epoch": 0.9025348219805797,
+      "grad_norm": 0.0,
+      "learning_rate": 4.941045789708144e-07,
+      "loss": 0.7993,
+      "step": 24817
+    },
+    {
+      "epoch": 0.9025711895843183,
+      "grad_norm": 0.0,
+      "learning_rate": 4.937389771488155e-07,
+      "loss": 0.9308,
+      "step": 24818
+    },
+    {
+      "epoch": 0.9026075571880569,
+      "grad_norm": 0.0,
+      "learning_rate": 4.933735072126988e-07,
+      "loss": 0.8226,
+      "step": 24819
+    },
+    {
+      "epoch": 0.9026439247917954,
+      "grad_norm": 0.0,
+      "learning_rate": 4.930081691675393e-07,
+      "loss": 0.9399,
+      "step": 24820
+    },
+    {
+      "epoch": 0.902680292395534,
+      "grad_norm": 0.0,
+      "learning_rate": 4.926429630184015e-07,
+      "loss": 0.9598,
+      "step": 24821
+    },
+    {
+      "epoch": 0.9027166599992726,
+      "grad_norm": 0.0,
+      "learning_rate": 4.922778887703539e-07,
+      "loss": 0.9696,
+      "step": 24822
+    },
+    {
+      "epoch": 0.9027530276030112,
+      "grad_norm": 0.0,
+      "learning_rate": 4.919129464284611e-07,
+      "loss": 0.9852,
+      "step": 24823
+    },
+    {
+      "epoch": 0.9027893952067498,
+      "grad_norm": 0.0,
+      "learning_rate": 4.915481359977847e-07,
+      "loss": 0.8222,
+      "step": 24824
+    },
+    {
+      "epoch": 0.9028257628104884,
+      "grad_norm": 0.0,
+      "learning_rate": 4.911834574833874e-07,
+      "loss": 0.8889,
+      "step": 24825
+    },
+    {
+      "epoch": 0.902862130414227,
+      "grad_norm": 0.0,
+      "learning_rate": 4.908189108903294e-07,
+      "loss": 0.8606,
+      "step": 24826
+    },
+    {
+      "epoch": 0.9028984980179656,
+      "grad_norm": 0.0,
+      "learning_rate": 4.904544962236668e-07,
+      "loss": 0.9823,
+      "step": 24827
+    },
+    {
+      "epoch": 0.9029348656217042,
+      "grad_norm": 0.0,
+      "learning_rate": 4.900902134884556e-07,
+      "loss": 0.9128,
+      "step": 24828
+    },
+    {
+      "epoch": 0.9029712332254428,
+      "grad_norm": 0.0,
+      "learning_rate": 4.897260626897493e-07,
+      "loss": 0.8876,
+      "step": 24829
+    },
+    {
+      "epoch": 0.9030076008291814,
+      "grad_norm": 0.0,
+      "learning_rate": 4.89362043832603e-07,
+      "loss": 0.9542,
+      "step": 24830
+    },
+    {
+      "epoch": 0.90304396843292,
+      "grad_norm": 0.0,
+      "learning_rate": 4.889981569220614e-07,
+      "loss": 0.8868,
+      "step": 24831
+    },
+    {
+      "epoch": 0.9030803360366585,
+      "grad_norm": 0.0,
+      "learning_rate": 4.886344019631773e-07,
+      "loss": 0.7698,
+      "step": 24832
+    },
+    {
+      "epoch": 0.9031167036403971,
+      "grad_norm": 0.0,
+      "learning_rate": 4.882707789609964e-07,
+      "loss": 0.85,
+      "step": 24833
+    },
+    {
+      "epoch": 0.9031530712441357,
+      "grad_norm": 0.0,
+      "learning_rate": 4.879072879205638e-07,
+      "loss": 0.9454,
+      "step": 24834
+    },
+    {
+      "epoch": 0.9031894388478743,
+      "grad_norm": 0.0,
+      "learning_rate": 4.875439288469208e-07,
+      "loss": 0.8431,
+      "step": 24835
+    },
+    {
+      "epoch": 0.9032258064516129,
+      "grad_norm": 0.0,
+      "learning_rate": 4.871807017451091e-07,
+      "loss": 0.9653,
+      "step": 24836
+    },
+    {
+      "epoch": 0.9032621740553515,
+      "grad_norm": 0.0,
+      "learning_rate": 4.868176066201691e-07,
+      "loss": 0.92,
+      "step": 24837
+    },
+    {
+      "epoch": 0.9032985416590901,
+      "grad_norm": 0.0,
+      "learning_rate": 4.864546434771356e-07,
+      "loss": 0.9249,
+      "step": 24838
+    },
+    {
+      "epoch": 0.9033349092628287,
+      "grad_norm": 0.0,
+      "learning_rate": 4.860918123210467e-07,
+      "loss": 0.8589,
+      "step": 24839
+    },
+    {
+      "epoch": 0.9033712768665673,
+      "grad_norm": 0.0,
+      "learning_rate": 4.857291131569364e-07,
+      "loss": 0.8268,
+      "step": 24840
+    },
+    {
+      "epoch": 0.9034076444703059,
+      "grad_norm": 0.0,
+      "learning_rate": 4.853665459898349e-07,
+      "loss": 0.8658,
+      "step": 24841
+    },
+    {
+      "epoch": 0.9034440120740445,
+      "grad_norm": 0.0,
+      "learning_rate": 4.850041108247727e-07,
+      "loss": 0.7537,
+      "step": 24842
+    },
+    {
+      "epoch": 0.903480379677783,
+      "grad_norm": 0.0,
+      "learning_rate": 4.846418076667802e-07,
+      "loss": 0.9947,
+      "step": 24843
+    },
+    {
+      "epoch": 0.9035167472815216,
+      "grad_norm": 0.0,
+      "learning_rate": 4.842796365208791e-07,
+      "loss": 0.9175,
+      "step": 24844
+    },
+    {
+      "epoch": 0.9035531148852602,
+      "grad_norm": 0.0,
+      "learning_rate": 4.839175973920995e-07,
+      "loss": 0.9595,
+      "step": 24845
+    },
+    {
+      "epoch": 0.9035894824889988,
+      "grad_norm": 0.0,
+      "learning_rate": 4.83555690285461e-07,
+      "loss": 0.9107,
+      "step": 24846
+    },
+    {
+      "epoch": 0.9036258500927374,
+      "grad_norm": 0.0,
+      "learning_rate": 4.831939152059862e-07,
+      "loss": 0.9499,
+      "step": 24847
+    },
+    {
+      "epoch": 0.903662217696476,
+      "grad_norm": 0.0,
+      "learning_rate": 4.828322721586909e-07,
+      "loss": 0.9551,
+      "step": 24848
+    },
+    {
+      "epoch": 0.9036985853002145,
+      "grad_norm": 0.0,
+      "learning_rate": 4.824707611485969e-07,
+      "loss": 0.9906,
+      "step": 24849
+    },
+    {
+      "epoch": 0.9037349529039531,
+      "grad_norm": 0.0,
+      "learning_rate": 4.821093821807155e-07,
+      "loss": 0.8264,
+      "step": 24850
+    },
+    {
+      "epoch": 0.9037713205076917,
+      "grad_norm": 0.0,
+      "learning_rate": 4.817481352600639e-07,
+      "loss": 0.9361,
+      "step": 24851
+    },
+    {
+      "epoch": 0.9038076881114303,
+      "grad_norm": 0.0,
+      "learning_rate": 4.813870203916527e-07,
+      "loss": 0.872,
+      "step": 24852
+    },
+    {
+      "epoch": 0.9038440557151689,
+      "grad_norm": 0.0,
+      "learning_rate": 4.810260375804898e-07,
+      "loss": 0.9007,
+      "step": 24853
+    },
+    {
+      "epoch": 0.9038804233189075,
+      "grad_norm": 0.0,
+      "learning_rate": 4.80665186831586e-07,
+      "loss": 0.9766,
+      "step": 24854
+    },
+    {
+      "epoch": 0.9039167909226461,
+      "grad_norm": 0.0,
+      "learning_rate": 4.803044681499458e-07,
+      "loss": 0.9689,
+      "step": 24855
+    },
+    {
+      "epoch": 0.9039531585263847,
+      "grad_norm": 0.0,
+      "learning_rate": 4.799438815405766e-07,
+      "loss": 0.8745,
+      "step": 24856
+    },
+    {
+      "epoch": 0.9039895261301233,
+      "grad_norm": 0.0,
+      "learning_rate": 4.795834270084754e-07,
+      "loss": 0.9596,
+      "step": 24857
+    },
+    {
+      "epoch": 0.9040258937338619,
+      "grad_norm": 0.0,
+      "learning_rate": 4.792231045586493e-07,
+      "loss": 0.7741,
+      "step": 24858
+    },
+    {
+      "epoch": 0.9040622613376005,
+      "grad_norm": 0.0,
+      "learning_rate": 4.788629141960932e-07,
+      "loss": 1.0374,
+      "step": 24859
+    },
+    {
+      "epoch": 0.9040986289413391,
+      "grad_norm": 0.0,
+      "learning_rate": 4.785028559258054e-07,
+      "loss": 0.9774,
+      "step": 24860
+    },
+    {
+      "epoch": 0.9041349965450777,
+      "grad_norm": 0.0,
+      "learning_rate": 4.781429297527806e-07,
+      "loss": 0.9458,
+      "step": 24861
+    },
+    {
+      "epoch": 0.9041713641488163,
+      "grad_norm": 0.0,
+      "learning_rate": 4.777831356820151e-07,
+      "loss": 0.8541,
+      "step": 24862
+    },
+    {
+      "epoch": 0.9042077317525549,
+      "grad_norm": 0.0,
+      "learning_rate": 4.774234737184957e-07,
+      "loss": 0.8694,
+      "step": 24863
+    },
+    {
+      "epoch": 0.9042440993562935,
+      "grad_norm": 0.0,
+      "learning_rate": 4.770639438672164e-07,
+      "loss": 0.9342,
+      "step": 24864
+    },
+    {
+      "epoch": 0.9042804669600321,
+      "grad_norm": 0.0,
+      "learning_rate": 4.767045461331632e-07,
+      "loss": 0.8954,
+      "step": 24865
+    },
+    {
+      "epoch": 0.9043168345637705,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7634528052132313e-07,
+      "loss": 0.8496,
+      "step": 24866
+    },
+    {
+      "epoch": 0.9043532021675091,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7598614703667887e-07,
+      "loss": 0.9578,
+      "step": 24867
+    },
+    {
+      "epoch": 0.9043895697712477,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7562714568421654e-07,
+      "loss": 0.8938,
+      "step": 24868
+    },
+    {
+      "epoch": 0.9044259373749863,
+      "grad_norm": 0.0,
+      "learning_rate": 4.752682764689109e-07,
+      "loss": 0.8516,
+      "step": 24869
+    },
+    {
+      "epoch": 0.9044623049787249,
+      "grad_norm": 0.0,
+      "learning_rate": 4.749095393957459e-07,
+      "loss": 0.8744,
+      "step": 24870
+    },
+    {
+      "epoch": 0.9044986725824635,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7455093446969636e-07,
+      "loss": 0.9071,
+      "step": 24871
+    },
+    {
+      "epoch": 0.9045350401862021,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7419246169573827e-07,
+      "loss": 0.9015,
+      "step": 24872
+    },
+    {
+      "epoch": 0.9045714077899407,
+      "grad_norm": 0.0,
+      "learning_rate": 4.738341210788444e-07,
+      "loss": 0.9533,
+      "step": 24873
+    },
+    {
+      "epoch": 0.9046077753936793,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7347591262398405e-07,
+      "loss": 0.9291,
+      "step": 24874
+    },
+    {
+      "epoch": 0.9046441429974179,
+      "grad_norm": 0.0,
+      "learning_rate": 4.731178363361333e-07,
+      "loss": 0.8897,
+      "step": 24875
+    },
+    {
+      "epoch": 0.9046805106011565,
+      "grad_norm": 0.0,
+      "learning_rate": 4.727598922202514e-07,
+      "loss": 0.9799,
+      "step": 24876
+    },
+    {
+      "epoch": 0.9047168782048951,
+      "grad_norm": 0.0,
+      "learning_rate": 4.724020802813112e-07,
+      "loss": 0.9456,
+      "step": 24877
+    },
+    {
+      "epoch": 0.9047532458086337,
+      "grad_norm": 0.0,
+      "learning_rate": 4.72044400524273e-07,
+      "loss": 0.8922,
+      "step": 24878
+    },
+    {
+      "epoch": 0.9047896134123723,
+      "grad_norm": 0.0,
+      "learning_rate": 4.716868529541008e-07,
+      "loss": 0.9393,
+      "step": 24879
+    },
+    {
+      "epoch": 0.9048259810161109,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7132943757575375e-07,
+      "loss": 0.9597,
+      "step": 24880
+    },
+    {
+      "epoch": 0.9048623486198495,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7097215439419364e-07,
+      "loss": 0.8964,
+      "step": 24881
+    },
+    {
+      "epoch": 0.904898716223588,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7061500341437306e-07,
+      "loss": 0.9137,
+      "step": 24882
+    },
+    {
+      "epoch": 0.9049350838273266,
+      "grad_norm": 0.0,
+      "learning_rate": 4.702579846412492e-07,
+      "loss": 0.8302,
+      "step": 24883
+    },
+    {
+      "epoch": 0.9049714514310652,
+      "grad_norm": 0.0,
+      "learning_rate": 4.699010980797747e-07,
+      "loss": 0.9792,
+      "step": 24884
+    },
+    {
+      "epoch": 0.9050078190348038,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6954434373490234e-07,
+      "loss": 0.9286,
+      "step": 24885
+    },
+    {
+      "epoch": 0.9050441866385424,
+      "grad_norm": 0.0,
+      "learning_rate": 4.691877216115781e-07,
+      "loss": 0.9731,
+      "step": 24886
+    },
+    {
+      "epoch": 0.905080554242281,
+      "grad_norm": 0.0,
+      "learning_rate": 4.688312317147525e-07,
+      "loss": 0.7979,
+      "step": 24887
+    },
+    {
+      "epoch": 0.9051169218460196,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6847487404937166e-07,
+      "loss": 0.8457,
+      "step": 24888
+    },
+    {
+      "epoch": 0.9051532894497581,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6811864862037816e-07,
+      "loss": 0.9155,
+      "step": 24889
+    },
+    {
+      "epoch": 0.9051896570534967,
+      "grad_norm": 0.0,
+      "learning_rate": 4.677625554327148e-07,
+      "loss": 0.9652,
+      "step": 24890
+    },
+    {
+      "epoch": 0.9052260246572353,
+      "grad_norm": 0.0,
+      "learning_rate": 4.674065944913209e-07,
+      "loss": 0.9637,
+      "step": 24891
+    },
+    {
+      "epoch": 0.9052623922609739,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6705076580113473e-07,
+      "loss": 0.8623,
+      "step": 24892
+    },
+    {
+      "epoch": 0.9052987598647125,
+      "grad_norm": 0.0,
+      "learning_rate": 4.666950693670935e-07,
+      "loss": 0.891,
+      "step": 24893
+    },
+    {
+      "epoch": 0.9053351274684511,
+      "grad_norm": 0.0,
+      "learning_rate": 4.663395051941344e-07,
+      "loss": 0.9496,
+      "step": 24894
+    },
+    {
+      "epoch": 0.9053714950721897,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6598407328718673e-07,
+      "loss": 0.8928,
+      "step": 24895
+    },
+    {
+      "epoch": 0.9054078626759283,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6562877365118333e-07,
+      "loss": 0.909,
+      "step": 24896
+    },
+    {
+      "epoch": 0.9054442302796669,
+      "grad_norm": 0.0,
+      "learning_rate": 4.652736062910534e-07,
+      "loss": 0.9054,
+      "step": 24897
+    },
+    {
+      "epoch": 0.9054805978834055,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6491857121172326e-07,
+      "loss": 0.8666,
+      "step": 24898
+    },
+    {
+      "epoch": 0.905516965487144,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6456366841811875e-07,
+      "loss": 0.8515,
+      "step": 24899
+    },
+    {
+      "epoch": 0.9055533330908826,
+      "grad_norm": 0.0,
+      "learning_rate": 4.642088979151671e-07,
+      "loss": 0.8071,
+      "step": 24900
+    },
+    {
+      "epoch": 0.9055897006946212,
+      "grad_norm": 0.0,
+      "learning_rate": 4.638542597077855e-07,
+      "loss": 0.8447,
+      "step": 24901
+    },
+    {
+      "epoch": 0.9056260682983598,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6349975380089675e-07,
+      "loss": 0.9093,
+      "step": 24902
+    },
+    {
+      "epoch": 0.9056624359020984,
+      "grad_norm": 0.0,
+      "learning_rate": 4.631453801994179e-07,
+      "loss": 0.8971,
+      "step": 24903
+    },
+    {
+      "epoch": 0.905698803505837,
+      "grad_norm": 0.0,
+      "learning_rate": 4.627911389082662e-07,
+      "loss": 0.9163,
+      "step": 24904
+    },
+    {
+      "epoch": 0.9057351711095756,
+      "grad_norm": 0.0,
+      "learning_rate": 4.624370299323544e-07,
+      "loss": 0.7498,
+      "step": 24905
+    },
+    {
+      "epoch": 0.9057715387133142,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6208305327659853e-07,
+      "loss": 0.9551,
+      "step": 24906
+    },
+    {
+      "epoch": 0.9058079063170528,
+      "grad_norm": 0.0,
+      "learning_rate": 4.61729208945908e-07,
+      "loss": 0.8161,
+      "step": 24907
+    },
+    {
+      "epoch": 0.9058442739207914,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6137549694519114e-07,
+      "loss": 0.8969,
+      "step": 24908
+    },
+    {
+      "epoch": 0.90588064152453,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6102191727935507e-07,
+      "loss": 0.8604,
+      "step": 24909
+    },
+    {
+      "epoch": 0.9059170091282686,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6066846995330817e-07,
+      "loss": 1.0045,
+      "step": 24910
+    },
+    {
+      "epoch": 0.9059533767320072,
+      "grad_norm": 0.0,
+      "learning_rate": 4.603151549719498e-07,
+      "loss": 0.9675,
+      "step": 24911
+    },
+    {
+      "epoch": 0.9059897443357458,
+      "grad_norm": 0.0,
+      "learning_rate": 4.599619723401827e-07,
+      "loss": 0.9027,
+      "step": 24912
+    },
+    {
+      "epoch": 0.9060261119394843,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5960892206290964e-07,
+      "loss": 0.9085,
+      "step": 24913
+    },
+    {
+      "epoch": 0.9060624795432229,
+      "grad_norm": 0.0,
+      "learning_rate": 4.592560041450256e-07,
+      "loss": 0.8699,
+      "step": 24914
+    },
+    {
+      "epoch": 0.9060988471469615,
+      "grad_norm": 0.0,
+      "learning_rate": 4.589032185914277e-07,
+      "loss": 0.9007,
+      "step": 24915
+    },
+    {
+      "epoch": 0.9061352147507,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5855056540701217e-07,
+      "loss": 0.9026,
+      "step": 24916
+    },
+    {
+      "epoch": 0.9061715823544386,
+      "grad_norm": 0.0,
+      "learning_rate": 4.581980445966694e-07,
+      "loss": 0.8787,
+      "step": 24917
+    },
+    {
+      "epoch": 0.9062079499581772,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5784565616528885e-07,
+      "loss": 1.0103,
+      "step": 24918
+    },
+    {
+      "epoch": 0.9062443175619158,
+      "grad_norm": 0.0,
+      "learning_rate": 4.574934001177622e-07,
+      "loss": 0.9401,
+      "step": 24919
+    },
+    {
+      "epoch": 0.9062806851656544,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5714127645897666e-07,
+      "loss": 0.9239,
+      "step": 24920
+    },
+    {
+      "epoch": 0.906317052769393,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5678928519381607e-07,
+      "loss": 0.6973,
+      "step": 24921
+    },
+    {
+      "epoch": 0.9063534203731316,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5643742632716426e-07,
+      "loss": 0.9406,
+      "step": 24922
+    },
+    {
+      "epoch": 0.9063897879768702,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5608569986390184e-07,
+      "loss": 0.8956,
+      "step": 24923
+    },
+    {
+      "epoch": 0.9064261555806088,
+      "grad_norm": 0.0,
+      "learning_rate": 4.557341058089093e-07,
+      "loss": 0.8663,
+      "step": 24924
+    },
+    {
+      "epoch": 0.9064625231843474,
+      "grad_norm": 0.0,
+      "learning_rate": 4.55382644167065e-07,
+      "loss": 0.886,
+      "step": 24925
+    },
+    {
+      "epoch": 0.906498890788086,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5503131494324395e-07,
+      "loss": 0.8921,
+      "step": 24926
+    },
+    {
+      "epoch": 0.9065352583918246,
+      "grad_norm": 0.0,
+      "learning_rate": 4.546801181423222e-07,
+      "loss": 1.0004,
+      "step": 24927
+    },
+    {
+      "epoch": 0.9065716259955632,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5432905376916804e-07,
+      "loss": 0.9323,
+      "step": 24928
+    },
+    {
+      "epoch": 0.9066079935993018,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5397812182865763e-07,
+      "loss": 0.8488,
+      "step": 24929
+    },
+    {
+      "epoch": 0.9066443612030404,
+      "grad_norm": 0.0,
+      "learning_rate": 4.53627322325656e-07,
+      "loss": 0.9777,
+      "step": 24930
+    },
+    {
+      "epoch": 0.906680728806779,
+      "grad_norm": 0.0,
+      "learning_rate": 4.532766552650292e-07,
+      "loss": 0.9182,
+      "step": 24931
+    },
+    {
+      "epoch": 0.9067170964105176,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5292612065164555e-07,
+      "loss": 0.8765,
+      "step": 24932
+    },
+    {
+      "epoch": 0.906753464014256,
+      "grad_norm": 0.0,
+      "learning_rate": 4.525757184903656e-07,
+      "loss": 0.9995,
+      "step": 24933
+    },
+    {
+      "epoch": 0.9067898316179946,
+      "grad_norm": 0.0,
+      "learning_rate": 4.522254487860522e-07,
+      "loss": 0.9691,
+      "step": 24934
+    },
+    {
+      "epoch": 0.9068261992217332,
+      "grad_norm": 0.0,
+      "learning_rate": 4.518753115435637e-07,
+      "loss": 1.0425,
+      "step": 24935
+    },
+    {
+      "epoch": 0.9068625668254718,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5152530676775827e-07,
+      "loss": 0.9117,
+      "step": 24936
+    },
+    {
+      "epoch": 0.9068989344292104,
+      "grad_norm": 0.0,
+      "learning_rate": 4.511754344634911e-07,
+      "loss": 0.8381,
+      "step": 24937
+    },
+    {
+      "epoch": 0.906935302032949,
+      "grad_norm": 0.0,
+      "learning_rate": 4.508256946356171e-07,
+      "loss": 0.8168,
+      "step": 24938
+    },
+    {
+      "epoch": 0.9069716696366876,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5047608728898905e-07,
+      "loss": 1.0391,
+      "step": 24939
+    },
+    {
+      "epoch": 0.9070080372404262,
+      "grad_norm": 0.0,
+      "learning_rate": 4.501266124284553e-07,
+      "loss": 0.9856,
+      "step": 24940
+    },
+    {
+      "epoch": 0.9070444048441648,
+      "grad_norm": 0.0,
+      "learning_rate": 4.497772700588654e-07,
+      "loss": 0.9404,
+      "step": 24941
+    },
+    {
+      "epoch": 0.9070807724479034,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4942806018506757e-07,
+      "loss": 0.9053,
+      "step": 24942
+    },
+    {
+      "epoch": 0.907117140051642,
+      "grad_norm": 0.0,
+      "learning_rate": 4.490789828119013e-07,
+      "loss": 0.979,
+      "step": 24943
+    },
+    {
+      "epoch": 0.9071535076553806,
+      "grad_norm": 0.0,
+      "learning_rate": 4.487300379442161e-07,
+      "loss": 0.9361,
+      "step": 24944
+    },
+    {
+      "epoch": 0.9071898752591192,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4838122558684914e-07,
+      "loss": 0.915,
+      "step": 24945
+    },
+    {
+      "epoch": 0.9072262428628578,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4803254574464105e-07,
+      "loss": 0.9252,
+      "step": 24946
+    },
+    {
+      "epoch": 0.9072626104665964,
+      "grad_norm": 0.0,
+      "learning_rate": 4.476839984224279e-07,
+      "loss": 0.8714,
+      "step": 24947
+    },
+    {
+      "epoch": 0.907298978070335,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4733558362504816e-07,
+      "loss": 0.8807,
+      "step": 24948
+    },
+    {
+      "epoch": 0.9073353456740736,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4698730135733335e-07,
+      "loss": 0.8952,
+      "step": 24949
+    },
+    {
+      "epoch": 0.9073717132778121,
+      "grad_norm": 0.0,
+      "learning_rate": 4.466391516241142e-07,
+      "loss": 0.9143,
+      "step": 24950
+    },
+    {
+      "epoch": 0.9074080808815507,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4629113443022343e-07,
+      "loss": 0.8602,
+      "step": 24951
+    },
+    {
+      "epoch": 0.9074444484852893,
+      "grad_norm": 0.0,
+      "learning_rate": 4.459432497804883e-07,
+      "loss": 0.751,
+      "step": 24952
+    },
+    {
+      "epoch": 0.9074808160890279,
+      "grad_norm": 0.0,
+      "learning_rate": 4.45595497679735e-07,
+      "loss": 0.9117,
+      "step": 24953
+    },
+    {
+      "epoch": 0.9075171836927665,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4524787813278737e-07,
+      "loss": 0.8612,
+      "step": 24954
+    },
+    {
+      "epoch": 0.9075535512965051,
+      "grad_norm": 0.0,
+      "learning_rate": 4.449003911444716e-07,
+      "loss": 0.9036,
+      "step": 24955
+    },
+    {
+      "epoch": 0.9075899189002437,
+      "grad_norm": 0.0,
+      "learning_rate": 4.445530367196027e-07,
+      "loss": 0.8905,
+      "step": 24956
+    },
+    {
+      "epoch": 0.9076262865039822,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4420581486300464e-07,
+      "loss": 1.0348,
+      "step": 24957
+    },
+    {
+      "epoch": 0.9076626541077208,
+      "grad_norm": 0.0,
+      "learning_rate": 4.438587255794935e-07,
+      "loss": 0.9342,
+      "step": 24958
+    },
+    {
+      "epoch": 0.9076990217114594,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4351176887388324e-07,
+      "loss": 0.9218,
+      "step": 24959
+    },
+    {
+      "epoch": 0.907735389315198,
+      "grad_norm": 0.0,
+      "learning_rate": 4.431649447509878e-07,
+      "loss": 0.9619,
+      "step": 24960
+    },
+    {
+      "epoch": 0.9077717569189366,
+      "grad_norm": 0.0,
+      "learning_rate": 4.428182532156222e-07,
+      "loss": 0.9405,
+      "step": 24961
+    },
+    {
+      "epoch": 0.9078081245226752,
+      "grad_norm": 0.0,
+      "learning_rate": 4.424716942725904e-07,
+      "loss": 0.8797,
+      "step": 24962
+    },
+    {
+      "epoch": 0.9078444921264138,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4212526792670406e-07,
+      "loss": 0.8359,
+      "step": 24963
+    },
+    {
+      "epoch": 0.9078808597301524,
+      "grad_norm": 0.0,
+      "learning_rate": 4.417789741827694e-07,
+      "loss": 0.9479,
+      "step": 24964
+    },
+    {
+      "epoch": 0.907917227333891,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4143281304559027e-07,
+      "loss": 1.0027,
+      "step": 24965
+    },
+    {
+      "epoch": 0.9079535949376295,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4108678451996845e-07,
+      "loss": 0.957,
+      "step": 24966
+    },
+    {
+      "epoch": 0.9079899625413681,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4074088861070674e-07,
+      "loss": 0.9458,
+      "step": 24967
+    },
+    {
+      "epoch": 0.9080263301451067,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4039512532260134e-07,
+      "loss": 0.9253,
+      "step": 24968
+    },
+    {
+      "epoch": 0.9080626977488453,
+      "grad_norm": 0.0,
+      "learning_rate": 4.400494946604483e-07,
+      "loss": 0.9507,
+      "step": 24969
+    },
+    {
+      "epoch": 0.9080990653525839,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3970399662904726e-07,
+      "loss": 1.0198,
+      "step": 24970
+    },
+    {
+      "epoch": 0.9081354329563225,
+      "grad_norm": 0.0,
+      "learning_rate": 4.393586312331888e-07,
+      "loss": 0.8973,
+      "step": 24971
+    },
+    {
+      "epoch": 0.9081718005600611,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3901339847766343e-07,
+      "loss": 0.9298,
+      "step": 24972
+    },
+    {
+      "epoch": 0.9082081681637997,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3866829836726186e-07,
+      "loss": 0.9617,
+      "step": 24973
+    },
+    {
+      "epoch": 0.9082445357675383,
+      "grad_norm": 0.0,
+      "learning_rate": 4.383233309067736e-07,
+      "loss": 0.9466,
+      "step": 24974
+    },
+    {
+      "epoch": 0.9082809033712769,
+      "grad_norm": 0.0,
+      "learning_rate": 4.379784961009803e-07,
+      "loss": 0.8654,
+      "step": 24975
+    },
+    {
+      "epoch": 0.9083172709750155,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3763379395467043e-07,
+      "loss": 0.9244,
+      "step": 24976
+    },
+    {
+      "epoch": 0.9083536385787541,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3728922447262454e-07,
+      "loss": 0.8835,
+      "step": 24977
+    },
+    {
+      "epoch": 0.9083900061824927,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3694478765962336e-07,
+      "loss": 0.8605,
+      "step": 24978
+    },
+    {
+      "epoch": 0.9084263737862313,
+      "grad_norm": 0.0,
+      "learning_rate": 4.366004835204429e-07,
+      "loss": 0.8923,
+      "step": 24979
+    },
+    {
+      "epoch": 0.9084627413899699,
+      "grad_norm": 0.0,
+      "learning_rate": 4.36256312059864e-07,
+      "loss": 0.8588,
+      "step": 24980
+    },
+    {
+      "epoch": 0.9084991089937084,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3591227328265815e-07,
+      "loss": 0.8786,
+      "step": 24981
+    },
+    {
+      "epoch": 0.908535476597447,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3556836719360064e-07,
+      "loss": 0.852,
+      "step": 24982
+    },
+    {
+      "epoch": 0.9085718442011855,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3522459379746197e-07,
+      "loss": 0.8899,
+      "step": 24983
+    },
+    {
+      "epoch": 0.9086082118049241,
+      "grad_norm": 0.0,
+      "learning_rate": 4.348809530990117e-07,
+      "loss": 0.8607,
+      "step": 24984
+    },
+    {
+      "epoch": 0.9086445794086627,
+      "grad_norm": 0.0,
+      "learning_rate": 4.34537445103016e-07,
+      "loss": 0.9354,
+      "step": 24985
+    },
+    {
+      "epoch": 0.9086809470124013,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3419406981424217e-07,
+      "loss": 0.9579,
+      "step": 24986
+    },
+    {
+      "epoch": 0.9087173146161399,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3385082723745533e-07,
+      "loss": 0.971,
+      "step": 24987
+    },
+    {
+      "epoch": 0.9087536822198785,
+      "grad_norm": 0.0,
+      "learning_rate": 4.335077173774127e-07,
+      "loss": 0.8954,
+      "step": 24988
+    },
+    {
+      "epoch": 0.9087900498236171,
+      "grad_norm": 0.0,
+      "learning_rate": 4.331647402388794e-07,
+      "loss": 0.9557,
+      "step": 24989
+    },
+    {
+      "epoch": 0.9088264174273557,
+      "grad_norm": 0.0,
+      "learning_rate": 4.328218958266106e-07,
+      "loss": 0.884,
+      "step": 24990
+    },
+    {
+      "epoch": 0.9088627850310943,
+      "grad_norm": 0.0,
+      "learning_rate": 4.324791841453646e-07,
+      "loss": 0.7947,
+      "step": 24991
+    },
+    {
+      "epoch": 0.9088991526348329,
+      "grad_norm": 0.0,
+      "learning_rate": 4.321366051998943e-07,
+      "loss": 0.9479,
+      "step": 24992
+    },
+    {
+      "epoch": 0.9089355202385715,
+      "grad_norm": 0.0,
+      "learning_rate": 4.317941589949559e-07,
+      "loss": 0.8925,
+      "step": 24993
+    },
+    {
+      "epoch": 0.9089718878423101,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3145184553529564e-07,
+      "loss": 0.9567,
+      "step": 24994
+    },
+    {
+      "epoch": 0.9090082554460487,
+      "grad_norm": 0.0,
+      "learning_rate": 4.311096648256663e-07,
+      "loss": 0.8969,
+      "step": 24995
+    },
+    {
+      "epoch": 0.9090446230497873,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3076761687081416e-07,
+      "loss": 1.0147,
+      "step": 24996
+    },
+    {
+      "epoch": 0.9090809906535259,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3042570167548426e-07,
+      "loss": 0.919,
+      "step": 24997
+    },
+    {
+      "epoch": 0.9091173582572645,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3008391924441837e-07,
+      "loss": 0.8508,
+      "step": 24998
+    },
+    {
+      "epoch": 0.9091537258610031,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2974226958236163e-07,
+      "loss": 0.8492,
+      "step": 24999
+    },
+    {
+      "epoch": 0.9091900934647416,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2940075269405244e-07,
+      "loss": 0.8117,
+      "step": 25000
+    },
+    {
+      "epoch": 0.9092264610684802,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2905936858422926e-07,
+      "loss": 0.9439,
+      "step": 25001
+    },
+    {
+      "epoch": 0.9092628286722187,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2871811725762824e-07,
+      "loss": 0.9163,
+      "step": 25002
+    },
+    {
+      "epoch": 0.9092991962759573,
+      "grad_norm": 0.0,
+      "learning_rate": 4.283769987189834e-07,
+      "loss": 0.9426,
+      "step": 25003
+    },
+    {
+      "epoch": 0.9093355638796959,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2803601297302544e-07,
+      "loss": 0.9278,
+      "step": 25004
+    },
+    {
+      "epoch": 0.9093719314834345,
+      "grad_norm": 0.0,
+      "learning_rate": 4.276951600244883e-07,
+      "loss": 0.8802,
+      "step": 25005
+    },
+    {
+      "epoch": 0.9094082990871731,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2735443987809934e-07,
+      "loss": 0.9057,
+      "step": 25006
+    },
+    {
+      "epoch": 0.9094446666909117,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2701385253858697e-07,
+      "loss": 0.9875,
+      "step": 25007
+    },
+    {
+      "epoch": 0.9094810342946503,
+      "grad_norm": 0.0,
+      "learning_rate": 4.266733980106741e-07,
+      "loss": 0.8533,
+      "step": 25008
+    },
+    {
+      "epoch": 0.9095174018983889,
+      "grad_norm": 0.0,
+      "learning_rate": 4.263330762990847e-07,
+      "loss": 0.7667,
+      "step": 25009
+    },
+    {
+      "epoch": 0.9095537695021275,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2599288740854173e-07,
+      "loss": 0.8325,
+      "step": 25010
+    },
+    {
+      "epoch": 0.9095901371058661,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2565283134376134e-07,
+      "loss": 0.9388,
+      "step": 25011
+    },
+    {
+      "epoch": 0.9096265047096047,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2531290810946646e-07,
+      "loss": 0.921,
+      "step": 25012
+    },
+    {
+      "epoch": 0.9096628723133433,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2497311771036884e-07,
+      "loss": 0.9702,
+      "step": 25013
+    },
+    {
+      "epoch": 0.9096992399170819,
+      "grad_norm": 0.0,
+      "learning_rate": 4.246334601511837e-07,
+      "loss": 0.8318,
+      "step": 25014
+    },
+    {
+      "epoch": 0.9097356075208205,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2429393543662487e-07,
+      "loss": 0.8627,
+      "step": 25015
+    },
+    {
+      "epoch": 0.9097719751245591,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2395454357140096e-07,
+      "loss": 0.926,
+      "step": 25016
+    },
+    {
+      "epoch": 0.9098083427282976,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2361528456021925e-07,
+      "loss": 0.927,
+      "step": 25017
+    },
+    {
+      "epoch": 0.9098447103320362,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2327615840779046e-07,
+      "loss": 1.0571,
+      "step": 25018
+    },
+    {
+      "epoch": 0.9098810779357748,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2293716511881743e-07,
+      "loss": 0.9732,
+      "step": 25019
+    },
+    {
+      "epoch": 0.9099174455395134,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2259830469800314e-07,
+      "loss": 0.8513,
+      "step": 25020
+    },
+    {
+      "epoch": 0.909953813143252,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2225957715004704e-07,
+      "loss": 0.9049,
+      "step": 25021
+    },
+    {
+      "epoch": 0.9099901807469906,
+      "grad_norm": 0.0,
+      "learning_rate": 4.219209824796533e-07,
+      "loss": 0.9311,
+      "step": 25022
+    },
+    {
+      "epoch": 0.9100265483507292,
+      "grad_norm": 0.0,
+      "learning_rate": 4.215825206915136e-07,
+      "loss": 0.9485,
+      "step": 25023
+    },
+    {
+      "epoch": 0.9100629159544678,
+      "grad_norm": 0.0,
+      "learning_rate": 4.212441917903276e-07,
+      "loss": 0.9067,
+      "step": 25024
+    },
+    {
+      "epoch": 0.9100992835582064,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2090599578078815e-07,
+      "loss": 0.8724,
+      "step": 25025
+    },
+    {
+      "epoch": 0.910135651161945,
+      "grad_norm": 0.0,
+      "learning_rate": 4.205679326675871e-07,
+      "loss": 0.9349,
+      "step": 25026
+    },
+    {
+      "epoch": 0.9101720187656835,
+      "grad_norm": 0.0,
+      "learning_rate": 4.202300024554151e-07,
+      "loss": 0.8917,
+      "step": 25027
+    },
+    {
+      "epoch": 0.9102083863694221,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1989220514895957e-07,
+      "loss": 0.8724,
+      "step": 25028
+    },
+    {
+      "epoch": 0.9102447539731607,
+      "grad_norm": 0.0,
+      "learning_rate": 4.195545407529067e-07,
+      "loss": 0.8332,
+      "step": 25029
+    },
+    {
+      "epoch": 0.9102811215768993,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1921700927194053e-07,
+      "loss": 0.8321,
+      "step": 25030
+    },
+    {
+      "epoch": 0.9103174891806379,
+      "grad_norm": 0.0,
+      "learning_rate": 4.188796107107473e-07,
+      "loss": 0.92,
+      "step": 25031
+    },
+    {
+      "epoch": 0.9103538567843765,
+      "grad_norm": 0.0,
+      "learning_rate": 4.185423450740045e-07,
+      "loss": 0.8732,
+      "step": 25032
+    },
+    {
+      "epoch": 0.9103902243881151,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1820521236639266e-07,
+      "loss": 0.9302,
+      "step": 25033
+    },
+    {
+      "epoch": 0.9104265919918536,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1786821259258816e-07,
+      "loss": 0.8655,
+      "step": 25034
+    },
+    {
+      "epoch": 0.9104629595955922,
+      "grad_norm": 0.0,
+      "learning_rate": 4.175313457572683e-07,
+      "loss": 0.9123,
+      "step": 25035
+    },
+    {
+      "epoch": 0.9104993271993308,
+      "grad_norm": 0.0,
+      "learning_rate": 4.171946118651027e-07,
+      "loss": 0.9957,
+      "step": 25036
+    },
+    {
+      "epoch": 0.9105356948030694,
+      "grad_norm": 0.0,
+      "learning_rate": 4.168580109207676e-07,
+      "loss": 0.9974,
+      "step": 25037
+    },
+    {
+      "epoch": 0.910572062406808,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1652154292893045e-07,
+      "loss": 0.9392,
+      "step": 25038
+    },
+    {
+      "epoch": 0.9106084300105466,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1618520789426075e-07,
+      "loss": 0.9272,
+      "step": 25039
+    },
+    {
+      "epoch": 0.9106447976142852,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1584900582142153e-07,
+      "loss": 0.94,
+      "step": 25040
+    },
+    {
+      "epoch": 0.9106811652180238,
+      "grad_norm": 0.0,
+      "learning_rate": 4.155129367150812e-07,
+      "loss": 0.9074,
+      "step": 25041
+    },
+    {
+      "epoch": 0.9107175328217624,
+      "grad_norm": 0.0,
+      "learning_rate": 4.151770005798994e-07,
+      "loss": 0.8947,
+      "step": 25042
+    },
+    {
+      "epoch": 0.910753900425501,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1484119742053685e-07,
+      "loss": 0.9046,
+      "step": 25043
+    },
+    {
+      "epoch": 0.9107902680292396,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1450552724165426e-07,
+      "loss": 0.8028,
+      "step": 25044
+    },
+    {
+      "epoch": 0.9108266356329782,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1416999004790794e-07,
+      "loss": 0.8819,
+      "step": 25045
+    },
+    {
+      "epoch": 0.9108630032367168,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1383458584395187e-07,
+      "loss": 0.9053,
+      "step": 25046
+    },
+    {
+      "epoch": 0.9108993708404554,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1349931463444016e-07,
+      "loss": 0.8837,
+      "step": 25047
+    },
+    {
+      "epoch": 0.910935738444194,
+      "grad_norm": 0.0,
+      "learning_rate": 4.131641764240235e-07,
+      "loss": 0.8782,
+      "step": 25048
+    },
+    {
+      "epoch": 0.9109721060479326,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1282917121735155e-07,
+      "loss": 0.8824,
+      "step": 25049
+    },
+    {
+      "epoch": 0.9110084736516711,
+      "grad_norm": 0.0,
+      "learning_rate": 4.124942990190728e-07,
+      "loss": 0.861,
+      "step": 25050
+    },
+    {
+      "epoch": 0.9110448412554096,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1215955983383236e-07,
+      "loss": 0.9703,
+      "step": 25051
+    },
+    {
+      "epoch": 0.9110812088591482,
+      "grad_norm": 0.0,
+      "learning_rate": 4.118249536662755e-07,
+      "loss": 0.8752,
+      "step": 25052
+    },
+    {
+      "epoch": 0.9111175764628868,
+      "grad_norm": 0.0,
+      "learning_rate": 4.114904805210429e-07,
+      "loss": 0.9865,
+      "step": 25053
+    },
+    {
+      "epoch": 0.9111539440666254,
+      "grad_norm": 0.0,
+      "learning_rate": 4.111561404027775e-07,
+      "loss": 0.8376,
+      "step": 25054
+    },
+    {
+      "epoch": 0.911190311670364,
+      "grad_norm": 0.0,
+      "learning_rate": 4.108219333161123e-07,
+      "loss": 0.9114,
+      "step": 25055
+    },
+    {
+      "epoch": 0.9112266792741026,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1048785926569023e-07,
+      "loss": 0.9759,
+      "step": 25056
+    },
+    {
+      "epoch": 0.9112630468778412,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1015391825614316e-07,
+      "loss": 0.9337,
+      "step": 25057
+    },
+    {
+      "epoch": 0.9112994144815798,
+      "grad_norm": 0.0,
+      "learning_rate": 4.09820110292104e-07,
+      "loss": 0.9343,
+      "step": 25058
+    },
+    {
+      "epoch": 0.9113357820853184,
+      "grad_norm": 0.0,
+      "learning_rate": 4.094864353782024e-07,
+      "loss": 0.8159,
+      "step": 25059
+    },
+    {
+      "epoch": 0.911372149689057,
+      "grad_norm": 0.0,
+      "learning_rate": 4.091528935190725e-07,
+      "loss": 0.9822,
+      "step": 25060
+    },
+    {
+      "epoch": 0.9114085172927956,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0881948471933607e-07,
+      "loss": 0.8949,
+      "step": 25061
+    },
+    {
+      "epoch": 0.9114448848965342,
+      "grad_norm": 0.0,
+      "learning_rate": 4.084862089836217e-07,
+      "loss": 0.9293,
+      "step": 25062
+    },
+    {
+      "epoch": 0.9114812525002728,
+      "grad_norm": 0.0,
+      "learning_rate": 4.081530663165534e-07,
+      "loss": 0.8685,
+      "step": 25063
+    },
+    {
+      "epoch": 0.9115176201040114,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0782005672275193e-07,
+      "loss": 0.8431,
+      "step": 25064
+    },
+    {
+      "epoch": 0.91155398770775,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0748718020683697e-07,
+      "loss": 0.9902,
+      "step": 25065
+    },
+    {
+      "epoch": 0.9115903553114886,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0715443677342815e-07,
+      "loss": 0.9313,
+      "step": 25066
+    },
+    {
+      "epoch": 0.9116267229152271,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0682182642714177e-07,
+      "loss": 0.8501,
+      "step": 25067
+    },
+    {
+      "epoch": 0.9116630905189657,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0648934917258964e-07,
+      "loss": 0.8942,
+      "step": 25068
+    },
+    {
+      "epoch": 0.9116994581227043,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0615700501438703e-07,
+      "loss": 0.9557,
+      "step": 25069
+    },
+    {
+      "epoch": 0.9117358257264429,
+      "grad_norm": 0.0,
+      "learning_rate": 4.058247939571447e-07,
+      "loss": 0.9124,
+      "step": 25070
+    },
+    {
+      "epoch": 0.9117721933301814,
+      "grad_norm": 0.0,
+      "learning_rate": 4.054927160054711e-07,
+      "loss": 0.874,
+      "step": 25071
+    },
+    {
+      "epoch": 0.91180856093392,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0516077116397155e-07,
+      "loss": 0.8525,
+      "step": 25072
+    },
+    {
+      "epoch": 0.9118449285376586,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0482895943725565e-07,
+      "loss": 0.8672,
+      "step": 25073
+    },
+    {
+      "epoch": 0.9118812961413972,
+      "grad_norm": 0.0,
+      "learning_rate": 4.044972808299219e-07,
+      "loss": 0.836,
+      "step": 25074
+    },
+    {
+      "epoch": 0.9119176637451358,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0416573534657553e-07,
+      "loss": 1.0169,
+      "step": 25075
+    },
+    {
+      "epoch": 0.9119540313488744,
+      "grad_norm": 0.0,
+      "learning_rate": 4.038343229918151e-07,
+      "loss": 0.9412,
+      "step": 25076
+    },
+    {
+      "epoch": 0.911990398952613,
+      "grad_norm": 0.0,
+      "learning_rate": 4.03503043770237e-07,
+      "loss": 0.9001,
+      "step": 25077
+    },
+    {
+      "epoch": 0.9120267665563516,
+      "grad_norm": 0.0,
+      "learning_rate": 4.031718976864385e-07,
+      "loss": 0.9001,
+      "step": 25078
+    },
+    {
+      "epoch": 0.9120631341600902,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0284088474501604e-07,
+      "loss": 0.8166,
+      "step": 25079
+    },
+    {
+      "epoch": 0.9120995017638288,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0251000495055703e-07,
+      "loss": 0.8807,
+      "step": 25080
+    },
+    {
+      "epoch": 0.9121358693675674,
+      "grad_norm": 0.0,
+      "learning_rate": 4.021792583076556e-07,
+      "loss": 0.9702,
+      "step": 25081
+    },
+    {
+      "epoch": 0.912172236971306,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0184864482090025e-07,
+      "loss": 0.8925,
+      "step": 25082
+    },
+    {
+      "epoch": 0.9122086045750446,
+      "grad_norm": 0.0,
+      "learning_rate": 4.015181644948762e-07,
+      "loss": 0.9312,
+      "step": 25083
+    },
+    {
+      "epoch": 0.9122449721787831,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0118781733416877e-07,
+      "loss": 0.8821,
+      "step": 25084
+    },
+    {
+      "epoch": 0.9122813397825217,
+      "grad_norm": 0.0,
+      "learning_rate": 4.008576033433609e-07,
+      "loss": 0.8235,
+      "step": 25085
+    },
+    {
+      "epoch": 0.9123177073862603,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0052752252703773e-07,
+      "loss": 0.932,
+      "step": 25086
+    },
+    {
+      "epoch": 0.9123540749899989,
+      "grad_norm": 0.0,
+      "learning_rate": 4.001975748897713e-07,
+      "loss": 0.9784,
+      "step": 25087
+    },
+    {
+      "epoch": 0.9123904425937375,
+      "grad_norm": 0.0,
+      "learning_rate": 3.998677604361456e-07,
+      "loss": 0.8716,
+      "step": 25088
+    },
+    {
+      "epoch": 0.9124268101974761,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9953807917073263e-07,
+      "loss": 0.8087,
+      "step": 25089
+    },
+    {
+      "epoch": 0.9124631778012147,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9920853109810865e-07,
+      "loss": 0.8223,
+      "step": 25090
+    },
+    {
+      "epoch": 0.9124995454049533,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9887911622284225e-07,
+      "loss": 0.9736,
+      "step": 25091
+    },
+    {
+      "epoch": 0.9125359130086919,
+      "grad_norm": 0.0,
+      "learning_rate": 3.985498345495098e-07,
+      "loss": 0.9098,
+      "step": 25092
+    },
+    {
+      "epoch": 0.9125722806124305,
+      "grad_norm": 0.0,
+      "learning_rate": 3.982206860826721e-07,
+      "loss": 0.981,
+      "step": 25093
+    },
+    {
+      "epoch": 0.912608648216169,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9789167082690093e-07,
+      "loss": 0.8508,
+      "step": 25094
+    },
+    {
+      "epoch": 0.9126450158199076,
+      "grad_norm": 0.0,
+      "learning_rate": 3.975627887867595e-07,
+      "loss": 1.1064,
+      "step": 25095
+    },
+    {
+      "epoch": 0.9126813834236462,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9723403996680955e-07,
+      "loss": 1.0009,
+      "step": 25096
+    },
+    {
+      "epoch": 0.9127177510273848,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9690542437161193e-07,
+      "loss": 0.8353,
+      "step": 25097
+    },
+    {
+      "epoch": 0.9127541186311234,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9657694200572747e-07,
+      "loss": 0.9094,
+      "step": 25098
+    },
+    {
+      "epoch": 0.912790486234862,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9624859287371363e-07,
+      "loss": 0.8695,
+      "step": 25099
+    },
+    {
+      "epoch": 0.9128268538386006,
+      "grad_norm": 0.0,
+      "learning_rate": 3.959203769801234e-07,
+      "loss": 0.8654,
+      "step": 25100
+    },
+    {
+      "epoch": 0.9128632214423391,
+      "grad_norm": 0.0,
+      "learning_rate": 3.955922943295121e-07,
+      "loss": 0.9969,
+      "step": 25101
+    },
+    {
+      "epoch": 0.9128995890460777,
+      "grad_norm": 0.0,
+      "learning_rate": 3.952643449264304e-07,
+      "loss": 0.9255,
+      "step": 25102
+    },
+    {
+      "epoch": 0.9129359566498163,
+      "grad_norm": 0.0,
+      "learning_rate": 3.949365287754292e-07,
+      "loss": 0.9849,
+      "step": 25103
+    },
+    {
+      "epoch": 0.9129723242535549,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9460884588105376e-07,
+      "loss": 0.9639,
+      "step": 25104
+    },
+    {
+      "epoch": 0.9130086918572935,
+      "grad_norm": 0.0,
+      "learning_rate": 3.942812962478548e-07,
+      "loss": 0.9498,
+      "step": 25105
+    },
+    {
+      "epoch": 0.9130450594610321,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9395387988037216e-07,
+      "loss": 0.9905,
+      "step": 25106
+    },
+    {
+      "epoch": 0.9130814270647707,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9362659678315205e-07,
+      "loss": 1.028,
+      "step": 25107
+    },
+    {
+      "epoch": 0.9131177946685093,
+      "grad_norm": 0.0,
+      "learning_rate": 3.932994469607321e-07,
+      "loss": 0.9259,
+      "step": 25108
+    },
+    {
+      "epoch": 0.9131541622722479,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9297243041765297e-07,
+      "loss": 0.8859,
+      "step": 25109
+    },
+    {
+      "epoch": 0.9131905298759865,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9264554715844895e-07,
+      "loss": 0.9368,
+      "step": 25110
+    },
+    {
+      "epoch": 0.9132268974797251,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9231879718765966e-07,
+      "loss": 0.9641,
+      "step": 25111
+    },
+    {
+      "epoch": 0.9132632650834637,
+      "grad_norm": 0.0,
+      "learning_rate": 3.919921805098137e-07,
+      "loss": 0.9035,
+      "step": 25112
+    },
+    {
+      "epoch": 0.9132996326872023,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9166569712944524e-07,
+      "loss": 0.9118,
+      "step": 25113
+    },
+    {
+      "epoch": 0.9133360002909409,
+      "grad_norm": 0.0,
+      "learning_rate": 3.913393470510818e-07,
+      "loss": 0.9756,
+      "step": 25114
+    },
+    {
+      "epoch": 0.9133723678946795,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9101313027925303e-07,
+      "loss": 0.8852,
+      "step": 25115
+    },
+    {
+      "epoch": 0.9134087354984181,
+      "grad_norm": 0.0,
+      "learning_rate": 3.90687046818482e-07,
+      "loss": 0.9482,
+      "step": 25116
+    },
+    {
+      "epoch": 0.9134451031021567,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9036109667329623e-07,
+      "loss": 0.97,
+      "step": 25117
+    },
+    {
+      "epoch": 0.9134814707058951,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9003527984821545e-07,
+      "loss": 0.898,
+      "step": 25118
+    },
+    {
+      "epoch": 0.9135178383096337,
+      "grad_norm": 0.0,
+      "learning_rate": 3.897095963477604e-07,
+      "loss": 0.8766,
+      "step": 25119
+    },
+    {
+      "epoch": 0.9135542059133723,
+      "grad_norm": 0.0,
+      "learning_rate": 3.893840461764475e-07,
+      "loss": 0.9702,
+      "step": 25120
+    },
+    {
+      "epoch": 0.9135905735171109,
+      "grad_norm": 0.0,
+      "learning_rate": 3.890586293387988e-07,
+      "loss": 0.9863,
+      "step": 25121
+    },
+    {
+      "epoch": 0.9136269411208495,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8873334583932386e-07,
+      "loss": 0.9011,
+      "step": 25122
+    },
+    {
+      "epoch": 0.9136633087245881,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8840819568253585e-07,
+      "loss": 0.9524,
+      "step": 25123
+    },
+    {
+      "epoch": 0.9136996763283267,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8808317887295e-07,
+      "loss": 0.939,
+      "step": 25124
+    },
+    {
+      "epoch": 0.9137360439320653,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8775829541506937e-07,
+      "loss": 0.9416,
+      "step": 25125
+    },
+    {
+      "epoch": 0.9137724115358039,
+      "grad_norm": 0.0,
+      "learning_rate": 3.874335453134059e-07,
+      "loss": 0.8653,
+      "step": 25126
+    },
+    {
+      "epoch": 0.9138087791395425,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8710892857246384e-07,
+      "loss": 1.0225,
+      "step": 25127
+    },
+    {
+      "epoch": 0.9138451467432811,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8678444519674617e-07,
+      "loss": 0.8063,
+      "step": 25128
+    },
+    {
+      "epoch": 0.9138815143470197,
+      "grad_norm": 0.0,
+      "learning_rate": 3.864600951907538e-07,
+      "loss": 0.8895,
+      "step": 25129
+    },
+    {
+      "epoch": 0.9139178819507583,
+      "grad_norm": 0.0,
+      "learning_rate": 3.861358785589897e-07,
+      "loss": 0.7504,
+      "step": 25130
+    },
+    {
+      "epoch": 0.9139542495544969,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8581179530594924e-07,
+      "loss": 0.8104,
+      "step": 25131
+    },
+    {
+      "epoch": 0.9139906171582355,
+      "grad_norm": 0.0,
+      "learning_rate": 3.854878454361288e-07,
+      "loss": 0.9028,
+      "step": 25132
+    },
+    {
+      "epoch": 0.9140269847619741,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8516402895402373e-07,
+      "loss": 0.8673,
+      "step": 25133
+    },
+    {
+      "epoch": 0.9140633523657127,
+      "grad_norm": 0.0,
+      "learning_rate": 3.84840345864127e-07,
+      "loss": 0.783,
+      "step": 25134
+    },
+    {
+      "epoch": 0.9140997199694512,
+      "grad_norm": 0.0,
+      "learning_rate": 3.845167961709262e-07,
+      "loss": 0.8469,
+      "step": 25135
+    },
+    {
+      "epoch": 0.9141360875731898,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8419337987891323e-07,
+      "loss": 0.9117,
+      "step": 25136
+    },
+    {
+      "epoch": 0.9141724551769284,
+      "grad_norm": 0.0,
+      "learning_rate": 3.838700969925746e-07,
+      "loss": 0.9154,
+      "step": 25137
+    },
+    {
+      "epoch": 0.914208822780667,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8354694751639555e-07,
+      "loss": 0.9618,
+      "step": 25138
+    },
+    {
+      "epoch": 0.9142451903844055,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8322393145485694e-07,
+      "loss": 0.9139,
+      "step": 25139
+    },
+    {
+      "epoch": 0.9142815579881441,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8290104881244405e-07,
+      "loss": 0.9222,
+      "step": 25140
+    },
+    {
+      "epoch": 0.9143179255918827,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8257829959363224e-07,
+      "loss": 0.7284,
+      "step": 25141
+    },
+    {
+      "epoch": 0.9143542931956213,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8225568380290125e-07,
+      "loss": 0.8396,
+      "step": 25142
+    },
+    {
+      "epoch": 0.9143906607993599,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8193320144472745e-07,
+      "loss": 0.928,
+      "step": 25143
+    },
+    {
+      "epoch": 0.9144270284030985,
+      "grad_norm": 0.0,
+      "learning_rate": 3.81610852523584e-07,
+      "loss": 0.8802,
+      "step": 25144
+    },
+    {
+      "epoch": 0.9144633960068371,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8128863704394394e-07,
+      "loss": 0.921,
+      "step": 25145
+    },
+    {
+      "epoch": 0.9144997636105757,
+      "grad_norm": 0.0,
+      "learning_rate": 3.80966555010277e-07,
+      "loss": 0.9149,
+      "step": 25146
+    },
+    {
+      "epoch": 0.9145361312143143,
+      "grad_norm": 0.0,
+      "learning_rate": 3.806446064270508e-07,
+      "loss": 0.9094,
+      "step": 25147
+    },
+    {
+      "epoch": 0.9145724988180529,
+      "grad_norm": 0.0,
+      "learning_rate": 3.803227912987317e-07,
+      "loss": 0.9076,
+      "step": 25148
+    },
+    {
+      "epoch": 0.9146088664217915,
+      "grad_norm": 0.0,
+      "learning_rate": 3.800011096297862e-07,
+      "loss": 0.8983,
+      "step": 25149
+    },
+    {
+      "epoch": 0.9146452340255301,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7967956142467624e-07,
+      "loss": 0.9165,
+      "step": 25150
+    },
+    {
+      "epoch": 0.9146816016292686,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7935814668786373e-07,
+      "loss": 0.933,
+      "step": 25151
+    },
+    {
+      "epoch": 0.9147179692330072,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7903686542380525e-07,
+      "loss": 1.0248,
+      "step": 25152
+    },
+    {
+      "epoch": 0.9147543368367458,
+      "grad_norm": 0.0,
+      "learning_rate": 3.787157176369627e-07,
+      "loss": 1.0057,
+      "step": 25153
+    },
+    {
+      "epoch": 0.9147907044404844,
+      "grad_norm": 0.0,
+      "learning_rate": 3.783947033317858e-07,
+      "loss": 0.938,
+      "step": 25154
+    },
+    {
+      "epoch": 0.914827072044223,
+      "grad_norm": 0.0,
+      "learning_rate": 3.780738225127323e-07,
+      "loss": 0.9745,
+      "step": 25155
+    },
+    {
+      "epoch": 0.9148634396479616,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7775307518425284e-07,
+      "loss": 0.8702,
+      "step": 25156
+    },
+    {
+      "epoch": 0.9148998072517002,
+      "grad_norm": 0.0,
+      "learning_rate": 3.774324613507985e-07,
+      "loss": 0.9688,
+      "step": 25157
+    },
+    {
+      "epoch": 0.9149361748554388,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7711198101681335e-07,
+      "loss": 0.935,
+      "step": 25158
+    },
+    {
+      "epoch": 0.9149725424591774,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7679163418674946e-07,
+      "loss": 0.9566,
+      "step": 25159
+    },
+    {
+      "epoch": 0.915008910062916,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7647142086504664e-07,
+      "loss": 0.9386,
+      "step": 25160
+    },
+    {
+      "epoch": 0.9150452776666546,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7615134105614795e-07,
+      "loss": 0.8674,
+      "step": 25161
+    },
+    {
+      "epoch": 0.9150816452703932,
+      "grad_norm": 0.0,
+      "learning_rate": 3.758313947644954e-07,
+      "loss": 0.8914,
+      "step": 25162
+    },
+    {
+      "epoch": 0.9151180128741317,
+      "grad_norm": 0.0,
+      "learning_rate": 3.755115819945288e-07,
+      "loss": 0.974,
+      "step": 25163
+    },
+    {
+      "epoch": 0.9151543804778703,
+      "grad_norm": 0.0,
+      "learning_rate": 3.751919027506823e-07,
+      "loss": 0.8102,
+      "step": 25164
+    },
+    {
+      "epoch": 0.9151907480816089,
+      "grad_norm": 0.0,
+      "learning_rate": 3.748723570373913e-07,
+      "loss": 0.8577,
+      "step": 25165
+    },
+    {
+      "epoch": 0.9152271156853475,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7455294485909344e-07,
+      "loss": 0.8841,
+      "step": 25166
+    },
+    {
+      "epoch": 0.9152634832890861,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7423366622021395e-07,
+      "loss": 0.903,
+      "step": 25167
+    },
+    {
+      "epoch": 0.9152998508928246,
+      "grad_norm": 0.0,
+      "learning_rate": 3.73914521125186e-07,
+      "loss": 1.02,
+      "step": 25168
+    },
+    {
+      "epoch": 0.9153362184965632,
+      "grad_norm": 0.0,
+      "learning_rate": 3.73595509578436e-07,
+      "loss": 0.926,
+      "step": 25169
+    },
+    {
+      "epoch": 0.9153725861003018,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7327663158439056e-07,
+      "loss": 0.8614,
+      "step": 25170
+    },
+    {
+      "epoch": 0.9154089537040404,
+      "grad_norm": 0.0,
+      "learning_rate": 3.729578871474726e-07,
+      "loss": 0.9586,
+      "step": 25171
+    },
+    {
+      "epoch": 0.915445321307779,
+      "grad_norm": 0.0,
+      "learning_rate": 3.726392762721076e-07,
+      "loss": 0.8541,
+      "step": 25172
+    },
+    {
+      "epoch": 0.9154816889115176,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7232079896270977e-07,
+      "loss": 0.949,
+      "step": 25173
+    },
+    {
+      "epoch": 0.9155180565152562,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7200245522370226e-07,
+      "loss": 0.9106,
+      "step": 25174
+    },
+    {
+      "epoch": 0.9155544241189948,
+      "grad_norm": 0.0,
+      "learning_rate": 3.716842450595015e-07,
+      "loss": 0.9612,
+      "step": 25175
+    },
+    {
+      "epoch": 0.9155907917227334,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7136616847452066e-07,
+      "loss": 0.8317,
+      "step": 25176
+    },
+    {
+      "epoch": 0.915627159326472,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7104822547317065e-07,
+      "loss": 0.8532,
+      "step": 25177
+    },
+    {
+      "epoch": 0.9156635269302106,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7073041605986793e-07,
+      "loss": 0.8026,
+      "step": 25178
+    },
+    {
+      "epoch": 0.9156998945339492,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7041274023901785e-07,
+      "loss": 0.9104,
+      "step": 25179
+    },
+    {
+      "epoch": 0.9157362621376878,
+      "grad_norm": 0.0,
+      "learning_rate": 3.700951980150269e-07,
+      "loss": 0.9987,
+      "step": 25180
+    },
+    {
+      "epoch": 0.9157726297414264,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6977778939230267e-07,
+      "loss": 0.8727,
+      "step": 25181
+    },
+    {
+      "epoch": 0.915808997345165,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6946051437524834e-07,
+      "loss": 0.9448,
+      "step": 25182
+    },
+    {
+      "epoch": 0.9158453649489036,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6914337296826585e-07,
+      "loss": 0.9543,
+      "step": 25183
+    },
+    {
+      "epoch": 0.9158817325526422,
+      "grad_norm": 0.0,
+      "learning_rate": 3.68826365175754e-07,
+      "loss": 0.973,
+      "step": 25184
+    },
+    {
+      "epoch": 0.9159181001563806,
+      "grad_norm": 0.0,
+      "learning_rate": 3.685094910021125e-07,
+      "loss": 0.9811,
+      "step": 25185
+    },
+    {
+      "epoch": 0.9159544677601192,
+      "grad_norm": 0.0,
+      "learning_rate": 3.681927504517346e-07,
+      "loss": 0.9285,
+      "step": 25186
+    },
+    {
+      "epoch": 0.9159908353638578,
+      "grad_norm": 0.0,
+      "learning_rate": 3.678761435290179e-07,
+      "loss": 0.9339,
+      "step": 25187
+    },
+    {
+      "epoch": 0.9160272029675964,
+      "grad_norm": 0.0,
+      "learning_rate": 3.675596702383533e-07,
+      "loss": 0.8412,
+      "step": 25188
+    },
+    {
+      "epoch": 0.916063570571335,
+      "grad_norm": 0.0,
+      "learning_rate": 3.672433305841328e-07,
+      "loss": 0.9027,
+      "step": 25189
+    },
+    {
+      "epoch": 0.9160999381750736,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6692712457074177e-07,
+      "loss": 0.8845,
+      "step": 25190
+    },
+    {
+      "epoch": 0.9161363057788122,
+      "grad_norm": 0.0,
+      "learning_rate": 3.666110522025712e-07,
+      "loss": 0.8562,
+      "step": 25191
+    },
+    {
+      "epoch": 0.9161726733825508,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6629511348400315e-07,
+      "loss": 0.9284,
+      "step": 25192
+    },
+    {
+      "epoch": 0.9162090409862894,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6597930841942184e-07,
+      "loss": 1.0622,
+      "step": 25193
+    },
+    {
+      "epoch": 0.916245408590028,
+      "grad_norm": 0.0,
+      "learning_rate": 3.656636370132094e-07,
+      "loss": 0.9337,
+      "step": 25194
+    },
+    {
+      "epoch": 0.9162817761937666,
+      "grad_norm": 0.0,
+      "learning_rate": 3.653480992697445e-07,
+      "loss": 0.778,
+      "step": 25195
+    },
+    {
+      "epoch": 0.9163181437975052,
+      "grad_norm": 0.0,
+      "learning_rate": 3.650326951934047e-07,
+      "loss": 0.9258,
+      "step": 25196
+    },
+    {
+      "epoch": 0.9163545114012438,
+      "grad_norm": 0.0,
+      "learning_rate": 3.647174247885654e-07,
+      "loss": 1.0107,
+      "step": 25197
+    },
+    {
+      "epoch": 0.9163908790049824,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6440228805960433e-07,
+      "loss": 0.9,
+      "step": 25198
+    },
+    {
+      "epoch": 0.916427246608721,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6408728501088673e-07,
+      "loss": 0.9346,
+      "step": 25199
+    },
+    {
+      "epoch": 0.9164636142124596,
+      "grad_norm": 0.0,
+      "learning_rate": 3.637724156467881e-07,
+      "loss": 0.9062,
+      "step": 25200
+    },
+    {
+      "epoch": 0.9164999818161982,
+      "grad_norm": 0.0,
+      "learning_rate": 3.634576799716749e-07,
+      "loss": 0.8281,
+      "step": 25201
+    },
+    {
+      "epoch": 0.9165363494199367,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6314307798991364e-07,
+      "loss": 0.8298,
+      "step": 25202
+    },
+    {
+      "epoch": 0.9165727170236753,
+      "grad_norm": 0.0,
+      "learning_rate": 3.628286097058686e-07,
+      "loss": 0.9281,
+      "step": 25203
+    },
+    {
+      "epoch": 0.9166090846274139,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6251427512390523e-07,
+      "loss": 0.9641,
+      "step": 25204
+    },
+    {
+      "epoch": 0.9166454522311525,
+      "grad_norm": 0.0,
+      "learning_rate": 3.622000742483811e-07,
+      "loss": 0.9715,
+      "step": 25205
+    },
+    {
+      "epoch": 0.916681819834891,
+      "grad_norm": 0.0,
+      "learning_rate": 3.618860070836572e-07,
+      "loss": 1.005,
+      "step": 25206
+    },
+    {
+      "epoch": 0.9167181874386297,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6157207363408995e-07,
+      "loss": 0.8413,
+      "step": 25207
+    },
+    {
+      "epoch": 0.9167545550423682,
+      "grad_norm": 0.0,
+      "learning_rate": 3.61258273904036e-07,
+      "loss": 0.9071,
+      "step": 25208
+    },
+    {
+      "epoch": 0.9167909226461068,
+      "grad_norm": 0.0,
+      "learning_rate": 3.609446078978451e-07,
+      "loss": 0.9243,
+      "step": 25209
+    },
+    {
+      "epoch": 0.9168272902498454,
+      "grad_norm": 0.0,
+      "learning_rate": 3.606310756198761e-07,
+      "loss": 0.8105,
+      "step": 25210
+    },
+    {
+      "epoch": 0.916863657853584,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6031767707447096e-07,
+      "loss": 0.8673,
+      "step": 25211
+    },
+    {
+      "epoch": 0.9169000254573226,
+      "grad_norm": 0.0,
+      "learning_rate": 3.600044122659818e-07,
+      "loss": 0.8657,
+      "step": 25212
+    },
+    {
+      "epoch": 0.9169363930610612,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5969128119875517e-07,
+      "loss": 0.7847,
+      "step": 25213
+    },
+    {
+      "epoch": 0.9169727606647998,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5937828387713316e-07,
+      "loss": 0.8744,
+      "step": 25214
+    },
+    {
+      "epoch": 0.9170091282685384,
+      "grad_norm": 0.0,
+      "learning_rate": 3.590654203054589e-07,
+      "loss": 0.9488,
+      "step": 25215
+    },
+    {
+      "epoch": 0.917045495872277,
+      "grad_norm": 0.0,
+      "learning_rate": 3.587526904880745e-07,
+      "loss": 0.8402,
+      "step": 25216
+    },
+    {
+      "epoch": 0.9170818634760156,
+      "grad_norm": 0.0,
+      "learning_rate": 3.584400944293176e-07,
+      "loss": 0.8751,
+      "step": 25217
+    },
+    {
+      "epoch": 0.9171182310797542,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5812763213352255e-07,
+      "loss": 0.8472,
+      "step": 25218
+    },
+    {
+      "epoch": 0.9171545986834927,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5781530360502914e-07,
+      "loss": 0.9503,
+      "step": 25219
+    },
+    {
+      "epoch": 0.9171909662872313,
+      "grad_norm": 0.0,
+      "learning_rate": 3.575031088481662e-07,
+      "loss": 0.9083,
+      "step": 25220
+    },
+    {
+      "epoch": 0.9172273338909699,
+      "grad_norm": 0.0,
+      "learning_rate": 3.57191047867268e-07,
+      "loss": 0.9788,
+      "step": 25221
+    },
+    {
+      "epoch": 0.9172637014947085,
+      "grad_norm": 0.0,
+      "learning_rate": 3.568791206666611e-07,
+      "loss": 0.9061,
+      "step": 25222
+    },
+    {
+      "epoch": 0.9173000690984471,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5656732725067645e-07,
+      "loss": 0.879,
+      "step": 25223
+    },
+    {
+      "epoch": 0.9173364367021857,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5625566762363615e-07,
+      "loss": 0.8387,
+      "step": 25224
+    },
+    {
+      "epoch": 0.9173728043059243,
+      "grad_norm": 0.0,
+      "learning_rate": 3.559441417898668e-07,
+      "loss": 0.8794,
+      "step": 25225
+    },
+    {
+      "epoch": 0.9174091719096629,
+      "grad_norm": 0.0,
+      "learning_rate": 3.556327497536893e-07,
+      "loss": 0.8955,
+      "step": 25226
+    },
+    {
+      "epoch": 0.9174455395134015,
+      "grad_norm": 0.0,
+      "learning_rate": 3.553214915194236e-07,
+      "loss": 0.963,
+      "step": 25227
+    },
+    {
+      "epoch": 0.9174819071171401,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5501036709138735e-07,
+      "loss": 0.8606,
+      "step": 25228
+    },
+    {
+      "epoch": 0.9175182747208787,
+      "grad_norm": 0.0,
+      "learning_rate": 3.546993764738982e-07,
+      "loss": 0.9215,
+      "step": 25229
+    },
+    {
+      "epoch": 0.9175546423246173,
+      "grad_norm": 0.0,
+      "learning_rate": 3.543885196712715e-07,
+      "loss": 0.7745,
+      "step": 25230
+    },
+    {
+      "epoch": 0.9175910099283558,
+      "grad_norm": 0.0,
+      "learning_rate": 3.540777966878184e-07,
+      "loss": 0.9285,
+      "step": 25231
+    },
+    {
+      "epoch": 0.9176273775320944,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5376720752784976e-07,
+      "loss": 0.9273,
+      "step": 25232
+    },
+    {
+      "epoch": 0.917663745135833,
+      "grad_norm": 0.0,
+      "learning_rate": 3.534567521956744e-07,
+      "loss": 0.8792,
+      "step": 25233
+    },
+    {
+      "epoch": 0.9177001127395716,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5314643069560006e-07,
+      "loss": 0.9706,
+      "step": 25234
+    },
+    {
+      "epoch": 0.9177364803433102,
+      "grad_norm": 0.0,
+      "learning_rate": 3.528362430319321e-07,
+      "loss": 0.8509,
+      "step": 25235
+    },
+    {
+      "epoch": 0.9177728479470487,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5252618920897376e-07,
+      "loss": 0.9016,
+      "step": 25236
+    },
+    {
+      "epoch": 0.9178092155507873,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5221626923102716e-07,
+      "loss": 0.8117,
+      "step": 25237
+    },
+    {
+      "epoch": 0.9178455831545259,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5190648310239107e-07,
+      "loss": 0.8687,
+      "step": 25238
+    },
+    {
+      "epoch": 0.9178819507582645,
+      "grad_norm": 0.0,
+      "learning_rate": 3.515968308273643e-07,
+      "loss": 0.9849,
+      "step": 25239
+    },
+    {
+      "epoch": 0.9179183183620031,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5128731241024115e-07,
+      "loss": 1.0615,
+      "step": 25240
+    },
+    {
+      "epoch": 0.9179546859657417,
+      "grad_norm": 0.0,
+      "learning_rate": 3.50977927855316e-07,
+      "loss": 0.8416,
+      "step": 25241
+    },
+    {
+      "epoch": 0.9179910535694803,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5066867716688325e-07,
+      "loss": 0.9289,
+      "step": 25242
+    },
+    {
+      "epoch": 0.9180274211732189,
+      "grad_norm": 0.0,
+      "learning_rate": 3.503595603492327e-07,
+      "loss": 0.9001,
+      "step": 25243
+    },
+    {
+      "epoch": 0.9180637887769575,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5005057740665095e-07,
+      "loss": 0.7706,
+      "step": 25244
+    },
+    {
+      "epoch": 0.9181001563806961,
+      "grad_norm": 0.0,
+      "learning_rate": 3.497417283434268e-07,
+      "loss": 0.9002,
+      "step": 25245
+    },
+    {
+      "epoch": 0.9181365239844347,
+      "grad_norm": 0.0,
+      "learning_rate": 3.494330131638435e-07,
+      "loss": 0.8474,
+      "step": 25246
+    },
+    {
+      "epoch": 0.9181728915881733,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4912443187218424e-07,
+      "loss": 0.8921,
+      "step": 25247
+    },
+    {
+      "epoch": 0.9182092591919119,
+      "grad_norm": 0.0,
+      "learning_rate": 3.488159844727312e-07,
+      "loss": 0.8118,
+      "step": 25248
+    },
+    {
+      "epoch": 0.9182456267956505,
+      "grad_norm": 0.0,
+      "learning_rate": 3.485076709697643e-07,
+      "loss": 0.8241,
+      "step": 25249
+    },
+    {
+      "epoch": 0.9182819943993891,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4819949136755906e-07,
+      "loss": 0.8547,
+      "step": 25250
+    },
+    {
+      "epoch": 0.9183183620031277,
+      "grad_norm": 0.0,
+      "learning_rate": 3.478914456703908e-07,
+      "loss": 0.8707,
+      "step": 25251
+    },
+    {
+      "epoch": 0.9183547296068661,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4758353388253617e-07,
+      "loss": 0.9472,
+      "step": 25252
+    },
+    {
+      "epoch": 0.9183910972106047,
+      "grad_norm": 0.0,
+      "learning_rate": 3.472757560082629e-07,
+      "loss": 0.8573,
+      "step": 25253
+    },
+    {
+      "epoch": 0.9184274648143433,
+      "grad_norm": 0.0,
+      "learning_rate": 3.469681120518442e-07,
+      "loss": 1.0089,
+      "step": 25254
+    },
+    {
+      "epoch": 0.9184638324180819,
+      "grad_norm": 0.0,
+      "learning_rate": 3.466606020175467e-07,
+      "loss": 0.9702,
+      "step": 25255
+    },
+    {
+      "epoch": 0.9185002000218205,
+      "grad_norm": 0.0,
+      "learning_rate": 3.46353225909638e-07,
+      "loss": 0.9166,
+      "step": 25256
+    },
+    {
+      "epoch": 0.9185365676255591,
+      "grad_norm": 0.0,
+      "learning_rate": 3.460459837323804e-07,
+      "loss": 0.7955,
+      "step": 25257
+    },
+    {
+      "epoch": 0.9185729352292977,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4573887549003817e-07,
+      "loss": 0.9538,
+      "step": 25258
+    },
+    {
+      "epoch": 0.9186093028330363,
+      "grad_norm": 0.0,
+      "learning_rate": 3.454319011868712e-07,
+      "loss": 1.0042,
+      "step": 25259
+    },
+    {
+      "epoch": 0.9186456704367749,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4512506082713836e-07,
+      "loss": 0.9139,
+      "step": 25260
+    },
+    {
+      "epoch": 0.9186820380405135,
+      "grad_norm": 0.0,
+      "learning_rate": 3.448183544150974e-07,
+      "loss": 0.9595,
+      "step": 25261
+    },
+    {
+      "epoch": 0.9187184056442521,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4451178195500255e-07,
+      "loss": 0.9305,
+      "step": 25262
+    },
+    {
+      "epoch": 0.9187547732479907,
+      "grad_norm": 0.0,
+      "learning_rate": 3.442053434511072e-07,
+      "loss": 0.8685,
+      "step": 25263
+    },
+    {
+      "epoch": 0.9187911408517293,
+      "grad_norm": 0.0,
+      "learning_rate": 3.438990389076624e-07,
+      "loss": 0.818,
+      "step": 25264
+    },
+    {
+      "epoch": 0.9188275084554679,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4359286832892025e-07,
+      "loss": 0.909,
+      "step": 25265
+    },
+    {
+      "epoch": 0.9188638760592065,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4328683171912403e-07,
+      "loss": 1.0505,
+      "step": 25266
+    },
+    {
+      "epoch": 0.9189002436629451,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4298092908252367e-07,
+      "loss": 0.911,
+      "step": 25267
+    },
+    {
+      "epoch": 0.9189366112666837,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4267516042336026e-07,
+      "loss": 0.9252,
+      "step": 25268
+    },
+    {
+      "epoch": 0.9189729788704222,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4236952574587814e-07,
+      "loss": 0.8897,
+      "step": 25269
+    },
+    {
+      "epoch": 0.9190093464741608,
+      "grad_norm": 0.0,
+      "learning_rate": 3.42064025054315e-07,
+      "loss": 0.866,
+      "step": 25270
+    },
+    {
+      "epoch": 0.9190457140778994,
+      "grad_norm": 0.0,
+      "learning_rate": 3.41758658352912e-07,
+      "loss": 0.8805,
+      "step": 25271
+    },
+    {
+      "epoch": 0.919082081681638,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4145342564590345e-07,
+      "loss": 0.9661,
+      "step": 25272
+    },
+    {
+      "epoch": 0.9191184492853766,
+      "grad_norm": 0.0,
+      "learning_rate": 3.41148326937526e-07,
+      "loss": 1.0786,
+      "step": 25273
+    },
+    {
+      "epoch": 0.9191548168891152,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4084336223201063e-07,
+      "loss": 0.8748,
+      "step": 25274
+    },
+    {
+      "epoch": 0.9191911844928538,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4053853153358964e-07,
+      "loss": 0.9542,
+      "step": 25275
+    },
+    {
+      "epoch": 0.9192275520965923,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4023383484649064e-07,
+      "loss": 0.9139,
+      "step": 25276
+    },
+    {
+      "epoch": 0.919263919700331,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3992927217494254e-07,
+      "loss": 0.9442,
+      "step": 25277
+    },
+    {
+      "epoch": 0.9193002873040695,
+      "grad_norm": 0.0,
+      "learning_rate": 3.396248435231697e-07,
+      "loss": 1.0103,
+      "step": 25278
+    },
+    {
+      "epoch": 0.9193366549078081,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3932054889539543e-07,
+      "loss": 0.9786,
+      "step": 25279
+    },
+    {
+      "epoch": 0.9193730225115467,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3901638829584193e-07,
+      "loss": 0.9786,
+      "step": 25280
+    },
+    {
+      "epoch": 0.9194093901152853,
+      "grad_norm": 0.0,
+      "learning_rate": 3.387123617287291e-07,
+      "loss": 0.9928,
+      "step": 25281
+    },
+    {
+      "epoch": 0.9194457577190239,
+      "grad_norm": 0.0,
+      "learning_rate": 3.384084691982759e-07,
+      "loss": 0.9503,
+      "step": 25282
+    },
+    {
+      "epoch": 0.9194821253227625,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3810471070869434e-07,
+      "loss": 0.9921,
+      "step": 25283
+    },
+    {
+      "epoch": 0.9195184929265011,
+      "grad_norm": 0.0,
+      "learning_rate": 3.378010862642045e-07,
+      "loss": 0.9586,
+      "step": 25284
+    },
+    {
+      "epoch": 0.9195548605302397,
+      "grad_norm": 0.0,
+      "learning_rate": 3.374975958690141e-07,
+      "loss": 0.8753,
+      "step": 25285
+    },
+    {
+      "epoch": 0.9195912281339782,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3719423952733533e-07,
+      "loss": 0.9115,
+      "step": 25286
+    },
+    {
+      "epoch": 0.9196275957377168,
+      "grad_norm": 0.0,
+      "learning_rate": 3.36891017243377e-07,
+      "loss": 0.9547,
+      "step": 25287
+    },
+    {
+      "epoch": 0.9196639633414554,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3658792902134584e-07,
+      "loss": 0.9488,
+      "step": 25288
+    },
+    {
+      "epoch": 0.919700330945194,
+      "grad_norm": 0.0,
+      "learning_rate": 3.362849748654451e-07,
+      "loss": 0.9048,
+      "step": 25289
+    },
+    {
+      "epoch": 0.9197366985489326,
+      "grad_norm": 0.0,
+      "learning_rate": 3.359821547798825e-07,
+      "loss": 0.9398,
+      "step": 25290
+    },
+    {
+      "epoch": 0.9197730661526712,
+      "grad_norm": 0.0,
+      "learning_rate": 3.356794687688536e-07,
+      "loss": 0.918,
+      "step": 25291
+    },
+    {
+      "epoch": 0.9198094337564098,
+      "grad_norm": 0.0,
+      "learning_rate": 3.353769168365606e-07,
+      "loss": 0.9464,
+      "step": 25292
+    },
+    {
+      "epoch": 0.9198458013601484,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3507449898720124e-07,
+      "loss": 0.7964,
+      "step": 25293
+    },
+    {
+      "epoch": 0.919882168963887,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3477221522496995e-07,
+      "loss": 0.9264,
+      "step": 25294
+    },
+    {
+      "epoch": 0.9199185365676256,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3447006555406117e-07,
+      "loss": 0.8228,
+      "step": 25295
+    },
+    {
+      "epoch": 0.9199549041713642,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3416804997866483e-07,
+      "loss": 0.8493,
+      "step": 25296
+    },
+    {
+      "epoch": 0.9199912717751028,
+      "grad_norm": 0.0,
+      "learning_rate": 3.338661685029765e-07,
+      "loss": 0.8483,
+      "step": 25297
+    },
+    {
+      "epoch": 0.9200276393788414,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3356442113117727e-07,
+      "loss": 0.9167,
+      "step": 25298
+    },
+    {
+      "epoch": 0.92006400698258,
+      "grad_norm": 0.0,
+      "learning_rate": 3.332628078674582e-07,
+      "loss": 0.7638,
+      "step": 25299
+    },
+    {
+      "epoch": 0.9201003745863185,
+      "grad_norm": 0.0,
+      "learning_rate": 3.329613287160027e-07,
+      "loss": 0.9383,
+      "step": 25300
+    },
+    {
+      "epoch": 0.9201367421900571,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3265998368099295e-07,
+      "loss": 0.978,
+      "step": 25301
+    },
+    {
+      "epoch": 0.9201731097937957,
+      "grad_norm": 0.0,
+      "learning_rate": 3.323587727666078e-07,
+      "loss": 0.8592,
+      "step": 25302
+    },
+    {
+      "epoch": 0.9202094773975342,
+      "grad_norm": 0.0,
+      "learning_rate": 3.320576959770316e-07,
+      "loss": 0.854,
+      "step": 25303
+    },
+    {
+      "epoch": 0.9202458450012728,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3175675331643454e-07,
+      "loss": 0.9309,
+      "step": 25304
+    },
+    {
+      "epoch": 0.9202822126050114,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3145594478899646e-07,
+      "loss": 0.9245,
+      "step": 25305
+    },
+    {
+      "epoch": 0.92031858020875,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3115527039888963e-07,
+      "loss": 0.7641,
+      "step": 25306
+    },
+    {
+      "epoch": 0.9203549478124886,
+      "grad_norm": 0.0,
+      "learning_rate": 3.308547301502851e-07,
+      "loss": 0.8971,
+      "step": 25307
+    },
+    {
+      "epoch": 0.9203913154162272,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3055432404735075e-07,
+      "loss": 0.809,
+      "step": 25308
+    },
+    {
+      "epoch": 0.9204276830199658,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3025405209425763e-07,
+      "loss": 0.9436,
+      "step": 25309
+    },
+    {
+      "epoch": 0.9204640506237044,
+      "grad_norm": 0.0,
+      "learning_rate": 3.299539142951702e-07,
+      "loss": 0.7892,
+      "step": 25310
+    },
+    {
+      "epoch": 0.920500418227443,
+      "grad_norm": 0.0,
+      "learning_rate": 3.296539106542518e-07,
+      "loss": 0.9471,
+      "step": 25311
+    },
+    {
+      "epoch": 0.9205367858311816,
+      "grad_norm": 0.0,
+      "learning_rate": 3.293540411756657e-07,
+      "loss": 1.0033,
+      "step": 25312
+    },
+    {
+      "epoch": 0.9205731534349202,
+      "grad_norm": 0.0,
+      "learning_rate": 3.290543058635709e-07,
+      "loss": 0.9377,
+      "step": 25313
+    },
+    {
+      "epoch": 0.9206095210386588,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2875470472212623e-07,
+      "loss": 0.9622,
+      "step": 25314
+    },
+    {
+      "epoch": 0.9206458886423974,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2845523775548727e-07,
+      "loss": 0.9957,
+      "step": 25315
+    },
+    {
+      "epoch": 0.920682256246136,
+      "grad_norm": 0.0,
+      "learning_rate": 3.281559049678118e-07,
+      "loss": 0.9157,
+      "step": 25316
+    },
+    {
+      "epoch": 0.9207186238498746,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2785670636324874e-07,
+      "loss": 0.9049,
+      "step": 25317
+    },
+    {
+      "epoch": 0.9207549914536132,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2755764194595144e-07,
+      "loss": 0.8033,
+      "step": 25318
+    },
+    {
+      "epoch": 0.9207913590573518,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2725871172006876e-07,
+      "loss": 0.8851,
+      "step": 25319
+    },
+    {
+      "epoch": 0.9208277266610903,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2695991568974737e-07,
+      "loss": 0.8988,
+      "step": 25320
+    },
+    {
+      "epoch": 0.9208640942648288,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2666125385913295e-07,
+      "loss": 0.9449,
+      "step": 25321
+    },
+    {
+      "epoch": 0.9209004618685674,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2636272623236984e-07,
+      "loss": 0.9823,
+      "step": 25322
+    },
+    {
+      "epoch": 0.920936829472306,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2606433281359705e-07,
+      "loss": 0.8968,
+      "step": 25323
+    },
+    {
+      "epoch": 0.9209731970760446,
+      "grad_norm": 0.0,
+      "learning_rate": 3.257660736069579e-07,
+      "loss": 0.9363,
+      "step": 25324
+    },
+    {
+      "epoch": 0.9210095646797832,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2546794861658793e-07,
+      "loss": 0.908,
+      "step": 25325
+    },
+    {
+      "epoch": 0.9210459322835218,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2516995784662277e-07,
+      "loss": 0.9136,
+      "step": 25326
+    },
+    {
+      "epoch": 0.9210822998872604,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2487210130119795e-07,
+      "loss": 0.9241,
+      "step": 25327
+    },
+    {
+      "epoch": 0.921118667490999,
+      "grad_norm": 0.0,
+      "learning_rate": 3.245743789844469e-07,
+      "loss": 0.8995,
+      "step": 25328
+    },
+    {
+      "epoch": 0.9211550350947376,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2427679090049734e-07,
+      "loss": 0.8579,
+      "step": 25329
+    },
+    {
+      "epoch": 0.9211914026984762,
+      "grad_norm": 0.0,
+      "learning_rate": 3.239793370534805e-07,
+      "loss": 0.8005,
+      "step": 25330
+    },
+    {
+      "epoch": 0.9212277703022148,
+      "grad_norm": 0.0,
+      "learning_rate": 3.236820174475197e-07,
+      "loss": 0.881,
+      "step": 25331
+    },
+    {
+      "epoch": 0.9212641379059534,
+      "grad_norm": 0.0,
+      "learning_rate": 3.23384832086745e-07,
+      "loss": 0.9479,
+      "step": 25332
+    },
+    {
+      "epoch": 0.921300505509692,
+      "grad_norm": 0.0,
+      "learning_rate": 3.230877809752753e-07,
+      "loss": 0.9655,
+      "step": 25333
+    },
+    {
+      "epoch": 0.9213368731134306,
+      "grad_norm": 0.0,
+      "learning_rate": 3.227908641172317e-07,
+      "loss": 0.9684,
+      "step": 25334
+    },
+    {
+      "epoch": 0.9213732407171692,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2249408151673653e-07,
+      "loss": 0.8428,
+      "step": 25335
+    },
+    {
+      "epoch": 0.9214096083209077,
+      "grad_norm": 0.0,
+      "learning_rate": 3.221974331779032e-07,
+      "loss": 1.0176,
+      "step": 25336
+    },
+    {
+      "epoch": 0.9214459759246463,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2190091910485054e-07,
+      "loss": 0.9091,
+      "step": 25337
+    },
+    {
+      "epoch": 0.9214823435283849,
+      "grad_norm": 0.0,
+      "learning_rate": 3.21604539301692e-07,
+      "loss": 1.0495,
+      "step": 25338
+    },
+    {
+      "epoch": 0.9215187111321235,
+      "grad_norm": 0.0,
+      "learning_rate": 3.213082937725376e-07,
+      "loss": 0.8552,
+      "step": 25339
+    },
+    {
+      "epoch": 0.9215550787358621,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2101218252149845e-07,
+      "loss": 0.9422,
+      "step": 25340
+    },
+    {
+      "epoch": 0.9215914463396007,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2071620555268246e-07,
+      "loss": 0.8768,
+      "step": 25341
+    },
+    {
+      "epoch": 0.9216278139433393,
+      "grad_norm": 0.0,
+      "learning_rate": 3.204203628701963e-07,
+      "loss": 0.843,
+      "step": 25342
+    },
+    {
+      "epoch": 0.9216641815470779,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2012465447814447e-07,
+      "loss": 0.9243,
+      "step": 25343
+    },
+    {
+      "epoch": 0.9217005491508165,
+      "grad_norm": 0.0,
+      "learning_rate": 3.198290803806292e-07,
+      "loss": 0.9467,
+      "step": 25344
+    },
+    {
+      "epoch": 0.921736916754555,
+      "grad_norm": 0.0,
+      "learning_rate": 3.195336405817506e-07,
+      "loss": 0.9695,
+      "step": 25345
+    },
+    {
+      "epoch": 0.9217732843582936,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1923833508560765e-07,
+      "loss": 0.939,
+      "step": 25346
+    },
+    {
+      "epoch": 0.9218096519620322,
+      "grad_norm": 0.0,
+      "learning_rate": 3.189431638962981e-07,
+      "loss": 0.886,
+      "step": 25347
+    },
+    {
+      "epoch": 0.9218460195657708,
+      "grad_norm": 0.0,
+      "learning_rate": 3.186481270179165e-07,
+      "loss": 0.8631,
+      "step": 25348
+    },
+    {
+      "epoch": 0.9218823871695094,
+      "grad_norm": 0.0,
+      "learning_rate": 3.183532244545562e-07,
+      "loss": 0.9769,
+      "step": 25349
+    },
+    {
+      "epoch": 0.921918754773248,
+      "grad_norm": 0.0,
+      "learning_rate": 3.180584562103073e-07,
+      "loss": 0.9232,
+      "step": 25350
+    },
+    {
+      "epoch": 0.9219551223769866,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1776382228926207e-07,
+      "loss": 0.8067,
+      "step": 25351
+    },
+    {
+      "epoch": 0.9219914899807252,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1746932269550614e-07,
+      "loss": 1.0129,
+      "step": 25352
+    },
+    {
+      "epoch": 0.9220278575844637,
+      "grad_norm": 0.0,
+      "learning_rate": 3.171749574331251e-07,
+      "loss": 0.9016,
+      "step": 25353
+    },
+    {
+      "epoch": 0.9220642251882023,
+      "grad_norm": 0.0,
+      "learning_rate": 3.168807265062035e-07,
+      "loss": 1.0162,
+      "step": 25354
+    },
+    {
+      "epoch": 0.9221005927919409,
+      "grad_norm": 0.0,
+      "learning_rate": 3.165866299188236e-07,
+      "loss": 0.8947,
+      "step": 25355
+    },
+    {
+      "epoch": 0.9221369603956795,
+      "grad_norm": 0.0,
+      "learning_rate": 3.162926676750655e-07,
+      "loss": 0.9464,
+      "step": 25356
+    },
+    {
+      "epoch": 0.9221733279994181,
+      "grad_norm": 0.0,
+      "learning_rate": 3.159988397790059e-07,
+      "loss": 0.7784,
+      "step": 25357
+    },
+    {
+      "epoch": 0.9222096956031567,
+      "grad_norm": 0.0,
+      "learning_rate": 3.157051462347238e-07,
+      "loss": 0.8668,
+      "step": 25358
+    },
+    {
+      "epoch": 0.9222460632068953,
+      "grad_norm": 0.0,
+      "learning_rate": 3.154115870462915e-07,
+      "loss": 0.8688,
+      "step": 25359
+    },
+    {
+      "epoch": 0.9222824308106339,
+      "grad_norm": 0.0,
+      "learning_rate": 3.151181622177835e-07,
+      "loss": 0.8046,
+      "step": 25360
+    },
+    {
+      "epoch": 0.9223187984143725,
+      "grad_norm": 0.0,
+      "learning_rate": 3.148248717532709e-07,
+      "loss": 0.8755,
+      "step": 25361
+    },
+    {
+      "epoch": 0.9223551660181111,
+      "grad_norm": 0.0,
+      "learning_rate": 3.145317156568206e-07,
+      "loss": 0.8582,
+      "step": 25362
+    },
+    {
+      "epoch": 0.9223915336218497,
+      "grad_norm": 0.0,
+      "learning_rate": 3.142386939325004e-07,
+      "loss": 1.0149,
+      "step": 25363
+    },
+    {
+      "epoch": 0.9224279012255883,
+      "grad_norm": 0.0,
+      "learning_rate": 3.139458065843781e-07,
+      "loss": 0.8592,
+      "step": 25364
+    },
+    {
+      "epoch": 0.9224642688293269,
+      "grad_norm": 0.0,
+      "learning_rate": 3.136530536165128e-07,
+      "loss": 0.9397,
+      "step": 25365
+    },
+    {
+      "epoch": 0.9225006364330655,
+      "grad_norm": 0.0,
+      "learning_rate": 3.133604350329689e-07,
+      "loss": 0.9194,
+      "step": 25366
+    },
+    {
+      "epoch": 0.922537004036804,
+      "grad_norm": 0.0,
+      "learning_rate": 3.130679508378054e-07,
+      "loss": 0.8432,
+      "step": 25367
+    },
+    {
+      "epoch": 0.9225733716405426,
+      "grad_norm": 0.0,
+      "learning_rate": 3.127756010350813e-07,
+      "loss": 1.0336,
+      "step": 25368
+    },
+    {
+      "epoch": 0.9226097392442812,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1248338562884896e-07,
+      "loss": 0.8555,
+      "step": 25369
+    },
+    {
+      "epoch": 0.9226461068480197,
+      "grad_norm": 0.0,
+      "learning_rate": 3.121913046231673e-07,
+      "loss": 1.0195,
+      "step": 25370
+    },
+    {
+      "epoch": 0.9226824744517583,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1189935802208525e-07,
+      "loss": 0.8832,
+      "step": 25371
+    },
+    {
+      "epoch": 0.9227188420554969,
+      "grad_norm": 0.0,
+      "learning_rate": 3.11607545829653e-07,
+      "loss": 0.8964,
+      "step": 25372
+    },
+    {
+      "epoch": 0.9227552096592355,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1131586804992174e-07,
+      "loss": 0.9364,
+      "step": 25373
+    },
+    {
+      "epoch": 0.9227915772629741,
+      "grad_norm": 0.0,
+      "learning_rate": 3.110243246869349e-07,
+      "loss": 0.9724,
+      "step": 25374
+    },
+    {
+      "epoch": 0.9228279448667127,
+      "grad_norm": 0.0,
+      "learning_rate": 3.107329157447403e-07,
+      "loss": 0.9491,
+      "step": 25375
+    },
+    {
+      "epoch": 0.9228643124704513,
+      "grad_norm": 0.0,
+      "learning_rate": 3.104416412273781e-07,
+      "loss": 0.8856,
+      "step": 25376
+    },
+    {
+      "epoch": 0.9229006800741899,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1015050113889056e-07,
+      "loss": 0.8755,
+      "step": 25377
+    },
+    {
+      "epoch": 0.9229370476779285,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0985949548331564e-07,
+      "loss": 0.8726,
+      "step": 25378
+    },
+    {
+      "epoch": 0.9229734152816671,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0956862426469337e-07,
+      "loss": 0.8114,
+      "step": 25379
+    },
+    {
+      "epoch": 0.9230097828854057,
+      "grad_norm": 0.0,
+      "learning_rate": 3.092778874870572e-07,
+      "loss": 0.9748,
+      "step": 25380
+    },
+    {
+      "epoch": 0.9230461504891443,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0898728515444064e-07,
+      "loss": 0.9269,
+      "step": 25381
+    },
+    {
+      "epoch": 0.9230825180928829,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0869681727087484e-07,
+      "loss": 0.8348,
+      "step": 25382
+    },
+    {
+      "epoch": 0.9231188856966215,
+      "grad_norm": 0.0,
+      "learning_rate": 3.084064838403922e-07,
+      "loss": 0.8788,
+      "step": 25383
+    },
+    {
+      "epoch": 0.9231552533003601,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0811628486701827e-07,
+      "loss": 0.9197,
+      "step": 25384
+    },
+    {
+      "epoch": 0.9231916209040987,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0782622035477993e-07,
+      "loss": 0.8865,
+      "step": 25385
+    },
+    {
+      "epoch": 0.9232279885078373,
+      "grad_norm": 0.0,
+      "learning_rate": 3.075362903077017e-07,
+      "loss": 0.929,
+      "step": 25386
+    },
+    {
+      "epoch": 0.9232643561115758,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0724649472980485e-07,
+      "loss": 0.8665,
+      "step": 25387
+    },
+    {
+      "epoch": 0.9233007237153144,
+      "grad_norm": 0.0,
+      "learning_rate": 3.069568336251105e-07,
+      "loss": 0.8156,
+      "step": 25388
+    },
+    {
+      "epoch": 0.923337091319053,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0666730699764e-07,
+      "loss": 0.9173,
+      "step": 25389
+    },
+    {
+      "epoch": 0.9233734589227915,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0637791485140567e-07,
+      "loss": 0.8427,
+      "step": 25390
+    },
+    {
+      "epoch": 0.9234098265265301,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0608865719042313e-07,
+      "loss": 0.8034,
+      "step": 25391
+    },
+    {
+      "epoch": 0.9234461941302687,
+      "grad_norm": 0.0,
+      "learning_rate": 3.057995340187081e-07,
+      "loss": 1.0139,
+      "step": 25392
+    },
+    {
+      "epoch": 0.9234825617340073,
+      "grad_norm": 0.0,
+      "learning_rate": 3.055105453402707e-07,
+      "loss": 0.9562,
+      "step": 25393
+    },
+    {
+      "epoch": 0.9235189293377459,
+      "grad_norm": 0.0,
+      "learning_rate": 3.052216911591188e-07,
+      "loss": 0.8435,
+      "step": 25394
+    },
+    {
+      "epoch": 0.9235552969414845,
+      "grad_norm": 0.0,
+      "learning_rate": 3.049329714792604e-07,
+      "loss": 0.9085,
+      "step": 25395
+    },
+    {
+      "epoch": 0.9235916645452231,
+      "grad_norm": 0.0,
+      "learning_rate": 3.046443863047033e-07,
+      "loss": 1.0035,
+      "step": 25396
+    },
+    {
+      "epoch": 0.9236280321489617,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0435593563944765e-07,
+      "loss": 0.7575,
+      "step": 25397
+    },
+    {
+      "epoch": 0.9236643997527003,
+      "grad_norm": 0.0,
+      "learning_rate": 3.040676194874981e-07,
+      "loss": 0.9094,
+      "step": 25398
+    },
+    {
+      "epoch": 0.9237007673564389,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0377943785285355e-07,
+      "loss": 0.9664,
+      "step": 25399
+    },
+    {
+      "epoch": 0.9237371349601775,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0349139073951205e-07,
+      "loss": 0.9735,
+      "step": 25400
+    },
+    {
+      "epoch": 0.9237735025639161,
+      "grad_norm": 0.0,
+      "learning_rate": 3.032034781514681e-07,
+      "loss": 0.949,
+      "step": 25401
+    },
+    {
+      "epoch": 0.9238098701676547,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0291570009272187e-07,
+      "loss": 0.9566,
+      "step": 25402
+    },
+    {
+      "epoch": 0.9238462377713933,
+      "grad_norm": 0.0,
+      "learning_rate": 3.026280565672579e-07,
+      "loss": 1.0105,
+      "step": 25403
+    },
+    {
+      "epoch": 0.9238826053751318,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0234054757907303e-07,
+      "loss": 0.9713,
+      "step": 25404
+    },
+    {
+      "epoch": 0.9239189729788704,
+      "grad_norm": 0.0,
+      "learning_rate": 3.020531731321541e-07,
+      "loss": 0.9977,
+      "step": 25405
+    },
+    {
+      "epoch": 0.923955340582609,
+      "grad_norm": 0.0,
+      "learning_rate": 3.017659332304879e-07,
+      "loss": 0.9113,
+      "step": 25406
+    },
+    {
+      "epoch": 0.9239917081863476,
+      "grad_norm": 0.0,
+      "learning_rate": 3.014788278780578e-07,
+      "loss": 1.0263,
+      "step": 25407
+    },
+    {
+      "epoch": 0.9240280757900862,
+      "grad_norm": 0.0,
+      "learning_rate": 3.011918570788497e-07,
+      "loss": 0.8859,
+      "step": 25408
+    },
+    {
+      "epoch": 0.9240644433938248,
+      "grad_norm": 0.0,
+      "learning_rate": 3.009050208368447e-07,
+      "loss": 0.792,
+      "step": 25409
+    },
+    {
+      "epoch": 0.9241008109975634,
+      "grad_norm": 0.0,
+      "learning_rate": 3.006183191560197e-07,
+      "loss": 0.8605,
+      "step": 25410
+    },
+    {
+      "epoch": 0.924137178601302,
+      "grad_norm": 0.0,
+      "learning_rate": 3.003317520403537e-07,
+      "loss": 0.9473,
+      "step": 25411
+    },
+    {
+      "epoch": 0.9241735462050406,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0004531949382355e-07,
+      "loss": 0.9113,
+      "step": 25412
+    },
+    {
+      "epoch": 0.9242099138087791,
+      "grad_norm": 0.0,
+      "learning_rate": 2.997590215204027e-07,
+      "loss": 0.8723,
+      "step": 25413
+    },
+    {
+      "epoch": 0.9242462814125177,
+      "grad_norm": 0.0,
+      "learning_rate": 2.994728581240602e-07,
+      "loss": 0.9089,
+      "step": 25414
+    },
+    {
+      "epoch": 0.9242826490162563,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9918682930877184e-07,
+      "loss": 0.7326,
+      "step": 25415
+    },
+    {
+      "epoch": 0.9243190166199949,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9890093507849996e-07,
+      "loss": 0.925,
+      "step": 25416
+    },
+    {
+      "epoch": 0.9243553842237335,
+      "grad_norm": 0.0,
+      "learning_rate": 2.986151754372146e-07,
+      "loss": 1.0009,
+      "step": 25417
+    },
+    {
+      "epoch": 0.9243917518274721,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9832955038887945e-07,
+      "loss": 0.767,
+      "step": 25418
+    },
+    {
+      "epoch": 0.9244281194312107,
+      "grad_norm": 0.0,
+      "learning_rate": 2.980440599374568e-07,
+      "loss": 0.9819,
+      "step": 25419
+    },
+    {
+      "epoch": 0.9244644870349493,
+      "grad_norm": 0.0,
+      "learning_rate": 2.977587040869057e-07,
+      "loss": 0.8137,
+      "step": 25420
+    },
+    {
+      "epoch": 0.9245008546386878,
+      "grad_norm": 0.0,
+      "learning_rate": 2.974734828411907e-07,
+      "loss": 0.9396,
+      "step": 25421
+    },
+    {
+      "epoch": 0.9245372222424264,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9718839620426097e-07,
+      "loss": 0.7938,
+      "step": 25422
+    },
+    {
+      "epoch": 0.924573589846165,
+      "grad_norm": 0.0,
+      "learning_rate": 2.969034441800789e-07,
+      "loss": 0.7895,
+      "step": 25423
+    },
+    {
+      "epoch": 0.9246099574499036,
+      "grad_norm": 0.0,
+      "learning_rate": 2.966186267725935e-07,
+      "loss": 0.9148,
+      "step": 25424
+    },
+    {
+      "epoch": 0.9246463250536422,
+      "grad_norm": 0.0,
+      "learning_rate": 2.963339439857571e-07,
+      "loss": 0.8361,
+      "step": 25425
+    },
+    {
+      "epoch": 0.9246826926573808,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9604939582351887e-07,
+      "loss": 0.8974,
+      "step": 25426
+    },
+    {
+      "epoch": 0.9247190602611194,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9576498228982785e-07,
+      "loss": 0.8286,
+      "step": 25427
+    },
+    {
+      "epoch": 0.924755427864858,
+      "grad_norm": 0.0,
+      "learning_rate": 2.954807033886309e-07,
+      "loss": 0.9392,
+      "step": 25428
+    },
+    {
+      "epoch": 0.9247917954685966,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9519655912386704e-07,
+      "loss": 0.9322,
+      "step": 25429
+    },
+    {
+      "epoch": 0.9248281630723352,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9491254949948423e-07,
+      "loss": 0.9232,
+      "step": 25430
+    },
+    {
+      "epoch": 0.9248645306760738,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9462867451941934e-07,
+      "loss": 0.9962,
+      "step": 25431
+    },
+    {
+      "epoch": 0.9249008982798124,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9434493418761036e-07,
+      "loss": 0.8952,
+      "step": 25432
+    },
+    {
+      "epoch": 0.924937265883551,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9406132850799517e-07,
+      "loss": 1.0045,
+      "step": 25433
+    },
+    {
+      "epoch": 0.9249736334872896,
+      "grad_norm": 0.0,
+      "learning_rate": 2.937778574845107e-07,
+      "loss": 0.958,
+      "step": 25434
+    },
+    {
+      "epoch": 0.9250100010910282,
+      "grad_norm": 0.0,
+      "learning_rate": 2.934945211210838e-07,
+      "loss": 1.0108,
+      "step": 25435
+    },
+    {
+      "epoch": 0.9250463686947668,
+      "grad_norm": 0.0,
+      "learning_rate": 2.932113194216502e-07,
+      "loss": 0.8536,
+      "step": 25436
+    },
+    {
+      "epoch": 0.9250827362985052,
+      "grad_norm": 0.0,
+      "learning_rate": 2.929282523901378e-07,
+      "loss": 0.9392,
+      "step": 25437
+    },
+    {
+      "epoch": 0.9251191039022438,
+      "grad_norm": 0.0,
+      "learning_rate": 2.926453200304724e-07,
+      "loss": 0.9983,
+      "step": 25438
+    },
+    {
+      "epoch": 0.9251554715059824,
+      "grad_norm": 0.0,
+      "learning_rate": 2.923625223465798e-07,
+      "loss": 0.8716,
+      "step": 25439
+    },
+    {
+      "epoch": 0.925191839109721,
+      "grad_norm": 0.0,
+      "learning_rate": 2.920798593423846e-07,
+      "loss": 0.9666,
+      "step": 25440
+    },
+    {
+      "epoch": 0.9252282067134596,
+      "grad_norm": 0.0,
+      "learning_rate": 2.917973310218081e-07,
+      "loss": 0.8746,
+      "step": 25441
+    },
+    {
+      "epoch": 0.9252645743171982,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9151493738876934e-07,
+      "loss": 0.8905,
+      "step": 25442
+    },
+    {
+      "epoch": 0.9253009419209368,
+      "grad_norm": 0.0,
+      "learning_rate": 2.912326784471864e-07,
+      "loss": 0.7982,
+      "step": 25443
+    },
+    {
+      "epoch": 0.9253373095246754,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9095055420097497e-07,
+      "loss": 0.9895,
+      "step": 25444
+    },
+    {
+      "epoch": 0.925373677128414,
+      "grad_norm": 0.0,
+      "learning_rate": 2.906685646540486e-07,
+      "loss": 0.8893,
+      "step": 25445
+    },
+    {
+      "epoch": 0.9254100447321526,
+      "grad_norm": 0.0,
+      "learning_rate": 2.903867098103208e-07,
+      "loss": 0.9053,
+      "step": 25446
+    },
+    {
+      "epoch": 0.9254464123358912,
+      "grad_norm": 0.0,
+      "learning_rate": 2.901049896737018e-07,
+      "loss": 1.0164,
+      "step": 25447
+    },
+    {
+      "epoch": 0.9254827799396298,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8982340424809854e-07,
+      "loss": 1.0243,
+      "step": 25448
+    },
+    {
+      "epoch": 0.9255191475433684,
+      "grad_norm": 0.0,
+      "learning_rate": 2.895419535374189e-07,
+      "loss": 0.8487,
+      "step": 25449
+    },
+    {
+      "epoch": 0.925555515147107,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8926063754556753e-07,
+      "loss": 1.0472,
+      "step": 25450
+    },
+    {
+      "epoch": 0.9255918827508456,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8897945627644696e-07,
+      "loss": 0.8572,
+      "step": 25451
+    },
+    {
+      "epoch": 0.9256282503545842,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8869840973395737e-07,
+      "loss": 0.9262,
+      "step": 25452
+    },
+    {
+      "epoch": 0.9256646179583228,
+      "grad_norm": 0.0,
+      "learning_rate": 2.884174979219989e-07,
+      "loss": 0.9765,
+      "step": 25453
+    },
+    {
+      "epoch": 0.9257009855620613,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8813672084446966e-07,
+      "loss": 0.7846,
+      "step": 25454
+    },
+    {
+      "epoch": 0.9257373531657999,
+      "grad_norm": 0.0,
+      "learning_rate": 2.878560785052642e-07,
+      "loss": 0.9382,
+      "step": 25455
+    },
+    {
+      "epoch": 0.9257737207695385,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8757557090827614e-07,
+      "loss": 0.7915,
+      "step": 25456
+    },
+    {
+      "epoch": 0.925810088373277,
+      "grad_norm": 0.0,
+      "learning_rate": 2.872951980573968e-07,
+      "loss": 0.7672,
+      "step": 25457
+    },
+    {
+      "epoch": 0.9258464559770156,
+      "grad_norm": 0.0,
+      "learning_rate": 2.870149599565142e-07,
+      "loss": 0.9182,
+      "step": 25458
+    },
+    {
+      "epoch": 0.9258828235807542,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8673485660951963e-07,
+      "loss": 0.9143,
+      "step": 25459
+    },
+    {
+      "epoch": 0.9259191911844928,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8645488802029774e-07,
+      "loss": 0.9085,
+      "step": 25460
+    },
+    {
+      "epoch": 0.9259555587882314,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8617505419273217e-07,
+      "loss": 0.941,
+      "step": 25461
+    },
+    {
+      "epoch": 0.92599192639197,
+      "grad_norm": 0.0,
+      "learning_rate": 2.858953551307053e-07,
+      "loss": 1.0062,
+      "step": 25462
+    },
+    {
+      "epoch": 0.9260282939957086,
+      "grad_norm": 0.0,
+      "learning_rate": 2.856157908380997e-07,
+      "loss": 0.8641,
+      "step": 25463
+    },
+    {
+      "epoch": 0.9260646615994472,
+      "grad_norm": 0.0,
+      "learning_rate": 2.853363613187898e-07,
+      "loss": 1.0631,
+      "step": 25464
+    },
+    {
+      "epoch": 0.9261010292031858,
+      "grad_norm": 0.0,
+      "learning_rate": 2.85057066576655e-07,
+      "loss": 0.8223,
+      "step": 25465
+    },
+    {
+      "epoch": 0.9261373968069244,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8477790661557205e-07,
+      "loss": 0.9341,
+      "step": 25466
+    },
+    {
+      "epoch": 0.926173764410663,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8449888143940893e-07,
+      "loss": 0.9469,
+      "step": 25467
+    },
+    {
+      "epoch": 0.9262101320144016,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8421999105204044e-07,
+      "loss": 0.9317,
+      "step": 25468
+    },
+    {
+      "epoch": 0.9262464996181402,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8394123545733455e-07,
+      "loss": 0.8856,
+      "step": 25469
+    },
+    {
+      "epoch": 0.9262828672218788,
+      "grad_norm": 0.0,
+      "learning_rate": 2.836626146591581e-07,
+      "loss": 0.9559,
+      "step": 25470
+    },
+    {
+      "epoch": 0.9263192348256173,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8338412866137696e-07,
+      "loss": 0.9198,
+      "step": 25471
+    },
+    {
+      "epoch": 0.9263556024293559,
+      "grad_norm": 0.0,
+      "learning_rate": 2.831057774678558e-07,
+      "loss": 0.8381,
+      "step": 25472
+    },
+    {
+      "epoch": 0.9263919700330945,
+      "grad_norm": 0.0,
+      "learning_rate": 2.82827561082456e-07,
+      "loss": 0.9125,
+      "step": 25473
+    },
+    {
+      "epoch": 0.9264283376368331,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8254947950903553e-07,
+      "loss": 0.872,
+      "step": 25474
+    },
+    {
+      "epoch": 0.9264647052405717,
+      "grad_norm": 0.0,
+      "learning_rate": 2.822715327514536e-07,
+      "loss": 1.0048,
+      "step": 25475
+    },
+    {
+      "epoch": 0.9265010728443103,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8199372081356926e-07,
+      "loss": 0.8395,
+      "step": 25476
+    },
+    {
+      "epoch": 0.9265374404480489,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8171604369923057e-07,
+      "loss": 0.8791,
+      "step": 25477
+    },
+    {
+      "epoch": 0.9265738080517875,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8143850141229554e-07,
+      "loss": 0.8918,
+      "step": 25478
+    },
+    {
+      "epoch": 0.9266101756555261,
+      "grad_norm": 0.0,
+      "learning_rate": 2.811610939566123e-07,
+      "loss": 0.9494,
+      "step": 25479
+    },
+    {
+      "epoch": 0.9266465432592647,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8088382133602877e-07,
+      "loss": 0.9358,
+      "step": 25480
+    },
+    {
+      "epoch": 0.9266829108630033,
+      "grad_norm": 0.0,
+      "learning_rate": 2.806066835543919e-07,
+      "loss": 0.8914,
+      "step": 25481
+    },
+    {
+      "epoch": 0.9267192784667418,
+      "grad_norm": 0.0,
+      "learning_rate": 2.803296806155498e-07,
+      "loss": 0.9362,
+      "step": 25482
+    },
+    {
+      "epoch": 0.9267556460704804,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8005281252334037e-07,
+      "loss": 0.7971,
+      "step": 25483
+    },
+    {
+      "epoch": 0.926792013674219,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7977607928160955e-07,
+      "loss": 0.8521,
+      "step": 25484
+    },
+    {
+      "epoch": 0.9268283812779576,
+      "grad_norm": 0.0,
+      "learning_rate": 2.794994808941931e-07,
+      "loss": 0.9277,
+      "step": 25485
+    },
+    {
+      "epoch": 0.9268647488816962,
+      "grad_norm": 0.0,
+      "learning_rate": 2.792230173649302e-07,
+      "loss": 0.917,
+      "step": 25486
+    },
+    {
+      "epoch": 0.9269011164854348,
+      "grad_norm": 0.0,
+      "learning_rate": 2.789466886976566e-07,
+      "loss": 0.9049,
+      "step": 25487
+    },
+    {
+      "epoch": 0.9269374840891733,
+      "grad_norm": 0.0,
+      "learning_rate": 2.786704948962049e-07,
+      "loss": 0.9099,
+      "step": 25488
+    },
+    {
+      "epoch": 0.9269738516929119,
+      "grad_norm": 0.0,
+      "learning_rate": 2.783944359644064e-07,
+      "loss": 0.9155,
+      "step": 25489
+    },
+    {
+      "epoch": 0.9270102192966505,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7811851190609253e-07,
+      "loss": 0.978,
+      "step": 25490
+    },
+    {
+      "epoch": 0.9270465869003891,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7784272272509015e-07,
+      "loss": 1.0591,
+      "step": 25491
+    },
+    {
+      "epoch": 0.9270829545041277,
+      "grad_norm": 0.0,
+      "learning_rate": 2.775670684252274e-07,
+      "loss": 0.9313,
+      "step": 25492
+    },
+    {
+      "epoch": 0.9271193221078663,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7729154901032673e-07,
+      "loss": 0.9121,
+      "step": 25493
+    },
+    {
+      "epoch": 0.9271556897116049,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7701616448420954e-07,
+      "loss": 0.9257,
+      "step": 25494
+    },
+    {
+      "epoch": 0.9271920573153435,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7674091485070056e-07,
+      "loss": 0.8779,
+      "step": 25495
+    },
+    {
+      "epoch": 0.9272284249190821,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7646580011361443e-07,
+      "loss": 0.8544,
+      "step": 25496
+    },
+    {
+      "epoch": 0.9272647925228207,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7619082027676934e-07,
+      "loss": 0.8075,
+      "step": 25497
+    },
+    {
+      "epoch": 0.9273011601265593,
+      "grad_norm": 0.0,
+      "learning_rate": 2.759159753439811e-07,
+      "loss": 0.8281,
+      "step": 25498
+    },
+    {
+      "epoch": 0.9273375277302979,
+      "grad_norm": 0.0,
+      "learning_rate": 2.75641265319061e-07,
+      "loss": 0.8878,
+      "step": 25499
+    },
+    {
+      "epoch": 0.9273738953340365,
+      "grad_norm": 0.0,
+      "learning_rate": 2.753666902058216e-07,
+      "loss": 0.9497,
+      "step": 25500
+    },
+    {
+      "epoch": 0.9274102629377751,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7509225000807325e-07,
+      "loss": 0.9055,
+      "step": 25501
+    },
+    {
+      "epoch": 0.9274466305415137,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7481794472962066e-07,
+      "loss": 0.849,
+      "step": 25502
+    },
+    {
+      "epoch": 0.9274829981452523,
+      "grad_norm": 0.0,
+      "learning_rate": 2.745437743742707e-07,
+      "loss": 0.941,
+      "step": 25503
+    },
+    {
+      "epoch": 0.9275193657489909,
+      "grad_norm": 0.0,
+      "learning_rate": 2.742697389458282e-07,
+      "loss": 0.9842,
+      "step": 25504
+    },
+    {
+      "epoch": 0.9275557333527293,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7399583844809343e-07,
+      "loss": 0.853,
+      "step": 25505
+    },
+    {
+      "epoch": 0.9275921009564679,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7372207288486773e-07,
+      "loss": 0.9225,
+      "step": 25506
+    },
+    {
+      "epoch": 0.9276284685602065,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7344844225994707e-07,
+      "loss": 0.9032,
+      "step": 25507
+    },
+    {
+      "epoch": 0.9276648361639451,
+      "grad_norm": 0.0,
+      "learning_rate": 2.731749465771316e-07,
+      "loss": 0.8713,
+      "step": 25508
+    },
+    {
+      "epoch": 0.9277012037676837,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7290158584021067e-07,
+      "loss": 0.9314,
+      "step": 25509
+    },
+    {
+      "epoch": 0.9277375713714223,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7262836005298e-07,
+      "loss": 0.8532,
+      "step": 25510
+    },
+    {
+      "epoch": 0.9277739389751609,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7235526921922994e-07,
+      "loss": 0.8679,
+      "step": 25511
+    },
+    {
+      "epoch": 0.9278103065788995,
+      "grad_norm": 0.0,
+      "learning_rate": 2.720823133427486e-07,
+      "loss": 0.8914,
+      "step": 25512
+    },
+    {
+      "epoch": 0.9278466741826381,
+      "grad_norm": 0.0,
+      "learning_rate": 2.718094924273218e-07,
+      "loss": 0.9129,
+      "step": 25513
+    },
+    {
+      "epoch": 0.9278830417863767,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7153680647673875e-07,
+      "loss": 0.9457,
+      "step": 25514
+    },
+    {
+      "epoch": 0.9279194093901153,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7126425549477755e-07,
+      "loss": 0.8521,
+      "step": 25515
+    },
+    {
+      "epoch": 0.9279557769938539,
+      "grad_norm": 0.0,
+      "learning_rate": 2.709918394852218e-07,
+      "loss": 0.9336,
+      "step": 25516
+    },
+    {
+      "epoch": 0.9279921445975925,
+      "grad_norm": 0.0,
+      "learning_rate": 2.707195584518507e-07,
+      "loss": 0.8023,
+      "step": 25517
+    },
+    {
+      "epoch": 0.9280285122013311,
+      "grad_norm": 0.0,
+      "learning_rate": 2.704474123984413e-07,
+      "loss": 0.9179,
+      "step": 25518
+    },
+    {
+      "epoch": 0.9280648798050697,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7017540132876943e-07,
+      "loss": 0.8709,
+      "step": 25519
+    },
+    {
+      "epoch": 0.9281012474088083,
+      "grad_norm": 0.0,
+      "learning_rate": 2.699035252466109e-07,
+      "loss": 0.8754,
+      "step": 25520
+    },
+    {
+      "epoch": 0.9281376150125468,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6963178415573386e-07,
+      "loss": 0.9096,
+      "step": 25521
+    },
+    {
+      "epoch": 0.9281739826162854,
+      "grad_norm": 0.0,
+      "learning_rate": 2.693601780599109e-07,
+      "loss": 0.9124,
+      "step": 25522
+    },
+    {
+      "epoch": 0.928210350220024,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6908870696291e-07,
+      "loss": 0.8351,
+      "step": 25523
+    },
+    {
+      "epoch": 0.9282467178237626,
+      "grad_norm": 0.0,
+      "learning_rate": 2.68817370868496e-07,
+      "loss": 0.9388,
+      "step": 25524
+    },
+    {
+      "epoch": 0.9282830854275012,
+      "grad_norm": 0.0,
+      "learning_rate": 2.685461697804348e-07,
+      "loss": 0.8636,
+      "step": 25525
+    },
+    {
+      "epoch": 0.9283194530312397,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6827510370248777e-07,
+      "loss": 0.9051,
+      "step": 25526
+    },
+    {
+      "epoch": 0.9283558206349783,
+      "grad_norm": 0.0,
+      "learning_rate": 2.680041726384175e-07,
+      "loss": 0.8867,
+      "step": 25527
+    },
+    {
+      "epoch": 0.9283921882387169,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6773337659197983e-07,
+      "loss": 0.9155,
+      "step": 25528
+    },
+    {
+      "epoch": 0.9284285558424555,
+      "grad_norm": 0.0,
+      "learning_rate": 2.67462715566934e-07,
+      "loss": 0.9685,
+      "step": 25529
+    },
+    {
+      "epoch": 0.9284649234461941,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6719218956703376e-07,
+      "loss": 0.8182,
+      "step": 25530
+    },
+    {
+      "epoch": 0.9285012910499327,
+      "grad_norm": 0.0,
+      "learning_rate": 2.669217985960337e-07,
+      "loss": 0.8925,
+      "step": 25531
+    },
+    {
+      "epoch": 0.9285376586536713,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6665154265768324e-07,
+      "loss": 0.8879,
+      "step": 25532
+    },
+    {
+      "epoch": 0.9285740262574099,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6638142175573476e-07,
+      "loss": 0.8448,
+      "step": 25533
+    },
+    {
+      "epoch": 0.9286103938611485,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6611143589393206e-07,
+      "loss": 0.9186,
+      "step": 25534
+    },
+    {
+      "epoch": 0.9286467614648871,
+      "grad_norm": 0.0,
+      "learning_rate": 2.658415850760232e-07,
+      "loss": 0.8045,
+      "step": 25535
+    },
+    {
+      "epoch": 0.9286831290686257,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6557186930575186e-07,
+      "loss": 0.8228,
+      "step": 25536
+    },
+    {
+      "epoch": 0.9287194966723643,
+      "grad_norm": 0.0,
+      "learning_rate": 2.653022885868584e-07,
+      "loss": 0.9455,
+      "step": 25537
+    },
+    {
+      "epoch": 0.9287558642761028,
+      "grad_norm": 0.0,
+      "learning_rate": 2.650328429230842e-07,
+      "loss": 1.0575,
+      "step": 25538
+    },
+    {
+      "epoch": 0.9287922318798414,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6476353231816744e-07,
+      "loss": 0.8748,
+      "step": 25539
+    },
+    {
+      "epoch": 0.92882859948358,
+      "grad_norm": 0.0,
+      "learning_rate": 2.64494356775844e-07,
+      "loss": 0.7458,
+      "step": 25540
+    },
+    {
+      "epoch": 0.9288649670873186,
+      "grad_norm": 0.0,
+      "learning_rate": 2.642253162998487e-07,
+      "loss": 0.7952,
+      "step": 25541
+    },
+    {
+      "epoch": 0.9289013346910572,
+      "grad_norm": 0.0,
+      "learning_rate": 2.639564108939141e-07,
+      "loss": 0.9478,
+      "step": 25542
+    },
+    {
+      "epoch": 0.9289377022947958,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6368764056177053e-07,
+      "loss": 0.873,
+      "step": 25543
+    },
+    {
+      "epoch": 0.9289740698985344,
+      "grad_norm": 0.0,
+      "learning_rate": 2.634190053071473e-07,
+      "loss": 0.8914,
+      "step": 25544
+    },
+    {
+      "epoch": 0.929010437502273,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6315050513377017e-07,
+      "loss": 0.9608,
+      "step": 25545
+    },
+    {
+      "epoch": 0.9290468051060116,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6288214004536625e-07,
+      "loss": 0.7611,
+      "step": 25546
+    },
+    {
+      "epoch": 0.9290831727097502,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6261391004565597e-07,
+      "loss": 0.9095,
+      "step": 25547
+    },
+    {
+      "epoch": 0.9291195403134888,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6234581513836286e-07,
+      "loss": 0.8568,
+      "step": 25548
+    },
+    {
+      "epoch": 0.9291559079172274,
+      "grad_norm": 0.0,
+      "learning_rate": 2.620778553272052e-07,
+      "loss": 0.9135,
+      "step": 25549
+    },
+    {
+      "epoch": 0.929192275520966,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6181003061590216e-07,
+      "loss": 0.9183,
+      "step": 25550
+    },
+    {
+      "epoch": 0.9292286431247045,
+      "grad_norm": 0.0,
+      "learning_rate": 2.615423410081663e-07,
+      "loss": 0.949,
+      "step": 25551
+    },
+    {
+      "epoch": 0.9292650107284431,
+      "grad_norm": 0.0,
+      "learning_rate": 2.612747865077148e-07,
+      "loss": 0.8794,
+      "step": 25552
+    },
+    {
+      "epoch": 0.9293013783321817,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6100736711825783e-07,
+      "loss": 0.9192,
+      "step": 25553
+    },
+    {
+      "epoch": 0.9293377459359203,
+      "grad_norm": 0.0,
+      "learning_rate": 2.607400828435058e-07,
+      "loss": 0.885,
+      "step": 25554
+    },
+    {
+      "epoch": 0.9293741135396588,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6047293368716697e-07,
+      "loss": 0.8709,
+      "step": 25555
+    },
+    {
+      "epoch": 0.9294104811433974,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6020591965294826e-07,
+      "loss": 0.9911,
+      "step": 25556
+    },
+    {
+      "epoch": 0.929446848747136,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5993904074455233e-07,
+      "loss": 0.8532,
+      "step": 25557
+    },
+    {
+      "epoch": 0.9294832163508746,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5967229696568285e-07,
+      "loss": 0.8579,
+      "step": 25558
+    },
+    {
+      "epoch": 0.9295195839546132,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5940568832004133e-07,
+      "loss": 0.8204,
+      "step": 25559
+    },
+    {
+      "epoch": 0.9295559515583518,
+      "grad_norm": 0.0,
+      "learning_rate": 2.591392148113259e-07,
+      "loss": 0.8418,
+      "step": 25560
+    },
+    {
+      "epoch": 0.9295923191620904,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5887287644323243e-07,
+      "loss": 0.917,
+      "step": 25561
+    },
+    {
+      "epoch": 0.929628686765829,
+      "grad_norm": 0.0,
+      "learning_rate": 2.586066732194581e-07,
+      "loss": 0.7787,
+      "step": 25562
+    },
+    {
+      "epoch": 0.9296650543695676,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5834060514369543e-07,
+      "loss": 0.9224,
+      "step": 25563
+    },
+    {
+      "epoch": 0.9297014219733062,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5807467221963256e-07,
+      "loss": 0.9626,
+      "step": 25564
+    },
+    {
+      "epoch": 0.9297377895770448,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5780887445096436e-07,
+      "loss": 0.8376,
+      "step": 25565
+    },
+    {
+      "epoch": 0.9297741571807834,
+      "grad_norm": 0.0,
+      "learning_rate": 2.575432118413745e-07,
+      "loss": 0.9455,
+      "step": 25566
+    },
+    {
+      "epoch": 0.929810524784522,
+      "grad_norm": 0.0,
+      "learning_rate": 2.572776843945501e-07,
+      "loss": 0.9601,
+      "step": 25567
+    },
+    {
+      "epoch": 0.9298468923882606,
+      "grad_norm": 0.0,
+      "learning_rate": 2.570122921141738e-07,
+      "loss": 0.8954,
+      "step": 25568
+    },
+    {
+      "epoch": 0.9298832599919992,
+      "grad_norm": 0.0,
+      "learning_rate": 2.567470350039303e-07,
+      "loss": 0.9347,
+      "step": 25569
+    },
+    {
+      "epoch": 0.9299196275957378,
+      "grad_norm": 0.0,
+      "learning_rate": 2.564819130674956e-07,
+      "loss": 0.9458,
+      "step": 25570
+    },
+    {
+      "epoch": 0.9299559951994764,
+      "grad_norm": 0.0,
+      "learning_rate": 2.562169263085512e-07,
+      "loss": 0.9686,
+      "step": 25571
+    },
+    {
+      "epoch": 0.9299923628032148,
+      "grad_norm": 0.0,
+      "learning_rate": 2.55952074730772e-07,
+      "loss": 0.9614,
+      "step": 25572
+    },
+    {
+      "epoch": 0.9300287304069534,
+      "grad_norm": 0.0,
+      "learning_rate": 2.556873583378339e-07,
+      "loss": 0.8321,
+      "step": 25573
+    },
+    {
+      "epoch": 0.930065098010692,
+      "grad_norm": 0.0,
+      "learning_rate": 2.554227771334061e-07,
+      "loss": 0.9175,
+      "step": 25574
+    },
+    {
+      "epoch": 0.9301014656144306,
+      "grad_norm": 0.0,
+      "learning_rate": 2.551583311211647e-07,
+      "loss": 0.8006,
+      "step": 25575
+    },
+    {
+      "epoch": 0.9301378332181692,
+      "grad_norm": 0.0,
+      "learning_rate": 2.548940203047723e-07,
+      "loss": 0.9314,
+      "step": 25576
+    },
+    {
+      "epoch": 0.9301742008219078,
+      "grad_norm": 0.0,
+      "learning_rate": 2.546298446879003e-07,
+      "loss": 0.9757,
+      "step": 25577
+    },
+    {
+      "epoch": 0.9302105684256464,
+      "grad_norm": 0.0,
+      "learning_rate": 2.543658042742125e-07,
+      "loss": 0.9894,
+      "step": 25578
+    },
+    {
+      "epoch": 0.930246936029385,
+      "grad_norm": 0.0,
+      "learning_rate": 2.541018990673716e-07,
+      "loss": 0.7403,
+      "step": 25579
+    },
+    {
+      "epoch": 0.9302833036331236,
+      "grad_norm": 0.0,
+      "learning_rate": 2.53838129071039e-07,
+      "loss": 0.9158,
+      "step": 25580
+    },
+    {
+      "epoch": 0.9303196712368622,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5357449428887514e-07,
+      "loss": 0.846,
+      "step": 25581
+    },
+    {
+      "epoch": 0.9303560388406008,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5331099472453715e-07,
+      "loss": 0.971,
+      "step": 25582
+    },
+    {
+      "epoch": 0.9303924064443394,
+      "grad_norm": 0.0,
+      "learning_rate": 2.530476303816787e-07,
+      "loss": 0.8914,
+      "step": 25583
+    },
+    {
+      "epoch": 0.930428774048078,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5278440126395574e-07,
+      "loss": 0.9132,
+      "step": 25584
+    },
+    {
+      "epoch": 0.9304651416518166,
+      "grad_norm": 0.0,
+      "learning_rate": 2.525213073750199e-07,
+      "loss": 1.0094,
+      "step": 25585
+    },
+    {
+      "epoch": 0.9305015092555552,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5225834871852153e-07,
+      "loss": 0.9789,
+      "step": 25586
+    },
+    {
+      "epoch": 0.9305378768592938,
+      "grad_norm": 0.0,
+      "learning_rate": 2.519955252981088e-07,
+      "loss": 0.9027,
+      "step": 25587
+    },
+    {
+      "epoch": 0.9305742444630324,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5173283711742657e-07,
+      "loss": 0.8657,
+      "step": 25588
+    },
+    {
+      "epoch": 0.9306106120667709,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5147028418011974e-07,
+      "loss": 0.9737,
+      "step": 25589
+    },
+    {
+      "epoch": 0.9306469796705095,
+      "grad_norm": 0.0,
+      "learning_rate": 2.512078664898321e-07,
+      "loss": 0.9432,
+      "step": 25590
+    },
+    {
+      "epoch": 0.9306833472742481,
+      "grad_norm": 0.0,
+      "learning_rate": 2.50945584050204e-07,
+      "loss": 0.877,
+      "step": 25591
+    },
+    {
+      "epoch": 0.9307197148779867,
+      "grad_norm": 0.0,
+      "learning_rate": 2.506834368648736e-07,
+      "loss": 0.9418,
+      "step": 25592
+    },
+    {
+      "epoch": 0.9307560824817253,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5042142493747703e-07,
+      "loss": 0.9142,
+      "step": 25593
+    },
+    {
+      "epoch": 0.9307924500854639,
+      "grad_norm": 0.0,
+      "learning_rate": 2.501595482716523e-07,
+      "loss": 1.0564,
+      "step": 25594
+    },
+    {
+      "epoch": 0.9308288176892024,
+      "grad_norm": 0.0,
+      "learning_rate": 2.498978068710278e-07,
+      "loss": 0.7616,
+      "step": 25595
+    },
+    {
+      "epoch": 0.930865185292941,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4963620073923947e-07,
+      "loss": 0.9757,
+      "step": 25596
+    },
+    {
+      "epoch": 0.9309015528966796,
+      "grad_norm": 0.0,
+      "learning_rate": 2.493747298799154e-07,
+      "loss": 1.0926,
+      "step": 25597
+    },
+    {
+      "epoch": 0.9309379205004182,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4911339429668167e-07,
+      "loss": 0.88,
+      "step": 25598
+    },
+    {
+      "epoch": 0.9309742881041568,
+      "grad_norm": 0.0,
+      "learning_rate": 2.488521939931643e-07,
+      "loss": 0.9215,
+      "step": 25599
+    },
+    {
+      "epoch": 0.9310106557078954,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4859112897299034e-07,
+      "loss": 0.9395,
+      "step": 25600
+    },
+    {
+      "epoch": 0.931047023311634,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4833019923977685e-07,
+      "loss": 0.861,
+      "step": 25601
+    },
+    {
+      "epoch": 0.9310833909153726,
+      "grad_norm": 0.0,
+      "learning_rate": 2.480694047971455e-07,
+      "loss": 0.8765,
+      "step": 25602
+    },
+    {
+      "epoch": 0.9311197585191112,
+      "grad_norm": 0.0,
+      "learning_rate": 2.478087456487166e-07,
+      "loss": 0.9925,
+      "step": 25603
+    },
+    {
+      "epoch": 0.9311561261228498,
+      "grad_norm": 0.0,
+      "learning_rate": 2.47548221798104e-07,
+      "loss": 0.7782,
+      "step": 25604
+    },
+    {
+      "epoch": 0.9311924937265884,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4728783324892256e-07,
+      "loss": 0.8757,
+      "step": 25605
+    },
+    {
+      "epoch": 0.9312288613303269,
+      "grad_norm": 0.0,
+      "learning_rate": 2.47027580004785e-07,
+      "loss": 0.8852,
+      "step": 25606
+    },
+    {
+      "epoch": 0.9312652289340655,
+      "grad_norm": 0.0,
+      "learning_rate": 2.467674620693039e-07,
+      "loss": 0.9138,
+      "step": 25607
+    },
+    {
+      "epoch": 0.9313015965378041,
+      "grad_norm": 0.0,
+      "learning_rate": 2.465074794460842e-07,
+      "loss": 0.9893,
+      "step": 25608
+    },
+    {
+      "epoch": 0.9313379641415427,
+      "grad_norm": 0.0,
+      "learning_rate": 2.462476321387364e-07,
+      "loss": 0.9066,
+      "step": 25609
+    },
+    {
+      "epoch": 0.9313743317452813,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4598792015086305e-07,
+      "loss": 0.9601,
+      "step": 25610
+    },
+    {
+      "epoch": 0.9314106993490199,
+      "grad_norm": 0.0,
+      "learning_rate": 2.45728343486068e-07,
+      "loss": 0.9023,
+      "step": 25611
+    },
+    {
+      "epoch": 0.9314470669527585,
+      "grad_norm": 0.0,
+      "learning_rate": 2.454689021479528e-07,
+      "loss": 0.8676,
+      "step": 25612
+    },
+    {
+      "epoch": 0.9314834345564971,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4520959614011796e-07,
+      "loss": 0.9502,
+      "step": 25613
+    },
+    {
+      "epoch": 0.9315198021602357,
+      "grad_norm": 0.0,
+      "learning_rate": 2.449504254661572e-07,
+      "loss": 0.8926,
+      "step": 25614
+    },
+    {
+      "epoch": 0.9315561697639743,
+      "grad_norm": 0.0,
+      "learning_rate": 2.44691390129671e-07,
+      "loss": 0.9841,
+      "step": 25615
+    },
+    {
+      "epoch": 0.9315925373677129,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4443249013424986e-07,
+      "loss": 0.9682,
+      "step": 25616
+    },
+    {
+      "epoch": 0.9316289049714515,
+      "grad_norm": 0.0,
+      "learning_rate": 2.441737254834864e-07,
+      "loss": 0.977,
+      "step": 25617
+    },
+    {
+      "epoch": 0.93166527257519,
+      "grad_norm": 0.0,
+      "learning_rate": 2.439150961809711e-07,
+      "loss": 0.8808,
+      "step": 25618
+    },
+    {
+      "epoch": 0.9317016401789286,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4365660223029106e-07,
+      "loss": 0.8957,
+      "step": 25619
+    },
+    {
+      "epoch": 0.9317380077826672,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4339824363503574e-07,
+      "loss": 0.8899,
+      "step": 25620
+    },
+    {
+      "epoch": 0.9317743753864058,
+      "grad_norm": 0.0,
+      "learning_rate": 2.431400203987844e-07,
+      "loss": 0.8219,
+      "step": 25621
+    },
+    {
+      "epoch": 0.9318107429901443,
+      "grad_norm": 0.0,
+      "learning_rate": 2.428819325251219e-07,
+      "loss": 0.9177,
+      "step": 25622
+    },
+    {
+      "epoch": 0.9318471105938829,
+      "grad_norm": 0.0,
+      "learning_rate": 2.426239800176311e-07,
+      "loss": 1.0534,
+      "step": 25623
+    },
+    {
+      "epoch": 0.9318834781976215,
+      "grad_norm": 0.0,
+      "learning_rate": 2.423661628798868e-07,
+      "loss": 0.8049,
+      "step": 25624
+    },
+    {
+      "epoch": 0.9319198458013601,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4210848111546836e-07,
+      "loss": 0.8101,
+      "step": 25625
+    },
+    {
+      "epoch": 0.9319562134050987,
+      "grad_norm": 0.0,
+      "learning_rate": 2.418509347279507e-07,
+      "loss": 0.9792,
+      "step": 25626
+    },
+    {
+      "epoch": 0.9319925810088373,
+      "grad_norm": 0.0,
+      "learning_rate": 2.415935237209055e-07,
+      "loss": 0.8536,
+      "step": 25627
+    },
+    {
+      "epoch": 0.9320289486125759,
+      "grad_norm": 0.0,
+      "learning_rate": 2.413362480979053e-07,
+      "loss": 0.8953,
+      "step": 25628
+    },
+    {
+      "epoch": 0.9320653162163145,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4107910786251854e-07,
+      "loss": 0.9299,
+      "step": 25629
+    },
+    {
+      "epoch": 0.9321016838200531,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4082210301831334e-07,
+      "loss": 0.8739,
+      "step": 25630
+    },
+    {
+      "epoch": 0.9321380514237917,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4056523356885463e-07,
+      "loss": 1.0287,
+      "step": 25631
+    },
+    {
+      "epoch": 0.9321744190275303,
+      "grad_norm": 0.0,
+      "learning_rate": 2.403084995177074e-07,
+      "loss": 1.0626,
+      "step": 25632
+    },
+    {
+      "epoch": 0.9322107866312689,
+      "grad_norm": 0.0,
+      "learning_rate": 2.40051900868431e-07,
+      "loss": 1.0768,
+      "step": 25633
+    },
+    {
+      "epoch": 0.9322471542350075,
+      "grad_norm": 0.0,
+      "learning_rate": 2.397954376245881e-07,
+      "loss": 1.0696,
+      "step": 25634
+    },
+    {
+      "epoch": 0.9322835218387461,
+      "grad_norm": 0.0,
+      "learning_rate": 2.395391097897359e-07,
+      "loss": 1.0574,
+      "step": 25635
+    },
+    {
+      "epoch": 0.9323198894424847,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3928291736742934e-07,
+      "loss": 0.8223,
+      "step": 25636
+    },
+    {
+      "epoch": 0.9323562570462233,
+      "grad_norm": 0.0,
+      "learning_rate": 2.390268603612234e-07,
+      "loss": 0.9301,
+      "step": 25637
+    },
+    {
+      "epoch": 0.9323926246499619,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3877093877467174e-07,
+      "loss": 0.8352,
+      "step": 25638
+    },
+    {
+      "epoch": 0.9324289922537004,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3851515261132494e-07,
+      "loss": 0.8895,
+      "step": 25639
+    },
+    {
+      "epoch": 0.932465359857439,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3825950187472802e-07,
+      "loss": 0.8994,
+      "step": 25640
+    },
+    {
+      "epoch": 0.9325017274611775,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3800398656843248e-07,
+      "loss": 0.9145,
+      "step": 25641
+    },
+    {
+      "epoch": 0.9325380950649161,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3774860669598e-07,
+      "loss": 0.8192,
+      "step": 25642
+    },
+    {
+      "epoch": 0.9325744626686547,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3749336226091546e-07,
+      "loss": 0.8509,
+      "step": 25643
+    },
+    {
+      "epoch": 0.9326108302723933,
+      "grad_norm": 0.0,
+      "learning_rate": 2.372382532667783e-07,
+      "loss": 0.9186,
+      "step": 25644
+    },
+    {
+      "epoch": 0.9326471978761319,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3698327971711012e-07,
+      "loss": 0.9459,
+      "step": 25645
+    },
+    {
+      "epoch": 0.9326835654798705,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3672844161544584e-07,
+      "loss": 0.8003,
+      "step": 25646
+    },
+    {
+      "epoch": 0.9327199330836091,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3647373896532378e-07,
+      "loss": 0.8823,
+      "step": 25647
+    },
+    {
+      "epoch": 0.9327563006873477,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3621917177027442e-07,
+      "loss": 0.8663,
+      "step": 25648
+    },
+    {
+      "epoch": 0.9327926682910863,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3596474003383163e-07,
+      "loss": 0.9556,
+      "step": 25649
+    },
+    {
+      "epoch": 0.9328290358948249,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3571044375952367e-07,
+      "loss": 0.8395,
+      "step": 25650
+    },
+    {
+      "epoch": 0.9328654034985635,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3545628295088107e-07,
+      "loss": 0.8723,
+      "step": 25651
+    },
+    {
+      "epoch": 0.9329017711023021,
+      "grad_norm": 0.0,
+      "learning_rate": 2.352022576114288e-07,
+      "loss": 0.8252,
+      "step": 25652
+    },
+    {
+      "epoch": 0.9329381387060407,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3494836774468953e-07,
+      "loss": 0.893,
+      "step": 25653
+    },
+    {
+      "epoch": 0.9329745063097793,
+      "grad_norm": 0.0,
+      "learning_rate": 2.346946133541872e-07,
+      "loss": 0.8585,
+      "step": 25654
+    },
+    {
+      "epoch": 0.9330108739135179,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3444099444344227e-07,
+      "loss": 0.8453,
+      "step": 25655
+    },
+    {
+      "epoch": 0.9330472415172564,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3418751101597192e-07,
+      "loss": 0.9946,
+      "step": 25656
+    },
+    {
+      "epoch": 0.933083609120995,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3393416307529558e-07,
+      "loss": 0.9636,
+      "step": 25657
+    },
+    {
+      "epoch": 0.9331199767247336,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3368095062492713e-07,
+      "loss": 0.8462,
+      "step": 25658
+    },
+    {
+      "epoch": 0.9331563443284722,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3342787366837594e-07,
+      "loss": 0.8285,
+      "step": 25659
+    },
+    {
+      "epoch": 0.9331927119322108,
+      "grad_norm": 0.0,
+      "learning_rate": 2.331749322091581e-07,
+      "loss": 0.9041,
+      "step": 25660
+    },
+    {
+      "epoch": 0.9332290795359494,
+      "grad_norm": 0.0,
+      "learning_rate": 2.329221262507808e-07,
+      "loss": 0.8372,
+      "step": 25661
+    },
+    {
+      "epoch": 0.933265447139688,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3266945579675126e-07,
+      "loss": 0.9118,
+      "step": 25662
+    },
+    {
+      "epoch": 0.9333018147434265,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3241692085057332e-07,
+      "loss": 0.8819,
+      "step": 25663
+    },
+    {
+      "epoch": 0.9333381823471651,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3216452141575418e-07,
+      "loss": 0.9666,
+      "step": 25664
+    },
+    {
+      "epoch": 0.9333745499509037,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3191225749579215e-07,
+      "loss": 0.8736,
+      "step": 25665
+    },
+    {
+      "epoch": 0.9334109175546423,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3166012909419e-07,
+      "loss": 0.917,
+      "step": 25666
+    },
+    {
+      "epoch": 0.9334472851583809,
+      "grad_norm": 0.0,
+      "learning_rate": 2.314081362144427e-07,
+      "loss": 0.8715,
+      "step": 25667
+    },
+    {
+      "epoch": 0.9334836527621195,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3115627886004855e-07,
+      "loss": 1.0061,
+      "step": 25668
+    },
+    {
+      "epoch": 0.9335200203658581,
+      "grad_norm": 0.0,
+      "learning_rate": 2.309045570345003e-07,
+      "loss": 0.834,
+      "step": 25669
+    },
+    {
+      "epoch": 0.9335563879695967,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3065297074129077e-07,
+      "loss": 0.9229,
+      "step": 25670
+    },
+    {
+      "epoch": 0.9335927555733353,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3040151998391048e-07,
+      "loss": 0.8946,
+      "step": 25671
+    },
+    {
+      "epoch": 0.9336291231770739,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3015020476584881e-07,
+      "loss": 0.9149,
+      "step": 25672
+    },
+    {
+      "epoch": 0.9336654907808124,
+      "grad_norm": 0.0,
+      "learning_rate": 2.298990250905897e-07,
+      "loss": 0.846,
+      "step": 25673
+    },
+    {
+      "epoch": 0.933701858384551,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2964798096162144e-07,
+      "loss": 0.9405,
+      "step": 25674
+    },
+    {
+      "epoch": 0.9337382259882896,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2939707238242349e-07,
+      "loss": 1.1121,
+      "step": 25675
+    },
+    {
+      "epoch": 0.9337745935920282,
+      "grad_norm": 0.0,
+      "learning_rate": 2.291462993564797e-07,
+      "loss": 0.8696,
+      "step": 25676
+    },
+    {
+      "epoch": 0.9338109611957668,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2889566188726842e-07,
+      "loss": 0.9981,
+      "step": 25677
+    },
+    {
+      "epoch": 0.9338473287995054,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2864515997826465e-07,
+      "loss": 0.8574,
+      "step": 25678
+    },
+    {
+      "epoch": 0.933883696403244,
+      "grad_norm": 0.0,
+      "learning_rate": 2.283947936329467e-07,
+      "loss": 0.713,
+      "step": 25679
+    },
+    {
+      "epoch": 0.9339200640069826,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2814456285478736e-07,
+      "loss": 0.8597,
+      "step": 25680
+    },
+    {
+      "epoch": 0.9339564316107212,
+      "grad_norm": 0.0,
+      "learning_rate": 2.278944676472572e-07,
+      "loss": 0.9146,
+      "step": 25681
+    },
+    {
+      "epoch": 0.9339927992144598,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2764450801382566e-07,
+      "loss": 0.8962,
+      "step": 25682
+    },
+    {
+      "epoch": 0.9340291668181984,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2739468395796215e-07,
+      "loss": 0.8576,
+      "step": 25683
+    },
+    {
+      "epoch": 0.934065534421937,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2714499548313173e-07,
+      "loss": 0.8004,
+      "step": 25684
+    },
+    {
+      "epoch": 0.9341019020256756,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2689544259279938e-07,
+      "loss": 0.8875,
+      "step": 25685
+    },
+    {
+      "epoch": 0.9341382696294142,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2664602529042567e-07,
+      "loss": 0.9083,
+      "step": 25686
+    },
+    {
+      "epoch": 0.9341746372331527,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2639674357947227e-07,
+      "loss": 0.9014,
+      "step": 25687
+    },
+    {
+      "epoch": 0.9342110048368913,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2614759746339644e-07,
+      "loss": 0.9925,
+      "step": 25688
+    },
+    {
+      "epoch": 0.9342473724406299,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2589858694565536e-07,
+      "loss": 0.9605,
+      "step": 25689
+    },
+    {
+      "epoch": 0.9342837400443684,
+      "grad_norm": 0.0,
+      "learning_rate": 2.256497120297052e-07,
+      "loss": 0.7362,
+      "step": 25690
+    },
+    {
+      "epoch": 0.934320107648107,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2540097271899653e-07,
+      "loss": 0.8785,
+      "step": 25691
+    },
+    {
+      "epoch": 0.9343564752518456,
+      "grad_norm": 0.0,
+      "learning_rate": 2.251523690169799e-07,
+      "loss": 0.8668,
+      "step": 25692
+    },
+    {
+      "epoch": 0.9343928428555842,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2490390092710702e-07,
+      "loss": 0.9058,
+      "step": 25693
+    },
+    {
+      "epoch": 0.9344292104593228,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2465556845282177e-07,
+      "loss": 0.8994,
+      "step": 25694
+    },
+    {
+      "epoch": 0.9344655780630614,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2440737159757253e-07,
+      "loss": 0.8318,
+      "step": 25695
+    },
+    {
+      "epoch": 0.9345019456668,
+      "grad_norm": 0.0,
+      "learning_rate": 2.24159310364801e-07,
+      "loss": 0.8916,
+      "step": 25696
+    },
+    {
+      "epoch": 0.9345383132705386,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2391138475794881e-07,
+      "loss": 0.9232,
+      "step": 25697
+    },
+    {
+      "epoch": 0.9345746808742772,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2366359478045663e-07,
+      "loss": 0.9685,
+      "step": 25698
+    },
+    {
+      "epoch": 0.9346110484780158,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2341594043576053e-07,
+      "loss": 0.9333,
+      "step": 25699
+    },
+    {
+      "epoch": 0.9346474160817544,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2316842172729668e-07,
+      "loss": 0.9174,
+      "step": 25700
+    },
+    {
+      "epoch": 0.934683783685493,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2292103865849902e-07,
+      "loss": 0.9609,
+      "step": 25701
+    },
+    {
+      "epoch": 0.9347201512892316,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2267379123280143e-07,
+      "loss": 0.9322,
+      "step": 25702
+    },
+    {
+      "epoch": 0.9347565188929702,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2242667945363226e-07,
+      "loss": 0.8466,
+      "step": 25703
+    },
+    {
+      "epoch": 0.9347928864967088,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2217970332441995e-07,
+      "loss": 0.8982,
+      "step": 25704
+    },
+    {
+      "epoch": 0.9348292541004474,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2193286284859062e-07,
+      "loss": 0.9191,
+      "step": 25705
+    },
+    {
+      "epoch": 0.9348656217041859,
+      "grad_norm": 0.0,
+      "learning_rate": 2.216861580295715e-07,
+      "loss": 0.9783,
+      "step": 25706
+    },
+    {
+      "epoch": 0.9349019893079245,
+      "grad_norm": 0.0,
+      "learning_rate": 2.214395888707821e-07,
+      "loss": 0.892,
+      "step": 25707
+    },
+    {
+      "epoch": 0.934938356911663,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2119315537564524e-07,
+      "loss": 0.8772,
+      "step": 25708
+    },
+    {
+      "epoch": 0.9349747245154016,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2094685754757817e-07,
+      "loss": 0.9497,
+      "step": 25709
+    },
+    {
+      "epoch": 0.9350110921191402,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2070069538999926e-07,
+      "loss": 0.9338,
+      "step": 25710
+    },
+    {
+      "epoch": 0.9350474597228788,
+      "grad_norm": 0.0,
+      "learning_rate": 2.204546689063225e-07,
+      "loss": 0.8784,
+      "step": 25711
+    },
+    {
+      "epoch": 0.9350838273266174,
+      "grad_norm": 0.0,
+      "learning_rate": 2.20208778099964e-07,
+      "loss": 0.9746,
+      "step": 25712
+    },
+    {
+      "epoch": 0.935120194930356,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1996302297432993e-07,
+      "loss": 0.9123,
+      "step": 25713
+    },
+    {
+      "epoch": 0.9351565625340946,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1971740353283422e-07,
+      "loss": 0.929,
+      "step": 25714
+    },
+    {
+      "epoch": 0.9351929301378332,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1947191977888416e-07,
+      "loss": 0.875,
+      "step": 25715
+    },
+    {
+      "epoch": 0.9352292977415718,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1922657171588367e-07,
+      "loss": 0.7506,
+      "step": 25716
+    },
+    {
+      "epoch": 0.9352656653453104,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1898135934723674e-07,
+      "loss": 0.9134,
+      "step": 25717
+    },
+    {
+      "epoch": 0.935302032949049,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1873628267634617e-07,
+      "loss": 0.9736,
+      "step": 25718
+    },
+    {
+      "epoch": 0.9353384005527876,
+      "grad_norm": 0.0,
+      "learning_rate": 2.184913417066137e-07,
+      "loss": 0.9012,
+      "step": 25719
+    },
+    {
+      "epoch": 0.9353747681565262,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1824653644143323e-07,
+      "loss": 0.8087,
+      "step": 25720
+    },
+    {
+      "epoch": 0.9354111357602648,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1800186688420434e-07,
+      "loss": 0.988,
+      "step": 25721
+    },
+    {
+      "epoch": 0.9354475033640034,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1775733303832202e-07,
+      "loss": 0.8432,
+      "step": 25722
+    },
+    {
+      "epoch": 0.9354838709677419,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1751293490717586e-07,
+      "loss": 0.8636,
+      "step": 25723
+    },
+    {
+      "epoch": 0.9355202385714805,
+      "grad_norm": 0.0,
+      "learning_rate": 2.172686724941586e-07,
+      "loss": 0.8854,
+      "step": 25724
+    },
+    {
+      "epoch": 0.9355566061752191,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1702454580265985e-07,
+      "loss": 0.918,
+      "step": 25725
+    },
+    {
+      "epoch": 0.9355929737789577,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1678055483606354e-07,
+      "loss": 0.9275,
+      "step": 25726
+    },
+    {
+      "epoch": 0.9356293413826963,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1653669959775692e-07,
+      "loss": 0.9118,
+      "step": 25727
+    },
+    {
+      "epoch": 0.9356657089864349,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1629298009112287e-07,
+      "loss": 0.8466,
+      "step": 25728
+    },
+    {
+      "epoch": 0.9357020765901735,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1604939631954314e-07,
+      "loss": 0.9184,
+      "step": 25729
+    },
+    {
+      "epoch": 0.9357384441939121,
+      "grad_norm": 0.0,
+      "learning_rate": 2.15805948286395e-07,
+      "loss": 0.8667,
+      "step": 25730
+    },
+    {
+      "epoch": 0.9357748117976507,
+      "grad_norm": 0.0,
+      "learning_rate": 2.155626359950591e-07,
+      "loss": 0.9386,
+      "step": 25731
+    },
+    {
+      "epoch": 0.9358111794013892,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1531945944890832e-07,
+      "loss": 0.8013,
+      "step": 25732
+    },
+    {
+      "epoch": 0.9358475470051278,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1507641865131767e-07,
+      "loss": 0.8945,
+      "step": 25733
+    },
+    {
+      "epoch": 0.9358839146088664,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1483351360565784e-07,
+      "loss": 0.9018,
+      "step": 25734
+    },
+    {
+      "epoch": 0.935920282212605,
+      "grad_norm": 0.0,
+      "learning_rate": 2.145907443153006e-07,
+      "loss": 0.9277,
+      "step": 25735
+    },
+    {
+      "epoch": 0.9359566498163436,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1434811078361317e-07,
+      "loss": 0.8517,
+      "step": 25736
+    },
+    {
+      "epoch": 0.9359930174200822,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1410561301396071e-07,
+      "loss": 0.9347,
+      "step": 25737
+    },
+    {
+      "epoch": 0.9360293850238208,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1386325100971052e-07,
+      "loss": 0.9524,
+      "step": 25738
+    },
+    {
+      "epoch": 0.9360657526275594,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1362102477422097e-07,
+      "loss": 0.9803,
+      "step": 25739
+    },
+    {
+      "epoch": 0.9361021202312979,
+      "grad_norm": 0.0,
+      "learning_rate": 2.13378934310855e-07,
+      "loss": 0.8971,
+      "step": 25740
+    },
+    {
+      "epoch": 0.9361384878350365,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1313697962297207e-07,
+      "loss": 0.9757,
+      "step": 25741
+    },
+    {
+      "epoch": 0.9361748554387751,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1289516071392736e-07,
+      "loss": 0.9369,
+      "step": 25742
+    },
+    {
+      "epoch": 0.9362112230425137,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1265347758707478e-07,
+      "loss": 0.7864,
+      "step": 25743
+    },
+    {
+      "epoch": 0.9362475906462523,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1241193024577055e-07,
+      "loss": 0.9738,
+      "step": 25744
+    },
+    {
+      "epoch": 0.9362839582499909,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1217051869336314e-07,
+      "loss": 0.8708,
+      "step": 25745
+    },
+    {
+      "epoch": 0.9363203258537295,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1192924293320317e-07,
+      "loss": 0.8235,
+      "step": 25746
+    },
+    {
+      "epoch": 0.9363566934574681,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1168810296863684e-07,
+      "loss": 0.8975,
+      "step": 25747
+    },
+    {
+      "epoch": 0.9363930610612067,
+      "grad_norm": 0.0,
+      "learning_rate": 2.114470988030104e-07,
+      "loss": 0.8784,
+      "step": 25748
+    },
+    {
+      "epoch": 0.9364294286649453,
+      "grad_norm": 0.0,
+      "learning_rate": 2.112062304396667e-07,
+      "loss": 0.779,
+      "step": 25749
+    },
+    {
+      "epoch": 0.9364657962686839,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1096549788194975e-07,
+      "loss": 0.89,
+      "step": 25750
+    },
+    {
+      "epoch": 0.9365021638724225,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1072490113319687e-07,
+      "loss": 0.9721,
+      "step": 25751
+    },
+    {
+      "epoch": 0.9365385314761611,
+      "grad_norm": 0.0,
+      "learning_rate": 2.104844401967465e-07,
+      "loss": 0.9352,
+      "step": 25752
+    },
+    {
+      "epoch": 0.9365748990798997,
+      "grad_norm": 0.0,
+      "learning_rate": 2.10244115075936e-07,
+      "loss": 1.0157,
+      "step": 25753
+    },
+    {
+      "epoch": 0.9366112666836383,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1000392577409822e-07,
+      "loss": 0.9287,
+      "step": 25754
+    },
+    {
+      "epoch": 0.9366476342873769,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0976387229456496e-07,
+      "loss": 0.9855,
+      "step": 25755
+    },
+    {
+      "epoch": 0.9366840018911154,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0952395464066688e-07,
+      "loss": 0.98,
+      "step": 25756
+    },
+    {
+      "epoch": 0.9367203694948539,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0928417281573577e-07,
+      "loss": 0.9447,
+      "step": 25757
+    },
+    {
+      "epoch": 0.9367567370985925,
+      "grad_norm": 0.0,
+      "learning_rate": 2.090445268230934e-07,
+      "loss": 1.0312,
+      "step": 25758
+    },
+    {
+      "epoch": 0.9367931047023311,
+      "grad_norm": 0.0,
+      "learning_rate": 2.088050166660671e-07,
+      "loss": 1.0049,
+      "step": 25759
+    },
+    {
+      "epoch": 0.9368294723060697,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0856564234797984e-07,
+      "loss": 1.0706,
+      "step": 25760
+    },
+    {
+      "epoch": 0.9368658399098083,
+      "grad_norm": 0.0,
+      "learning_rate": 2.083264038721511e-07,
+      "loss": 0.8848,
+      "step": 25761
+    },
+    {
+      "epoch": 0.9369022075135469,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0808730124190047e-07,
+      "loss": 0.8895,
+      "step": 25762
+    },
+    {
+      "epoch": 0.9369385751172855,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0784833446054753e-07,
+      "loss": 0.982,
+      "step": 25763
+    },
+    {
+      "epoch": 0.9369749427210241,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0760950353140408e-07,
+      "loss": 1.0423,
+      "step": 25764
+    },
+    {
+      "epoch": 0.9370113103247627,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0737080845778634e-07,
+      "loss": 0.9662,
+      "step": 25765
+    },
+    {
+      "epoch": 0.9370476779285013,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0713224924300391e-07,
+      "loss": 0.9607,
+      "step": 25766
+    },
+    {
+      "epoch": 0.9370840455322399,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0689382589036856e-07,
+      "loss": 0.9144,
+      "step": 25767
+    },
+    {
+      "epoch": 0.9371204131359785,
+      "grad_norm": 0.0,
+      "learning_rate": 2.066555384031843e-07,
+      "loss": 0.9391,
+      "step": 25768
+    },
+    {
+      "epoch": 0.9371567807397171,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0641738678476188e-07,
+      "loss": 1.0341,
+      "step": 25769
+    },
+    {
+      "epoch": 0.9371931483434557,
+      "grad_norm": 0.0,
+      "learning_rate": 2.061793710384019e-07,
+      "loss": 0.9405,
+      "step": 25770
+    },
+    {
+      "epoch": 0.9372295159471943,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0594149116740737e-07,
+      "loss": 0.961,
+      "step": 25771
+    },
+    {
+      "epoch": 0.9372658835509329,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0570374717507891e-07,
+      "loss": 0.9314,
+      "step": 25772
+    },
+    {
+      "epoch": 0.9373022511546715,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0546613906471612e-07,
+      "loss": 0.9192,
+      "step": 25773
+    },
+    {
+      "epoch": 0.93733861875841,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0522866683961196e-07,
+      "loss": 1.0138,
+      "step": 25774
+    },
+    {
+      "epoch": 0.9373749863621486,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0499133050306374e-07,
+      "loss": 0.9474,
+      "step": 25775
+    },
+    {
+      "epoch": 0.9374113539658872,
+      "grad_norm": 0.0,
+      "learning_rate": 2.047541300583644e-07,
+      "loss": 0.9998,
+      "step": 25776
+    },
+    {
+      "epoch": 0.9374477215696257,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0451706550880245e-07,
+      "loss": 0.9761,
+      "step": 25777
+    },
+    {
+      "epoch": 0.9374840891733643,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0428013685766857e-07,
+      "loss": 0.8573,
+      "step": 25778
+    },
+    {
+      "epoch": 0.9375204567771029,
+      "grad_norm": 0.0,
+      "learning_rate": 2.040433441082501e-07,
+      "loss": 0.9073,
+      "step": 25779
+    },
+    {
+      "epoch": 0.9375568243808415,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0380668726383112e-07,
+      "loss": 0.9312,
+      "step": 25780
+    },
+    {
+      "epoch": 0.9375931919845801,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0357016632769567e-07,
+      "loss": 0.9303,
+      "step": 25781
+    },
+    {
+      "epoch": 0.9376295595883187,
+      "grad_norm": 0.0,
+      "learning_rate": 2.033337813031244e-07,
+      "loss": 0.9071,
+      "step": 25782
+    },
+    {
+      "epoch": 0.9376659271920573,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0309753219339812e-07,
+      "loss": 0.928,
+      "step": 25783
+    },
+    {
+      "epoch": 0.9377022947957959,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0286141900179303e-07,
+      "loss": 0.7888,
+      "step": 25784
+    },
+    {
+      "epoch": 0.9377386623995345,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0262544173158428e-07,
+      "loss": 0.9162,
+      "step": 25785
+    },
+    {
+      "epoch": 0.9377750300032731,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0238960038604927e-07,
+      "loss": 0.9702,
+      "step": 25786
+    },
+    {
+      "epoch": 0.9378113976070117,
+      "grad_norm": 0.0,
+      "learning_rate": 2.021538949684554e-07,
+      "loss": 0.9469,
+      "step": 25787
+    },
+    {
+      "epoch": 0.9378477652107503,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0191832548207558e-07,
+      "loss": 0.9509,
+      "step": 25788
+    },
+    {
+      "epoch": 0.9378841328144889,
+      "grad_norm": 0.0,
+      "learning_rate": 2.016828919301783e-07,
+      "loss": 0.803,
+      "step": 25789
+    },
+    {
+      "epoch": 0.9379205004182274,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0144759431602768e-07,
+      "loss": 0.8027,
+      "step": 25790
+    },
+    {
+      "epoch": 0.937956868021966,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0121243264288993e-07,
+      "loss": 1.0,
+      "step": 25791
+    },
+    {
+      "epoch": 0.9379932356257046,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0097740691402802e-07,
+      "loss": 0.9375,
+      "step": 25792
+    },
+    {
+      "epoch": 0.9380296032294432,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0074251713270042e-07,
+      "loss": 0.907,
+      "step": 25793
+    },
+    {
+      "epoch": 0.9380659708331818,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0050776330216571e-07,
+      "loss": 0.9378,
+      "step": 25794
+    },
+    {
+      "epoch": 0.9381023384369204,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0027314542568454e-07,
+      "loss": 0.9048,
+      "step": 25795
+    },
+    {
+      "epoch": 0.938138706040659,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0003866350650768e-07,
+      "loss": 0.9083,
+      "step": 25796
+    },
+    {
+      "epoch": 0.9381750736443976,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9980431754789142e-07,
+      "loss": 0.9522,
+      "step": 25797
+    },
+    {
+      "epoch": 0.9382114412481362,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9957010755308538e-07,
+      "loss": 0.9671,
+      "step": 25798
+    },
+    {
+      "epoch": 0.9382478088518748,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9933603352533914e-07,
+      "loss": 0.8085,
+      "step": 25799
+    },
+    {
+      "epoch": 0.9382841764556134,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9910209546789905e-07,
+      "loss": 1.0215,
+      "step": 25800
+    },
+    {
+      "epoch": 0.938320544059352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9886829338401248e-07,
+      "loss": 0.9121,
+      "step": 25801
+    },
+    {
+      "epoch": 0.9383569116630905,
+      "grad_norm": 0.0,
+      "learning_rate": 1.986346272769224e-07,
+      "loss": 0.9124,
+      "step": 25802
+    },
+    {
+      "epoch": 0.9383932792668291,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9840109714987066e-07,
+      "loss": 0.9766,
+      "step": 25803
+    },
+    {
+      "epoch": 0.9384296468705677,
+      "grad_norm": 0.0,
+      "learning_rate": 1.981677030060958e-07,
+      "loss": 0.8742,
+      "step": 25804
+    },
+    {
+      "epoch": 0.9384660144743063,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9793444484883962e-07,
+      "loss": 0.9124,
+      "step": 25805
+    },
+    {
+      "epoch": 0.9385023820780449,
+      "grad_norm": 0.0,
+      "learning_rate": 1.977013226813329e-07,
+      "loss": 0.7918,
+      "step": 25806
+    },
+    {
+      "epoch": 0.9385387496817834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9746833650681418e-07,
+      "loss": 0.9262,
+      "step": 25807
+    },
+    {
+      "epoch": 0.938575117285522,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9723548632851418e-07,
+      "loss": 0.9353,
+      "step": 25808
+    },
+    {
+      "epoch": 0.9386114848892606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9700277214966368e-07,
+      "loss": 0.9609,
+      "step": 25809
+    },
+    {
+      "epoch": 0.9386478524929992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9677019397349006e-07,
+      "loss": 0.9753,
+      "step": 25810
+    },
+    {
+      "epoch": 0.9386842200967378,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9653775180322298e-07,
+      "loss": 0.9583,
+      "step": 25811
+    },
+    {
+      "epoch": 0.9387205877004764,
+      "grad_norm": 0.0,
+      "learning_rate": 1.963054456420843e-07,
+      "loss": 0.9667,
+      "step": 25812
+    },
+    {
+      "epoch": 0.938756955304215,
+      "grad_norm": 0.0,
+      "learning_rate": 1.96073275493297e-07,
+      "loss": 0.8724,
+      "step": 25813
+    },
+    {
+      "epoch": 0.9387933229079536,
+      "grad_norm": 0.0,
+      "learning_rate": 1.958412413600841e-07,
+      "loss": 0.9874,
+      "step": 25814
+    },
+    {
+      "epoch": 0.9388296905116922,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9560934324566294e-07,
+      "loss": 0.9169,
+      "step": 25815
+    },
+    {
+      "epoch": 0.9388660581154308,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9537758115325212e-07,
+      "loss": 0.9265,
+      "step": 25816
+    },
+    {
+      "epoch": 0.9389024257191694,
+      "grad_norm": 0.0,
+      "learning_rate": 1.951459550860657e-07,
+      "loss": 0.9718,
+      "step": 25817
+    },
+    {
+      "epoch": 0.938938793322908,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9491446504732004e-07,
+      "loss": 1.0557,
+      "step": 25818
+    },
+    {
+      "epoch": 0.9389751609266466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9468311104022253e-07,
+      "loss": 0.7713,
+      "step": 25819
+    },
+    {
+      "epoch": 0.9390115285303852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9445189306798507e-07,
+      "loss": 1.0759,
+      "step": 25820
+    },
+    {
+      "epoch": 0.9390478961341238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.942208111338162e-07,
+      "loss": 0.9492,
+      "step": 25821
+    },
+    {
+      "epoch": 0.9390842637378624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9398986524092e-07,
+      "loss": 0.8923,
+      "step": 25822
+    },
+    {
+      "epoch": 0.939120631341601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9375905539250172e-07,
+      "loss": 0.8245,
+      "step": 25823
+    },
+    {
+      "epoch": 0.9391569989453394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9352838159176434e-07,
+      "loss": 0.9209,
+      "step": 25824
+    },
+    {
+      "epoch": 0.939193366549078,
+      "grad_norm": 0.0,
+      "learning_rate": 1.932978438419053e-07,
+      "loss": 0.9441,
+      "step": 25825
+    },
+    {
+      "epoch": 0.9392297341528166,
+      "grad_norm": 0.0,
+      "learning_rate": 1.930674421461265e-07,
+      "loss": 0.8144,
+      "step": 25826
+    },
+    {
+      "epoch": 0.9392661017565552,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9283717650762203e-07,
+      "loss": 0.9041,
+      "step": 25827
+    },
+    {
+      "epoch": 0.9393024693602938,
+      "grad_norm": 0.0,
+      "learning_rate": 1.926070469295871e-07,
+      "loss": 0.8964,
+      "step": 25828
+    },
+    {
+      "epoch": 0.9393388369640324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9237705341521474e-07,
+      "loss": 0.9139,
+      "step": 25829
+    },
+    {
+      "epoch": 0.939375204567771,
+      "grad_norm": 0.0,
+      "learning_rate": 1.921471959676957e-07,
+      "loss": 0.9378,
+      "step": 25830
+    },
+    {
+      "epoch": 0.9394115721715096,
+      "grad_norm": 0.0,
+      "learning_rate": 1.919174745902197e-07,
+      "loss": 0.8812,
+      "step": 25831
+    },
+    {
+      "epoch": 0.9394479397752482,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9168788928597083e-07,
+      "loss": 0.974,
+      "step": 25832
+    },
+    {
+      "epoch": 0.9394843073789868,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9145844005813762e-07,
+      "loss": 0.8696,
+      "step": 25833
+    },
+    {
+      "epoch": 0.9395206749827254,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9122912690990093e-07,
+      "loss": 0.964,
+      "step": 25834
+    },
+    {
+      "epoch": 0.939557042586464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9099994984444481e-07,
+      "loss": 0.8354,
+      "step": 25835
+    },
+    {
+      "epoch": 0.9395934101902026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9077090886494565e-07,
+      "loss": 0.9634,
+      "step": 25836
+    },
+    {
+      "epoch": 0.9396297777939412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9054200397458422e-07,
+      "loss": 0.8584,
+      "step": 25837
+    },
+    {
+      "epoch": 0.9396661453976798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9031323517653355e-07,
+      "loss": 0.9383,
+      "step": 25838
+    },
+    {
+      "epoch": 0.9397025130014184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9008460247396886e-07,
+      "loss": 0.8404,
+      "step": 25839
+    },
+    {
+      "epoch": 0.939738880605157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.898561058700621e-07,
+      "loss": 0.9402,
+      "step": 25840
+    },
+    {
+      "epoch": 0.9397752482088955,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8962774536798289e-07,
+      "loss": 0.8781,
+      "step": 25841
+    },
+    {
+      "epoch": 0.9398116158126341,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8939952097089986e-07,
+      "loss": 0.836,
+      "step": 25842
+    },
+    {
+      "epoch": 0.9398479834163727,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8917143268197934e-07,
+      "loss": 0.7794,
+      "step": 25843
+    },
+    {
+      "epoch": 0.9398843510201113,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8894348050438437e-07,
+      "loss": 1.0994,
+      "step": 25844
+    },
+    {
+      "epoch": 0.9399207186238498,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8871566444127908e-07,
+      "loss": 0.9257,
+      "step": 25845
+    },
+    {
+      "epoch": 0.9399570862275884,
+      "grad_norm": 0.0,
+      "learning_rate": 1.884879844958243e-07,
+      "loss": 0.823,
+      "step": 25846
+    },
+    {
+      "epoch": 0.939993453831327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8826044067117745e-07,
+      "loss": 0.9424,
+      "step": 25847
+    },
+    {
+      "epoch": 0.9400298214350656,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8803303297049492e-07,
+      "loss": 0.912,
+      "step": 25848
+    },
+    {
+      "epoch": 0.9400661890388042,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8780576139693419e-07,
+      "loss": 0.8769,
+      "step": 25849
+    },
+    {
+      "epoch": 0.9401025566425428,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8757862595364716e-07,
+      "loss": 0.9501,
+      "step": 25850
+    },
+    {
+      "epoch": 0.9401389242462814,
+      "grad_norm": 0.0,
+      "learning_rate": 1.873516266437836e-07,
+      "loss": 0.9106,
+      "step": 25851
+    },
+    {
+      "epoch": 0.94017529185002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8712476347049423e-07,
+      "loss": 0.9871,
+      "step": 25852
+    },
+    {
+      "epoch": 0.9402116594537586,
+      "grad_norm": 0.0,
+      "learning_rate": 1.868980364369255e-07,
+      "loss": 0.9382,
+      "step": 25853
+    },
+    {
+      "epoch": 0.9402480270574972,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8667144554622486e-07,
+      "loss": 0.9467,
+      "step": 25854
+    },
+    {
+      "epoch": 0.9402843946612358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.864449908015331e-07,
+      "loss": 0.9456,
+      "step": 25855
+    },
+    {
+      "epoch": 0.9403207622649744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8621867220599443e-07,
+      "loss": 0.8877,
+      "step": 25856
+    },
+    {
+      "epoch": 0.940357129868713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8599248976274741e-07,
+      "loss": 0.9121,
+      "step": 25857
+    },
+    {
+      "epoch": 0.9403934974724515,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8576644347493067e-07,
+      "loss": 0.9455,
+      "step": 25858
+    },
+    {
+      "epoch": 0.9404298650761901,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8554053334567945e-07,
+      "loss": 1.0425,
+      "step": 25859
+    },
+    {
+      "epoch": 0.9404662326799287,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8531475937812905e-07,
+      "loss": 0.9823,
+      "step": 25860
+    },
+    {
+      "epoch": 0.9405026002836673,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8508912157541025e-07,
+      "loss": 0.7992,
+      "step": 25861
+    },
+    {
+      "epoch": 0.9405389678874059,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8486361994065505e-07,
+      "loss": 0.8777,
+      "step": 25862
+    },
+    {
+      "epoch": 0.9405753354911445,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8463825447699092e-07,
+      "loss": 0.9106,
+      "step": 25863
+    },
+    {
+      "epoch": 0.9406117030948831,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8441302518754533e-07,
+      "loss": 0.8277,
+      "step": 25864
+    },
+    {
+      "epoch": 0.9406480706986217,
+      "grad_norm": 0.0,
+      "learning_rate": 1.841879320754425e-07,
+      "loss": 0.8693,
+      "step": 25865
+    },
+    {
+      "epoch": 0.9406844383023603,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8396297514380545e-07,
+      "loss": 0.8416,
+      "step": 25866
+    },
+    {
+      "epoch": 0.9407208059060989,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8373815439575283e-07,
+      "loss": 0.8759,
+      "step": 25867
+    },
+    {
+      "epoch": 0.9407571735098375,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8351346983440876e-07,
+      "loss": 0.8406,
+      "step": 25868
+    },
+    {
+      "epoch": 0.940793541113576,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8328892146288635e-07,
+      "loss": 0.8965,
+      "step": 25869
+    },
+    {
+      "epoch": 0.9408299087173146,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8306450928430197e-07,
+      "loss": 0.9497,
+      "step": 25870
+    },
+    {
+      "epoch": 0.9408662763210532,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8284023330176982e-07,
+      "loss": 0.8008,
+      "step": 25871
+    },
+    {
+      "epoch": 0.9409026439247918,
+      "grad_norm": 0.0,
+      "learning_rate": 1.826160935183996e-07,
+      "loss": 0.8991,
+      "step": 25872
+    },
+    {
+      "epoch": 0.9409390115285304,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8239208993730327e-07,
+      "loss": 0.9774,
+      "step": 25873
+    },
+    {
+      "epoch": 0.940975379132269,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8216822256158507e-07,
+      "loss": 0.9154,
+      "step": 25874
+    },
+    {
+      "epoch": 0.9410117467360075,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8194449139435576e-07,
+      "loss": 0.895,
+      "step": 25875
+    },
+    {
+      "epoch": 0.9410481143397461,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8172089643871517e-07,
+      "loss": 0.8978,
+      "step": 25876
+    },
+    {
+      "epoch": 0.9410844819434847,
+      "grad_norm": 0.0,
+      "learning_rate": 1.814974376977674e-07,
+      "loss": 0.9043,
+      "step": 25877
+    },
+    {
+      "epoch": 0.9411208495472233,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8127411517461114e-07,
+      "loss": 0.965,
+      "step": 25878
+    },
+    {
+      "epoch": 0.9411572171509619,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8105092887234722e-07,
+      "loss": 0.9273,
+      "step": 25879
+    },
+    {
+      "epoch": 0.9411935847547005,
+      "grad_norm": 0.0,
+      "learning_rate": 1.808278787940687e-07,
+      "loss": 0.9749,
+      "step": 25880
+    },
+    {
+      "epoch": 0.9412299523584391,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8060496494287205e-07,
+      "loss": 0.8604,
+      "step": 25881
+    },
+    {
+      "epoch": 0.9412663199621777,
+      "grad_norm": 0.0,
+      "learning_rate": 1.803821873218503e-07,
+      "loss": 0.9909,
+      "step": 25882
+    },
+    {
+      "epoch": 0.9413026875659163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8015954593409323e-07,
+      "loss": 0.9536,
+      "step": 25883
+    },
+    {
+      "epoch": 0.9413390551696549,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7993704078268948e-07,
+      "loss": 0.9216,
+      "step": 25884
+    },
+    {
+      "epoch": 0.9413754227733935,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7971467187072765e-07,
+      "loss": 0.7867,
+      "step": 25885
+    },
+    {
+      "epoch": 0.9414117903771321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.794924392012909e-07,
+      "loss": 0.9158,
+      "step": 25886
+    },
+    {
+      "epoch": 0.9414481579808707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7927034277746225e-07,
+      "loss": 1.0512,
+      "step": 25887
+    },
+    {
+      "epoch": 0.9414845255846093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7904838260232594e-07,
+      "loss": 0.8593,
+      "step": 25888
+    },
+    {
+      "epoch": 0.9415208931883479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7882655867895726e-07,
+      "loss": 0.9993,
+      "step": 25889
+    },
+    {
+      "epoch": 0.9415572607920865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.78604871010436e-07,
+      "loss": 0.9464,
+      "step": 25890
+    },
+    {
+      "epoch": 0.9415936283958249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7838331959983857e-07,
+      "loss": 0.9678,
+      "step": 25891
+    },
+    {
+      "epoch": 0.9416299959995635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7816190445023695e-07,
+      "loss": 0.7678,
+      "step": 25892
+    },
+    {
+      "epoch": 0.9416663636033021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7794062556470205e-07,
+      "loss": 0.956,
+      "step": 25893
+    },
+    {
+      "epoch": 0.9417027312070407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.777194829463069e-07,
+      "loss": 0.8873,
+      "step": 25894
+    },
+    {
+      "epoch": 0.9417390988107793,
+      "grad_norm": 0.0,
+      "learning_rate": 1.774984765981169e-07,
+      "loss": 0.7882,
+      "step": 25895
+    },
+    {
+      "epoch": 0.9417754664145179,
+      "grad_norm": 0.0,
+      "learning_rate": 1.772776065232007e-07,
+      "loss": 0.813,
+      "step": 25896
+    },
+    {
+      "epoch": 0.9418118340182565,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7705687272462026e-07,
+      "loss": 0.8807,
+      "step": 25897
+    },
+    {
+      "epoch": 0.9418482016219951,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7683627520543866e-07,
+      "loss": 0.9918,
+      "step": 25898
+    },
+    {
+      "epoch": 0.9418845692257337,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7661581396871575e-07,
+      "loss": 0.9095,
+      "step": 25899
+    },
+    {
+      "epoch": 0.9419209368294723,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7639548901751237e-07,
+      "loss": 0.9329,
+      "step": 25900
+    },
+    {
+      "epoch": 0.9419573044332109,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7617530035488272e-07,
+      "loss": 0.957,
+      "step": 25901
+    },
+    {
+      "epoch": 0.9419936720369495,
+      "grad_norm": 0.0,
+      "learning_rate": 1.759552479838833e-07,
+      "loss": 0.996,
+      "step": 25902
+    },
+    {
+      "epoch": 0.9420300396406881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7573533190756498e-07,
+      "loss": 0.8454,
+      "step": 25903
+    },
+    {
+      "epoch": 0.9420664072444267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7551555212898198e-07,
+      "loss": 0.8224,
+      "step": 25904
+    },
+    {
+      "epoch": 0.9421027748481653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7529590865117963e-07,
+      "loss": 0.9284,
+      "step": 25905
+    },
+    {
+      "epoch": 0.9421391424519039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7507640147720883e-07,
+      "loss": 0.7977,
+      "step": 25906
+    },
+    {
+      "epoch": 0.9421755100556425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7485703061011271e-07,
+      "loss": 0.8217,
+      "step": 25907
+    },
+    {
+      "epoch": 0.942211877659381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.746377960529344e-07,
+      "loss": 0.9325,
+      "step": 25908
+    },
+    {
+      "epoch": 0.9422482452631196,
+      "grad_norm": 0.0,
+      "learning_rate": 1.74418697808717e-07,
+      "loss": 0.92,
+      "step": 25909
+    },
+    {
+      "epoch": 0.9422846128668582,
+      "grad_norm": 0.0,
+      "learning_rate": 1.741997358805003e-07,
+      "loss": 0.9504,
+      "step": 25910
+    },
+    {
+      "epoch": 0.9423209804705968,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7398091027131968e-07,
+      "loss": 0.9913,
+      "step": 25911
+    },
+    {
+      "epoch": 0.9423573480743354,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7376222098421268e-07,
+      "loss": 1.0482,
+      "step": 25912
+    },
+    {
+      "epoch": 0.942393715678074,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7354366802221468e-07,
+      "loss": 0.8673,
+      "step": 25913
+    },
+    {
+      "epoch": 0.9424300832818125,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7332525138835543e-07,
+      "loss": 0.9037,
+      "step": 25914
+    },
+    {
+      "epoch": 0.9424664508855511,
+      "grad_norm": 0.0,
+      "learning_rate": 1.73106971085667e-07,
+      "loss": 0.8949,
+      "step": 25915
+    },
+    {
+      "epoch": 0.9425028184892897,
+      "grad_norm": 0.0,
+      "learning_rate": 1.728888271171747e-07,
+      "loss": 0.9443,
+      "step": 25916
+    },
+    {
+      "epoch": 0.9425391860930283,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7267081948590948e-07,
+      "loss": 0.9598,
+      "step": 25917
+    },
+    {
+      "epoch": 0.9425755536967669,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7245294819489223e-07,
+      "loss": 0.9599,
+      "step": 25918
+    },
+    {
+      "epoch": 0.9426119213005055,
+      "grad_norm": 0.0,
+      "learning_rate": 1.722352132471472e-07,
+      "loss": 0.8573,
+      "step": 25919
+    },
+    {
+      "epoch": 0.9426482889042441,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7201761464569422e-07,
+      "loss": 0.9815,
+      "step": 25920
+    },
+    {
+      "epoch": 0.9426846565079827,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7180015239355418e-07,
+      "loss": 0.9682,
+      "step": 25921
+    },
+    {
+      "epoch": 0.9427210241117213,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7158282649374026e-07,
+      "loss": 0.8695,
+      "step": 25922
+    },
+    {
+      "epoch": 0.9427573917154599,
+      "grad_norm": 0.0,
+      "learning_rate": 1.713656369492722e-07,
+      "loss": 0.9622,
+      "step": 25923
+    },
+    {
+      "epoch": 0.9427937593191985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.711485837631599e-07,
+      "loss": 0.8236,
+      "step": 25924
+    },
+    {
+      "epoch": 0.942830126922937,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7093166693841646e-07,
+      "loss": 0.889,
+      "step": 25925
+    },
+    {
+      "epoch": 0.9428664945266756,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7071488647804945e-07,
+      "loss": 0.8918,
+      "step": 25926
+    },
+    {
+      "epoch": 0.9429028621304142,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7049824238506764e-07,
+      "loss": 0.9111,
+      "step": 25927
+    },
+    {
+      "epoch": 0.9429392297341528,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7028173466247744e-07,
+      "loss": 0.8655,
+      "step": 25928
+    },
+    {
+      "epoch": 0.9429755973378914,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7006536331327982e-07,
+      "loss": 0.8955,
+      "step": 25929
+    },
+    {
+      "epoch": 0.94301196494163,
+      "grad_norm": 0.0,
+      "learning_rate": 1.698491283404813e-07,
+      "loss": 0.8918,
+      "step": 25930
+    },
+    {
+      "epoch": 0.9430483325453686,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6963302974707607e-07,
+      "loss": 0.8928,
+      "step": 25931
+    },
+    {
+      "epoch": 0.9430847001491072,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6941706753606623e-07,
+      "loss": 0.9224,
+      "step": 25932
+    },
+    {
+      "epoch": 0.9431210677528458,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6920124171044717e-07,
+      "loss": 0.916,
+      "step": 25933
+    },
+    {
+      "epoch": 0.9431574353565844,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6898555227321312e-07,
+      "loss": 0.861,
+      "step": 25934
+    },
+    {
+      "epoch": 0.943193802960323,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6876999922735393e-07,
+      "loss": 1.0078,
+      "step": 25935
+    },
+    {
+      "epoch": 0.9432301705640616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.68554582575865e-07,
+      "loss": 0.9187,
+      "step": 25936
+    },
+    {
+      "epoch": 0.9432665381678002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6833930232173056e-07,
+      "loss": 1.054,
+      "step": 25937
+    },
+    {
+      "epoch": 0.9433029057715387,
+      "grad_norm": 0.0,
+      "learning_rate": 1.681241584679394e-07,
+      "loss": 0.9829,
+      "step": 25938
+    },
+    {
+      "epoch": 0.9433392733752773,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6790915101747685e-07,
+      "loss": 0.8693,
+      "step": 25939
+    },
+    {
+      "epoch": 0.9433756409790159,
+      "grad_norm": 0.0,
+      "learning_rate": 1.676942799733239e-07,
+      "loss": 0.8259,
+      "step": 25940
+    },
+    {
+      "epoch": 0.9434120085827545,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6747954533846257e-07,
+      "loss": 0.8685,
+      "step": 25941
+    },
+    {
+      "epoch": 0.943448376186493,
+      "grad_norm": 0.0,
+      "learning_rate": 1.672649471158727e-07,
+      "loss": 0.9389,
+      "step": 25942
+    },
+    {
+      "epoch": 0.9434847437902316,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6705048530852973e-07,
+      "loss": 0.8957,
+      "step": 25943
+    },
+    {
+      "epoch": 0.9435211113939702,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6683615991941127e-07,
+      "loss": 0.8627,
+      "step": 25944
+    },
+    {
+      "epoch": 0.9435574789977088,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6662197095148936e-07,
+      "loss": 0.9841,
+      "step": 25945
+    },
+    {
+      "epoch": 0.9435938466014474,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6640791840773608e-07,
+      "loss": 0.9548,
+      "step": 25946
+    },
+    {
+      "epoch": 0.943630214205186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6619400229112016e-07,
+      "loss": 0.9353,
+      "step": 25947
+    },
+    {
+      "epoch": 0.9436665818089246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6598022260461034e-07,
+      "loss": 0.8815,
+      "step": 25948
+    },
+    {
+      "epoch": 0.9437029494126632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6576657935117312e-07,
+      "loss": 0.8975,
+      "step": 25949
+    },
+    {
+      "epoch": 0.9437393170164018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6555307253377062e-07,
+      "loss": 0.8784,
+      "step": 25950
+    },
+    {
+      "epoch": 0.9437756846201404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.65339702155366e-07,
+      "loss": 0.8543,
+      "step": 25951
+    },
+    {
+      "epoch": 0.943812052223879,
+      "grad_norm": 0.0,
+      "learning_rate": 1.651264682189202e-07,
+      "loss": 0.9117,
+      "step": 25952
+    },
+    {
+      "epoch": 0.9438484198276176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6491337072739087e-07,
+      "loss": 0.8541,
+      "step": 25953
+    },
+    {
+      "epoch": 0.9438847874313562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6470040968373234e-07,
+      "loss": 0.9732,
+      "step": 25954
+    },
+    {
+      "epoch": 0.9439211550350948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6448758509090446e-07,
+      "loss": 0.9641,
+      "step": 25955
+    },
+    {
+      "epoch": 0.9439575226388334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6427489695185372e-07,
+      "loss": 0.9068,
+      "step": 25956
+    },
+    {
+      "epoch": 0.943993890242572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6406234526953558e-07,
+      "loss": 0.9208,
+      "step": 25957
+    },
+    {
+      "epoch": 0.9440302578463106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6384993004689544e-07,
+      "loss": 0.9074,
+      "step": 25958
+    },
+    {
+      "epoch": 0.944066625450049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6363765128688313e-07,
+      "loss": 0.9528,
+      "step": 25959
+    },
+    {
+      "epoch": 0.9441029930537876,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6342550899244191e-07,
+      "loss": 0.9745,
+      "step": 25960
+    },
+    {
+      "epoch": 0.9441393606575262,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6321350316651497e-07,
+      "loss": 0.8417,
+      "step": 25961
+    },
+    {
+      "epoch": 0.9441757282612648,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6300163381204547e-07,
+      "loss": 0.7612,
+      "step": 25962
+    },
+    {
+      "epoch": 0.9442120958650034,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6278990093197e-07,
+      "loss": 0.944,
+      "step": 25963
+    },
+    {
+      "epoch": 0.944248463468742,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6257830452922841e-07,
+      "loss": 0.9228,
+      "step": 25964
+    },
+    {
+      "epoch": 0.9442848310724806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.62366844606755e-07,
+      "loss": 0.8941,
+      "step": 25965
+    },
+    {
+      "epoch": 0.9443211986762192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.621555211674841e-07,
+      "loss": 0.9133,
+      "step": 25966
+    },
+    {
+      "epoch": 0.9443575662799578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6194433421434675e-07,
+      "loss": 0.849,
+      "step": 25967
+    },
+    {
+      "epoch": 0.9443939338836964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6173328375027497e-07,
+      "loss": 0.8908,
+      "step": 25968
+    },
+    {
+      "epoch": 0.944430301487435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6152236977819312e-07,
+      "loss": 0.9482,
+      "step": 25969
+    },
+    {
+      "epoch": 0.9444666690911736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6131159230103e-07,
+      "loss": 0.8752,
+      "step": 25970
+    },
+    {
+      "epoch": 0.9445030366949122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6110095132170988e-07,
+      "loss": 0.9592,
+      "step": 25971
+    },
+    {
+      "epoch": 0.9445394042986508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6089044684315492e-07,
+      "loss": 0.8626,
+      "step": 25972
+    },
+    {
+      "epoch": 0.9445757719023894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6068007886828275e-07,
+      "loss": 0.8279,
+      "step": 25973
+    },
+    {
+      "epoch": 0.944612139506128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6046984740001658e-07,
+      "loss": 0.9829,
+      "step": 25974
+    },
+    {
+      "epoch": 0.9446485071098665,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6025975244126968e-07,
+      "loss": 0.9606,
+      "step": 25975
+    },
+    {
+      "epoch": 0.9446848747136051,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6004979399495857e-07,
+      "loss": 0.8371,
+      "step": 25976
+    },
+    {
+      "epoch": 0.9447212423173437,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5983997206399647e-07,
+      "loss": 0.9591,
+      "step": 25977
+    },
+    {
+      "epoch": 0.9447576099210823,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5963028665129222e-07,
+      "loss": 0.9744,
+      "step": 25978
+    },
+    {
+      "epoch": 0.9447939775248209,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5942073775975563e-07,
+      "loss": 0.8534,
+      "step": 25979
+    },
+    {
+      "epoch": 0.9448303451285595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5921132539229444e-07,
+      "loss": 0.923,
+      "step": 25980
+    },
+    {
+      "epoch": 0.944866712732298,
+      "grad_norm": 0.0,
+      "learning_rate": 1.590020495518141e-07,
+      "loss": 1.0304,
+      "step": 25981
+    },
+    {
+      "epoch": 0.9449030803360366,
+      "grad_norm": 0.0,
+      "learning_rate": 1.587929102412189e-07,
+      "loss": 1.013,
+      "step": 25982
+    },
+    {
+      "epoch": 0.9449394479397752,
+      "grad_norm": 0.0,
+      "learning_rate": 1.585839074634077e-07,
+      "loss": 0.7228,
+      "step": 25983
+    },
+    {
+      "epoch": 0.9449758155435138,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5837504122128254e-07,
+      "loss": 0.9296,
+      "step": 25984
+    },
+    {
+      "epoch": 0.9450121831472524,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5816631151774008e-07,
+      "loss": 0.9182,
+      "step": 25985
+    },
+    {
+      "epoch": 0.945048550750991,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5795771835567575e-07,
+      "loss": 0.9437,
+      "step": 25986
+    },
+    {
+      "epoch": 0.9450849183547296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5774926173798499e-07,
+      "loss": 0.9987,
+      "step": 25987
+    },
+    {
+      "epoch": 0.9451212859584682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5754094166755772e-07,
+      "loss": 0.8602,
+      "step": 25988
+    },
+    {
+      "epoch": 0.9451576535622068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.573327581472861e-07,
+      "loss": 0.9891,
+      "step": 25989
+    },
+    {
+      "epoch": 0.9451940211659454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.571247111800578e-07,
+      "loss": 0.8065,
+      "step": 25990
+    },
+    {
+      "epoch": 0.945230388769684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5691680076875825e-07,
+      "loss": 0.8289,
+      "step": 25991
+    },
+    {
+      "epoch": 0.9452667563734225,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5670902691627187e-07,
+      "loss": 0.8896,
+      "step": 25992
+    },
+    {
+      "epoch": 0.9453031239771611,
+      "grad_norm": 0.0,
+      "learning_rate": 1.56501389625483e-07,
+      "loss": 0.9135,
+      "step": 25993
+    },
+    {
+      "epoch": 0.9453394915808997,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5629388889927156e-07,
+      "loss": 0.8898,
+      "step": 25994
+    },
+    {
+      "epoch": 0.9453758591846383,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5608652474051523e-07,
+      "loss": 0.9498,
+      "step": 25995
+    },
+    {
+      "epoch": 0.9454122267883769,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5587929715209282e-07,
+      "loss": 0.8958,
+      "step": 25996
+    },
+    {
+      "epoch": 0.9454485943921155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.556722061368776e-07,
+      "loss": 0.9723,
+      "step": 25997
+    },
+    {
+      "epoch": 0.9454849619958541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5546525169774173e-07,
+      "loss": 0.8487,
+      "step": 25998
+    },
+    {
+      "epoch": 0.9455213295995927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5525843383755955e-07,
+      "loss": 0.8366,
+      "step": 25999
+    },
+    {
+      "epoch": 0.9455576972033313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.550517525591988e-07,
+      "loss": 1.0121,
+      "step": 26000
+    },
+    {
+      "epoch": 0.9455940648070699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5484520786552605e-07,
+      "loss": 0.8984,
+      "step": 26001
+    },
+    {
+      "epoch": 0.9456304324108085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5463879975940786e-07,
+      "loss": 0.8385,
+      "step": 26002
+    },
+    {
+      "epoch": 0.9456668000145471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.544325282437087e-07,
+      "loss": 1.0755,
+      "step": 26003
+    },
+    {
+      "epoch": 0.9457031676182857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.542263933212884e-07,
+      "loss": 0.9448,
+      "step": 26004
+    },
+    {
+      "epoch": 0.9457395352220243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5402039499500587e-07,
+      "loss": 0.9359,
+      "step": 26005
+    },
+    {
+      "epoch": 0.9457759028257628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5381453326772322e-07,
+      "loss": 0.9686,
+      "step": 26006
+    },
+    {
+      "epoch": 0.9458122704295014,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5360880814229262e-07,
+      "loss": 1.0251,
+      "step": 26007
+    },
+    {
+      "epoch": 0.94584863803324,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5340321962157068e-07,
+      "loss": 0.839,
+      "step": 26008
+    },
+    {
+      "epoch": 0.9458850056369785,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5319776770840732e-07,
+      "loss": 0.9507,
+      "step": 26009
+    },
+    {
+      "epoch": 0.9459213732407171,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5299245240565474e-07,
+      "loss": 0.9478,
+      "step": 26010
+    },
+    {
+      "epoch": 0.9459577408444557,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5278727371616065e-07,
+      "loss": 0.8197,
+      "step": 26011
+    },
+    {
+      "epoch": 0.9459941084481943,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5258223164277164e-07,
+      "loss": 0.8136,
+      "step": 26012
+    },
+    {
+      "epoch": 0.9460304760519329,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5237732618833323e-07,
+      "loss": 0.9561,
+      "step": 26013
+    },
+    {
+      "epoch": 0.9460668436556715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.521725573556876e-07,
+      "loss": 0.8825,
+      "step": 26014
+    },
+    {
+      "epoch": 0.9461032112594101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5196792514767354e-07,
+      "loss": 0.9235,
+      "step": 26015
+    },
+    {
+      "epoch": 0.9461395788631487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.517634295671344e-07,
+      "loss": 0.8647,
+      "step": 26016
+    },
+    {
+      "epoch": 0.9461759464668873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5155907061690344e-07,
+      "loss": 0.9415,
+      "step": 26017
+    },
+    {
+      "epoch": 0.9462123140706259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.513548482998173e-07,
+      "loss": 0.9764,
+      "step": 26018
+    },
+    {
+      "epoch": 0.9462486816743645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.511507626187103e-07,
+      "loss": 0.9289,
+      "step": 26019
+    },
+    {
+      "epoch": 0.9462850492781031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.509468135764114e-07,
+      "loss": 0.8513,
+      "step": 26020
+    },
+    {
+      "epoch": 0.9463214168818417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5074300117575157e-07,
+      "loss": 0.896,
+      "step": 26021
+    },
+    {
+      "epoch": 0.9463577844855803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5053932541955973e-07,
+      "loss": 0.8892,
+      "step": 26022
+    },
+    {
+      "epoch": 0.9463941520893189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5033578631066026e-07,
+      "loss": 0.8129,
+      "step": 26023
+    },
+    {
+      "epoch": 0.9464305196930575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5013238385187535e-07,
+      "loss": 0.8602,
+      "step": 26024
+    },
+    {
+      "epoch": 0.9464668872967961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4992911804602938e-07,
+      "loss": 0.8836,
+      "step": 26025
+    },
+    {
+      "epoch": 0.9465032549005346,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4972598889594125e-07,
+      "loss": 1.0021,
+      "step": 26026
+    },
+    {
+      "epoch": 0.9465396225042731,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4952299640442869e-07,
+      "loss": 0.951,
+      "step": 26027
+    },
+    {
+      "epoch": 0.9465759901080117,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4932014057430832e-07,
+      "loss": 0.9247,
+      "step": 26028
+    },
+    {
+      "epoch": 0.9466123577117503,
+      "grad_norm": 0.0,
+      "learning_rate": 1.491174214083968e-07,
+      "loss": 0.9263,
+      "step": 26029
+    },
+    {
+      "epoch": 0.9466487253154889,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4891483890950298e-07,
+      "loss": 0.8823,
+      "step": 26030
+    },
+    {
+      "epoch": 0.9466850929192275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4871239308043905e-07,
+      "loss": 1.0099,
+      "step": 26031
+    },
+    {
+      "epoch": 0.9467214605229661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.485100839240139e-07,
+      "loss": 0.9898,
+      "step": 26032
+    },
+    {
+      "epoch": 0.9467578281267047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4830791144303413e-07,
+      "loss": 1.0631,
+      "step": 26033
+    },
+    {
+      "epoch": 0.9467941957304433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.481058756403031e-07,
+      "loss": 0.8423,
+      "step": 26034
+    },
+    {
+      "epoch": 0.9468305633341819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.479039765186263e-07,
+      "loss": 0.8296,
+      "step": 26035
+    },
+    {
+      "epoch": 0.9468669309379205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4770221408080265e-07,
+      "loss": 0.9875,
+      "step": 26036
+    },
+    {
+      "epoch": 0.9469032985416591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.475005883296321e-07,
+      "loss": 1.0204,
+      "step": 26037
+    },
+    {
+      "epoch": 0.9469396661453977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4729909926791352e-07,
+      "loss": 0.9028,
+      "step": 26038
+    },
+    {
+      "epoch": 0.9469760337491363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4709774689843915e-07,
+      "loss": 0.7617,
+      "step": 26039
+    },
+    {
+      "epoch": 0.9470124013528749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.468965312240045e-07,
+      "loss": 0.8213,
+      "step": 26040
+    },
+    {
+      "epoch": 0.9470487689566135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4669545224740178e-07,
+      "loss": 0.854,
+      "step": 26041
+    },
+    {
+      "epoch": 0.9470851365603521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.464945099714188e-07,
+      "loss": 0.9868,
+      "step": 26042
+    },
+    {
+      "epoch": 0.9471215041640906,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4629370439884326e-07,
+      "loss": 0.9269,
+      "step": 26043
+    },
+    {
+      "epoch": 0.9471578717678292,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4609303553246302e-07,
+      "loss": 0.8631,
+      "step": 26044
+    },
+    {
+      "epoch": 0.9471942393715678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4589250337506134e-07,
+      "loss": 0.8688,
+      "step": 26045
+    },
+    {
+      "epoch": 0.9472306069753064,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4569210792942045e-07,
+      "loss": 0.8082,
+      "step": 26046
+    },
+    {
+      "epoch": 0.947266974579045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4549184919831815e-07,
+      "loss": 1.0262,
+      "step": 26047
+    },
+    {
+      "epoch": 0.9473033421827836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4529172718453665e-07,
+      "loss": 0.9671,
+      "step": 26048
+    },
+    {
+      "epoch": 0.9473397097865222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.450917418908493e-07,
+      "loss": 0.9878,
+      "step": 26049
+    },
+    {
+      "epoch": 0.9473760773902608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4489189332003272e-07,
+      "loss": 0.8688,
+      "step": 26050
+    },
+    {
+      "epoch": 0.9474124449939993,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4469218147485808e-07,
+      "loss": 0.844,
+      "step": 26051
+    },
+    {
+      "epoch": 0.9474488125977379,
+      "grad_norm": 0.0,
+      "learning_rate": 1.444926063580965e-07,
+      "loss": 0.906,
+      "step": 26052
+    },
+    {
+      "epoch": 0.9474851802014765,
+      "grad_norm": 0.0,
+      "learning_rate": 1.442931679725157e-07,
+      "loss": 0.8777,
+      "step": 26053
+    },
+    {
+      "epoch": 0.9475215478052151,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4409386632088574e-07,
+      "loss": 0.8737,
+      "step": 26054
+    },
+    {
+      "epoch": 0.9475579154089537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4389470140596884e-07,
+      "loss": 0.6821,
+      "step": 26055
+    },
+    {
+      "epoch": 0.9475942830126923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4369567323052946e-07,
+      "loss": 0.9036,
+      "step": 26056
+    },
+    {
+      "epoch": 0.9476306506164309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4349678179732763e-07,
+      "loss": 0.9348,
+      "step": 26057
+    },
+    {
+      "epoch": 0.9476670182201695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4329802710912332e-07,
+      "loss": 0.8709,
+      "step": 26058
+    },
+    {
+      "epoch": 0.9477033858239081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.430994091686744e-07,
+      "loss": 0.8781,
+      "step": 26059
+    },
+    {
+      "epoch": 0.9477397534276466,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4290092797873634e-07,
+      "loss": 0.8988,
+      "step": 26060
+    },
+    {
+      "epoch": 0.9477761210313852,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4270258354206257e-07,
+      "loss": 0.806,
+      "step": 26061
+    },
+    {
+      "epoch": 0.9478124886351238,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4250437586140419e-07,
+      "loss": 0.9194,
+      "step": 26062
+    },
+    {
+      "epoch": 0.9478488562388624,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4230630493951126e-07,
+      "loss": 0.9351,
+      "step": 26063
+    },
+    {
+      "epoch": 0.947885223842601,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4210837077913376e-07,
+      "loss": 0.7681,
+      "step": 26064
+    },
+    {
+      "epoch": 0.9479215914463396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4191057338301506e-07,
+      "loss": 0.8102,
+      "step": 26065
+    },
+    {
+      "epoch": 0.9479579590500782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4171291275389965e-07,
+      "loss": 0.8431,
+      "step": 26066
+    },
+    {
+      "epoch": 0.9479943266538168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4151538889453197e-07,
+      "loss": 0.9371,
+      "step": 26067
+    },
+    {
+      "epoch": 0.9480306942575554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4131800180764988e-07,
+      "loss": 0.9373,
+      "step": 26068
+    },
+    {
+      "epoch": 0.948067061861294,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4112075149599336e-07,
+      "loss": 0.8821,
+      "step": 26069
+    },
+    {
+      "epoch": 0.9481034294650326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4092363796229802e-07,
+      "loss": 0.9052,
+      "step": 26070
+    },
+    {
+      "epoch": 0.9481397970687712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4072666120929946e-07,
+      "loss": 0.8675,
+      "step": 26071
+    },
+    {
+      "epoch": 0.9481761646725098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4052982123972881e-07,
+      "loss": 0.9398,
+      "step": 26072
+    },
+    {
+      "epoch": 0.9482125322762484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4033311805631943e-07,
+      "loss": 1.025,
+      "step": 26073
+    },
+    {
+      "epoch": 0.948248899879987,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4013655166179917e-07,
+      "loss": 0.9426,
+      "step": 26074
+    },
+    {
+      "epoch": 0.9482852674837255,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3994012205889474e-07,
+      "loss": 0.8397,
+      "step": 26075
+    },
+    {
+      "epoch": 0.948321635087464,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3974382925033058e-07,
+      "loss": 0.891,
+      "step": 26076
+    },
+    {
+      "epoch": 0.9483580026912026,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3954767323883233e-07,
+      "loss": 0.8057,
+      "step": 26077
+    },
+    {
+      "epoch": 0.9483943702949412,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3935165402711782e-07,
+      "loss": 0.9624,
+      "step": 26078
+    },
+    {
+      "epoch": 0.9484307378986798,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3915577161791037e-07,
+      "loss": 1.0054,
+      "step": 26079
+    },
+    {
+      "epoch": 0.9484671055024184,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3896002601392677e-07,
+      "loss": 0.8655,
+      "step": 26080
+    },
+    {
+      "epoch": 0.948503473106157,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3876441721788037e-07,
+      "loss": 0.8951,
+      "step": 26081
+    },
+    {
+      "epoch": 0.9485398407098956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3856894523248676e-07,
+      "loss": 0.8981,
+      "step": 26082
+    },
+    {
+      "epoch": 0.9485762083136342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3837361006045714e-07,
+      "loss": 0.8995,
+      "step": 26083
+    },
+    {
+      "epoch": 0.9486125759173728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3817841170450262e-07,
+      "loss": 0.9159,
+      "step": 26084
+    },
+    {
+      "epoch": 0.9486489435211114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.379833501673289e-07,
+      "loss": 0.9257,
+      "step": 26085
+    },
+    {
+      "epoch": 0.94868531112485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3778842545164596e-07,
+      "loss": 0.9536,
+      "step": 26086
+    },
+    {
+      "epoch": 0.9487216787285886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3759363756015497e-07,
+      "loss": 0.9546,
+      "step": 26087
+    },
+    {
+      "epoch": 0.9487580463323272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3739898649555938e-07,
+      "loss": 0.9348,
+      "step": 26088
+    },
+    {
+      "epoch": 0.9487944139360658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3720447226055923e-07,
+      "loss": 0.8363,
+      "step": 26089
+    },
+    {
+      "epoch": 0.9488307815398044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3701009485785454e-07,
+      "loss": 0.8785,
+      "step": 26090
+    },
+    {
+      "epoch": 0.948867149143543,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3681585429013989e-07,
+      "loss": 0.8486,
+      "step": 26091
+    },
+    {
+      "epoch": 0.9489035167472816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3662175056011196e-07,
+      "loss": 0.8422,
+      "step": 26092
+    },
+    {
+      "epoch": 0.9489398843510201,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3642778367046193e-07,
+      "loss": 0.996,
+      "step": 26093
+    },
+    {
+      "epoch": 0.9489762519547587,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3623395362388324e-07,
+      "loss": 1.0163,
+      "step": 26094
+    },
+    {
+      "epoch": 0.9490126195584973,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3604026042306263e-07,
+      "loss": 0.993,
+      "step": 26095
+    },
+    {
+      "epoch": 0.9490489871622358,
+      "grad_norm": 0.0,
+      "learning_rate": 1.35846704070689e-07,
+      "loss": 0.9502,
+      "step": 26096
+    },
+    {
+      "epoch": 0.9490853547659744,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3565328456944583e-07,
+      "loss": 0.9677,
+      "step": 26097
+    },
+    {
+      "epoch": 0.949121722369713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3546000192201868e-07,
+      "loss": 0.9032,
+      "step": 26098
+    },
+    {
+      "epoch": 0.9491580899734516,
+      "grad_norm": 0.0,
+      "learning_rate": 1.352668561310888e-07,
+      "loss": 0.8997,
+      "step": 26099
+    },
+    {
+      "epoch": 0.9491944575771902,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3507384719933402e-07,
+      "loss": 0.8454,
+      "step": 26100
+    },
+    {
+      "epoch": 0.9492308251809288,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3488097512943332e-07,
+      "loss": 0.9635,
+      "step": 26101
+    },
+    {
+      "epoch": 0.9492671927846674,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3468823992406232e-07,
+      "loss": 0.8731,
+      "step": 26102
+    },
+    {
+      "epoch": 0.949303560388406,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3449564158589445e-07,
+      "loss": 0.9184,
+      "step": 26103
+    },
+    {
+      "epoch": 0.9493399279921446,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3430318011760203e-07,
+      "loss": 0.8107,
+      "step": 26104
+    },
+    {
+      "epoch": 0.9493762955958832,
+      "grad_norm": 0.0,
+      "learning_rate": 1.341108555218562e-07,
+      "loss": 0.9312,
+      "step": 26105
+    },
+    {
+      "epoch": 0.9494126631996218,
+      "grad_norm": 0.0,
+      "learning_rate": 1.339186678013238e-07,
+      "loss": 0.9935,
+      "step": 26106
+    },
+    {
+      "epoch": 0.9494490308033604,
+      "grad_norm": 0.0,
+      "learning_rate": 1.337266169586726e-07,
+      "loss": 0.8886,
+      "step": 26107
+    },
+    {
+      "epoch": 0.949485398407099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3353470299656612e-07,
+      "loss": 1.0186,
+      "step": 26108
+    },
+    {
+      "epoch": 0.9495217660108376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3334292591766551e-07,
+      "loss": 0.8497,
+      "step": 26109
+    },
+    {
+      "epoch": 0.9495581336145761,
+      "grad_norm": 0.0,
+      "learning_rate": 1.331512857246342e-07,
+      "loss": 0.8477,
+      "step": 26110
+    },
+    {
+      "epoch": 0.9495945012183147,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3295978242012897e-07,
+      "loss": 0.9727,
+      "step": 26111
+    },
+    {
+      "epoch": 0.9496308688220533,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3276841600680767e-07,
+      "loss": 0.9099,
+      "step": 26112
+    },
+    {
+      "epoch": 0.9496672364257919,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3257718648732486e-07,
+      "loss": 0.8475,
+      "step": 26113
+    },
+    {
+      "epoch": 0.9497036040295305,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3238609386433288e-07,
+      "loss": 0.9235,
+      "step": 26114
+    },
+    {
+      "epoch": 0.9497399716332691,
+      "grad_norm": 0.0,
+      "learning_rate": 1.32195138140484e-07,
+      "loss": 0.9171,
+      "step": 26115
+    },
+    {
+      "epoch": 0.9497763392370077,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3200431931842617e-07,
+      "loss": 0.9363,
+      "step": 26116
+    },
+    {
+      "epoch": 0.9498127068407463,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3181363740080942e-07,
+      "loss": 0.9347,
+      "step": 26117
+    },
+    {
+      "epoch": 0.9498490744444849,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3162309239027615e-07,
+      "loss": 0.8329,
+      "step": 26118
+    },
+    {
+      "epoch": 0.9498854420482234,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3143268428947086e-07,
+      "loss": 0.8702,
+      "step": 26119
+    },
+    {
+      "epoch": 0.949921809651962,
+      "grad_norm": 0.0,
+      "learning_rate": 1.312424131010359e-07,
+      "loss": 0.8625,
+      "step": 26120
+    },
+    {
+      "epoch": 0.9499581772557006,
+      "grad_norm": 0.0,
+      "learning_rate": 1.310522788276103e-07,
+      "loss": 0.8332,
+      "step": 26121
+    },
+    {
+      "epoch": 0.9499945448594392,
+      "grad_norm": 0.0,
+      "learning_rate": 1.30862281471833e-07,
+      "loss": 0.9807,
+      "step": 26122
+    },
+    {
+      "epoch": 0.9500309124631778,
+      "grad_norm": 0.0,
+      "learning_rate": 1.306724210363375e-07,
+      "loss": 0.8197,
+      "step": 26123
+    },
+    {
+      "epoch": 0.9500672800669164,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3048269752376052e-07,
+      "loss": 0.9076,
+      "step": 26124
+    },
+    {
+      "epoch": 0.950103647670655,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3029311093673224e-07,
+      "loss": 0.9883,
+      "step": 26125
+    },
+    {
+      "epoch": 0.9501400152743936,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3010366127788499e-07,
+      "loss": 0.9043,
+      "step": 26126
+    },
+    {
+      "epoch": 0.9501763828781321,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2991434854984442e-07,
+      "loss": 0.8959,
+      "step": 26127
+    },
+    {
+      "epoch": 0.9502127504818707,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2972517275523956e-07,
+      "loss": 0.9104,
+      "step": 26128
+    },
+    {
+      "epoch": 0.9502491180856093,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2953613389669272e-07,
+      "loss": 0.9396,
+      "step": 26129
+    },
+    {
+      "epoch": 0.9502854856893479,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2934723197682853e-07,
+      "loss": 0.9794,
+      "step": 26130
+    },
+    {
+      "epoch": 0.9503218532930865,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2915846699826707e-07,
+      "loss": 0.9968,
+      "step": 26131
+    },
+    {
+      "epoch": 0.9503582208968251,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2896983896362737e-07,
+      "loss": 0.9539,
+      "step": 26132
+    },
+    {
+      "epoch": 0.9503945885005637,
+      "grad_norm": 0.0,
+      "learning_rate": 1.28781347875524e-07,
+      "loss": 0.8757,
+      "step": 26133
+    },
+    {
+      "epoch": 0.9504309561043023,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2859299373657595e-07,
+      "loss": 0.8979,
+      "step": 26134
+    },
+    {
+      "epoch": 0.9504673237080409,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2840477654939343e-07,
+      "loss": 0.8598,
+      "step": 26135
+    },
+    {
+      "epoch": 0.9505036913117795,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2821669631658873e-07,
+      "loss": 0.9146,
+      "step": 26136
+    },
+    {
+      "epoch": 0.9505400589155181,
+      "grad_norm": 0.0,
+      "learning_rate": 1.28028753040772e-07,
+      "loss": 0.8605,
+      "step": 26137
+    },
+    {
+      "epoch": 0.9505764265192567,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2784094672454894e-07,
+      "loss": 0.8769,
+      "step": 26138
+    },
+    {
+      "epoch": 0.9506127941229953,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2765327737052634e-07,
+      "loss": 1.0134,
+      "step": 26139
+    },
+    {
+      "epoch": 0.9506491617267339,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2746574498130772e-07,
+      "loss": 0.8723,
+      "step": 26140
+    },
+    {
+      "epoch": 0.9506855293304725,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2727834955949425e-07,
+      "loss": 0.8385,
+      "step": 26141
+    },
+    {
+      "epoch": 0.950721896934211,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2709109110768613e-07,
+      "loss": 0.9231,
+      "step": 26142
+    },
+    {
+      "epoch": 0.9507582645379496,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2690396962848127e-07,
+      "loss": 0.8916,
+      "step": 26143
+    },
+    {
+      "epoch": 0.9507946321416881,
+      "grad_norm": 0.0,
+      "learning_rate": 1.267169851244765e-07,
+      "loss": 0.913,
+      "step": 26144
+    },
+    {
+      "epoch": 0.9508309997454267,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2653013759826416e-07,
+      "loss": 1.0074,
+      "step": 26145
+    },
+    {
+      "epoch": 0.9508673673491653,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2634342705243774e-07,
+      "loss": 0.8744,
+      "step": 26146
+    },
+    {
+      "epoch": 0.9509037349529039,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2615685348958852e-07,
+      "loss": 0.8715,
+      "step": 26147
+    },
+    {
+      "epoch": 0.9509401025566425,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2597041691230217e-07,
+      "loss": 0.8981,
+      "step": 26148
+    },
+    {
+      "epoch": 0.9509764701603811,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2578411732316664e-07,
+      "loss": 0.9714,
+      "step": 26149
+    },
+    {
+      "epoch": 0.9510128377641197,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2559795472476877e-07,
+      "loss": 0.8743,
+      "step": 26150
+    },
+    {
+      "epoch": 0.9510492053678583,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2541192911968757e-07,
+      "loss": 0.8688,
+      "step": 26151
+    },
+    {
+      "epoch": 0.9510855729715969,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2522604051050657e-07,
+      "loss": 0.8308,
+      "step": 26152
+    },
+    {
+      "epoch": 0.9511219405753355,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2504028889980368e-07,
+      "loss": 0.9197,
+      "step": 26153
+    },
+    {
+      "epoch": 0.9511583081790741,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2485467429015574e-07,
+      "loss": 0.9684,
+      "step": 26154
+    },
+    {
+      "epoch": 0.9511946757828127,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2466919668413847e-07,
+      "loss": 0.8592,
+      "step": 26155
+    },
+    {
+      "epoch": 0.9512310433865513,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2448385608432422e-07,
+      "loss": 0.8041,
+      "step": 26156
+    },
+    {
+      "epoch": 0.9512674109902899,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2429865249328655e-07,
+      "loss": 0.8783,
+      "step": 26157
+    },
+    {
+      "epoch": 0.9513037785940285,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2411358591359224e-07,
+      "loss": 0.8808,
+      "step": 26158
+    },
+    {
+      "epoch": 0.9513401461977671,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2392865634780927e-07,
+      "loss": 0.9185,
+      "step": 26159
+    },
+    {
+      "epoch": 0.9513765138015056,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2374386379850445e-07,
+      "loss": 0.9724,
+      "step": 26160
+    },
+    {
+      "epoch": 0.9514128814052442,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2355920826824018e-07,
+      "loss": 0.9302,
+      "step": 26161
+    },
+    {
+      "epoch": 0.9514492490089828,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2337468975958e-07,
+      "loss": 0.8137,
+      "step": 26162
+    },
+    {
+      "epoch": 0.9514856166127214,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2319030827508294e-07,
+      "loss": 1.0158,
+      "step": 26163
+    },
+    {
+      "epoch": 0.95152198421646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2300606381730696e-07,
+      "loss": 0.8653,
+      "step": 26164
+    },
+    {
+      "epoch": 0.9515583518201985,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2282195638880778e-07,
+      "loss": 0.9969,
+      "step": 26165
+    },
+    {
+      "epoch": 0.9515947194239371,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2263798599214117e-07,
+      "loss": 0.9946,
+      "step": 26166
+    },
+    {
+      "epoch": 0.9516310870276757,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2245415262985726e-07,
+      "loss": 0.9172,
+      "step": 26167
+    },
+    {
+      "epoch": 0.9516674546314143,
+      "grad_norm": 0.0,
+      "learning_rate": 1.222704563045074e-07,
+      "loss": 0.9045,
+      "step": 26168
+    },
+    {
+      "epoch": 0.9517038222351529,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2208689701864062e-07,
+      "loss": 1.0103,
+      "step": 26169
+    },
+    {
+      "epoch": 0.9517401898388915,
+      "grad_norm": 0.0,
+      "learning_rate": 1.219034747748038e-07,
+      "loss": 0.8448,
+      "step": 26170
+    },
+    {
+      "epoch": 0.9517765574426301,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2172018957554044e-07,
+      "loss": 0.8585,
+      "step": 26171
+    },
+    {
+      "epoch": 0.9518129250463687,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2153704142339405e-07,
+      "loss": 0.7332,
+      "step": 26172
+    },
+    {
+      "epoch": 0.9518492926501073,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2135403032090488e-07,
+      "loss": 0.9917,
+      "step": 26173
+    },
+    {
+      "epoch": 0.9518856602538459,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2117115627061417e-07,
+      "loss": 0.9573,
+      "step": 26174
+    },
+    {
+      "epoch": 0.9519220278575845,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2098841927505657e-07,
+      "loss": 0.9328,
+      "step": 26175
+    },
+    {
+      "epoch": 0.9519583954613231,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2080581933676781e-07,
+      "loss": 0.9206,
+      "step": 26176
+    },
+    {
+      "epoch": 0.9519947630650616,
+      "grad_norm": 0.0,
+      "learning_rate": 1.206233564582826e-07,
+      "loss": 0.9338,
+      "step": 26177
+    },
+    {
+      "epoch": 0.9520311306688002,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2044103064212998e-07,
+      "loss": 0.9427,
+      "step": 26178
+    },
+    {
+      "epoch": 0.9520674982725388,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2025884189084236e-07,
+      "loss": 0.934,
+      "step": 26179
+    },
+    {
+      "epoch": 0.9521038658762774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.200767902069444e-07,
+      "loss": 0.8929,
+      "step": 26180
+    },
+    {
+      "epoch": 0.952140233480016,
+      "grad_norm": 0.0,
+      "learning_rate": 1.198948755929641e-07,
+      "loss": 0.9439,
+      "step": 26181
+    },
+    {
+      "epoch": 0.9521766010837546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1971309805142384e-07,
+      "loss": 1.0527,
+      "step": 26182
+    },
+    {
+      "epoch": 0.9522129686874932,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1953145758484608e-07,
+      "loss": 0.9619,
+      "step": 26183
+    },
+    {
+      "epoch": 0.9522493362912318,
+      "grad_norm": 0.0,
+      "learning_rate": 1.193499541957499e-07,
+      "loss": 0.9759,
+      "step": 26184
+    },
+    {
+      "epoch": 0.9522857038949704,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1916858788665664e-07,
+      "loss": 1.0122,
+      "step": 26185
+    },
+    {
+      "epoch": 0.952322071498709,
+      "grad_norm": 0.0,
+      "learning_rate": 1.189873586600776e-07,
+      "loss": 0.9333,
+      "step": 26186
+    },
+    {
+      "epoch": 0.9523584391024476,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1880626651853189e-07,
+      "loss": 0.9591,
+      "step": 26187
+    },
+    {
+      "epoch": 0.9523948067061861,
+      "grad_norm": 0.0,
+      "learning_rate": 1.186253114645286e-07,
+      "loss": 1.009,
+      "step": 26188
+    },
+    {
+      "epoch": 0.9524311743099247,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1844449350057907e-07,
+      "loss": 0.8793,
+      "step": 26189
+    },
+    {
+      "epoch": 0.9524675419136633,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1826381262919129e-07,
+      "loss": 0.821,
+      "step": 26190
+    },
+    {
+      "epoch": 0.9525039095174019,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1808326885287436e-07,
+      "loss": 0.8937,
+      "step": 26191
+    },
+    {
+      "epoch": 0.9525402771211405,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1790286217413071e-07,
+      "loss": 0.9246,
+      "step": 26192
+    },
+    {
+      "epoch": 0.9525766447248791,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1772259259546504e-07,
+      "loss": 0.8885,
+      "step": 26193
+    },
+    {
+      "epoch": 0.9526130123286176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1754246011937531e-07,
+      "loss": 0.8957,
+      "step": 26194
+    },
+    {
+      "epoch": 0.9526493799323562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1736246474836399e-07,
+      "loss": 0.9872,
+      "step": 26195
+    },
+    {
+      "epoch": 0.9526857475360948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1718260648492685e-07,
+      "loss": 0.992,
+      "step": 26196
+    },
+    {
+      "epoch": 0.9527221151398334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1700288533155857e-07,
+      "loss": 0.9244,
+      "step": 26197
+    },
+    {
+      "epoch": 0.952758482743572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.168233012907538e-07,
+      "loss": 0.9136,
+      "step": 26198
+    },
+    {
+      "epoch": 0.9527948503473106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1664385436500281e-07,
+      "loss": 0.9652,
+      "step": 26199
+    },
+    {
+      "epoch": 0.9528312179510492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1646454455679579e-07,
+      "loss": 0.9314,
+      "step": 26200
+    },
+    {
+      "epoch": 0.9528675855547878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1628537186862077e-07,
+      "loss": 0.9791,
+      "step": 26201
+    },
+    {
+      "epoch": 0.9529039531585264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1610633630296243e-07,
+      "loss": 0.8044,
+      "step": 26202
+    },
+    {
+      "epoch": 0.952940320762265,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1592743786230432e-07,
+      "loss": 0.8992,
+      "step": 26203
+    },
+    {
+      "epoch": 0.9529766883660036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1574867654913113e-07,
+      "loss": 0.9679,
+      "step": 26204
+    },
+    {
+      "epoch": 0.9530130559697422,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1557005236592089e-07,
+      "loss": 0.9153,
+      "step": 26205
+    },
+    {
+      "epoch": 0.9530494235734808,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1539156531515161e-07,
+      "loss": 1.0565,
+      "step": 26206
+    },
+    {
+      "epoch": 0.9530857911772194,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1521321539929909e-07,
+      "loss": 0.8885,
+      "step": 26207
+    },
+    {
+      "epoch": 0.953122158780958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1503500262084021e-07,
+      "loss": 0.889,
+      "step": 26208
+    },
+    {
+      "epoch": 0.9531585263846966,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1485692698224416e-07,
+      "loss": 0.9167,
+      "step": 26209
+    },
+    {
+      "epoch": 0.9531948939884352,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1467898848598335e-07,
+      "loss": 0.9384,
+      "step": 26210
+    },
+    {
+      "epoch": 0.9532312615921736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1450118713452695e-07,
+      "loss": 0.947,
+      "step": 26211
+    },
+    {
+      "epoch": 0.9532676291959122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1432352293034077e-07,
+      "loss": 1.0048,
+      "step": 26212
+    },
+    {
+      "epoch": 0.9533039967996508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1414599587588837e-07,
+      "loss": 0.903,
+      "step": 26213
+    },
+    {
+      "epoch": 0.9533403644033894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1396860597363668e-07,
+      "loss": 0.9411,
+      "step": 26214
+    },
+    {
+      "epoch": 0.953376732007128,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1379135322604263e-07,
+      "loss": 0.8558,
+      "step": 26215
+    },
+    {
+      "epoch": 0.9534130996108666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1361423763556645e-07,
+      "loss": 0.9425,
+      "step": 26216
+    },
+    {
+      "epoch": 0.9534494672146052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.134372592046662e-07,
+      "loss": 0.9263,
+      "step": 26217
+    },
+    {
+      "epoch": 0.9534858348183438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1326041793579656e-07,
+      "loss": 0.8917,
+      "step": 26218
+    },
+    {
+      "epoch": 0.9535222024220824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1308371383141226e-07,
+      "loss": 0.9057,
+      "step": 26219
+    },
+    {
+      "epoch": 0.953558570025821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1290714689396243e-07,
+      "loss": 1.0042,
+      "step": 26220
+    },
+    {
+      "epoch": 0.9535949376295596,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1273071712589845e-07,
+      "loss": 0.8182,
+      "step": 26221
+    },
+    {
+      "epoch": 0.9536313052332982,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1255442452966724e-07,
+      "loss": 0.8916,
+      "step": 26222
+    },
+    {
+      "epoch": 0.9536676728370368,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1237826910771576e-07,
+      "loss": 0.8435,
+      "step": 26223
+    },
+    {
+      "epoch": 0.9537040404407754,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1220225086248649e-07,
+      "loss": 0.9878,
+      "step": 26224
+    },
+    {
+      "epoch": 0.953740408044514,
+      "grad_norm": 0.0,
+      "learning_rate": 1.120263697964219e-07,
+      "loss": 0.8979,
+      "step": 26225
+    },
+    {
+      "epoch": 0.9537767756482526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1185062591196227e-07,
+      "loss": 0.8163,
+      "step": 26226
+    },
+    {
+      "epoch": 0.9538131432519912,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1167501921154677e-07,
+      "loss": 0.8704,
+      "step": 26227
+    },
+    {
+      "epoch": 0.9538495108557297,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1149954969760901e-07,
+      "loss": 0.9669,
+      "step": 26228
+    },
+    {
+      "epoch": 0.9538858784594683,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1132421737258592e-07,
+      "loss": 0.8952,
+      "step": 26229
+    },
+    {
+      "epoch": 0.9539222460632069,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1114902223891e-07,
+      "loss": 0.912,
+      "step": 26230
+    },
+    {
+      "epoch": 0.9539586136669455,
+      "grad_norm": 0.0,
+      "learning_rate": 1.109739642990093e-07,
+      "loss": 1.0186,
+      "step": 26231
+    },
+    {
+      "epoch": 0.953994981270684,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1079904355531523e-07,
+      "loss": 0.9977,
+      "step": 26232
+    },
+    {
+      "epoch": 0.9540313488744226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1062426001025361e-07,
+      "loss": 0.9169,
+      "step": 26233
+    },
+    {
+      "epoch": 0.9540677164781612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1044961366624918e-07,
+      "loss": 0.9976,
+      "step": 26234
+    },
+    {
+      "epoch": 0.9541040840818998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1027510452572331e-07,
+      "loss": 0.8071,
+      "step": 26235
+    },
+    {
+      "epoch": 0.9541404516856384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1010073259109966e-07,
+      "loss": 0.9983,
+      "step": 26236
+    },
+    {
+      "epoch": 0.954176819289377,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0992649786479736e-07,
+      "loss": 0.8502,
+      "step": 26237
+    },
+    {
+      "epoch": 0.9542131868931156,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0975240034923229e-07,
+      "loss": 0.9065,
+      "step": 26238
+    },
+    {
+      "epoch": 0.9542495544968542,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0957844004681916e-07,
+      "loss": 0.9483,
+      "step": 26239
+    },
+    {
+      "epoch": 0.9542859221005928,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0940461695997385e-07,
+      "loss": 0.8716,
+      "step": 26240
+    },
+    {
+      "epoch": 0.9543222897043314,
+      "grad_norm": 0.0,
+      "learning_rate": 1.092309310911055e-07,
+      "loss": 0.9161,
+      "step": 26241
+    },
+    {
+      "epoch": 0.95435865730807,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0905738244262554e-07,
+      "loss": 0.8427,
+      "step": 26242
+    },
+    {
+      "epoch": 0.9543950249118086,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0888397101694092e-07,
+      "loss": 0.8477,
+      "step": 26243
+    },
+    {
+      "epoch": 0.9544313925155472,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0871069681645752e-07,
+      "loss": 0.9243,
+      "step": 26244
+    },
+    {
+      "epoch": 0.9544677601192857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0853755984357895e-07,
+      "loss": 1.0047,
+      "step": 26245
+    },
+    {
+      "epoch": 0.9545041277230243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0836456010070773e-07,
+      "loss": 0.8846,
+      "step": 26246
+    },
+    {
+      "epoch": 0.9545404953267629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0819169759024417e-07,
+      "loss": 0.9069,
+      "step": 26247
+    },
+    {
+      "epoch": 0.9545768629305015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0801897231458524e-07,
+      "loss": 1.0607,
+      "step": 26248
+    },
+    {
+      "epoch": 0.9546132305342401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0784638427612904e-07,
+      "loss": 1.0342,
+      "step": 26249
+    },
+    {
+      "epoch": 0.9546495981379787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0767393347726918e-07,
+      "loss": 0.9604,
+      "step": 26250
+    },
+    {
+      "epoch": 0.9546859657417173,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0750161992039709e-07,
+      "loss": 0.8419,
+      "step": 26251
+    },
+    {
+      "epoch": 0.9547223333454559,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0732944360790643e-07,
+      "loss": 0.8524,
+      "step": 26252
+    },
+    {
+      "epoch": 0.9547587009491945,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0715740454218304e-07,
+      "loss": 0.9037,
+      "step": 26253
+    },
+    {
+      "epoch": 0.9547950685529331,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0698550272561281e-07,
+      "loss": 0.9492,
+      "step": 26254
+    },
+    {
+      "epoch": 0.9548314361566717,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0681373816058382e-07,
+      "loss": 0.8031,
+      "step": 26255
+    },
+    {
+      "epoch": 0.9548678037604102,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0664211084947751e-07,
+      "loss": 0.9597,
+      "step": 26256
+    },
+    {
+      "epoch": 0.9549041713641488,
+      "grad_norm": 0.0,
+      "learning_rate": 1.064706207946753e-07,
+      "loss": 0.9569,
+      "step": 26257
+    },
+    {
+      "epoch": 0.9549405389678874,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0629926799855528e-07,
+      "loss": 0.7834,
+      "step": 26258
+    },
+    {
+      "epoch": 0.954976906571626,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0612805246349778e-07,
+      "loss": 0.8413,
+      "step": 26259
+    },
+    {
+      "epoch": 0.9550132741753646,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0595697419187312e-07,
+      "loss": 0.8327,
+      "step": 26260
+    },
+    {
+      "epoch": 0.9550496417791031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.057860331860594e-07,
+      "loss": 1.0323,
+      "step": 26261
+    },
+    {
+      "epoch": 0.9550860093828417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0561522944842584e-07,
+      "loss": 0.969,
+      "step": 26262
+    },
+    {
+      "epoch": 0.9551223769865803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0544456298134275e-07,
+      "loss": 0.9405,
+      "step": 26263
+    },
+    {
+      "epoch": 0.9551587445903189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0527403378717716e-07,
+      "loss": 0.9711,
+      "step": 26264
+    },
+    {
+      "epoch": 0.9551951121940575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0510364186829713e-07,
+      "loss": 0.9988,
+      "step": 26265
+    },
+    {
+      "epoch": 0.9552314797977961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0493338722706414e-07,
+      "loss": 0.761,
+      "step": 26266
+    },
+    {
+      "epoch": 0.9552678474015347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0476326986584074e-07,
+      "loss": 0.8876,
+      "step": 26267
+    },
+    {
+      "epoch": 0.9553042150052733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0459328978698835e-07,
+      "loss": 0.9711,
+      "step": 26268
+    },
+    {
+      "epoch": 0.9553405826090119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.04423446992864e-07,
+      "loss": 0.857,
+      "step": 26269
+    },
+    {
+      "epoch": 0.9553769502127505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0425374148582357e-07,
+      "loss": 0.8348,
+      "step": 26270
+    },
+    {
+      "epoch": 0.9554133178164891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0408417326822295e-07,
+      "loss": 0.9813,
+      "step": 26271
+    },
+    {
+      "epoch": 0.9554496854202277,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0391474234241471e-07,
+      "loss": 0.9964,
+      "step": 26272
+    },
+    {
+      "epoch": 0.9554860530239663,
+      "grad_norm": 0.0,
+      "learning_rate": 1.037454487107481e-07,
+      "loss": 0.8233,
+      "step": 26273
+    },
+    {
+      "epoch": 0.9555224206277049,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0357629237557232e-07,
+      "loss": 0.9544,
+      "step": 26274
+    },
+    {
+      "epoch": 0.9555587882314435,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0340727333923329e-07,
+      "loss": 0.8546,
+      "step": 26275
+    },
+    {
+      "epoch": 0.9555951558351821,
+      "grad_norm": 0.0,
+      "learning_rate": 1.03238391604078e-07,
+      "loss": 0.9441,
+      "step": 26276
+    },
+    {
+      "epoch": 0.9556315234389207,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0306964717244794e-07,
+      "loss": 0.8335,
+      "step": 26277
+    },
+    {
+      "epoch": 0.9556678910426591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0290104004668567e-07,
+      "loss": 0.8874,
+      "step": 26278
+    },
+    {
+      "epoch": 0.9557042586463977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.027325702291282e-07,
+      "loss": 0.979,
+      "step": 26279
+    },
+    {
+      "epoch": 0.9557406262501363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0256423772211477e-07,
+      "loss": 1.0059,
+      "step": 26280
+    },
+    {
+      "epoch": 0.9557769938538749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0239604252797908e-07,
+      "loss": 0.9142,
+      "step": 26281
+    },
+    {
+      "epoch": 0.9558133614576135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0222798464905593e-07,
+      "loss": 0.9346,
+      "step": 26282
+    },
+    {
+      "epoch": 0.9558497290613521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0206006408767566e-07,
+      "loss": 0.8619,
+      "step": 26283
+    },
+    {
+      "epoch": 0.9558860966650907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0189228084616976e-07,
+      "loss": 0.8928,
+      "step": 26284
+    },
+    {
+      "epoch": 0.9559224642688293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0172463492686412e-07,
+      "loss": 0.8771,
+      "step": 26285
+    },
+    {
+      "epoch": 0.9559588318725679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0155712633208581e-07,
+      "loss": 0.9282,
+      "step": 26286
+    },
+    {
+      "epoch": 0.9559951994763065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0138975506415849e-07,
+      "loss": 0.9794,
+      "step": 26287
+    },
+    {
+      "epoch": 0.9560315670800451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0122252112540365e-07,
+      "loss": 0.9669,
+      "step": 26288
+    },
+    {
+      "epoch": 0.9560679346837837,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0105542451814054e-07,
+      "loss": 1.0275,
+      "step": 26289
+    },
+    {
+      "epoch": 0.9561043022875223,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0088846524469064e-07,
+      "loss": 0.825,
+      "step": 26290
+    },
+    {
+      "epoch": 0.9561406698912609,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0072164330736767e-07,
+      "loss": 0.8701,
+      "step": 26291
+    },
+    {
+      "epoch": 0.9561770374949995,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0055495870848641e-07,
+      "loss": 0.9283,
+      "step": 26292
+    },
+    {
+      "epoch": 0.9562134050987381,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0038841145035948e-07,
+      "loss": 0.9804,
+      "step": 26293
+    },
+    {
+      "epoch": 0.9562497727024767,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0022200153529838e-07,
+      "loss": 0.9077,
+      "step": 26294
+    },
+    {
+      "epoch": 0.9562861403062152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0005572896561011e-07,
+      "loss": 0.8619,
+      "step": 26295
+    },
+    {
+      "epoch": 0.9563225079099538,
+      "grad_norm": 0.0,
+      "learning_rate": 9.988959374360285e-08,
+      "loss": 0.9595,
+      "step": 26296
+    },
+    {
+      "epoch": 0.9563588755136924,
+      "grad_norm": 0.0,
+      "learning_rate": 9.972359587158142e-08,
+      "loss": 0.8141,
+      "step": 26297
+    },
+    {
+      "epoch": 0.956395243117431,
+      "grad_norm": 0.0,
+      "learning_rate": 9.955773535184732e-08,
+      "loss": 0.9231,
+      "step": 26298
+    },
+    {
+      "epoch": 0.9564316107211696,
+      "grad_norm": 0.0,
+      "learning_rate": 9.939201218670424e-08,
+      "loss": 0.9258,
+      "step": 26299
+    },
+    {
+      "epoch": 0.9564679783249082,
+      "grad_norm": 0.0,
+      "learning_rate": 9.922642637844925e-08,
+      "loss": 0.8572,
+      "step": 26300
+    },
+    {
+      "epoch": 0.9565043459286467,
+      "grad_norm": 0.0,
+      "learning_rate": 9.906097792938052e-08,
+      "loss": 0.8015,
+      "step": 26301
+    },
+    {
+      "epoch": 0.9565407135323853,
+      "grad_norm": 0.0,
+      "learning_rate": 9.889566684179175e-08,
+      "loss": 0.8834,
+      "step": 26302
+    },
+    {
+      "epoch": 0.9565770811361239,
+      "grad_norm": 0.0,
+      "learning_rate": 9.873049311797889e-08,
+      "loss": 1.0118,
+      "step": 26303
+    },
+    {
+      "epoch": 0.9566134487398625,
+      "grad_norm": 0.0,
+      "learning_rate": 9.856545676023233e-08,
+      "loss": 0.923,
+      "step": 26304
+    },
+    {
+      "epoch": 0.9566498163436011,
+      "grad_norm": 0.0,
+      "learning_rate": 9.840055777084246e-08,
+      "loss": 0.8302,
+      "step": 26305
+    },
+    {
+      "epoch": 0.9566861839473397,
+      "grad_norm": 0.0,
+      "learning_rate": 9.823579615209522e-08,
+      "loss": 0.943,
+      "step": 26306
+    },
+    {
+      "epoch": 0.9567225515510783,
+      "grad_norm": 0.0,
+      "learning_rate": 9.807117190627769e-08,
+      "loss": 0.8737,
+      "step": 26307
+    },
+    {
+      "epoch": 0.9567589191548169,
+      "grad_norm": 0.0,
+      "learning_rate": 9.790668503567357e-08,
+      "loss": 0.9109,
+      "step": 26308
+    },
+    {
+      "epoch": 0.9567952867585555,
+      "grad_norm": 0.0,
+      "learning_rate": 9.774233554256551e-08,
+      "loss": 1.0178,
+      "step": 26309
+    },
+    {
+      "epoch": 0.9568316543622941,
+      "grad_norm": 0.0,
+      "learning_rate": 9.757812342923278e-08,
+      "loss": 0.7889,
+      "step": 26310
+    },
+    {
+      "epoch": 0.9568680219660327,
+      "grad_norm": 0.0,
+      "learning_rate": 9.741404869795356e-08,
+      "loss": 0.9563,
+      "step": 26311
+    },
+    {
+      "epoch": 0.9569043895697712,
+      "grad_norm": 0.0,
+      "learning_rate": 9.725011135100493e-08,
+      "loss": 0.9044,
+      "step": 26312
+    },
+    {
+      "epoch": 0.9569407571735098,
+      "grad_norm": 0.0,
+      "learning_rate": 9.70863113906606e-08,
+      "loss": 0.772,
+      "step": 26313
+    },
+    {
+      "epoch": 0.9569771247772484,
+      "grad_norm": 0.0,
+      "learning_rate": 9.692264881919433e-08,
+      "loss": 0.8233,
+      "step": 26314
+    },
+    {
+      "epoch": 0.957013492380987,
+      "grad_norm": 0.0,
+      "learning_rate": 9.675912363887429e-08,
+      "loss": 0.8868,
+      "step": 26315
+    },
+    {
+      "epoch": 0.9570498599847256,
+      "grad_norm": 0.0,
+      "learning_rate": 9.65957358519709e-08,
+      "loss": 0.9226,
+      "step": 26316
+    },
+    {
+      "epoch": 0.9570862275884642,
+      "grad_norm": 0.0,
+      "learning_rate": 9.643248546075123e-08,
+      "loss": 0.9021,
+      "step": 26317
+    },
+    {
+      "epoch": 0.9571225951922028,
+      "grad_norm": 0.0,
+      "learning_rate": 9.626937246747903e-08,
+      "loss": 0.8349,
+      "step": 26318
+    },
+    {
+      "epoch": 0.9571589627959414,
+      "grad_norm": 0.0,
+      "learning_rate": 9.610639687441692e-08,
+      "loss": 0.9779,
+      "step": 26319
+    },
+    {
+      "epoch": 0.95719533039968,
+      "grad_norm": 0.0,
+      "learning_rate": 9.594355868382865e-08,
+      "loss": 0.9799,
+      "step": 26320
+    },
+    {
+      "epoch": 0.9572316980034186,
+      "grad_norm": 0.0,
+      "learning_rate": 9.578085789796909e-08,
+      "loss": 0.9199,
+      "step": 26321
+    },
+    {
+      "epoch": 0.9572680656071572,
+      "grad_norm": 0.0,
+      "learning_rate": 9.561829451909976e-08,
+      "loss": 0.8163,
+      "step": 26322
+    },
+    {
+      "epoch": 0.9573044332108958,
+      "grad_norm": 0.0,
+      "learning_rate": 9.54558685494733e-08,
+      "loss": 0.8308,
+      "step": 26323
+    },
+    {
+      "epoch": 0.9573408008146344,
+      "grad_norm": 0.0,
+      "learning_rate": 9.529357999134459e-08,
+      "loss": 0.9887,
+      "step": 26324
+    },
+    {
+      "epoch": 0.957377168418373,
+      "grad_norm": 0.0,
+      "learning_rate": 9.513142884696292e-08,
+      "loss": 0.8492,
+      "step": 26325
+    },
+    {
+      "epoch": 0.9574135360221115,
+      "grad_norm": 0.0,
+      "learning_rate": 9.496941511858204e-08,
+      "loss": 0.8708,
+      "step": 26326
+    },
+    {
+      "epoch": 0.9574499036258501,
+      "grad_norm": 0.0,
+      "learning_rate": 9.480753880844462e-08,
+      "loss": 0.8614,
+      "step": 26327
+    },
+    {
+      "epoch": 0.9574862712295887,
+      "grad_norm": 0.0,
+      "learning_rate": 9.464579991880107e-08,
+      "loss": 0.9182,
+      "step": 26328
+    },
+    {
+      "epoch": 0.9575226388333272,
+      "grad_norm": 0.0,
+      "learning_rate": 9.448419845189294e-08,
+      "loss": 0.9608,
+      "step": 26329
+    },
+    {
+      "epoch": 0.9575590064370658,
+      "grad_norm": 0.0,
+      "learning_rate": 9.432273440996064e-08,
+      "loss": 0.8621,
+      "step": 26330
+    },
+    {
+      "epoch": 0.9575953740408044,
+      "grad_norm": 0.0,
+      "learning_rate": 9.416140779524796e-08,
+      "loss": 0.9539,
+      "step": 26331
+    },
+    {
+      "epoch": 0.957631741644543,
+      "grad_norm": 0.0,
+      "learning_rate": 9.400021860998976e-08,
+      "loss": 0.8259,
+      "step": 26332
+    },
+    {
+      "epoch": 0.9576681092482816,
+      "grad_norm": 0.0,
+      "learning_rate": 9.383916685642536e-08,
+      "loss": 1.0187,
+      "step": 26333
+    },
+    {
+      "epoch": 0.9577044768520202,
+      "grad_norm": 0.0,
+      "learning_rate": 9.36782525367852e-08,
+      "loss": 0.9494,
+      "step": 26334
+    },
+    {
+      "epoch": 0.9577408444557588,
+      "grad_norm": 0.0,
+      "learning_rate": 9.351747565330638e-08,
+      "loss": 0.8161,
+      "step": 26335
+    },
+    {
+      "epoch": 0.9577772120594974,
+      "grad_norm": 0.0,
+      "learning_rate": 9.3356836208216e-08,
+      "loss": 0.9119,
+      "step": 26336
+    },
+    {
+      "epoch": 0.957813579663236,
+      "grad_norm": 0.0,
+      "learning_rate": 9.319633420374342e-08,
+      "loss": 0.9007,
+      "step": 26337
+    },
+    {
+      "epoch": 0.9578499472669746,
+      "grad_norm": 0.0,
+      "learning_rate": 9.303596964211681e-08,
+      "loss": 0.9474,
+      "step": 26338
+    },
+    {
+      "epoch": 0.9578863148707132,
+      "grad_norm": 0.0,
+      "learning_rate": 9.287574252555997e-08,
+      "loss": 0.9292,
+      "step": 26339
+    },
+    {
+      "epoch": 0.9579226824744518,
+      "grad_norm": 0.0,
+      "learning_rate": 9.271565285629447e-08,
+      "loss": 0.8491,
+      "step": 26340
+    },
+    {
+      "epoch": 0.9579590500781904,
+      "grad_norm": 0.0,
+      "learning_rate": 9.255570063654406e-08,
+      "loss": 0.9284,
+      "step": 26341
+    },
+    {
+      "epoch": 0.957995417681929,
+      "grad_norm": 0.0,
+      "learning_rate": 9.239588586852587e-08,
+      "loss": 0.8216,
+      "step": 26342
+    },
+    {
+      "epoch": 0.9580317852856676,
+      "grad_norm": 0.0,
+      "learning_rate": 9.223620855445814e-08,
+      "loss": 0.9274,
+      "step": 26343
+    },
+    {
+      "epoch": 0.9580681528894062,
+      "grad_norm": 0.0,
+      "learning_rate": 9.207666869655463e-08,
+      "loss": 0.8773,
+      "step": 26344
+    },
+    {
+      "epoch": 0.9581045204931447,
+      "grad_norm": 0.0,
+      "learning_rate": 9.191726629703135e-08,
+      "loss": 0.7948,
+      "step": 26345
+    },
+    {
+      "epoch": 0.9581408880968832,
+      "grad_norm": 0.0,
+      "learning_rate": 9.175800135809765e-08,
+      "loss": 0.8956,
+      "step": 26346
+    },
+    {
+      "epoch": 0.9581772557006218,
+      "grad_norm": 0.0,
+      "learning_rate": 9.159887388196398e-08,
+      "loss": 1.0314,
+      "step": 26347
+    },
+    {
+      "epoch": 0.9582136233043604,
+      "grad_norm": 0.0,
+      "learning_rate": 9.143988387083746e-08,
+      "loss": 0.844,
+      "step": 26348
+    },
+    {
+      "epoch": 0.958249990908099,
+      "grad_norm": 0.0,
+      "learning_rate": 9.128103132692412e-08,
+      "loss": 0.9263,
+      "step": 26349
+    },
+    {
+      "epoch": 0.9582863585118376,
+      "grad_norm": 0.0,
+      "learning_rate": 9.112231625242774e-08,
+      "loss": 0.8815,
+      "step": 26350
+    },
+    {
+      "epoch": 0.9583227261155762,
+      "grad_norm": 0.0,
+      "learning_rate": 9.0963738649551e-08,
+      "loss": 0.9827,
+      "step": 26351
+    },
+    {
+      "epoch": 0.9583590937193148,
+      "grad_norm": 0.0,
+      "learning_rate": 9.080529852049324e-08,
+      "loss": 0.9751,
+      "step": 26352
+    },
+    {
+      "epoch": 0.9583954613230534,
+      "grad_norm": 0.0,
+      "learning_rate": 9.064699586745163e-08,
+      "loss": 0.8976,
+      "step": 26353
+    },
+    {
+      "epoch": 0.958431828926792,
+      "grad_norm": 0.0,
+      "learning_rate": 9.048883069262326e-08,
+      "loss": 0.9547,
+      "step": 26354
+    },
+    {
+      "epoch": 0.9584681965305306,
+      "grad_norm": 0.0,
+      "learning_rate": 9.033080299820419e-08,
+      "loss": 1.0746,
+      "step": 26355
+    },
+    {
+      "epoch": 0.9585045641342692,
+      "grad_norm": 0.0,
+      "learning_rate": 9.017291278638374e-08,
+      "loss": 0.9607,
+      "step": 26356
+    },
+    {
+      "epoch": 0.9585409317380078,
+      "grad_norm": 0.0,
+      "learning_rate": 9.001516005935462e-08,
+      "loss": 1.0246,
+      "step": 26357
+    },
+    {
+      "epoch": 0.9585772993417464,
+      "grad_norm": 0.0,
+      "learning_rate": 8.985754481930508e-08,
+      "loss": 0.9072,
+      "step": 26358
+    },
+    {
+      "epoch": 0.958613666945485,
+      "grad_norm": 0.0,
+      "learning_rate": 8.970006706842116e-08,
+      "loss": 0.9285,
+      "step": 26359
+    },
+    {
+      "epoch": 0.9586500345492236,
+      "grad_norm": 0.0,
+      "learning_rate": 8.954272680888775e-08,
+      "loss": 0.9012,
+      "step": 26360
+    },
+    {
+      "epoch": 0.9586864021529622,
+      "grad_norm": 0.0,
+      "learning_rate": 8.938552404288758e-08,
+      "loss": 0.8272,
+      "step": 26361
+    },
+    {
+      "epoch": 0.9587227697567007,
+      "grad_norm": 0.0,
+      "learning_rate": 8.922845877260222e-08,
+      "loss": 0.879,
+      "step": 26362
+    },
+    {
+      "epoch": 0.9587591373604393,
+      "grad_norm": 0.0,
+      "learning_rate": 8.907153100021104e-08,
+      "loss": 0.877,
+      "step": 26363
+    },
+    {
+      "epoch": 0.9587955049641779,
+      "grad_norm": 0.0,
+      "learning_rate": 8.891474072789008e-08,
+      "loss": 1.0045,
+      "step": 26364
+    },
+    {
+      "epoch": 0.9588318725679165,
+      "grad_norm": 0.0,
+      "learning_rate": 8.875808795781537e-08,
+      "loss": 0.8502,
+      "step": 26365
+    },
+    {
+      "epoch": 0.9588682401716551,
+      "grad_norm": 0.0,
+      "learning_rate": 8.860157269215963e-08,
+      "loss": 0.9072,
+      "step": 26366
+    },
+    {
+      "epoch": 0.9589046077753937,
+      "grad_norm": 0.0,
+      "learning_rate": 8.844519493309444e-08,
+      "loss": 0.9554,
+      "step": 26367
+    },
+    {
+      "epoch": 0.9589409753791323,
+      "grad_norm": 0.0,
+      "learning_rate": 8.82889546827903e-08,
+      "loss": 0.7527,
+      "step": 26368
+    },
+    {
+      "epoch": 0.9589773429828709,
+      "grad_norm": 0.0,
+      "learning_rate": 8.813285194341325e-08,
+      "loss": 0.9234,
+      "step": 26369
+    },
+    {
+      "epoch": 0.9590137105866094,
+      "grad_norm": 0.0,
+      "learning_rate": 8.797688671712934e-08,
+      "loss": 0.761,
+      "step": 26370
+    },
+    {
+      "epoch": 0.959050078190348,
+      "grad_norm": 0.0,
+      "learning_rate": 8.78210590061046e-08,
+      "loss": 0.9159,
+      "step": 26371
+    },
+    {
+      "epoch": 0.9590864457940866,
+      "grad_norm": 0.0,
+      "learning_rate": 8.766536881249732e-08,
+      "loss": 0.9651,
+      "step": 26372
+    },
+    {
+      "epoch": 0.9591228133978252,
+      "grad_norm": 0.0,
+      "learning_rate": 8.750981613847021e-08,
+      "loss": 1.071,
+      "step": 26373
+    },
+    {
+      "epoch": 0.9591591810015638,
+      "grad_norm": 0.0,
+      "learning_rate": 8.735440098618042e-08,
+      "loss": 0.9306,
+      "step": 26374
+    },
+    {
+      "epoch": 0.9591955486053024,
+      "grad_norm": 0.0,
+      "learning_rate": 8.719912335778403e-08,
+      "loss": 0.943,
+      "step": 26375
+    },
+    {
+      "epoch": 0.959231916209041,
+      "grad_norm": 0.0,
+      "learning_rate": 8.704398325543484e-08,
+      "loss": 0.9749,
+      "step": 26376
+    },
+    {
+      "epoch": 0.9592682838127796,
+      "grad_norm": 0.0,
+      "learning_rate": 8.68889806812867e-08,
+      "loss": 0.8442,
+      "step": 26377
+    },
+    {
+      "epoch": 0.9593046514165182,
+      "grad_norm": 0.0,
+      "learning_rate": 8.673411563748901e-08,
+      "loss": 0.834,
+      "step": 26378
+    },
+    {
+      "epoch": 0.9593410190202567,
+      "grad_norm": 0.0,
+      "learning_rate": 8.657938812619004e-08,
+      "loss": 0.792,
+      "step": 26379
+    },
+    {
+      "epoch": 0.9593773866239953,
+      "grad_norm": 0.0,
+      "learning_rate": 8.642479814953586e-08,
+      "loss": 0.9118,
+      "step": 26380
+    },
+    {
+      "epoch": 0.9594137542277339,
+      "grad_norm": 0.0,
+      "learning_rate": 8.627034570967364e-08,
+      "loss": 0.8875,
+      "step": 26381
+    },
+    {
+      "epoch": 0.9594501218314725,
+      "grad_norm": 0.0,
+      "learning_rate": 8.611603080874276e-08,
+      "loss": 0.8303,
+      "step": 26382
+    },
+    {
+      "epoch": 0.9594864894352111,
+      "grad_norm": 0.0,
+      "learning_rate": 8.596185344888819e-08,
+      "loss": 0.9646,
+      "step": 26383
+    },
+    {
+      "epoch": 0.9595228570389497,
+      "grad_norm": 0.0,
+      "learning_rate": 8.580781363224488e-08,
+      "loss": 0.8471,
+      "step": 26384
+    },
+    {
+      "epoch": 0.9595592246426883,
+      "grad_norm": 0.0,
+      "learning_rate": 8.565391136095335e-08,
+      "loss": 0.9078,
+      "step": 26385
+    },
+    {
+      "epoch": 0.9595955922464269,
+      "grad_norm": 0.0,
+      "learning_rate": 8.550014663714634e-08,
+      "loss": 0.8235,
+      "step": 26386
+    },
+    {
+      "epoch": 0.9596319598501655,
+      "grad_norm": 0.0,
+      "learning_rate": 8.53465194629577e-08,
+      "loss": 0.8383,
+      "step": 26387
+    },
+    {
+      "epoch": 0.9596683274539041,
+      "grad_norm": 0.0,
+      "learning_rate": 8.519302984051903e-08,
+      "loss": 0.9283,
+      "step": 26388
+    },
+    {
+      "epoch": 0.9597046950576427,
+      "grad_norm": 0.0,
+      "learning_rate": 8.503967777195976e-08,
+      "loss": 1.0091,
+      "step": 26389
+    },
+    {
+      "epoch": 0.9597410626613813,
+      "grad_norm": 0.0,
+      "learning_rate": 8.48864632594093e-08,
+      "loss": 0.8481,
+      "step": 26390
+    },
+    {
+      "epoch": 0.9597774302651199,
+      "grad_norm": 0.0,
+      "learning_rate": 8.473338630498928e-08,
+      "loss": 0.8275,
+      "step": 26391
+    },
+    {
+      "epoch": 0.9598137978688585,
+      "grad_norm": 0.0,
+      "learning_rate": 8.458044691082579e-08,
+      "loss": 1.045,
+      "step": 26392
+    },
+    {
+      "epoch": 0.959850165472597,
+      "grad_norm": 0.0,
+      "learning_rate": 8.442764507904156e-08,
+      "loss": 0.8862,
+      "step": 26393
+    },
+    {
+      "epoch": 0.9598865330763356,
+      "grad_norm": 0.0,
+      "learning_rate": 8.42749808117549e-08,
+      "loss": 0.8396,
+      "step": 26394
+    },
+    {
+      "epoch": 0.9599229006800742,
+      "grad_norm": 0.0,
+      "learning_rate": 8.4122454111083e-08,
+      "loss": 0.8036,
+      "step": 26395
+    },
+    {
+      "epoch": 0.9599592682838127,
+      "grad_norm": 0.0,
+      "learning_rate": 8.397006497914528e-08,
+      "loss": 0.9051,
+      "step": 26396
+    },
+    {
+      "epoch": 0.9599956358875513,
+      "grad_norm": 0.0,
+      "learning_rate": 8.381781341805339e-08,
+      "loss": 0.8669,
+      "step": 26397
+    },
+    {
+      "epoch": 0.9600320034912899,
+      "grad_norm": 0.0,
+      "learning_rate": 8.366569942991898e-08,
+      "loss": 0.8329,
+      "step": 26398
+    },
+    {
+      "epoch": 0.9600683710950285,
+      "grad_norm": 0.0,
+      "learning_rate": 8.351372301685478e-08,
+      "loss": 0.8534,
+      "step": 26399
+    },
+    {
+      "epoch": 0.9601047386987671,
+      "grad_norm": 0.0,
+      "learning_rate": 8.336188418096803e-08,
+      "loss": 0.9984,
+      "step": 26400
+    },
+    {
+      "epoch": 0.9601411063025057,
+      "grad_norm": 0.0,
+      "learning_rate": 8.321018292436478e-08,
+      "loss": 0.9073,
+      "step": 26401
+    },
+    {
+      "epoch": 0.9601774739062443,
+      "grad_norm": 0.0,
+      "learning_rate": 8.305861924915004e-08,
+      "loss": 0.9341,
+      "step": 26402
+    },
+    {
+      "epoch": 0.9602138415099829,
+      "grad_norm": 0.0,
+      "learning_rate": 8.290719315742657e-08,
+      "loss": 0.8264,
+      "step": 26403
+    },
+    {
+      "epoch": 0.9602502091137215,
+      "grad_norm": 0.0,
+      "learning_rate": 8.275590465129602e-08,
+      "loss": 0.8596,
+      "step": 26404
+    },
+    {
+      "epoch": 0.9602865767174601,
+      "grad_norm": 0.0,
+      "learning_rate": 8.260475373285559e-08,
+      "loss": 0.9573,
+      "step": 26405
+    },
+    {
+      "epoch": 0.9603229443211987,
+      "grad_norm": 0.0,
+      "learning_rate": 8.245374040420473e-08,
+      "loss": 0.9548,
+      "step": 26406
+    },
+    {
+      "epoch": 0.9603593119249373,
+      "grad_norm": 0.0,
+      "learning_rate": 8.23028646674362e-08,
+      "loss": 0.8768,
+      "step": 26407
+    },
+    {
+      "epoch": 0.9603956795286759,
+      "grad_norm": 0.0,
+      "learning_rate": 8.215212652464277e-08,
+      "loss": 0.8895,
+      "step": 26408
+    },
+    {
+      "epoch": 0.9604320471324145,
+      "grad_norm": 0.0,
+      "learning_rate": 8.200152597791944e-08,
+      "loss": 0.8164,
+      "step": 26409
+    },
+    {
+      "epoch": 0.9604684147361531,
+      "grad_norm": 0.0,
+      "learning_rate": 8.18510630293512e-08,
+      "loss": 0.8821,
+      "step": 26410
+    },
+    {
+      "epoch": 0.9605047823398917,
+      "grad_norm": 0.0,
+      "learning_rate": 8.17007376810286e-08,
+      "loss": 0.9334,
+      "step": 26411
+    },
+    {
+      "epoch": 0.9605411499436303,
+      "grad_norm": 0.0,
+      "learning_rate": 8.155054993503553e-08,
+      "loss": 0.8785,
+      "step": 26412
+    },
+    {
+      "epoch": 0.9605775175473688,
+      "grad_norm": 0.0,
+      "learning_rate": 8.140049979345587e-08,
+      "loss": 0.9801,
+      "step": 26413
+    },
+    {
+      "epoch": 0.9606138851511073,
+      "grad_norm": 0.0,
+      "learning_rate": 8.125058725837132e-08,
+      "loss": 0.8958,
+      "step": 26414
+    },
+    {
+      "epoch": 0.960650252754846,
+      "grad_norm": 0.0,
+      "learning_rate": 8.110081233186351e-08,
+      "loss": 0.9383,
+      "step": 26415
+    },
+    {
+      "epoch": 0.9606866203585845,
+      "grad_norm": 0.0,
+      "learning_rate": 8.095117501600746e-08,
+      "loss": 0.916,
+      "step": 26416
+    },
+    {
+      "epoch": 0.9607229879623231,
+      "grad_norm": 0.0,
+      "learning_rate": 8.080167531288153e-08,
+      "loss": 0.9256,
+      "step": 26417
+    },
+    {
+      "epoch": 0.9607593555660617,
+      "grad_norm": 0.0,
+      "learning_rate": 8.065231322455846e-08,
+      "loss": 0.9188,
+      "step": 26418
+    },
+    {
+      "epoch": 0.9607957231698003,
+      "grad_norm": 0.0,
+      "learning_rate": 8.050308875311108e-08,
+      "loss": 0.8885,
+      "step": 26419
+    },
+    {
+      "epoch": 0.9608320907735389,
+      "grad_norm": 0.0,
+      "learning_rate": 8.035400190060771e-08,
+      "loss": 1.0222,
+      "step": 26420
+    },
+    {
+      "epoch": 0.9608684583772775,
+      "grad_norm": 0.0,
+      "learning_rate": 8.020505266912004e-08,
+      "loss": 1.0035,
+      "step": 26421
+    },
+    {
+      "epoch": 0.9609048259810161,
+      "grad_norm": 0.0,
+      "learning_rate": 8.005624106071308e-08,
+      "loss": 0.9431,
+      "step": 26422
+    },
+    {
+      "epoch": 0.9609411935847547,
+      "grad_norm": 0.0,
+      "learning_rate": 7.990756707745073e-08,
+      "loss": 0.9165,
+      "step": 26423
+    },
+    {
+      "epoch": 0.9609775611884933,
+      "grad_norm": 0.0,
+      "learning_rate": 7.97590307213969e-08,
+      "loss": 0.9466,
+      "step": 26424
+    },
+    {
+      "epoch": 0.9610139287922319,
+      "grad_norm": 0.0,
+      "learning_rate": 7.961063199461106e-08,
+      "loss": 0.8732,
+      "step": 26425
+    },
+    {
+      "epoch": 0.9610502963959705,
+      "grad_norm": 0.0,
+      "learning_rate": 7.946237089915154e-08,
+      "loss": 0.9749,
+      "step": 26426
+    },
+    {
+      "epoch": 0.9610866639997091,
+      "grad_norm": 0.0,
+      "learning_rate": 7.931424743707672e-08,
+      "loss": 0.9186,
+      "step": 26427
+    },
+    {
+      "epoch": 0.9611230316034477,
+      "grad_norm": 0.0,
+      "learning_rate": 7.916626161044161e-08,
+      "loss": 0.8836,
+      "step": 26428
+    },
+    {
+      "epoch": 0.9611593992071863,
+      "grad_norm": 0.0,
+      "learning_rate": 7.901841342129901e-08,
+      "loss": 0.9987,
+      "step": 26429
+    },
+    {
+      "epoch": 0.9611957668109248,
+      "grad_norm": 0.0,
+      "learning_rate": 7.887070287169951e-08,
+      "loss": 0.864,
+      "step": 26430
+    },
+    {
+      "epoch": 0.9612321344146634,
+      "grad_norm": 0.0,
+      "learning_rate": 7.872312996369259e-08,
+      "loss": 0.7923,
+      "step": 26431
+    },
+    {
+      "epoch": 0.961268502018402,
+      "grad_norm": 0.0,
+      "learning_rate": 7.857569469932547e-08,
+      "loss": 0.9686,
+      "step": 26432
+    },
+    {
+      "epoch": 0.9613048696221406,
+      "grad_norm": 0.0,
+      "learning_rate": 7.842839708064432e-08,
+      "loss": 0.8944,
+      "step": 26433
+    },
+    {
+      "epoch": 0.9613412372258792,
+      "grad_norm": 0.0,
+      "learning_rate": 7.828123710969193e-08,
+      "loss": 0.8951,
+      "step": 26434
+    },
+    {
+      "epoch": 0.9613776048296178,
+      "grad_norm": 0.0,
+      "learning_rate": 7.813421478851114e-08,
+      "loss": 0.8357,
+      "step": 26435
+    },
+    {
+      "epoch": 0.9614139724333564,
+      "grad_norm": 0.0,
+      "learning_rate": 7.798733011914028e-08,
+      "loss": 0.8989,
+      "step": 26436
+    },
+    {
+      "epoch": 0.961450340037095,
+      "grad_norm": 0.0,
+      "learning_rate": 7.784058310361775e-08,
+      "loss": 0.9263,
+      "step": 26437
+    },
+    {
+      "epoch": 0.9614867076408335,
+      "grad_norm": 0.0,
+      "learning_rate": 7.769397374397969e-08,
+      "loss": 0.9509,
+      "step": 26438
+    },
+    {
+      "epoch": 0.9615230752445721,
+      "grad_norm": 0.0,
+      "learning_rate": 7.754750204226003e-08,
+      "loss": 1.0317,
+      "step": 26439
+    },
+    {
+      "epoch": 0.9615594428483107,
+      "grad_norm": 0.0,
+      "learning_rate": 7.740116800049047e-08,
+      "loss": 0.8692,
+      "step": 26440
+    },
+    {
+      "epoch": 0.9615958104520493,
+      "grad_norm": 0.0,
+      "learning_rate": 7.72549716207005e-08,
+      "loss": 0.8915,
+      "step": 26441
+    },
+    {
+      "epoch": 0.9616321780557879,
+      "grad_norm": 0.0,
+      "learning_rate": 7.710891290492073e-08,
+      "loss": 0.9973,
+      "step": 26442
+    },
+    {
+      "epoch": 0.9616685456595265,
+      "grad_norm": 0.0,
+      "learning_rate": 7.696299185517397e-08,
+      "loss": 0.98,
+      "step": 26443
+    },
+    {
+      "epoch": 0.9617049132632651,
+      "grad_norm": 0.0,
+      "learning_rate": 7.681720847348862e-08,
+      "loss": 0.923,
+      "step": 26444
+    },
+    {
+      "epoch": 0.9617412808670037,
+      "grad_norm": 0.0,
+      "learning_rate": 7.667156276188525e-08,
+      "loss": 0.9772,
+      "step": 26445
+    },
+    {
+      "epoch": 0.9617776484707422,
+      "grad_norm": 0.0,
+      "learning_rate": 7.65260547223834e-08,
+      "loss": 0.9166,
+      "step": 26446
+    },
+    {
+      "epoch": 0.9618140160744808,
+      "grad_norm": 0.0,
+      "learning_rate": 7.638068435700363e-08,
+      "loss": 0.9472,
+      "step": 26447
+    },
+    {
+      "epoch": 0.9618503836782194,
+      "grad_norm": 0.0,
+      "learning_rate": 7.623545166776324e-08,
+      "loss": 0.9669,
+      "step": 26448
+    },
+    {
+      "epoch": 0.961886751281958,
+      "grad_norm": 0.0,
+      "learning_rate": 7.609035665667397e-08,
+      "loss": 0.8771,
+      "step": 26449
+    },
+    {
+      "epoch": 0.9619231188856966,
+      "grad_norm": 0.0,
+      "learning_rate": 7.594539932575196e-08,
+      "loss": 1.09,
+      "step": 26450
+    },
+    {
+      "epoch": 0.9619594864894352,
+      "grad_norm": 0.0,
+      "learning_rate": 7.580057967700782e-08,
+      "loss": 0.92,
+      "step": 26451
+    },
+    {
+      "epoch": 0.9619958540931738,
+      "grad_norm": 0.0,
+      "learning_rate": 7.565589771244775e-08,
+      "loss": 1.0348,
+      "step": 26452
+    },
+    {
+      "epoch": 0.9620322216969124,
+      "grad_norm": 0.0,
+      "learning_rate": 7.551135343408345e-08,
+      "loss": 0.8031,
+      "step": 26453
+    },
+    {
+      "epoch": 0.962068589300651,
+      "grad_norm": 0.0,
+      "learning_rate": 7.536694684391776e-08,
+      "loss": 0.8533,
+      "step": 26454
+    },
+    {
+      "epoch": 0.9621049569043896,
+      "grad_norm": 0.0,
+      "learning_rate": 7.522267794395355e-08,
+      "loss": 0.946,
+      "step": 26455
+    },
+    {
+      "epoch": 0.9621413245081282,
+      "grad_norm": 0.0,
+      "learning_rate": 7.507854673619363e-08,
+      "loss": 0.8761,
+      "step": 26456
+    },
+    {
+      "epoch": 0.9621776921118668,
+      "grad_norm": 0.0,
+      "learning_rate": 7.493455322263754e-08,
+      "loss": 0.9345,
+      "step": 26457
+    },
+    {
+      "epoch": 0.9622140597156054,
+      "grad_norm": 0.0,
+      "learning_rate": 7.479069740528255e-08,
+      "loss": 0.999,
+      "step": 26458
+    },
+    {
+      "epoch": 0.962250427319344,
+      "grad_norm": 0.0,
+      "learning_rate": 7.464697928612485e-08,
+      "loss": 0.9098,
+      "step": 26459
+    },
+    {
+      "epoch": 0.9622867949230826,
+      "grad_norm": 0.0,
+      "learning_rate": 7.450339886715729e-08,
+      "loss": 0.9664,
+      "step": 26460
+    },
+    {
+      "epoch": 0.9623231625268212,
+      "grad_norm": 0.0,
+      "learning_rate": 7.435995615037384e-08,
+      "loss": 0.8409,
+      "step": 26461
+    },
+    {
+      "epoch": 0.9623595301305597,
+      "grad_norm": 0.0,
+      "learning_rate": 7.421665113776177e-08,
+      "loss": 0.8197,
+      "step": 26462
+    },
+    {
+      "epoch": 0.9623958977342982,
+      "grad_norm": 0.0,
+      "learning_rate": 7.407348383131286e-08,
+      "loss": 0.8377,
+      "step": 26463
+    },
+    {
+      "epoch": 0.9624322653380368,
+      "grad_norm": 0.0,
+      "learning_rate": 7.393045423300993e-08,
+      "loss": 0.8277,
+      "step": 26464
+    },
+    {
+      "epoch": 0.9624686329417754,
+      "grad_norm": 0.0,
+      "learning_rate": 7.378756234483808e-08,
+      "loss": 0.9156,
+      "step": 26465
+    },
+    {
+      "epoch": 0.962505000545514,
+      "grad_norm": 0.0,
+      "learning_rate": 7.364480816878239e-08,
+      "loss": 0.8058,
+      "step": 26466
+    },
+    {
+      "epoch": 0.9625413681492526,
+      "grad_norm": 0.0,
+      "learning_rate": 7.350219170682015e-08,
+      "loss": 0.8575,
+      "step": 26467
+    },
+    {
+      "epoch": 0.9625777357529912,
+      "grad_norm": 0.0,
+      "learning_rate": 7.335971296093092e-08,
+      "loss": 0.9965,
+      "step": 26468
+    },
+    {
+      "epoch": 0.9626141033567298,
+      "grad_norm": 0.0,
+      "learning_rate": 7.321737193309087e-08,
+      "loss": 0.9307,
+      "step": 26469
+    },
+    {
+      "epoch": 0.9626504709604684,
+      "grad_norm": 0.0,
+      "learning_rate": 7.30751686252773e-08,
+      "loss": 0.9197,
+      "step": 26470
+    },
+    {
+      "epoch": 0.962686838564207,
+      "grad_norm": 0.0,
+      "learning_rate": 7.293310303946088e-08,
+      "loss": 0.9795,
+      "step": 26471
+    },
+    {
+      "epoch": 0.9627232061679456,
+      "grad_norm": 0.0,
+      "learning_rate": 7.279117517761225e-08,
+      "loss": 0.914,
+      "step": 26472
+    },
+    {
+      "epoch": 0.9627595737716842,
+      "grad_norm": 0.0,
+      "learning_rate": 7.264938504170205e-08,
+      "loss": 0.8557,
+      "step": 26473
+    },
+    {
+      "epoch": 0.9627959413754228,
+      "grad_norm": 0.0,
+      "learning_rate": 7.250773263369537e-08,
+      "loss": 0.9124,
+      "step": 26474
+    },
+    {
+      "epoch": 0.9628323089791614,
+      "grad_norm": 0.0,
+      "learning_rate": 7.236621795555954e-08,
+      "loss": 0.902,
+      "step": 26475
+    },
+    {
+      "epoch": 0.9628686765829,
+      "grad_norm": 0.0,
+      "learning_rate": 7.222484100925741e-08,
+      "loss": 0.877,
+      "step": 26476
+    },
+    {
+      "epoch": 0.9629050441866386,
+      "grad_norm": 0.0,
+      "learning_rate": 7.208360179674855e-08,
+      "loss": 0.9728,
+      "step": 26477
+    },
+    {
+      "epoch": 0.9629414117903772,
+      "grad_norm": 0.0,
+      "learning_rate": 7.194250031999472e-08,
+      "loss": 0.7901,
+      "step": 26478
+    },
+    {
+      "epoch": 0.9629777793941158,
+      "grad_norm": 0.0,
+      "learning_rate": 7.180153658095213e-08,
+      "loss": 0.9404,
+      "step": 26479
+    },
+    {
+      "epoch": 0.9630141469978543,
+      "grad_norm": 0.0,
+      "learning_rate": 7.166071058157809e-08,
+      "loss": 0.8907,
+      "step": 26480
+    },
+    {
+      "epoch": 0.9630505146015929,
+      "grad_norm": 0.0,
+      "learning_rate": 7.152002232382327e-08,
+      "loss": 0.8586,
+      "step": 26481
+    },
+    {
+      "epoch": 0.9630868822053315,
+      "grad_norm": 0.0,
+      "learning_rate": 7.137947180964278e-08,
+      "loss": 0.773,
+      "step": 26482
+    },
+    {
+      "epoch": 0.96312324980907,
+      "grad_norm": 0.0,
+      "learning_rate": 7.123905904098616e-08,
+      "loss": 0.9588,
+      "step": 26483
+    },
+    {
+      "epoch": 0.9631596174128086,
+      "grad_norm": 0.0,
+      "learning_rate": 7.109878401979853e-08,
+      "loss": 0.9603,
+      "step": 26484
+    },
+    {
+      "epoch": 0.9631959850165472,
+      "grad_norm": 0.0,
+      "learning_rate": 7.095864674802944e-08,
+      "loss": 0.8356,
+      "step": 26485
+    },
+    {
+      "epoch": 0.9632323526202858,
+      "grad_norm": 0.0,
+      "learning_rate": 7.081864722762177e-08,
+      "loss": 0.908,
+      "step": 26486
+    },
+    {
+      "epoch": 0.9632687202240244,
+      "grad_norm": 0.0,
+      "learning_rate": 7.067878546051732e-08,
+      "loss": 0.8867,
+      "step": 26487
+    },
+    {
+      "epoch": 0.963305087827763,
+      "grad_norm": 0.0,
+      "learning_rate": 7.053906144865563e-08,
+      "loss": 0.8921,
+      "step": 26488
+    },
+    {
+      "epoch": 0.9633414554315016,
+      "grad_norm": 0.0,
+      "learning_rate": 7.03994751939785e-08,
+      "loss": 0.8423,
+      "step": 26489
+    },
+    {
+      "epoch": 0.9633778230352402,
+      "grad_norm": 0.0,
+      "learning_rate": 7.02600266984188e-08,
+      "loss": 0.8852,
+      "step": 26490
+    },
+    {
+      "epoch": 0.9634141906389788,
+      "grad_norm": 0.0,
+      "learning_rate": 7.01207159639128e-08,
+      "loss": 0.8669,
+      "step": 26491
+    },
+    {
+      "epoch": 0.9634505582427174,
+      "grad_norm": 0.0,
+      "learning_rate": 6.998154299239446e-08,
+      "loss": 0.8817,
+      "step": 26492
+    },
+    {
+      "epoch": 0.963486925846456,
+      "grad_norm": 0.0,
+      "learning_rate": 6.984250778579226e-08,
+      "loss": 0.9433,
+      "step": 26493
+    },
+    {
+      "epoch": 0.9635232934501946,
+      "grad_norm": 0.0,
+      "learning_rate": 6.970361034603579e-08,
+      "loss": 0.8926,
+      "step": 26494
+    },
+    {
+      "epoch": 0.9635596610539332,
+      "grad_norm": 0.0,
+      "learning_rate": 6.956485067505458e-08,
+      "loss": 0.9212,
+      "step": 26495
+    },
+    {
+      "epoch": 0.9635960286576718,
+      "grad_norm": 0.0,
+      "learning_rate": 6.942622877476935e-08,
+      "loss": 0.8279,
+      "step": 26496
+    },
+    {
+      "epoch": 0.9636323962614103,
+      "grad_norm": 0.0,
+      "learning_rate": 6.928774464710631e-08,
+      "loss": 0.9419,
+      "step": 26497
+    },
+    {
+      "epoch": 0.9636687638651489,
+      "grad_norm": 0.0,
+      "learning_rate": 6.914939829398615e-08,
+      "loss": 0.9501,
+      "step": 26498
+    },
+    {
+      "epoch": 0.9637051314688875,
+      "grad_norm": 0.0,
+      "learning_rate": 6.901118971732845e-08,
+      "loss": 0.8619,
+      "step": 26499
+    },
+    {
+      "epoch": 0.9637414990726261,
+      "grad_norm": 0.0,
+      "learning_rate": 6.887311891904946e-08,
+      "loss": 1.0619,
+      "step": 26500
+    },
+    {
+      "epoch": 0.9637778666763647,
+      "grad_norm": 0.0,
+      "learning_rate": 6.873518590106543e-08,
+      "loss": 0.9588,
+      "step": 26501
+    },
+    {
+      "epoch": 0.9638142342801033,
+      "grad_norm": 0.0,
+      "learning_rate": 6.859739066529148e-08,
+      "loss": 0.8588,
+      "step": 26502
+    },
+    {
+      "epoch": 0.9638506018838419,
+      "grad_norm": 0.0,
+      "learning_rate": 6.84597332136372e-08,
+      "loss": 0.9084,
+      "step": 26503
+    },
+    {
+      "epoch": 0.9638869694875805,
+      "grad_norm": 0.0,
+      "learning_rate": 6.832221354801216e-08,
+      "loss": 0.9133,
+      "step": 26504
+    },
+    {
+      "epoch": 0.963923337091319,
+      "grad_norm": 0.0,
+      "learning_rate": 6.818483167032597e-08,
+      "loss": 0.9538,
+      "step": 26505
+    },
+    {
+      "epoch": 0.9639597046950577,
+      "grad_norm": 0.0,
+      "learning_rate": 6.804758758248486e-08,
+      "loss": 0.8637,
+      "step": 26506
+    },
+    {
+      "epoch": 0.9639960722987962,
+      "grad_norm": 0.0,
+      "learning_rate": 6.791048128638956e-08,
+      "loss": 0.8252,
+      "step": 26507
+    },
+    {
+      "epoch": 0.9640324399025348,
+      "grad_norm": 0.0,
+      "learning_rate": 6.777351278394628e-08,
+      "loss": 0.9484,
+      "step": 26508
+    },
+    {
+      "epoch": 0.9640688075062734,
+      "grad_norm": 0.0,
+      "learning_rate": 6.763668207705354e-08,
+      "loss": 0.9704,
+      "step": 26509
+    },
+    {
+      "epoch": 0.964105175110012,
+      "grad_norm": 0.0,
+      "learning_rate": 6.74999891676087e-08,
+      "loss": 0.9913,
+      "step": 26510
+    },
+    {
+      "epoch": 0.9641415427137506,
+      "grad_norm": 0.0,
+      "learning_rate": 6.736343405751022e-08,
+      "loss": 0.8751,
+      "step": 26511
+    },
+    {
+      "epoch": 0.9641779103174892,
+      "grad_norm": 0.0,
+      "learning_rate": 6.722701674865107e-08,
+      "loss": 0.8736,
+      "step": 26512
+    },
+    {
+      "epoch": 0.9642142779212278,
+      "grad_norm": 0.0,
+      "learning_rate": 6.709073724292303e-08,
+      "loss": 0.9805,
+      "step": 26513
+    },
+    {
+      "epoch": 0.9642506455249663,
+      "grad_norm": 0.0,
+      "learning_rate": 6.695459554221906e-08,
+      "loss": 0.9796,
+      "step": 26514
+    },
+    {
+      "epoch": 0.9642870131287049,
+      "grad_norm": 0.0,
+      "learning_rate": 6.681859164842763e-08,
+      "loss": 0.8887,
+      "step": 26515
+    },
+    {
+      "epoch": 0.9643233807324435,
+      "grad_norm": 0.0,
+      "learning_rate": 6.668272556343391e-08,
+      "loss": 0.8827,
+      "step": 26516
+    },
+    {
+      "epoch": 0.9643597483361821,
+      "grad_norm": 0.0,
+      "learning_rate": 6.654699728912417e-08,
+      "loss": 0.8711,
+      "step": 26517
+    },
+    {
+      "epoch": 0.9643961159399207,
+      "grad_norm": 0.0,
+      "learning_rate": 6.641140682738135e-08,
+      "loss": 0.8339,
+      "step": 26518
+    },
+    {
+      "epoch": 0.9644324835436593,
+      "grad_norm": 0.0,
+      "learning_rate": 6.627595418008615e-08,
+      "loss": 0.938,
+      "step": 26519
+    },
+    {
+      "epoch": 0.9644688511473979,
+      "grad_norm": 0.0,
+      "learning_rate": 6.614063934911818e-08,
+      "loss": 0.9338,
+      "step": 26520
+    },
+    {
+      "epoch": 0.9645052187511365,
+      "grad_norm": 0.0,
+      "learning_rate": 6.600546233635485e-08,
+      "loss": 0.9211,
+      "step": 26521
+    },
+    {
+      "epoch": 0.9645415863548751,
+      "grad_norm": 0.0,
+      "learning_rate": 6.58704231436702e-08,
+      "loss": 0.9174,
+      "step": 26522
+    },
+    {
+      "epoch": 0.9645779539586137,
+      "grad_norm": 0.0,
+      "learning_rate": 6.57355217729394e-08,
+      "loss": 0.8569,
+      "step": 26523
+    },
+    {
+      "epoch": 0.9646143215623523,
+      "grad_norm": 0.0,
+      "learning_rate": 6.560075822603318e-08,
+      "loss": 0.9201,
+      "step": 26524
+    },
+    {
+      "epoch": 0.9646506891660909,
+      "grad_norm": 0.0,
+      "learning_rate": 6.546613250482225e-08,
+      "loss": 0.9647,
+      "step": 26525
+    },
+    {
+      "epoch": 0.9646870567698295,
+      "grad_norm": 0.0,
+      "learning_rate": 6.53316446111718e-08,
+      "loss": 1.0046,
+      "step": 26526
+    },
+    {
+      "epoch": 0.9647234243735681,
+      "grad_norm": 0.0,
+      "learning_rate": 6.519729454695034e-08,
+      "loss": 1.0309,
+      "step": 26527
+    },
+    {
+      "epoch": 0.9647597919773067,
+      "grad_norm": 0.0,
+      "learning_rate": 6.506308231402081e-08,
+      "loss": 0.9161,
+      "step": 26528
+    },
+    {
+      "epoch": 0.9647961595810453,
+      "grad_norm": 0.0,
+      "learning_rate": 6.492900791424506e-08,
+      "loss": 0.9918,
+      "step": 26529
+    },
+    {
+      "epoch": 0.9648325271847837,
+      "grad_norm": 0.0,
+      "learning_rate": 6.479507134948271e-08,
+      "loss": 0.8738,
+      "step": 26530
+    },
+    {
+      "epoch": 0.9648688947885223,
+      "grad_norm": 0.0,
+      "learning_rate": 6.466127262159339e-08,
+      "loss": 0.9602,
+      "step": 26531
+    },
+    {
+      "epoch": 0.9649052623922609,
+      "grad_norm": 0.0,
+      "learning_rate": 6.452761173243227e-08,
+      "loss": 0.9503,
+      "step": 26532
+    },
+    {
+      "epoch": 0.9649416299959995,
+      "grad_norm": 0.0,
+      "learning_rate": 6.439408868385344e-08,
+      "loss": 0.9824,
+      "step": 26533
+    },
+    {
+      "epoch": 0.9649779975997381,
+      "grad_norm": 0.0,
+      "learning_rate": 6.426070347770986e-08,
+      "loss": 0.902,
+      "step": 26534
+    },
+    {
+      "epoch": 0.9650143652034767,
+      "grad_norm": 0.0,
+      "learning_rate": 6.412745611585336e-08,
+      "loss": 0.8421,
+      "step": 26535
+    },
+    {
+      "epoch": 0.9650507328072153,
+      "grad_norm": 0.0,
+      "learning_rate": 6.399434660012915e-08,
+      "loss": 0.9833,
+      "step": 26536
+    },
+    {
+      "epoch": 0.9650871004109539,
+      "grad_norm": 0.0,
+      "learning_rate": 6.386137493238798e-08,
+      "loss": 1.0098,
+      "step": 26537
+    },
+    {
+      "epoch": 0.9651234680146925,
+      "grad_norm": 0.0,
+      "learning_rate": 6.372854111447169e-08,
+      "loss": 0.8935,
+      "step": 26538
+    },
+    {
+      "epoch": 0.9651598356184311,
+      "grad_norm": 0.0,
+      "learning_rate": 6.359584514822436e-08,
+      "loss": 0.9933,
+      "step": 26539
+    },
+    {
+      "epoch": 0.9651962032221697,
+      "grad_norm": 0.0,
+      "learning_rate": 6.346328703548787e-08,
+      "loss": 0.8576,
+      "step": 26540
+    },
+    {
+      "epoch": 0.9652325708259083,
+      "grad_norm": 0.0,
+      "learning_rate": 6.333086677809852e-08,
+      "loss": 1.0432,
+      "step": 26541
+    },
+    {
+      "epoch": 0.9652689384296469,
+      "grad_norm": 0.0,
+      "learning_rate": 6.319858437789595e-08,
+      "loss": 0.9319,
+      "step": 26542
+    },
+    {
+      "epoch": 0.9653053060333855,
+      "grad_norm": 0.0,
+      "learning_rate": 6.306643983671423e-08,
+      "loss": 0.9298,
+      "step": 26543
+    },
+    {
+      "epoch": 0.9653416736371241,
+      "grad_norm": 0.0,
+      "learning_rate": 6.293443315638748e-08,
+      "loss": 0.8804,
+      "step": 26544
+    },
+    {
+      "epoch": 0.9653780412408627,
+      "grad_norm": 0.0,
+      "learning_rate": 6.280256433874531e-08,
+      "loss": 0.8117,
+      "step": 26545
+    },
+    {
+      "epoch": 0.9654144088446013,
+      "grad_norm": 0.0,
+      "learning_rate": 6.267083338561963e-08,
+      "loss": 0.9839,
+      "step": 26546
+    },
+    {
+      "epoch": 0.9654507764483398,
+      "grad_norm": 0.0,
+      "learning_rate": 6.253924029883673e-08,
+      "loss": 0.805,
+      "step": 26547
+    },
+    {
+      "epoch": 0.9654871440520784,
+      "grad_norm": 0.0,
+      "learning_rate": 6.240778508022294e-08,
+      "loss": 0.8362,
+      "step": 26548
+    },
+    {
+      "epoch": 0.965523511655817,
+      "grad_norm": 0.0,
+      "learning_rate": 6.227646773160013e-08,
+      "loss": 0.9545,
+      "step": 26549
+    },
+    {
+      "epoch": 0.9655598792595556,
+      "grad_norm": 0.0,
+      "learning_rate": 6.214528825479239e-08,
+      "loss": 0.8294,
+      "step": 26550
+    },
+    {
+      "epoch": 0.9655962468632941,
+      "grad_norm": 0.0,
+      "learning_rate": 6.201424665161937e-08,
+      "loss": 0.8388,
+      "step": 26551
+    },
+    {
+      "epoch": 0.9656326144670327,
+      "grad_norm": 0.0,
+      "learning_rate": 6.188334292389741e-08,
+      "loss": 0.8233,
+      "step": 26552
+    },
+    {
+      "epoch": 0.9656689820707713,
+      "grad_norm": 0.0,
+      "learning_rate": 6.175257707344506e-08,
+      "loss": 0.9544,
+      "step": 26553
+    },
+    {
+      "epoch": 0.9657053496745099,
+      "grad_norm": 0.0,
+      "learning_rate": 6.16219491020742e-08,
+      "loss": 0.9511,
+      "step": 26554
+    },
+    {
+      "epoch": 0.9657417172782485,
+      "grad_norm": 0.0,
+      "learning_rate": 6.149145901159782e-08,
+      "loss": 0.9324,
+      "step": 26555
+    },
+    {
+      "epoch": 0.9657780848819871,
+      "grad_norm": 0.0,
+      "learning_rate": 6.136110680382779e-08,
+      "loss": 0.9256,
+      "step": 26556
+    },
+    {
+      "epoch": 0.9658144524857257,
+      "grad_norm": 0.0,
+      "learning_rate": 6.123089248056935e-08,
+      "loss": 0.8902,
+      "step": 26557
+    },
+    {
+      "epoch": 0.9658508200894643,
+      "grad_norm": 0.0,
+      "learning_rate": 6.110081604363327e-08,
+      "loss": 0.8822,
+      "step": 26558
+    },
+    {
+      "epoch": 0.9658871876932029,
+      "grad_norm": 0.0,
+      "learning_rate": 6.097087749482033e-08,
+      "loss": 0.8671,
+      "step": 26559
+    },
+    {
+      "epoch": 0.9659235552969415,
+      "grad_norm": 0.0,
+      "learning_rate": 6.084107683593576e-08,
+      "loss": 0.8859,
+      "step": 26560
+    },
+    {
+      "epoch": 0.9659599229006801,
+      "grad_norm": 0.0,
+      "learning_rate": 6.071141406877923e-08,
+      "loss": 0.8496,
+      "step": 26561
+    },
+    {
+      "epoch": 0.9659962905044187,
+      "grad_norm": 0.0,
+      "learning_rate": 6.058188919514928e-08,
+      "loss": 0.8408,
+      "step": 26562
+    },
+    {
+      "epoch": 0.9660326581081573,
+      "grad_norm": 0.0,
+      "learning_rate": 6.04525022168434e-08,
+      "loss": 0.9304,
+      "step": 26563
+    },
+    {
+      "epoch": 0.9660690257118958,
+      "grad_norm": 0.0,
+      "learning_rate": 6.03232531356568e-08,
+      "loss": 0.8341,
+      "step": 26564
+    },
+    {
+      "epoch": 0.9661053933156344,
+      "grad_norm": 0.0,
+      "learning_rate": 6.019414195338358e-08,
+      "loss": 0.883,
+      "step": 26565
+    },
+    {
+      "epoch": 0.966141760919373,
+      "grad_norm": 0.0,
+      "learning_rate": 6.006516867181234e-08,
+      "loss": 0.912,
+      "step": 26566
+    },
+    {
+      "epoch": 0.9661781285231116,
+      "grad_norm": 0.0,
+      "learning_rate": 5.993633329273496e-08,
+      "loss": 0.898,
+      "step": 26567
+    },
+    {
+      "epoch": 0.9662144961268502,
+      "grad_norm": 0.0,
+      "learning_rate": 5.98076358179378e-08,
+      "loss": 0.9437,
+      "step": 26568
+    },
+    {
+      "epoch": 0.9662508637305888,
+      "grad_norm": 0.0,
+      "learning_rate": 5.96790762492061e-08,
+      "loss": 1.0078,
+      "step": 26569
+    },
+    {
+      "epoch": 0.9662872313343274,
+      "grad_norm": 0.0,
+      "learning_rate": 5.955065458832399e-08,
+      "loss": 0.9523,
+      "step": 26570
+    },
+    {
+      "epoch": 0.966323598938066,
+      "grad_norm": 0.0,
+      "learning_rate": 5.942237083707225e-08,
+      "loss": 0.8362,
+      "step": 26571
+    },
+    {
+      "epoch": 0.9663599665418046,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9294224997231695e-08,
+      "loss": 0.8586,
+      "step": 26572
+    },
+    {
+      "epoch": 0.9663963341455432,
+      "grad_norm": 0.0,
+      "learning_rate": 5.916621707057979e-08,
+      "loss": 0.9719,
+      "step": 26573
+    },
+    {
+      "epoch": 0.9664327017492818,
+      "grad_norm": 0.0,
+      "learning_rate": 5.903834705889289e-08,
+      "loss": 0.8806,
+      "step": 26574
+    },
+    {
+      "epoch": 0.9664690693530203,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8910614963944015e-08,
+      "loss": 0.887,
+      "step": 26575
+    },
+    {
+      "epoch": 0.966505436956759,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8783020787506196e-08,
+      "loss": 0.8742,
+      "step": 26576
+    },
+    {
+      "epoch": 0.9665418045604975,
+      "grad_norm": 0.0,
+      "learning_rate": 5.865556453134802e-08,
+      "loss": 0.9598,
+      "step": 26577
+    },
+    {
+      "epoch": 0.9665781721642361,
+      "grad_norm": 0.0,
+      "learning_rate": 5.852824619724029e-08,
+      "loss": 0.9459,
+      "step": 26578
+    },
+    {
+      "epoch": 0.9666145397679747,
+      "grad_norm": 0.0,
+      "learning_rate": 5.840106578694826e-08,
+      "loss": 0.8843,
+      "step": 26579
+    },
+    {
+      "epoch": 0.9666509073717133,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8274023302234975e-08,
+      "loss": 0.9831,
+      "step": 26580
+    },
+    {
+      "epoch": 0.9666872749754518,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8147118744864563e-08,
+      "loss": 0.9857,
+      "step": 26581
+    },
+    {
+      "epoch": 0.9667236425791904,
+      "grad_norm": 0.0,
+      "learning_rate": 5.802035211659784e-08,
+      "loss": 0.9182,
+      "step": 26582
+    },
+    {
+      "epoch": 0.966760010182929,
+      "grad_norm": 0.0,
+      "learning_rate": 5.789372341919231e-08,
+      "loss": 0.9576,
+      "step": 26583
+    },
+    {
+      "epoch": 0.9667963777866676,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7767232654405424e-08,
+      "loss": 0.8954,
+      "step": 26584
+    },
+    {
+      "epoch": 0.9668327453904062,
+      "grad_norm": 0.0,
+      "learning_rate": 5.764087982399136e-08,
+      "loss": 0.9771,
+      "step": 26585
+    },
+    {
+      "epoch": 0.9668691129941448,
+      "grad_norm": 0.0,
+      "learning_rate": 5.751466492970537e-08,
+      "loss": 0.9323,
+      "step": 26586
+    },
+    {
+      "epoch": 0.9669054805978834,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7388587973294943e-08,
+      "loss": 0.9199,
+      "step": 26587
+    },
+    {
+      "epoch": 0.966941848201622,
+      "grad_norm": 0.0,
+      "learning_rate": 5.726264895651312e-08,
+      "loss": 0.877,
+      "step": 26588
+    },
+    {
+      "epoch": 0.9669782158053606,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7136847881102966e-08,
+      "loss": 0.9238,
+      "step": 26589
+    },
+    {
+      "epoch": 0.9670145834090992,
+      "grad_norm": 0.0,
+      "learning_rate": 5.7011184748814176e-08,
+      "loss": 0.9833,
+      "step": 26590
+    },
+    {
+      "epoch": 0.9670509510128378,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6885659561386476e-08,
+      "loss": 0.9038,
+      "step": 26591
+    },
+    {
+      "epoch": 0.9670873186165764,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6760272320562915e-08,
+      "loss": 0.8053,
+      "step": 26592
+    },
+    {
+      "epoch": 0.967123686220315,
+      "grad_norm": 0.0,
+      "learning_rate": 5.663502302808321e-08,
+      "loss": 0.8928,
+      "step": 26593
+    },
+    {
+      "epoch": 0.9671600538240536,
+      "grad_norm": 0.0,
+      "learning_rate": 5.650991168568487e-08,
+      "loss": 0.9421,
+      "step": 26594
+    },
+    {
+      "epoch": 0.9671964214277922,
+      "grad_norm": 0.0,
+      "learning_rate": 5.638493829510427e-08,
+      "loss": 0.9188,
+      "step": 26595
+    },
+    {
+      "epoch": 0.9672327890315308,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6260102858073375e-08,
+      "loss": 0.9326,
+      "step": 26596
+    },
+    {
+      "epoch": 0.9672691566352694,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6135405376325224e-08,
+      "loss": 0.9445,
+      "step": 26597
+    },
+    {
+      "epoch": 0.9673055242390078,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6010845851590664e-08,
+      "loss": 0.972,
+      "step": 26598
+    },
+    {
+      "epoch": 0.9673418918427464,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5886424285596095e-08,
+      "loss": 1.0313,
+      "step": 26599
+    },
+    {
+      "epoch": 0.967378259446485,
+      "grad_norm": 0.0,
+      "learning_rate": 5.576214068006902e-08,
+      "loss": 1.074,
+      "step": 26600
+    },
+    {
+      "epoch": 0.9674146270502236,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5637995036733615e-08,
+      "loss": 0.9037,
+      "step": 26601
+    },
+    {
+      "epoch": 0.9674509946539622,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5513987357310727e-08,
+      "loss": 1.0154,
+      "step": 26602
+    },
+    {
+      "epoch": 0.9674873622577008,
+      "grad_norm": 0.0,
+      "learning_rate": 5.539011764352231e-08,
+      "loss": 0.8055,
+      "step": 26603
+    },
+    {
+      "epoch": 0.9675237298614394,
+      "grad_norm": 0.0,
+      "learning_rate": 5.5266385897086996e-08,
+      "loss": 0.9594,
+      "step": 26604
+    },
+    {
+      "epoch": 0.967560097465178,
+      "grad_norm": 0.0,
+      "learning_rate": 5.514279211972118e-08,
+      "loss": 0.9674,
+      "step": 26605
+    },
+    {
+      "epoch": 0.9675964650689166,
+      "grad_norm": 0.0,
+      "learning_rate": 5.501933631313794e-08,
+      "loss": 0.9244,
+      "step": 26606
+    },
+    {
+      "epoch": 0.9676328326726552,
+      "grad_norm": 0.0,
+      "learning_rate": 5.489601847905257e-08,
+      "loss": 1.0007,
+      "step": 26607
+    },
+    {
+      "epoch": 0.9676692002763938,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4772838619173704e-08,
+      "loss": 0.843,
+      "step": 26608
+    },
+    {
+      "epoch": 0.9677055678801324,
+      "grad_norm": 0.0,
+      "learning_rate": 5.464979673521109e-08,
+      "loss": 0.8988,
+      "step": 26609
+    },
+    {
+      "epoch": 0.967741935483871,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4526892828872244e-08,
+      "loss": 0.9425,
+      "step": 26610
+    },
+    {
+      "epoch": 0.9677783030876096,
+      "grad_norm": 0.0,
+      "learning_rate": 5.440412690186248e-08,
+      "loss": 0.9262,
+      "step": 26611
+    },
+    {
+      "epoch": 0.9678146706913482,
+      "grad_norm": 0.0,
+      "learning_rate": 5.428149895588375e-08,
+      "loss": 0.9568,
+      "step": 26612
+    },
+    {
+      "epoch": 0.9678510382950868,
+      "grad_norm": 0.0,
+      "learning_rate": 5.415900899263915e-08,
+      "loss": 0.8534,
+      "step": 26613
+    },
+    {
+      "epoch": 0.9678874058988254,
+      "grad_norm": 0.0,
+      "learning_rate": 5.403665701382621e-08,
+      "loss": 0.9105,
+      "step": 26614
+    },
+    {
+      "epoch": 0.9679237735025639,
+      "grad_norm": 0.0,
+      "learning_rate": 5.391444302114357e-08,
+      "loss": 0.9411,
+      "step": 26615
+    },
+    {
+      "epoch": 0.9679601411063025,
+      "grad_norm": 0.0,
+      "learning_rate": 5.379236701628654e-08,
+      "loss": 0.9696,
+      "step": 26616
+    },
+    {
+      "epoch": 0.9679965087100411,
+      "grad_norm": 0.0,
+      "learning_rate": 5.367042900094821e-08,
+      "loss": 0.8683,
+      "step": 26617
+    },
+    {
+      "epoch": 0.9680328763137797,
+      "grad_norm": 0.0,
+      "learning_rate": 5.354862897682056e-08,
+      "loss": 0.9242,
+      "step": 26618
+    },
+    {
+      "epoch": 0.9680692439175183,
+      "grad_norm": 0.0,
+      "learning_rate": 5.342696694559335e-08,
+      "loss": 0.9506,
+      "step": 26619
+    },
+    {
+      "epoch": 0.9681056115212568,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3305442908955226e-08,
+      "loss": 0.8337,
+      "step": 26620
+    },
+    {
+      "epoch": 0.9681419791249954,
+      "grad_norm": 0.0,
+      "learning_rate": 5.3184056868590405e-08,
+      "loss": 0.9325,
+      "step": 26621
+    },
+    {
+      "epoch": 0.968178346728734,
+      "grad_norm": 0.0,
+      "learning_rate": 5.306280882618531e-08,
+      "loss": 0.8838,
+      "step": 26622
+    },
+    {
+      "epoch": 0.9682147143324726,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2941698783420814e-08,
+      "loss": 0.8767,
+      "step": 26623
+    },
+    {
+      "epoch": 0.9682510819362112,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2820726741976694e-08,
+      "loss": 0.8454,
+      "step": 26624
+    },
+    {
+      "epoch": 0.9682874495399498,
+      "grad_norm": 0.0,
+      "learning_rate": 5.26998927035316e-08,
+      "loss": 0.8836,
+      "step": 26625
+    },
+    {
+      "epoch": 0.9683238171436884,
+      "grad_norm": 0.0,
+      "learning_rate": 5.257919666976308e-08,
+      "loss": 0.9048,
+      "step": 26626
+    },
+    {
+      "epoch": 0.968360184747427,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2458638642344237e-08,
+      "loss": 0.9736,
+      "step": 26627
+    },
+    {
+      "epoch": 0.9683965523511656,
+      "grad_norm": 0.0,
+      "learning_rate": 5.233821862294819e-08,
+      "loss": 0.9978,
+      "step": 26628
+    },
+    {
+      "epoch": 0.9684329199549042,
+      "grad_norm": 0.0,
+      "learning_rate": 5.2217936613244705e-08,
+      "loss": 0.8799,
+      "step": 26629
+    },
+    {
+      "epoch": 0.9684692875586428,
+      "grad_norm": 0.0,
+      "learning_rate": 5.209779261490466e-08,
+      "loss": 0.9095,
+      "step": 26630
+    },
+    {
+      "epoch": 0.9685056551623813,
+      "grad_norm": 0.0,
+      "learning_rate": 5.19777866295923e-08,
+      "loss": 0.9747,
+      "step": 26631
+    },
+    {
+      "epoch": 0.9685420227661199,
+      "grad_norm": 0.0,
+      "learning_rate": 5.185791865897405e-08,
+      "loss": 0.8585,
+      "step": 26632
+    },
+    {
+      "epoch": 0.9685783903698585,
+      "grad_norm": 0.0,
+      "learning_rate": 5.173818870471303e-08,
+      "loss": 0.8106,
+      "step": 26633
+    },
+    {
+      "epoch": 0.9686147579735971,
+      "grad_norm": 0.0,
+      "learning_rate": 5.161859676846903e-08,
+      "loss": 0.7887,
+      "step": 26634
+    },
+    {
+      "epoch": 0.9686511255773357,
+      "grad_norm": 0.0,
+      "learning_rate": 5.149914285190294e-08,
+      "loss": 0.976,
+      "step": 26635
+    },
+    {
+      "epoch": 0.9686874931810743,
+      "grad_norm": 0.0,
+      "learning_rate": 5.137982695667121e-08,
+      "loss": 0.9232,
+      "step": 26636
+    },
+    {
+      "epoch": 0.9687238607848129,
+      "grad_norm": 0.0,
+      "learning_rate": 5.126064908442918e-08,
+      "loss": 0.9479,
+      "step": 26637
+    },
+    {
+      "epoch": 0.9687602283885515,
+      "grad_norm": 0.0,
+      "learning_rate": 5.114160923682887e-08,
+      "loss": 0.8452,
+      "step": 26638
+    },
+    {
+      "epoch": 0.9687965959922901,
+      "grad_norm": 0.0,
+      "learning_rate": 5.102270741552451e-08,
+      "loss": 0.9797,
+      "step": 26639
+    },
+    {
+      "epoch": 0.9688329635960287,
+      "grad_norm": 0.0,
+      "learning_rate": 5.090394362216367e-08,
+      "loss": 0.8486,
+      "step": 26640
+    },
+    {
+      "epoch": 0.9688693311997673,
+      "grad_norm": 0.0,
+      "learning_rate": 5.078531785839502e-08,
+      "loss": 0.9341,
+      "step": 26641
+    },
+    {
+      "epoch": 0.9689056988035059,
+      "grad_norm": 0.0,
+      "learning_rate": 5.066683012586393e-08,
+      "loss": 1.0168,
+      "step": 26642
+    },
+    {
+      "epoch": 0.9689420664072445,
+      "grad_norm": 0.0,
+      "learning_rate": 5.054848042621463e-08,
+      "loss": 0.8005,
+      "step": 26643
+    },
+    {
+      "epoch": 0.968978434010983,
+      "grad_norm": 0.0,
+      "learning_rate": 5.043026876108803e-08,
+      "loss": 0.9059,
+      "step": 26644
+    },
+    {
+      "epoch": 0.9690148016147216,
+      "grad_norm": 0.0,
+      "learning_rate": 5.031219513212504e-08,
+      "loss": 0.8435,
+      "step": 26645
+    },
+    {
+      "epoch": 0.9690511692184602,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0194259540964354e-08,
+      "loss": 0.9251,
+      "step": 26646
+    },
+    {
+      "epoch": 0.9690875368221988,
+      "grad_norm": 0.0,
+      "learning_rate": 5.007646198924132e-08,
+      "loss": 0.8908,
+      "step": 26647
+    },
+    {
+      "epoch": 0.9691239044259373,
+      "grad_norm": 0.0,
+      "learning_rate": 4.995880247858909e-08,
+      "loss": 0.8872,
+      "step": 26648
+    },
+    {
+      "epoch": 0.9691602720296759,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9841281010641893e-08,
+      "loss": 0.9373,
+      "step": 26649
+    },
+    {
+      "epoch": 0.9691966396334145,
+      "grad_norm": 0.0,
+      "learning_rate": 4.972389758702956e-08,
+      "loss": 0.9054,
+      "step": 26650
+    },
+    {
+      "epoch": 0.9692330072371531,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9606652209380765e-08,
+      "loss": 0.9247,
+      "step": 26651
+    },
+    {
+      "epoch": 0.9692693748408917,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9489544879322004e-08,
+      "loss": 0.9441,
+      "step": 26652
+    },
+    {
+      "epoch": 0.9693057424446303,
+      "grad_norm": 0.0,
+      "learning_rate": 4.937257559847752e-08,
+      "loss": 0.9604,
+      "step": 26653
+    },
+    {
+      "epoch": 0.9693421100483689,
+      "grad_norm": 0.0,
+      "learning_rate": 4.925574436847047e-08,
+      "loss": 0.8596,
+      "step": 26654
+    },
+    {
+      "epoch": 0.9693784776521075,
+      "grad_norm": 0.0,
+      "learning_rate": 4.913905119092177e-08,
+      "loss": 0.8327,
+      "step": 26655
+    },
+    {
+      "epoch": 0.9694148452558461,
+      "grad_norm": 0.0,
+      "learning_rate": 4.9022496067450133e-08,
+      "loss": 0.9627,
+      "step": 26656
+    },
+    {
+      "epoch": 0.9694512128595847,
+      "grad_norm": 0.0,
+      "learning_rate": 4.890607899967315e-08,
+      "loss": 0.8751,
+      "step": 26657
+    },
+    {
+      "epoch": 0.9694875804633233,
+      "grad_norm": 0.0,
+      "learning_rate": 4.878979998920508e-08,
+      "loss": 0.837,
+      "step": 26658
+    },
+    {
+      "epoch": 0.9695239480670619,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8673659037660193e-08,
+      "loss": 0.9506,
+      "step": 26659
+    },
+    {
+      "epoch": 0.9695603156708005,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8557656146648316e-08,
+      "loss": 0.8757,
+      "step": 26660
+    },
+    {
+      "epoch": 0.9695966832745391,
+      "grad_norm": 0.0,
+      "learning_rate": 4.844179131778037e-08,
+      "loss": 0.7949,
+      "step": 26661
+    },
+    {
+      "epoch": 0.9696330508782777,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8326064552662865e-08,
+      "loss": 0.9046,
+      "step": 26662
+    },
+    {
+      "epoch": 0.9696694184820163,
+      "grad_norm": 0.0,
+      "learning_rate": 4.8210475852902285e-08,
+      "loss": 0.9099,
+      "step": 26663
+    },
+    {
+      "epoch": 0.9697057860857549,
+      "grad_norm": 0.0,
+      "learning_rate": 4.809502522010068e-08,
+      "loss": 0.9563,
+      "step": 26664
+    },
+    {
+      "epoch": 0.9697421536894933,
+      "grad_norm": 0.0,
+      "learning_rate": 4.797971265586121e-08,
+      "loss": 0.9363,
+      "step": 26665
+    },
+    {
+      "epoch": 0.9697785212932319,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7864538161783714e-08,
+      "loss": 0.9299,
+      "step": 26666
+    },
+    {
+      "epoch": 0.9698148888969705,
+      "grad_norm": 0.0,
+      "learning_rate": 4.774950173946358e-08,
+      "loss": 0.7715,
+      "step": 26667
+    },
+    {
+      "epoch": 0.9698512565007091,
+      "grad_norm": 0.0,
+      "learning_rate": 4.763460339050063e-08,
+      "loss": 0.819,
+      "step": 26668
+    },
+    {
+      "epoch": 0.9698876241044477,
+      "grad_norm": 0.0,
+      "learning_rate": 4.751984311648694e-08,
+      "loss": 0.8965,
+      "step": 26669
+    },
+    {
+      "epoch": 0.9699239917081863,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7405220919013454e-08,
+      "loss": 0.8116,
+      "step": 26670
+    },
+    {
+      "epoch": 0.9699603593119249,
+      "grad_norm": 0.0,
+      "learning_rate": 4.729073679967222e-08,
+      "loss": 0.9738,
+      "step": 26671
+    },
+    {
+      "epoch": 0.9699967269156635,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7176390760051985e-08,
+      "loss": 1.0091,
+      "step": 26672
+    },
+    {
+      "epoch": 0.9700330945194021,
+      "grad_norm": 0.0,
+      "learning_rate": 4.706218280173702e-08,
+      "loss": 0.8718,
+      "step": 26673
+    },
+    {
+      "epoch": 0.9700694621231407,
+      "grad_norm": 0.0,
+      "learning_rate": 4.694811292631274e-08,
+      "loss": 0.8113,
+      "step": 26674
+    },
+    {
+      "epoch": 0.9701058297268793,
+      "grad_norm": 0.0,
+      "learning_rate": 4.683418113536231e-08,
+      "loss": 1.0045,
+      "step": 26675
+    },
+    {
+      "epoch": 0.9701421973306179,
+      "grad_norm": 0.0,
+      "learning_rate": 4.672038743046558e-08,
+      "loss": 0.9153,
+      "step": 26676
+    },
+    {
+      "epoch": 0.9701785649343565,
+      "grad_norm": 0.0,
+      "learning_rate": 4.660673181320241e-08,
+      "loss": 0.9384,
+      "step": 26677
+    },
+    {
+      "epoch": 0.9702149325380951,
+      "grad_norm": 0.0,
+      "learning_rate": 4.649321428514819e-08,
+      "loss": 1.0102,
+      "step": 26678
+    },
+    {
+      "epoch": 0.9702513001418337,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6379834847878335e-08,
+      "loss": 0.8836,
+      "step": 26679
+    },
+    {
+      "epoch": 0.9702876677455723,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6266593502966026e-08,
+      "loss": 0.9083,
+      "step": 26680
+    },
+    {
+      "epoch": 0.9703240353493109,
+      "grad_norm": 0.0,
+      "learning_rate": 4.615349025198335e-08,
+      "loss": 0.8471,
+      "step": 26681
+    },
+    {
+      "epoch": 0.9703604029530494,
+      "grad_norm": 0.0,
+      "learning_rate": 4.604052509649681e-08,
+      "loss": 0.9203,
+      "step": 26682
+    },
+    {
+      "epoch": 0.970396770556788,
+      "grad_norm": 0.0,
+      "learning_rate": 4.592769803807517e-08,
+      "loss": 0.9129,
+      "step": 26683
+    },
+    {
+      "epoch": 0.9704331381605266,
+      "grad_norm": 0.0,
+      "learning_rate": 4.581500907828496e-08,
+      "loss": 0.9996,
+      "step": 26684
+    },
+    {
+      "epoch": 0.9704695057642652,
+      "grad_norm": 0.0,
+      "learning_rate": 4.570245821868713e-08,
+      "loss": 0.7463,
+      "step": 26685
+    },
+    {
+      "epoch": 0.9705058733680038,
+      "grad_norm": 0.0,
+      "learning_rate": 4.559004546084489e-08,
+      "loss": 0.9229,
+      "step": 26686
+    },
+    {
+      "epoch": 0.9705422409717424,
+      "grad_norm": 0.0,
+      "learning_rate": 4.54777708063181e-08,
+      "loss": 0.9794,
+      "step": 26687
+    },
+    {
+      "epoch": 0.970578608575481,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5365634256662186e-08,
+      "loss": 0.9711,
+      "step": 26688
+    },
+    {
+      "epoch": 0.9706149761792195,
+      "grad_norm": 0.0,
+      "learning_rate": 4.525363581343478e-08,
+      "loss": 0.9217,
+      "step": 26689
+    },
+    {
+      "epoch": 0.9706513437829581,
+      "grad_norm": 0.0,
+      "learning_rate": 4.51417754781891e-08,
+      "loss": 1.01,
+      "step": 26690
+    },
+    {
+      "epoch": 0.9706877113866967,
+      "grad_norm": 0.0,
+      "learning_rate": 4.5030053252478334e-08,
+      "loss": 0.9039,
+      "step": 26691
+    },
+    {
+      "epoch": 0.9707240789904353,
+      "grad_norm": 0.0,
+      "learning_rate": 4.491846913785014e-08,
+      "loss": 0.8614,
+      "step": 26692
+    },
+    {
+      "epoch": 0.9707604465941739,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4807023135854386e-08,
+      "loss": 0.9106,
+      "step": 26693
+    },
+    {
+      "epoch": 0.9707968141979125,
+      "grad_norm": 0.0,
+      "learning_rate": 4.469571524803651e-08,
+      "loss": 0.8623,
+      "step": 26694
+    },
+    {
+      "epoch": 0.9708331818016511,
+      "grad_norm": 0.0,
+      "learning_rate": 4.458454547594082e-08,
+      "loss": 0.8188,
+      "step": 26695
+    },
+    {
+      "epoch": 0.9708695494053897,
+      "grad_norm": 0.0,
+      "learning_rate": 4.447351382110942e-08,
+      "loss": 0.8566,
+      "step": 26696
+    },
+    {
+      "epoch": 0.9709059170091283,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4362620285083314e-08,
+      "loss": 0.8036,
+      "step": 26697
+    },
+    {
+      "epoch": 0.9709422846128669,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4251864869400144e-08,
+      "loss": 0.8887,
+      "step": 26698
+    },
+    {
+      "epoch": 0.9709786522166054,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4141247575596456e-08,
+      "loss": 0.8345,
+      "step": 26699
+    },
+    {
+      "epoch": 0.971015019820344,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4030768405208814e-08,
+      "loss": 0.9424,
+      "step": 26700
+    },
+    {
+      "epoch": 0.9710513874240826,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3920427359768204e-08,
+      "loss": 0.9417,
+      "step": 26701
+    },
+    {
+      "epoch": 0.9710877550278212,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3810224440805626e-08,
+      "loss": 0.9573,
+      "step": 26702
+    },
+    {
+      "epoch": 0.9711241226315598,
+      "grad_norm": 0.0,
+      "learning_rate": 4.370015964985097e-08,
+      "loss": 0.9466,
+      "step": 26703
+    },
+    {
+      "epoch": 0.9711604902352984,
+      "grad_norm": 0.0,
+      "learning_rate": 4.359023298842968e-08,
+      "loss": 0.8933,
+      "step": 26704
+    },
+    {
+      "epoch": 0.971196857839037,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3480444458067204e-08,
+      "loss": 0.938,
+      "step": 26705
+    },
+    {
+      "epoch": 0.9712332254427756,
+      "grad_norm": 0.0,
+      "learning_rate": 4.337079406028788e-08,
+      "loss": 0.9077,
+      "step": 26706
+    },
+    {
+      "epoch": 0.9712695930465142,
+      "grad_norm": 0.0,
+      "learning_rate": 4.32612817966116e-08,
+      "loss": 0.8383,
+      "step": 26707
+    },
+    {
+      "epoch": 0.9713059606502528,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3151907668558256e-08,
+      "loss": 0.8841,
+      "step": 26708
+    },
+    {
+      "epoch": 0.9713423282539914,
+      "grad_norm": 0.0,
+      "learning_rate": 4.304267167764553e-08,
+      "loss": 0.8606,
+      "step": 26709
+    },
+    {
+      "epoch": 0.97137869585773,
+      "grad_norm": 0.0,
+      "learning_rate": 4.293357382538887e-08,
+      "loss": 0.9223,
+      "step": 26710
+    },
+    {
+      "epoch": 0.9714150634614686,
+      "grad_norm": 0.0,
+      "learning_rate": 4.282461411330152e-08,
+      "loss": 0.7321,
+      "step": 26711
+    },
+    {
+      "epoch": 0.9714514310652071,
+      "grad_norm": 0.0,
+      "learning_rate": 4.271579254289559e-08,
+      "loss": 0.8621,
+      "step": 26712
+    },
+    {
+      "epoch": 0.9714877986689457,
+      "grad_norm": 0.0,
+      "learning_rate": 4.260710911567989e-08,
+      "loss": 0.9641,
+      "step": 26713
+    },
+    {
+      "epoch": 0.9715241662726843,
+      "grad_norm": 0.0,
+      "learning_rate": 4.24985638331632e-08,
+      "loss": 1.0802,
+      "step": 26714
+    },
+    {
+      "epoch": 0.9715605338764228,
+      "grad_norm": 0.0,
+      "learning_rate": 4.239015669685098e-08,
+      "loss": 1.0249,
+      "step": 26715
+    },
+    {
+      "epoch": 0.9715969014801614,
+      "grad_norm": 0.0,
+      "learning_rate": 4.228188770824759e-08,
+      "loss": 0.8652,
+      "step": 26716
+    },
+    {
+      "epoch": 0.9716332690839,
+      "grad_norm": 0.0,
+      "learning_rate": 4.217375686885406e-08,
+      "loss": 0.9695,
+      "step": 26717
+    },
+    {
+      "epoch": 0.9716696366876386,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2065764180172496e-08,
+      "loss": 0.9302,
+      "step": 26718
+    },
+    {
+      "epoch": 0.9717060042913772,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1957909643699504e-08,
+      "loss": 1.001,
+      "step": 26719
+    },
+    {
+      "epoch": 0.9717423718951158,
+      "grad_norm": 0.0,
+      "learning_rate": 4.185019326093165e-08,
+      "loss": 0.919,
+      "step": 26720
+    },
+    {
+      "epoch": 0.9717787394988544,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1742615033363296e-08,
+      "loss": 0.8819,
+      "step": 26721
+    },
+    {
+      "epoch": 0.971815107102593,
+      "grad_norm": 0.0,
+      "learning_rate": 4.16351749624877e-08,
+      "loss": 0.9172,
+      "step": 26722
+    },
+    {
+      "epoch": 0.9718514747063316,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1527873049795885e-08,
+      "loss": 0.9034,
+      "step": 26723
+    },
+    {
+      "epoch": 0.9718878423100702,
+      "grad_norm": 0.0,
+      "learning_rate": 4.142070929677333e-08,
+      "loss": 0.8656,
+      "step": 26724
+    },
+    {
+      "epoch": 0.9719242099138088,
+      "grad_norm": 0.0,
+      "learning_rate": 4.1313683704911066e-08,
+      "loss": 0.8494,
+      "step": 26725
+    },
+    {
+      "epoch": 0.9719605775175474,
+      "grad_norm": 0.0,
+      "learning_rate": 4.120679627569124e-08,
+      "loss": 0.8812,
+      "step": 26726
+    },
+    {
+      "epoch": 0.971996945121286,
+      "grad_norm": 0.0,
+      "learning_rate": 4.110004701059711e-08,
+      "loss": 0.9529,
+      "step": 26727
+    },
+    {
+      "epoch": 0.9720333127250246,
+      "grad_norm": 0.0,
+      "learning_rate": 4.099343591111082e-08,
+      "loss": 0.7889,
+      "step": 26728
+    },
+    {
+      "epoch": 0.9720696803287632,
+      "grad_norm": 0.0,
+      "learning_rate": 4.088696297871009e-08,
+      "loss": 0.9298,
+      "step": 26729
+    },
+    {
+      "epoch": 0.9721060479325018,
+      "grad_norm": 0.0,
+      "learning_rate": 4.078062821487261e-08,
+      "loss": 0.8205,
+      "step": 26730
+    },
+    {
+      "epoch": 0.9721424155362404,
+      "grad_norm": 0.0,
+      "learning_rate": 4.067443162107387e-08,
+      "loss": 0.9587,
+      "step": 26731
+    },
+    {
+      "epoch": 0.9721787831399789,
+      "grad_norm": 0.0,
+      "learning_rate": 4.056837319878715e-08,
+      "loss": 0.9105,
+      "step": 26732
+    },
+    {
+      "epoch": 0.9722151507437174,
+      "grad_norm": 0.0,
+      "learning_rate": 4.046245294948237e-08,
+      "loss": 0.9252,
+      "step": 26733
+    },
+    {
+      "epoch": 0.972251518347456,
+      "grad_norm": 0.0,
+      "learning_rate": 4.0356670874631685e-08,
+      "loss": 1.0567,
+      "step": 26734
+    },
+    {
+      "epoch": 0.9722878859511946,
+      "grad_norm": 0.0,
+      "learning_rate": 4.02510269757006e-08,
+      "loss": 0.9072,
+      "step": 26735
+    },
+    {
+      "epoch": 0.9723242535549332,
+      "grad_norm": 0.0,
+      "learning_rate": 4.014552125415572e-08,
+      "loss": 0.9174,
+      "step": 26736
+    },
+    {
+      "epoch": 0.9723606211586718,
+      "grad_norm": 0.0,
+      "learning_rate": 4.004015371146031e-08,
+      "loss": 1.0118,
+      "step": 26737
+    },
+    {
+      "epoch": 0.9723969887624104,
+      "grad_norm": 0.0,
+      "learning_rate": 3.993492434907764e-08,
+      "loss": 0.8957,
+      "step": 26738
+    },
+    {
+      "epoch": 0.972433356366149,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9829833168465446e-08,
+      "loss": 0.9436,
+      "step": 26739
+    },
+    {
+      "epoch": 0.9724697239698876,
+      "grad_norm": 0.0,
+      "learning_rate": 3.972488017108256e-08,
+      "loss": 0.8606,
+      "step": 26740
+    },
+    {
+      "epoch": 0.9725060915736262,
+      "grad_norm": 0.0,
+      "learning_rate": 3.962006535838559e-08,
+      "loss": 0.9228,
+      "step": 26741
+    },
+    {
+      "epoch": 0.9725424591773648,
+      "grad_norm": 0.0,
+      "learning_rate": 3.951538873182781e-08,
+      "loss": 0.8311,
+      "step": 26742
+    },
+    {
+      "epoch": 0.9725788267811034,
+      "grad_norm": 0.0,
+      "learning_rate": 3.94108502928614e-08,
+      "loss": 0.9798,
+      "step": 26743
+    },
+    {
+      "epoch": 0.972615194384842,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9306450042938535e-08,
+      "loss": 0.8804,
+      "step": 26744
+    },
+    {
+      "epoch": 0.9726515619885806,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9202187983505836e-08,
+      "loss": 0.9061,
+      "step": 26745
+    },
+    {
+      "epoch": 0.9726879295923192,
+      "grad_norm": 0.0,
+      "learning_rate": 3.9098064116009915e-08,
+      "loss": 0.8853,
+      "step": 26746
+    },
+    {
+      "epoch": 0.9727242971960578,
+      "grad_norm": 0.0,
+      "learning_rate": 3.899407844189518e-08,
+      "loss": 0.879,
+      "step": 26747
+    },
+    {
+      "epoch": 0.9727606647997964,
+      "grad_norm": 0.0,
+      "learning_rate": 3.889023096260603e-08,
+      "loss": 0.9226,
+      "step": 26748
+    },
+    {
+      "epoch": 0.9727970324035349,
+      "grad_norm": 0.0,
+      "learning_rate": 3.87865216795813e-08,
+      "loss": 0.9633,
+      "step": 26749
+    },
+    {
+      "epoch": 0.9728334000072735,
+      "grad_norm": 0.0,
+      "learning_rate": 3.868295059426097e-08,
+      "loss": 0.9013,
+      "step": 26750
+    },
+    {
+      "epoch": 0.9728697676110121,
+      "grad_norm": 0.0,
+      "learning_rate": 3.857951770808166e-08,
+      "loss": 0.9037,
+      "step": 26751
+    },
+    {
+      "epoch": 0.9729061352147507,
+      "grad_norm": 0.0,
+      "learning_rate": 3.847622302247667e-08,
+      "loss": 0.9053,
+      "step": 26752
+    },
+    {
+      "epoch": 0.9729425028184893,
+      "grad_norm": 0.0,
+      "learning_rate": 3.837306653888151e-08,
+      "loss": 0.8102,
+      "step": 26753
+    },
+    {
+      "epoch": 0.9729788704222279,
+      "grad_norm": 0.0,
+      "learning_rate": 3.8270048258726154e-08,
+      "loss": 0.9015,
+      "step": 26754
+    },
+    {
+      "epoch": 0.9730152380259665,
+      "grad_norm": 0.0,
+      "learning_rate": 3.816716818344057e-08,
+      "loss": 0.951,
+      "step": 26755
+    },
+    {
+      "epoch": 0.973051605629705,
+      "grad_norm": 0.0,
+      "learning_rate": 3.80644263144514e-08,
+      "loss": 0.8354,
+      "step": 26756
+    },
+    {
+      "epoch": 0.9730879732334436,
+      "grad_norm": 0.0,
+      "learning_rate": 3.796182265318305e-08,
+      "loss": 0.9226,
+      "step": 26757
+    },
+    {
+      "epoch": 0.9731243408371822,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7859357201061044e-08,
+      "loss": 0.969,
+      "step": 26758
+    },
+    {
+      "epoch": 0.9731607084409208,
+      "grad_norm": 0.0,
+      "learning_rate": 3.775702995950647e-08,
+      "loss": 0.9478,
+      "step": 26759
+    },
+    {
+      "epoch": 0.9731970760446594,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7654840929937095e-08,
+      "loss": 0.7847,
+      "step": 26760
+    },
+    {
+      "epoch": 0.973233443648398,
+      "grad_norm": 0.0,
+      "learning_rate": 3.755279011377399e-08,
+      "loss": 0.8681,
+      "step": 26761
+    },
+    {
+      "epoch": 0.9732698112521366,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7450877512429376e-08,
+      "loss": 0.8287,
+      "step": 26762
+    },
+    {
+      "epoch": 0.9733061788558752,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7349103127319874e-08,
+      "loss": 0.8179,
+      "step": 26763
+    },
+    {
+      "epoch": 0.9733425464596138,
+      "grad_norm": 0.0,
+      "learning_rate": 3.724746695985659e-08,
+      "loss": 0.9761,
+      "step": 26764
+    },
+    {
+      "epoch": 0.9733789140633524,
+      "grad_norm": 0.0,
+      "learning_rate": 3.71459690114484e-08,
+      "loss": 0.9606,
+      "step": 26765
+    },
+    {
+      "epoch": 0.9734152816670909,
+      "grad_norm": 0.0,
+      "learning_rate": 3.7044609283504174e-08,
+      "loss": 0.9082,
+      "step": 26766
+    },
+    {
+      "epoch": 0.9734516492708295,
+      "grad_norm": 0.0,
+      "learning_rate": 3.694338777743167e-08,
+      "loss": 0.8206,
+      "step": 26767
+    },
+    {
+      "epoch": 0.9734880168745681,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6842304494633106e-08,
+      "loss": 0.8464,
+      "step": 26768
+    },
+    {
+      "epoch": 0.9735243844783067,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6741359436512914e-08,
+      "loss": 0.99,
+      "step": 26769
+    },
+    {
+      "epoch": 0.9735607520820453,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6640552604468856e-08,
+      "loss": 0.9483,
+      "step": 26770
+    },
+    {
+      "epoch": 0.9735971196857839,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6539883999902046e-08,
+      "loss": 0.835,
+      "step": 26771
+    },
+    {
+      "epoch": 0.9736334872895225,
+      "grad_norm": 0.0,
+      "learning_rate": 3.643935362420692e-08,
+      "loss": 0.9132,
+      "step": 26772
+    },
+    {
+      "epoch": 0.9736698548932611,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6338961478780135e-08,
+      "loss": 0.8793,
+      "step": 26773
+    },
+    {
+      "epoch": 0.9737062224969997,
+      "grad_norm": 0.0,
+      "learning_rate": 3.623870756501391e-08,
+      "loss": 0.9465,
+      "step": 26774
+    },
+    {
+      "epoch": 0.9737425901007383,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6138591884298243e-08,
+      "loss": 0.9654,
+      "step": 26775
+    },
+    {
+      "epoch": 0.9737789577044769,
+      "grad_norm": 0.0,
+      "learning_rate": 3.603861443802314e-08,
+      "loss": 0.7456,
+      "step": 26776
+    },
+    {
+      "epoch": 0.9738153253082155,
+      "grad_norm": 0.0,
+      "learning_rate": 3.593877522757527e-08,
+      "loss": 0.8857,
+      "step": 26777
+    },
+    {
+      "epoch": 0.9738516929119541,
+      "grad_norm": 0.0,
+      "learning_rate": 3.583907425433908e-08,
+      "loss": 0.9294,
+      "step": 26778
+    },
+    {
+      "epoch": 0.9738880605156927,
+      "grad_norm": 0.0,
+      "learning_rate": 3.573951151969901e-08,
+      "loss": 0.9775,
+      "step": 26779
+    },
+    {
+      "epoch": 0.9739244281194313,
+      "grad_norm": 0.0,
+      "learning_rate": 3.564008702503619e-08,
+      "loss": 0.8913,
+      "step": 26780
+    },
+    {
+      "epoch": 0.9739607957231698,
+      "grad_norm": 0.0,
+      "learning_rate": 3.554080077172839e-08,
+      "loss": 0.8047,
+      "step": 26781
+    },
+    {
+      "epoch": 0.9739971633269084,
+      "grad_norm": 0.0,
+      "learning_rate": 3.544165276115452e-08,
+      "loss": 0.9243,
+      "step": 26782
+    },
+    {
+      "epoch": 0.9740335309306469,
+      "grad_norm": 0.0,
+      "learning_rate": 3.534264299469126e-08,
+      "loss": 0.8488,
+      "step": 26783
+    },
+    {
+      "epoch": 0.9740698985343855,
+      "grad_norm": 0.0,
+      "learning_rate": 3.524377147370972e-08,
+      "loss": 0.8707,
+      "step": 26784
+    },
+    {
+      "epoch": 0.9741062661381241,
+      "grad_norm": 0.0,
+      "learning_rate": 3.514503819958215e-08,
+      "loss": 0.9573,
+      "step": 26785
+    },
+    {
+      "epoch": 0.9741426337418627,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5046443173679666e-08,
+      "loss": 0.8958,
+      "step": 26786
+    },
+    {
+      "epoch": 0.9741790013456013,
+      "grad_norm": 0.0,
+      "learning_rate": 3.494798639736896e-08,
+      "loss": 0.9156,
+      "step": 26787
+    },
+    {
+      "epoch": 0.9742153689493399,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4849667872016714e-08,
+      "loss": 0.8829,
+      "step": 26788
+    },
+    {
+      "epoch": 0.9742517365530785,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4751487598987387e-08,
+      "loss": 0.7751,
+      "step": 26789
+    },
+    {
+      "epoch": 0.9742881041568171,
+      "grad_norm": 0.0,
+      "learning_rate": 3.465344557964101e-08,
+      "loss": 0.9382,
+      "step": 26790
+    },
+    {
+      "epoch": 0.9743244717605557,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4555541815339824e-08,
+      "loss": 0.7812,
+      "step": 26791
+    },
+    {
+      "epoch": 0.9743608393642943,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4457776307441624e-08,
+      "loss": 0.7887,
+      "step": 26792
+    },
+    {
+      "epoch": 0.9743972069680329,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4360149057302006e-08,
+      "loss": 0.8837,
+      "step": 26793
+    },
+    {
+      "epoch": 0.9744335745717715,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4262660066275436e-08,
+      "loss": 0.9733,
+      "step": 26794
+    },
+    {
+      "epoch": 0.9744699421755101,
+      "grad_norm": 0.0,
+      "learning_rate": 3.416530933571638e-08,
+      "loss": 0.8977,
+      "step": 26795
+    },
+    {
+      "epoch": 0.9745063097792487,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4068096866972654e-08,
+      "loss": 0.8119,
+      "step": 26796
+    },
+    {
+      "epoch": 0.9745426773829873,
+      "grad_norm": 0.0,
+      "learning_rate": 3.39710226613954e-08,
+      "loss": 0.7785,
+      "step": 26797
+    },
+    {
+      "epoch": 0.9745790449867259,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3874086720329104e-08,
+      "loss": 0.9524,
+      "step": 26798
+    },
+    {
+      "epoch": 0.9746154125904645,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3777289045120454e-08,
+      "loss": 0.9883,
+      "step": 26799
+    },
+    {
+      "epoch": 0.974651780194203,
+      "grad_norm": 0.0,
+      "learning_rate": 3.368062963711061e-08,
+      "loss": 1.1032,
+      "step": 26800
+    },
+    {
+      "epoch": 0.9746881477979416,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3584108497642934e-08,
+      "loss": 0.8313,
+      "step": 26801
+    },
+    {
+      "epoch": 0.9747245154016801,
+      "grad_norm": 0.0,
+      "learning_rate": 3.348772562805525e-08,
+      "loss": 0.9745,
+      "step": 26802
+    },
+    {
+      "epoch": 0.9747608830054187,
+      "grad_norm": 0.0,
+      "learning_rate": 3.339148102968426e-08,
+      "loss": 0.8723,
+      "step": 26803
+    },
+    {
+      "epoch": 0.9747972506091573,
+      "grad_norm": 0.0,
+      "learning_rate": 3.329537470386557e-08,
+      "loss": 0.8836,
+      "step": 26804
+    },
+    {
+      "epoch": 0.9748336182128959,
+      "grad_norm": 0.0,
+      "learning_rate": 3.319940665193366e-08,
+      "loss": 0.9579,
+      "step": 26805
+    },
+    {
+      "epoch": 0.9748699858166345,
+      "grad_norm": 0.0,
+      "learning_rate": 3.310357687521748e-08,
+      "loss": 0.9005,
+      "step": 26806
+    },
+    {
+      "epoch": 0.9749063534203731,
+      "grad_norm": 0.0,
+      "learning_rate": 3.300788537504929e-08,
+      "loss": 0.8988,
+      "step": 26807
+    },
+    {
+      "epoch": 0.9749427210241117,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2912332152754686e-08,
+      "loss": 0.8613,
+      "step": 26808
+    },
+    {
+      "epoch": 0.9749790886278503,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2816917209660406e-08,
+      "loss": 0.7912,
+      "step": 26809
+    },
+    {
+      "epoch": 0.9750154562315889,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2721640547088705e-08,
+      "loss": 0.8322,
+      "step": 26810
+    },
+    {
+      "epoch": 0.9750518238353275,
+      "grad_norm": 0.0,
+      "learning_rate": 3.262650216636409e-08,
+      "loss": 0.9062,
+      "step": 26811
+    },
+    {
+      "epoch": 0.9750881914390661,
+      "grad_norm": 0.0,
+      "learning_rate": 3.25315020688044e-08,
+      "loss": 0.9967,
+      "step": 26812
+    },
+    {
+      "epoch": 0.9751245590428047,
+      "grad_norm": 0.0,
+      "learning_rate": 3.243664025572746e-08,
+      "loss": 0.8056,
+      "step": 26813
+    },
+    {
+      "epoch": 0.9751609266465433,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2341916728449995e-08,
+      "loss": 0.8707,
+      "step": 26814
+    },
+    {
+      "epoch": 0.9751972942502819,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2247331488286516e-08,
+      "loss": 0.8311,
+      "step": 26815
+    },
+    {
+      "epoch": 0.9752336618540204,
+      "grad_norm": 0.0,
+      "learning_rate": 3.21528845365493e-08,
+      "loss": 0.9429,
+      "step": 26816
+    },
+    {
+      "epoch": 0.975270029457759,
+      "grad_norm": 0.0,
+      "learning_rate": 3.205857587454841e-08,
+      "loss": 0.9314,
+      "step": 26817
+    },
+    {
+      "epoch": 0.9753063970614976,
+      "grad_norm": 0.0,
+      "learning_rate": 3.19644055035917e-08,
+      "loss": 0.9049,
+      "step": 26818
+    },
+    {
+      "epoch": 0.9753427646652362,
+      "grad_norm": 0.0,
+      "learning_rate": 3.187037342498589e-08,
+      "loss": 0.8503,
+      "step": 26819
+    },
+    {
+      "epoch": 0.9753791322689748,
+      "grad_norm": 0.0,
+      "learning_rate": 3.177647964003661e-08,
+      "loss": 0.9154,
+      "step": 26820
+    },
+    {
+      "epoch": 0.9754154998727134,
+      "grad_norm": 0.0,
+      "learning_rate": 3.168272415004503e-08,
+      "loss": 0.9383,
+      "step": 26821
+    },
+    {
+      "epoch": 0.975451867476452,
+      "grad_norm": 0.0,
+      "learning_rate": 3.158910695631234e-08,
+      "loss": 0.9005,
+      "step": 26822
+    },
+    {
+      "epoch": 0.9754882350801906,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1495628060136395e-08,
+      "loss": 0.8905,
+      "step": 26823
+    },
+    {
+      "epoch": 0.9755246026839292,
+      "grad_norm": 0.0,
+      "learning_rate": 3.140228746281726e-08,
+      "loss": 0.9244,
+      "step": 26824
+    },
+    {
+      "epoch": 0.9755609702876678,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1309085165646124e-08,
+      "loss": 0.955,
+      "step": 26825
+    },
+    {
+      "epoch": 0.9755973378914063,
+      "grad_norm": 0.0,
+      "learning_rate": 3.121602116991862e-08,
+      "loss": 0.8845,
+      "step": 26826
+    },
+    {
+      "epoch": 0.9756337054951449,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1123095476924827e-08,
+      "loss": 0.8777,
+      "step": 26827
+    },
+    {
+      "epoch": 0.9756700730988835,
+      "grad_norm": 0.0,
+      "learning_rate": 3.103030808795371e-08,
+      "loss": 0.8686,
+      "step": 26828
+    },
+    {
+      "epoch": 0.9757064407026221,
+      "grad_norm": 0.0,
+      "learning_rate": 3.093765900429313e-08,
+      "loss": 0.7592,
+      "step": 26829
+    },
+    {
+      "epoch": 0.9757428083063607,
+      "grad_norm": 0.0,
+      "learning_rate": 3.084514822722873e-08,
+      "loss": 0.9752,
+      "step": 26830
+    },
+    {
+      "epoch": 0.9757791759100993,
+      "grad_norm": 0.0,
+      "learning_rate": 3.075277575804392e-08,
+      "loss": 0.7775,
+      "step": 26831
+    },
+    {
+      "epoch": 0.9758155435138379,
+      "grad_norm": 0.0,
+      "learning_rate": 3.066054159801879e-08,
+      "loss": 1.0219,
+      "step": 26832
+    },
+    {
+      "epoch": 0.9758519111175764,
+      "grad_norm": 0.0,
+      "learning_rate": 3.056844574843454e-08,
+      "loss": 0.8751,
+      "step": 26833
+    },
+    {
+      "epoch": 0.975888278721315,
+      "grad_norm": 0.0,
+      "learning_rate": 3.047648821056792e-08,
+      "loss": 1.1096,
+      "step": 26834
+    },
+    {
+      "epoch": 0.9759246463250536,
+      "grad_norm": 0.0,
+      "learning_rate": 3.038466898569459e-08,
+      "loss": 0.9293,
+      "step": 26835
+    },
+    {
+      "epoch": 0.9759610139287922,
+      "grad_norm": 0.0,
+      "learning_rate": 3.029298807509018e-08,
+      "loss": 0.9001,
+      "step": 26836
+    },
+    {
+      "epoch": 0.9759973815325308,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0201445480024794e-08,
+      "loss": 0.9349,
+      "step": 26837
+    },
+    {
+      "epoch": 0.9760337491362694,
+      "grad_norm": 0.0,
+      "learning_rate": 3.011004120176853e-08,
+      "loss": 0.8869,
+      "step": 26838
+    },
+    {
+      "epoch": 0.976070116740008,
+      "grad_norm": 0.0,
+      "learning_rate": 3.001877524159147e-08,
+      "loss": 0.9866,
+      "step": 26839
+    },
+    {
+      "epoch": 0.9761064843437466,
+      "grad_norm": 0.0,
+      "learning_rate": 2.992764760075706e-08,
+      "loss": 0.8899,
+      "step": 26840
+    },
+    {
+      "epoch": 0.9761428519474852,
+      "grad_norm": 0.0,
+      "learning_rate": 2.983665828053206e-08,
+      "loss": 0.9986,
+      "step": 26841
+    },
+    {
+      "epoch": 0.9761792195512238,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9745807282176574e-08,
+      "loss": 0.9804,
+      "step": 26842
+    },
+    {
+      "epoch": 0.9762155871549624,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9655094606951818e-08,
+      "loss": 0.9957,
+      "step": 26843
+    },
+    {
+      "epoch": 0.976251954758701,
+      "grad_norm": 0.0,
+      "learning_rate": 2.956452025611789e-08,
+      "loss": 0.9113,
+      "step": 26844
+    },
+    {
+      "epoch": 0.9762883223624396,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9474084230928234e-08,
+      "loss": 0.791,
+      "step": 26845
+    },
+    {
+      "epoch": 0.9763246899661782,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9383786532640736e-08,
+      "loss": 0.8943,
+      "step": 26846
+    },
+    {
+      "epoch": 0.9763610575699168,
+      "grad_norm": 0.0,
+      "learning_rate": 2.929362716250661e-08,
+      "loss": 1.0231,
+      "step": 26847
+    },
+    {
+      "epoch": 0.9763974251736554,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9203606121777085e-08,
+      "loss": 0.8572,
+      "step": 26848
+    },
+    {
+      "epoch": 0.976433792777394,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9113723411700046e-08,
+      "loss": 0.8712,
+      "step": 26849
+    },
+    {
+      "epoch": 0.9764701603811324,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9023979033523386e-08,
+      "loss": 0.8981,
+      "step": 26850
+    },
+    {
+      "epoch": 0.976506527984871,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8934372988491667e-08,
+      "loss": 0.8271,
+      "step": 26851
+    },
+    {
+      "epoch": 0.9765428955886096,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8844905277849446e-08,
+      "loss": 0.9037,
+      "step": 26852
+    },
+    {
+      "epoch": 0.9765792631923482,
+      "grad_norm": 0.0,
+      "learning_rate": 2.875557590283684e-08,
+      "loss": 0.9643,
+      "step": 26853
+    },
+    {
+      "epoch": 0.9766156307960868,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8666384864692863e-08,
+      "loss": 0.9203,
+      "step": 26854
+    },
+    {
+      "epoch": 0.9766519983998254,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8577332164654303e-08,
+      "loss": 0.8565,
+      "step": 26855
+    },
+    {
+      "epoch": 0.976688366003564,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8488417803959056e-08,
+      "loss": 0.9111,
+      "step": 26856
+    },
+    {
+      "epoch": 0.9767247336073026,
+      "grad_norm": 0.0,
+      "learning_rate": 2.839964178383947e-08,
+      "loss": 0.918,
+      "step": 26857
+    },
+    {
+      "epoch": 0.9767611012110412,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8311004105525673e-08,
+      "loss": 0.8418,
+      "step": 26858
+    },
+    {
+      "epoch": 0.9767974688147798,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8222504770250014e-08,
+      "loss": 0.8573,
+      "step": 26859
+    },
+    {
+      "epoch": 0.9768338364185184,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8134143779238175e-08,
+      "loss": 0.8623,
+      "step": 26860
+    },
+    {
+      "epoch": 0.976870204022257,
+      "grad_norm": 0.0,
+      "learning_rate": 2.804592113371696e-08,
+      "loss": 0.7968,
+      "step": 26861
+    },
+    {
+      "epoch": 0.9769065716259956,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7957836834910935e-08,
+      "loss": 0.8598,
+      "step": 26862
+    },
+    {
+      "epoch": 0.9769429392297342,
+      "grad_norm": 0.0,
+      "learning_rate": 2.786989088404135e-08,
+      "loss": 0.9542,
+      "step": 26863
+    },
+    {
+      "epoch": 0.9769793068334728,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7782083282327232e-08,
+      "loss": 0.8324,
+      "step": 26864
+    },
+    {
+      "epoch": 0.9770156744372114,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7694414030989825e-08,
+      "loss": 0.8736,
+      "step": 26865
+    },
+    {
+      "epoch": 0.97705204204095,
+      "grad_norm": 0.0,
+      "learning_rate": 2.76068831312426e-08,
+      "loss": 0.9871,
+      "step": 26866
+    },
+    {
+      "epoch": 0.9770884096446885,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7519490584300146e-08,
+      "loss": 0.8563,
+      "step": 26867
+    },
+    {
+      "epoch": 0.9771247772484271,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7432236391375934e-08,
+      "loss": 0.8658,
+      "step": 26868
+    },
+    {
+      "epoch": 0.9771611448521657,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7345120553681214e-08,
+      "loss": 0.9224,
+      "step": 26869
+    },
+    {
+      "epoch": 0.9771975124559042,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7258143072422803e-08,
+      "loss": 0.9383,
+      "step": 26870
+    },
+    {
+      "epoch": 0.9772338800596428,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7171303948808624e-08,
+      "loss": 0.9521,
+      "step": 26871
+    },
+    {
+      "epoch": 0.9772702476633814,
+      "grad_norm": 0.0,
+      "learning_rate": 2.708460318404327e-08,
+      "loss": 0.9075,
+      "step": 26872
+    },
+    {
+      "epoch": 0.97730661526712,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6998040779329105e-08,
+      "loss": 0.8726,
+      "step": 26873
+    },
+    {
+      "epoch": 0.9773429828708586,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6911616735866285e-08,
+      "loss": 0.8697,
+      "step": 26874
+    },
+    {
+      "epoch": 0.9773793504745972,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6825331054856074e-08,
+      "loss": 0.9407,
+      "step": 26875
+    },
+    {
+      "epoch": 0.9774157180783358,
+      "grad_norm": 0.0,
+      "learning_rate": 2.673918373749307e-08,
+      "loss": 0.9169,
+      "step": 26876
+    },
+    {
+      "epoch": 0.9774520856820744,
+      "grad_norm": 0.0,
+      "learning_rate": 2.66531747849752e-08,
+      "loss": 0.9545,
+      "step": 26877
+    },
+    {
+      "epoch": 0.977488453285813,
+      "grad_norm": 0.0,
+      "learning_rate": 2.656730419849263e-08,
+      "loss": 0.9393,
+      "step": 26878
+    },
+    {
+      "epoch": 0.9775248208895516,
+      "grad_norm": 0.0,
+      "learning_rate": 2.648157197923995e-08,
+      "loss": 0.8839,
+      "step": 26879
+    },
+    {
+      "epoch": 0.9775611884932902,
+      "grad_norm": 0.0,
+      "learning_rate": 2.639597812840289e-08,
+      "loss": 0.8115,
+      "step": 26880
+    },
+    {
+      "epoch": 0.9775975560970288,
+      "grad_norm": 0.0,
+      "learning_rate": 2.631052264717271e-08,
+      "loss": 0.9077,
+      "step": 26881
+    },
+    {
+      "epoch": 0.9776339237007674,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6225205536732913e-08,
+      "loss": 0.988,
+      "step": 26882
+    },
+    {
+      "epoch": 0.977670291304506,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6140026798266994e-08,
+      "loss": 0.8664,
+      "step": 26883
+    },
+    {
+      "epoch": 0.9777066589082445,
+      "grad_norm": 0.0,
+      "learning_rate": 2.605498643295734e-08,
+      "loss": 0.9288,
+      "step": 26884
+    },
+    {
+      "epoch": 0.9777430265119831,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5970084441983013e-08,
+      "loss": 0.8003,
+      "step": 26885
+    },
+    {
+      "epoch": 0.9777793941157217,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5885320826523063e-08,
+      "loss": 1.0103,
+      "step": 26886
+    },
+    {
+      "epoch": 0.9778157617194603,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5800695587753223e-08,
+      "loss": 0.922,
+      "step": 26887
+    },
+    {
+      "epoch": 0.9778521293231989,
+      "grad_norm": 0.0,
+      "learning_rate": 2.571620872684588e-08,
+      "loss": 0.8987,
+      "step": 26888
+    },
+    {
+      "epoch": 0.9778884969269375,
+      "grad_norm": 0.0,
+      "learning_rate": 2.563186024497566e-08,
+      "loss": 0.9612,
+      "step": 26889
+    },
+    {
+      "epoch": 0.9779248645306761,
+      "grad_norm": 0.0,
+      "learning_rate": 2.554765014331162e-08,
+      "loss": 0.9022,
+      "step": 26890
+    },
+    {
+      "epoch": 0.9779612321344147,
+      "grad_norm": 0.0,
+      "learning_rate": 2.546357842302061e-08,
+      "loss": 0.8546,
+      "step": 26891
+    },
+    {
+      "epoch": 0.9779975997381533,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5379645085272798e-08,
+      "loss": 0.9062,
+      "step": 26892
+    },
+    {
+      "epoch": 0.9780339673418919,
+      "grad_norm": 0.0,
+      "learning_rate": 2.529585013122837e-08,
+      "loss": 0.8657,
+      "step": 26893
+    },
+    {
+      "epoch": 0.9780703349456304,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5212193562051958e-08,
+      "loss": 0.7489,
+      "step": 26894
+    },
+    {
+      "epoch": 0.978106702549369,
+      "grad_norm": 0.0,
+      "learning_rate": 2.512867537890484e-08,
+      "loss": 0.7971,
+      "step": 26895
+    },
+    {
+      "epoch": 0.9781430701531076,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5045295582944993e-08,
+      "loss": 0.9939,
+      "step": 26896
+    },
+    {
+      "epoch": 0.9781794377568462,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4962054175328153e-08,
+      "loss": 0.8784,
+      "step": 26897
+    },
+    {
+      "epoch": 0.9782158053605848,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4878951157210064e-08,
+      "loss": 0.9576,
+      "step": 26898
+    },
+    {
+      "epoch": 0.9782521729643234,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4795986529744243e-08,
+      "loss": 0.9263,
+      "step": 26899
+    },
+    {
+      "epoch": 0.9782885405680619,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4713160294080885e-08,
+      "loss": 0.9318,
+      "step": 26900
+    },
+    {
+      "epoch": 0.9783249081718005,
+      "grad_norm": 0.0,
+      "learning_rate": 2.463047245136907e-08,
+      "loss": 0.9155,
+      "step": 26901
+    },
+    {
+      "epoch": 0.9783612757755391,
+      "grad_norm": 0.0,
+      "learning_rate": 2.454792300275677e-08,
+      "loss": 0.7709,
+      "step": 26902
+    },
+    {
+      "epoch": 0.9783976433792777,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4465511949388622e-08,
+      "loss": 0.8611,
+      "step": 26903
+    },
+    {
+      "epoch": 0.9784340109830163,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4383239292408155e-08,
+      "loss": 0.9059,
+      "step": 26904
+    },
+    {
+      "epoch": 0.9784703785867549,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4301105032956683e-08,
+      "loss": 0.8516,
+      "step": 26905
+    },
+    {
+      "epoch": 0.9785067461904935,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4219109172173295e-08,
+      "loss": 0.7311,
+      "step": 26906
+    },
+    {
+      "epoch": 0.9785431137942321,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4137251711195965e-08,
+      "loss": 0.8072,
+      "step": 26907
+    },
+    {
+      "epoch": 0.9785794813979707,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4055532651160453e-08,
+      "loss": 0.9882,
+      "step": 26908
+    },
+    {
+      "epoch": 0.9786158490017093,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3973951993201404e-08,
+      "loss": 0.9882,
+      "step": 26909
+    },
+    {
+      "epoch": 0.9786522166054479,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3892509738447922e-08,
+      "loss": 0.8655,
+      "step": 26910
+    },
+    {
+      "epoch": 0.9786885842091865,
+      "grad_norm": 0.0,
+      "learning_rate": 2.381120588803354e-08,
+      "loss": 0.9765,
+      "step": 26911
+    },
+    {
+      "epoch": 0.9787249518129251,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3730040443082912e-08,
+      "loss": 0.9038,
+      "step": 26912
+    },
+    {
+      "epoch": 0.9787613194166637,
+      "grad_norm": 0.0,
+      "learning_rate": 2.364901340472403e-08,
+      "loss": 0.9321,
+      "step": 26913
+    },
+    {
+      "epoch": 0.9787976870204023,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3568124774080438e-08,
+      "loss": 1.0263,
+      "step": 26914
+    },
+    {
+      "epoch": 0.9788340546241409,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3487374552274568e-08,
+      "loss": 0.9746,
+      "step": 26915
+    },
+    {
+      "epoch": 0.9788704222278795,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3406762740425525e-08,
+      "loss": 0.8998,
+      "step": 26916
+    },
+    {
+      "epoch": 0.9789067898316179,
+      "grad_norm": 0.0,
+      "learning_rate": 2.332628933965353e-08,
+      "loss": 0.9501,
+      "step": 26917
+    },
+    {
+      "epoch": 0.9789431574353565,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3245954351074352e-08,
+      "loss": 0.8694,
+      "step": 26918
+    },
+    {
+      "epoch": 0.9789795250390951,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3165757775802654e-08,
+      "loss": 0.9725,
+      "step": 26919
+    },
+    {
+      "epoch": 0.9790158926428337,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3085699614949773e-08,
+      "loss": 0.9481,
+      "step": 26920
+    },
+    {
+      "epoch": 0.9790522602465723,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3005779869628153e-08,
+      "loss": 0.8495,
+      "step": 26921
+    },
+    {
+      "epoch": 0.9790886278503109,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2925998540945792e-08,
+      "loss": 0.8963,
+      "step": 26922
+    },
+    {
+      "epoch": 0.9791249954540495,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2846355630009586e-08,
+      "loss": 0.9501,
+      "step": 26923
+    },
+    {
+      "epoch": 0.9791613630577881,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2766851137924208e-08,
+      "loss": 0.8852,
+      "step": 26924
+    },
+    {
+      "epoch": 0.9791977306615267,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2687485065793214e-08,
+      "loss": 0.9543,
+      "step": 26925
+    },
+    {
+      "epoch": 0.9792340982652653,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2608257414717948e-08,
+      "loss": 0.8582,
+      "step": 26926
+    },
+    {
+      "epoch": 0.9792704658690039,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2529168185796426e-08,
+      "loss": 0.8691,
+      "step": 26927
+    },
+    {
+      "epoch": 0.9793068334727425,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2450217380126647e-08,
+      "loss": 0.8164,
+      "step": 26928
+    },
+    {
+      "epoch": 0.9793432010764811,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2371404998803304e-08,
+      "loss": 0.9159,
+      "step": 26929
+    },
+    {
+      "epoch": 0.9793795686802197,
+      "grad_norm": 0.0,
+      "learning_rate": 2.229273104292107e-08,
+      "loss": 0.8406,
+      "step": 26930
+    },
+    {
+      "epoch": 0.9794159362839583,
+      "grad_norm": 0.0,
+      "learning_rate": 2.221419551356907e-08,
+      "loss": 0.8731,
+      "step": 26931
+    },
+    {
+      "epoch": 0.9794523038876969,
+      "grad_norm": 0.0,
+      "learning_rate": 2.213579841183977e-08,
+      "loss": 0.9687,
+      "step": 26932
+    },
+    {
+      "epoch": 0.9794886714914355,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2057539738820078e-08,
+      "loss": 0.9025,
+      "step": 26933
+    },
+    {
+      "epoch": 0.979525039095174,
+      "grad_norm": 0.0,
+      "learning_rate": 2.197941949559468e-08,
+      "loss": 0.8985,
+      "step": 26934
+    },
+    {
+      "epoch": 0.9795614066989126,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1901437683247152e-08,
+      "loss": 1.0074,
+      "step": 26935
+    },
+    {
+      "epoch": 0.9795977743026512,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1823594302861074e-08,
+      "loss": 0.9828,
+      "step": 26936
+    },
+    {
+      "epoch": 0.9796341419063898,
+      "grad_norm": 0.0,
+      "learning_rate": 2.174588935551447e-08,
+      "loss": 0.8814,
+      "step": 26937
+    },
+    {
+      "epoch": 0.9796705095101284,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1668322842287593e-08,
+      "loss": 0.9085,
+      "step": 26938
+    },
+    {
+      "epoch": 0.979706877113867,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1590894764254023e-08,
+      "loss": 0.8598,
+      "step": 26939
+    },
+    {
+      "epoch": 0.9797432447176055,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1513605122490678e-08,
+      "loss": 0.9154,
+      "step": 26940
+    },
+    {
+      "epoch": 0.9797796123213441,
+      "grad_norm": 0.0,
+      "learning_rate": 2.14364539180667e-08,
+      "loss": 0.8309,
+      "step": 26941
+    },
+    {
+      "epoch": 0.9798159799250827,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1359441152055683e-08,
+      "loss": 0.9936,
+      "step": 26942
+    },
+    {
+      "epoch": 0.9798523475288213,
+      "grad_norm": 0.0,
+      "learning_rate": 2.128256682552454e-08,
+      "loss": 0.8766,
+      "step": 26943
+    },
+    {
+      "epoch": 0.9798887151325599,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1205830939539096e-08,
+      "loss": 0.9306,
+      "step": 26944
+    },
+    {
+      "epoch": 0.9799250827362985,
+      "grad_norm": 0.0,
+      "learning_rate": 2.112923349516516e-08,
+      "loss": 0.851,
+      "step": 26945
+    },
+    {
+      "epoch": 0.9799614503400371,
+      "grad_norm": 0.0,
+      "learning_rate": 2.105277449346521e-08,
+      "loss": 0.9019,
+      "step": 26946
+    },
+    {
+      "epoch": 0.9799978179437757,
+      "grad_norm": 0.0,
+      "learning_rate": 2.097645393549841e-08,
+      "loss": 0.8934,
+      "step": 26947
+    },
+    {
+      "epoch": 0.9800341855475143,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0900271822326123e-08,
+      "loss": 0.9629,
+      "step": 26948
+    },
+    {
+      "epoch": 0.9800705531512529,
+      "grad_norm": 0.0,
+      "learning_rate": 2.082422815500418e-08,
+      "loss": 0.918,
+      "step": 26949
+    },
+    {
+      "epoch": 0.9801069207549915,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0748322934586172e-08,
+      "loss": 0.9055,
+      "step": 26950
+    },
+    {
+      "epoch": 0.98014328835873,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0672556162127933e-08,
+      "loss": 0.936,
+      "step": 26951
+    },
+    {
+      "epoch": 0.9801796559624686,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0596927838678614e-08,
+      "loss": 0.9925,
+      "step": 26952
+    },
+    {
+      "epoch": 0.9802160235662072,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0521437965288493e-08,
+      "loss": 0.9715,
+      "step": 26953
+    },
+    {
+      "epoch": 0.9802523911699458,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0446086543003396e-08,
+      "loss": 0.8932,
+      "step": 26954
+    },
+    {
+      "epoch": 0.9802887587736844,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0370873572870263e-08,
+      "loss": 0.9237,
+      "step": 26955
+    },
+    {
+      "epoch": 0.980325126377423,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0295799055931598e-08,
+      "loss": 0.8654,
+      "step": 26956
+    },
+    {
+      "epoch": 0.9803614939811616,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0220862993229896e-08,
+      "loss": 0.8244,
+      "step": 26957
+    },
+    {
+      "epoch": 0.9803978615849002,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0146065385804325e-08,
+      "loss": 0.9685,
+      "step": 26958
+    },
+    {
+      "epoch": 0.9804342291886388,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0071406234691838e-08,
+      "loss": 0.8474,
+      "step": 26959
+    },
+    {
+      "epoch": 0.9804705967923774,
+      "grad_norm": 0.0,
+      "learning_rate": 1.999688554092938e-08,
+      "loss": 0.8773,
+      "step": 26960
+    },
+    {
+      "epoch": 0.980506964396116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9922503305550565e-08,
+      "loss": 0.9362,
+      "step": 26961
+    },
+    {
+      "epoch": 0.9805433319998546,
+      "grad_norm": 0.0,
+      "learning_rate": 1.98482595295868e-08,
+      "loss": 0.9733,
+      "step": 26962
+    },
+    {
+      "epoch": 0.9805796996035931,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9774154214068365e-08,
+      "loss": 0.8661,
+      "step": 26963
+    },
+    {
+      "epoch": 0.9806160672073317,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9700187360024437e-08,
+      "loss": 0.7807,
+      "step": 26964
+    },
+    {
+      "epoch": 0.9806524348110703,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9626358968479752e-08,
+      "loss": 0.9907,
+      "step": 26965
+    },
+    {
+      "epoch": 0.9806888024148089,
+      "grad_norm": 0.0,
+      "learning_rate": 1.955266904045794e-08,
+      "loss": 0.9637,
+      "step": 26966
+    },
+    {
+      "epoch": 0.9807251700185475,
+      "grad_norm": 0.0,
+      "learning_rate": 1.947911757698373e-08,
+      "loss": 0.8211,
+      "step": 26967
+    },
+    {
+      "epoch": 0.980761537622286,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9405704579075203e-08,
+      "loss": 0.9153,
+      "step": 26968
+    },
+    {
+      "epoch": 0.9807979052260246,
+      "grad_norm": 0.0,
+      "learning_rate": 1.933243004775265e-08,
+      "loss": 0.9415,
+      "step": 26969
+    },
+    {
+      "epoch": 0.9808342728297632,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9259293984031925e-08,
+      "loss": 0.851,
+      "step": 26970
+    },
+    {
+      "epoch": 0.9808706404335018,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9186296388927773e-08,
+      "loss": 0.77,
+      "step": 26971
+    },
+    {
+      "epoch": 0.9809070080372404,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9113437263452718e-08,
+      "loss": 0.913,
+      "step": 26972
+    },
+    {
+      "epoch": 0.980943375640979,
+      "grad_norm": 0.0,
+      "learning_rate": 1.904071660861706e-08,
+      "loss": 0.8516,
+      "step": 26973
+    },
+    {
+      "epoch": 0.9809797432447176,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8968134425431107e-08,
+      "loss": 0.9361,
+      "step": 26974
+    },
+    {
+      "epoch": 0.9810161108484562,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8895690714900717e-08,
+      "loss": 0.8682,
+      "step": 26975
+    },
+    {
+      "epoch": 0.9810524784521948,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8823385478030643e-08,
+      "loss": 0.8041,
+      "step": 26976
+    },
+    {
+      "epoch": 0.9810888460559334,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8751218715825635e-08,
+      "loss": 0.9208,
+      "step": 26977
+    },
+    {
+      "epoch": 0.981125213659672,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8679190429286008e-08,
+      "loss": 0.9182,
+      "step": 26978
+    },
+    {
+      "epoch": 0.9811615812634106,
+      "grad_norm": 0.0,
+      "learning_rate": 1.860730061940985e-08,
+      "loss": 0.8532,
+      "step": 26979
+    },
+    {
+      "epoch": 0.9811979488671492,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8535549287197472e-08,
+      "loss": 0.9891,
+      "step": 26980
+    },
+    {
+      "epoch": 0.9812343164708878,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8463936433641418e-08,
+      "loss": 0.8851,
+      "step": 26981
+    },
+    {
+      "epoch": 0.9812706840746264,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8392462059736438e-08,
+      "loss": 0.897,
+      "step": 26982
+    },
+    {
+      "epoch": 0.981307051678365,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8321126166472857e-08,
+      "loss": 0.7961,
+      "step": 26983
+    },
+    {
+      "epoch": 0.9813434192821036,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8249928754843216e-08,
+      "loss": 0.885,
+      "step": 26984
+    },
+    {
+      "epoch": 0.981379786885842,
+      "grad_norm": 0.0,
+      "learning_rate": 1.817886982583228e-08,
+      "loss": 0.9338,
+      "step": 26985
+    },
+    {
+      "epoch": 0.9814161544895806,
+      "grad_norm": 0.0,
+      "learning_rate": 1.810794938042815e-08,
+      "loss": 0.9688,
+      "step": 26986
+    },
+    {
+      "epoch": 0.9814525220933192,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8037167419613367e-08,
+      "loss": 1.0084,
+      "step": 26987
+    },
+    {
+      "epoch": 0.9814888896970578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7966523944370484e-08,
+      "loss": 0.9066,
+      "step": 26988
+    },
+    {
+      "epoch": 0.9815252573007964,
+      "grad_norm": 0.0,
+      "learning_rate": 1.789601895567872e-08,
+      "loss": 0.8429,
+      "step": 26989
+    },
+    {
+      "epoch": 0.981561624904535,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7825652454517282e-08,
+      "loss": 0.968,
+      "step": 26990
+    },
+    {
+      "epoch": 0.9815979925082736,
+      "grad_norm": 0.0,
+      "learning_rate": 1.775542444186318e-08,
+      "loss": 0.9087,
+      "step": 26991
+    },
+    {
+      "epoch": 0.9816343601120122,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7685334918688957e-08,
+      "loss": 0.9937,
+      "step": 26992
+    },
+    {
+      "epoch": 0.9816707277157508,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7615383885967173e-08,
+      "loss": 1.0176,
+      "step": 26993
+    },
+    {
+      "epoch": 0.9817070953194894,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7545571344669276e-08,
+      "loss": 0.8809,
+      "step": 26994
+    },
+    {
+      "epoch": 0.981743462923228,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7475897295762267e-08,
+      "loss": 1.0231,
+      "step": 26995
+    },
+    {
+      "epoch": 0.9817798305269666,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7406361740215372e-08,
+      "loss": 0.9218,
+      "step": 26996
+    },
+    {
+      "epoch": 0.9818161981307052,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7336964678991153e-08,
+      "loss": 0.9872,
+      "step": 26997
+    },
+    {
+      "epoch": 0.9818525657344438,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7267706113052172e-08,
+      "loss": 0.875,
+      "step": 26998
+    },
+    {
+      "epoch": 0.9818889333381824,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7198586043360997e-08,
+      "loss": 0.9581,
+      "step": 26999
+    },
+    {
+      "epoch": 0.981925300941921,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7129604470875748e-08,
+      "loss": 0.8683,
+      "step": 27000
+    },
+    {
+      "epoch": 0.9819616685456595,
+      "grad_norm": 0.0,
+      "learning_rate": 1.706076139655344e-08,
+      "loss": 0.9849,
+      "step": 27001
+    },
+    {
+      "epoch": 0.9819980361493981,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6992056821348856e-08,
+      "loss": 0.955,
+      "step": 27002
+    },
+    {
+      "epoch": 0.9820344037531367,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6923490746214576e-08,
+      "loss": 0.9011,
+      "step": 27003
+    },
+    {
+      "epoch": 0.9820707713568753,
+      "grad_norm": 0.0,
+      "learning_rate": 1.685506317210428e-08,
+      "loss": 0.8179,
+      "step": 27004
+    },
+    {
+      "epoch": 0.9821071389606139,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6786774099964986e-08,
+      "loss": 0.9534,
+      "step": 27005
+    },
+    {
+      "epoch": 0.9821435065643525,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6718623530744827e-08,
+      "loss": 0.9377,
+      "step": 27006
+    },
+    {
+      "epoch": 0.982179874168091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.665061146538971e-08,
+      "loss": 0.9657,
+      "step": 27007
+    },
+    {
+      "epoch": 0.9822162417718296,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6582737904842217e-08,
+      "loss": 0.9554,
+      "step": 27008
+    },
+    {
+      "epoch": 0.9822526093755682,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6515002850044927e-08,
+      "loss": 1.0155,
+      "step": 27009
+    },
+    {
+      "epoch": 0.9822889769793068,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6447406301938197e-08,
+      "loss": 0.9954,
+      "step": 27010
+    },
+    {
+      "epoch": 0.9823253445830454,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6379948261457946e-08,
+      "loss": 0.8839,
+      "step": 27011
+    },
+    {
+      "epoch": 0.982361712186784,
+      "grad_norm": 0.0,
+      "learning_rate": 1.63126287295412e-08,
+      "loss": 0.8854,
+      "step": 27012
+    },
+    {
+      "epoch": 0.9823980797905226,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6245447707121664e-08,
+      "loss": 0.9004,
+      "step": 27013
+    },
+    {
+      "epoch": 0.9824344473942612,
+      "grad_norm": 0.0,
+      "learning_rate": 1.617840519513081e-08,
+      "loss": 0.9109,
+      "step": 27014
+    },
+    {
+      "epoch": 0.9824708149979998,
+      "grad_norm": 0.0,
+      "learning_rate": 1.611150119450011e-08,
+      "loss": 0.8924,
+      "step": 27015
+    },
+    {
+      "epoch": 0.9825071826017384,
+      "grad_norm": 0.0,
+      "learning_rate": 1.604473570615772e-08,
+      "loss": 0.8437,
+      "step": 27016
+    },
+    {
+      "epoch": 0.982543550205477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5978108731028454e-08,
+      "loss": 0.9267,
+      "step": 27017
+    },
+    {
+      "epoch": 0.9825799178092155,
+      "grad_norm": 0.0,
+      "learning_rate": 1.591162027003712e-08,
+      "loss": 0.8919,
+      "step": 27018
+    },
+    {
+      "epoch": 0.9826162854129541,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5845270324106322e-08,
+      "loss": 0.9224,
+      "step": 27019
+    },
+    {
+      "epoch": 0.9826526530166927,
+      "grad_norm": 0.0,
+      "learning_rate": 1.577905889415754e-08,
+      "loss": 0.8403,
+      "step": 27020
+    },
+    {
+      "epoch": 0.9826890206204313,
+      "grad_norm": 0.0,
+      "learning_rate": 1.571298598110893e-08,
+      "loss": 0.9834,
+      "step": 27021
+    },
+    {
+      "epoch": 0.9827253882241699,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5647051585875316e-08,
+      "loss": 0.8433,
+      "step": 27022
+    },
+    {
+      "epoch": 0.9827617558279085,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5581255709374854e-08,
+      "loss": 0.7819,
+      "step": 27023
+    },
+    {
+      "epoch": 0.9827981234316471,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5515598352516814e-08,
+      "loss": 0.8157,
+      "step": 27024
+    },
+    {
+      "epoch": 0.9828344910353857,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5450079516213802e-08,
+      "loss": 0.9487,
+      "step": 27025
+    },
+    {
+      "epoch": 0.9828708586391243,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5384699201375087e-08,
+      "loss": 0.9563,
+      "step": 27026
+    },
+    {
+      "epoch": 0.9829072262428629,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5319457408907723e-08,
+      "loss": 0.8995,
+      "step": 27027
+    },
+    {
+      "epoch": 0.9829435938466015,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5254354139716542e-08,
+      "loss": 0.8541,
+      "step": 27028
+    },
+    {
+      "epoch": 0.9829799614503401,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5189389394704158e-08,
+      "loss": 0.9469,
+      "step": 27029
+    },
+    {
+      "epoch": 0.9830163290540787,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5124563174772068e-08,
+      "loss": 0.9697,
+      "step": 27030
+    },
+    {
+      "epoch": 0.9830526966578172,
+      "grad_norm": 0.0,
+      "learning_rate": 1.505987548082066e-08,
+      "loss": 0.9023,
+      "step": 27031
+    },
+    {
+      "epoch": 0.9830890642615558,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4995326313747007e-08,
+      "loss": 0.8439,
+      "step": 27032
+    },
+    {
+      "epoch": 0.9831254318652944,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4930915674445934e-08,
+      "loss": 0.922,
+      "step": 27033
+    },
+    {
+      "epoch": 0.983161799469033,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4866643563811178e-08,
+      "loss": 0.9121,
+      "step": 27034
+    },
+    {
+      "epoch": 0.9831981670727715,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4802509982734247e-08,
+      "loss": 0.9125,
+      "step": 27035
+    },
+    {
+      "epoch": 0.9832345346765101,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4738514932106651e-08,
+      "loss": 0.9033,
+      "step": 27036
+    },
+    {
+      "epoch": 0.9832709022802487,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4674658412814347e-08,
+      "loss": 0.8122,
+      "step": 27037
+    },
+    {
+      "epoch": 0.9833072698839873,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4610940425743292e-08,
+      "loss": 0.7335,
+      "step": 27038
+    },
+    {
+      "epoch": 0.9833436374877259,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4547360971778335e-08,
+      "loss": 0.753,
+      "step": 27039
+    },
+    {
+      "epoch": 0.9833800050914645,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4483920051800993e-08,
+      "loss": 0.8776,
+      "step": 27040
+    },
+    {
+      "epoch": 0.9834163726952031,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4420617666692783e-08,
+      "loss": 1.0056,
+      "step": 27041
+    },
+    {
+      "epoch": 0.9834527402989417,
+      "grad_norm": 0.0,
+      "learning_rate": 1.435745381732967e-08,
+      "loss": 0.9694,
+      "step": 27042
+    },
+    {
+      "epoch": 0.9834891079026803,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4294428504589841e-08,
+      "loss": 0.9247,
+      "step": 27043
+    },
+    {
+      "epoch": 0.9835254755064189,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4231541729345932e-08,
+      "loss": 0.9678,
+      "step": 27044
+    },
+    {
+      "epoch": 0.9835618431101575,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4168793492472798e-08,
+      "loss": 0.9422,
+      "step": 27045
+    },
+    {
+      "epoch": 0.9835982107138961,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4106183794838634e-08,
+      "loss": 0.8779,
+      "step": 27046
+    },
+    {
+      "epoch": 0.9836345783176347,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4043712637312745e-08,
+      "loss": 1.0379,
+      "step": 27047
+    },
+    {
+      "epoch": 0.9836709459213733,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3981380020763324e-08,
+      "loss": 0.8946,
+      "step": 27048
+    },
+    {
+      "epoch": 0.9837073135251119,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3919185946053016e-08,
+      "loss": 0.9466,
+      "step": 27049
+    },
+    {
+      "epoch": 0.9837436811288505,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3857130414046683e-08,
+      "loss": 0.8725,
+      "step": 27050
+    },
+    {
+      "epoch": 0.9837800487325891,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3795213425603638e-08,
+      "loss": 0.8159,
+      "step": 27051
+    },
+    {
+      "epoch": 0.9838164163363275,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3733434981583194e-08,
+      "loss": 0.9925,
+      "step": 27052
+    },
+    {
+      "epoch": 0.9838527839400661,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3671795082842444e-08,
+      "loss": 0.9406,
+      "step": 27053
+    },
+    {
+      "epoch": 0.9838891515438047,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3610293730236257e-08,
+      "loss": 0.9093,
+      "step": 27054
+    },
+    {
+      "epoch": 0.9839255191475433,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3548930924619508e-08,
+      "loss": 0.9274,
+      "step": 27055
+    },
+    {
+      "epoch": 0.9839618867512819,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3487706666842626e-08,
+      "loss": 0.8071,
+      "step": 27056
+    },
+    {
+      "epoch": 0.9839982543550205,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3426620957753821e-08,
+      "loss": 0.8513,
+      "step": 27057
+    },
+    {
+      "epoch": 0.9840346219587591,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3365673798202416e-08,
+      "loss": 0.903,
+      "step": 27058
+    },
+    {
+      "epoch": 0.9840709895624977,
+      "grad_norm": 0.0,
+      "learning_rate": 1.330486518903329e-08,
+      "loss": 0.8952,
+      "step": 27059
+    },
+    {
+      "epoch": 0.9841073571662363,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3244195131089099e-08,
+      "loss": 0.8606,
+      "step": 27060
+    },
+    {
+      "epoch": 0.9841437247699749,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3183663625212506e-08,
+      "loss": 0.9643,
+      "step": 27061
+    },
+    {
+      "epoch": 0.9841800923737135,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3123270672242838e-08,
+      "loss": 0.8588,
+      "step": 27062
+    },
+    {
+      "epoch": 0.9842164599774521,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3063016273018314e-08,
+      "loss": 0.9308,
+      "step": 27063
+    },
+    {
+      "epoch": 0.9842528275811907,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3002900428376043e-08,
+      "loss": 0.8981,
+      "step": 27064
+    },
+    {
+      "epoch": 0.9842891951849293,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2942923139147579e-08,
+      "loss": 0.9006,
+      "step": 27065
+    },
+    {
+      "epoch": 0.9843255627886679,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2883084406166702e-08,
+      "loss": 0.8624,
+      "step": 27066
+    },
+    {
+      "epoch": 0.9843619303924065,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2823384230262747e-08,
+      "loss": 0.9551,
+      "step": 27067
+    },
+    {
+      "epoch": 0.9843982979961451,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2763822612265053e-08,
+      "loss": 0.9746,
+      "step": 27068
+    },
+    {
+      "epoch": 0.9844346655998836,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2704399552998515e-08,
+      "loss": 0.965,
+      "step": 27069
+    },
+    {
+      "epoch": 0.9844710332036222,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2645115053289136e-08,
+      "loss": 1.0386,
+      "step": 27070
+    },
+    {
+      "epoch": 0.9845074008073608,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2585969113958485e-08,
+      "loss": 0.9391,
+      "step": 27071
+    },
+    {
+      "epoch": 0.9845437684110994,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2526961735827015e-08,
+      "loss": 0.8718,
+      "step": 27072
+    },
+    {
+      "epoch": 0.984580136014838,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2468092919712959e-08,
+      "loss": 0.8224,
+      "step": 27073
+    },
+    {
+      "epoch": 0.9846165036185766,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2409362666434555e-08,
+      "loss": 0.9142,
+      "step": 27074
+    },
+    {
+      "epoch": 0.9846528712223152,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2350770976805593e-08,
+      "loss": 0.9763,
+      "step": 27075
+    },
+    {
+      "epoch": 0.9846892388260537,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2292317851638757e-08,
+      "loss": 0.9692,
+      "step": 27076
+    },
+    {
+      "epoch": 0.9847256064297923,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2234003291745623e-08,
+      "loss": 0.9175,
+      "step": 27077
+    },
+    {
+      "epoch": 0.9847619740335309,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2175827297934428e-08,
+      "loss": 0.9598,
+      "step": 27078
+    },
+    {
+      "epoch": 0.9847983416372695,
+      "grad_norm": 0.0,
+      "learning_rate": 1.211778987101342e-08,
+      "loss": 0.9768,
+      "step": 27079
+    },
+    {
+      "epoch": 0.9848347092410081,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2059891011786396e-08,
+      "loss": 0.8182,
+      "step": 27080
+    },
+    {
+      "epoch": 0.9848710768447467,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2002130721057159e-08,
+      "loss": 0.9699,
+      "step": 27081
+    },
+    {
+      "epoch": 0.9849074444484853,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1944508999628402e-08,
+      "loss": 0.9933,
+      "step": 27082
+    },
+    {
+      "epoch": 0.9849438120522239,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1887025848297262e-08,
+      "loss": 0.8456,
+      "step": 27083
+    },
+    {
+      "epoch": 0.9849801796559625,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1829681267863102e-08,
+      "loss": 0.9319,
+      "step": 27084
+    },
+    {
+      "epoch": 0.985016547259701,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1772475259120842e-08,
+      "loss": 0.9388,
+      "step": 27085
+    },
+    {
+      "epoch": 0.9850529148634396,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1715407822863178e-08,
+      "loss": 0.9271,
+      "step": 27086
+    },
+    {
+      "epoch": 0.9850892824671782,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1658478959883924e-08,
+      "loss": 0.9478,
+      "step": 27087
+    },
+    {
+      "epoch": 0.9851256500709168,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1601688670970224e-08,
+      "loss": 0.9236,
+      "step": 27088
+    },
+    {
+      "epoch": 0.9851620176746554,
+      "grad_norm": 0.0,
+      "learning_rate": 1.154503695691256e-08,
+      "loss": 0.8061,
+      "step": 27089
+    },
+    {
+      "epoch": 0.985198385278394,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1488523818495856e-08,
+      "loss": 0.7785,
+      "step": 27090
+    },
+    {
+      "epoch": 0.9852347528821326,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1432149256502822e-08,
+      "loss": 0.9445,
+      "step": 27091
+    },
+    {
+      "epoch": 0.9852711204858712,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1375913271717276e-08,
+      "loss": 1.0174,
+      "step": 27092
+    },
+    {
+      "epoch": 0.9853074880896098,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1319815864919703e-08,
+      "loss": 0.894,
+      "step": 27093
+    },
+    {
+      "epoch": 0.9853438556933484,
+      "grad_norm": 0.0,
+      "learning_rate": 1.126385703688726e-08,
+      "loss": 0.9954,
+      "step": 27094
+    },
+    {
+      "epoch": 0.985380223297087,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1208036788395993e-08,
+      "loss": 0.7971,
+      "step": 27095
+    },
+    {
+      "epoch": 0.9854165909008256,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1152355120221947e-08,
+      "loss": 0.9431,
+      "step": 27096
+    },
+    {
+      "epoch": 0.9854529585045642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1096812033135618e-08,
+      "loss": 0.8501,
+      "step": 27097
+    },
+    {
+      "epoch": 0.9854893261083028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.104140752790861e-08,
+      "loss": 0.7627,
+      "step": 27098
+    },
+    {
+      "epoch": 0.9855256937120414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0986141605310308e-08,
+      "loss": 0.8819,
+      "step": 27099
+    },
+    {
+      "epoch": 0.98556206131578,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0931014266106766e-08,
+      "loss": 0.9608,
+      "step": 27100
+    },
+    {
+      "epoch": 0.9855984289195185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0876025511062926e-08,
+      "loss": 0.9308,
+      "step": 27101
+    },
+    {
+      "epoch": 0.985634796523257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0821175340940404e-08,
+      "loss": 0.8599,
+      "step": 27102
+    },
+    {
+      "epoch": 0.9856711641269956,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0766463756503031e-08,
+      "loss": 1.055,
+      "step": 27103
+    },
+    {
+      "epoch": 0.9857075317307342,
+      "grad_norm": 0.0,
+      "learning_rate": 1.071189075850687e-08,
+      "loss": 0.9985,
+      "step": 27104
+    },
+    {
+      "epoch": 0.9857438993344728,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0657456347710205e-08,
+      "loss": 0.8494,
+      "step": 27105
+    },
+    {
+      "epoch": 0.9857802669382114,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0603160524869093e-08,
+      "loss": 0.9199,
+      "step": 27106
+    },
+    {
+      "epoch": 0.98581663454195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.054900329073516e-08,
+      "loss": 0.9613,
+      "step": 27107
+    },
+    {
+      "epoch": 0.9858530021456886,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0494984646061135e-08,
+      "loss": 0.8247,
+      "step": 27108
+    },
+    {
+      "epoch": 0.9858893697494272,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0441104591595308e-08,
+      "loss": 0.9627,
+      "step": 27109
+    },
+    {
+      "epoch": 0.9859257373531658,
+      "grad_norm": 0.0,
+      "learning_rate": 1.038736312808597e-08,
+      "loss": 0.9613,
+      "step": 27110
+    },
+    {
+      "epoch": 0.9859621049569044,
+      "grad_norm": 0.0,
+      "learning_rate": 1.033376025627919e-08,
+      "loss": 0.9028,
+      "step": 27111
+    },
+    {
+      "epoch": 0.985998472560643,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0280295976917709e-08,
+      "loss": 0.7911,
+      "step": 27112
+    },
+    {
+      "epoch": 0.9860348401643816,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0226970290744264e-08,
+      "loss": 0.8344,
+      "step": 27113
+    },
+    {
+      "epoch": 0.9860712077681202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0173783198497156e-08,
+      "loss": 0.8596,
+      "step": 27114
+    },
+    {
+      "epoch": 0.9861075753718588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0120734700914681e-08,
+      "loss": 0.6842,
+      "step": 27115
+    },
+    {
+      "epoch": 0.9861439429755974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0067824798734027e-08,
+      "loss": 0.9264,
+      "step": 27116
+    },
+    {
+      "epoch": 0.986180310579336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0015053492687943e-08,
+      "loss": 0.8893,
+      "step": 27117
+    },
+    {
+      "epoch": 0.9862166781830746,
+      "grad_norm": 0.0,
+      "learning_rate": 9.962420783509175e-09,
+      "loss": 0.9681,
+      "step": 27118
+    },
+    {
+      "epoch": 0.986253045786813,
+      "grad_norm": 0.0,
+      "learning_rate": 9.90992667192825e-09,
+      "loss": 0.9761,
+      "step": 27119
+    },
+    {
+      "epoch": 0.9862894133905517,
+      "grad_norm": 0.0,
+      "learning_rate": 9.857571158673473e-09,
+      "loss": 0.8358,
+      "step": 27120
+    },
+    {
+      "epoch": 0.9863257809942902,
+      "grad_norm": 0.0,
+      "learning_rate": 9.805354244469823e-09,
+      "loss": 0.9828,
+      "step": 27121
+    },
+    {
+      "epoch": 0.9863621485980288,
+      "grad_norm": 0.0,
+      "learning_rate": 9.753275930042272e-09,
+      "loss": 0.8994,
+      "step": 27122
+    },
+    {
+      "epoch": 0.9863985162017674,
+      "grad_norm": 0.0,
+      "learning_rate": 9.701336216114688e-09,
+      "loss": 0.9234,
+      "step": 27123
+    },
+    {
+      "epoch": 0.986434883805506,
+      "grad_norm": 0.0,
+      "learning_rate": 9.649535103405383e-09,
+      "loss": 0.9836,
+      "step": 27124
+    },
+    {
+      "epoch": 0.9864712514092446,
+      "grad_norm": 0.0,
+      "learning_rate": 9.597872592636003e-09,
+      "loss": 0.9035,
+      "step": 27125
+    },
+    {
+      "epoch": 0.9865076190129832,
+      "grad_norm": 0.0,
+      "learning_rate": 9.546348684520423e-09,
+      "loss": 0.9032,
+      "step": 27126
+    },
+    {
+      "epoch": 0.9865439866167218,
+      "grad_norm": 0.0,
+      "learning_rate": 9.494963379774736e-09,
+      "loss": 0.8922,
+      "step": 27127
+    },
+    {
+      "epoch": 0.9865803542204604,
+      "grad_norm": 0.0,
+      "learning_rate": 9.443716679112813e-09,
+      "loss": 0.9769,
+      "step": 27128
+    },
+    {
+      "epoch": 0.986616721824199,
+      "grad_norm": 0.0,
+      "learning_rate": 9.39260858324298e-09,
+      "loss": 0.8446,
+      "step": 27129
+    },
+    {
+      "epoch": 0.9866530894279376,
+      "grad_norm": 0.0,
+      "learning_rate": 9.341639092876886e-09,
+      "loss": 0.9573,
+      "step": 27130
+    },
+    {
+      "epoch": 0.9866894570316762,
+      "grad_norm": 0.0,
+      "learning_rate": 9.290808208720636e-09,
+      "loss": 0.8855,
+      "step": 27131
+    },
+    {
+      "epoch": 0.9867258246354148,
+      "grad_norm": 0.0,
+      "learning_rate": 9.240115931479221e-09,
+      "loss": 0.8401,
+      "step": 27132
+    },
+    {
+      "epoch": 0.9867621922391534,
+      "grad_norm": 0.0,
+      "learning_rate": 9.189562261856521e-09,
+      "loss": 0.9744,
+      "step": 27133
+    },
+    {
+      "epoch": 0.986798559842892,
+      "grad_norm": 0.0,
+      "learning_rate": 9.139147200553088e-09,
+      "loss": 0.7754,
+      "step": 27134
+    },
+    {
+      "epoch": 0.9868349274466306,
+      "grad_norm": 0.0,
+      "learning_rate": 9.088870748269474e-09,
+      "loss": 0.8782,
+      "step": 27135
+    },
+    {
+      "epoch": 0.9868712950503691,
+      "grad_norm": 0.0,
+      "learning_rate": 9.038732905701786e-09,
+      "loss": 0.9194,
+      "step": 27136
+    },
+    {
+      "epoch": 0.9869076626541077,
+      "grad_norm": 0.0,
+      "learning_rate": 8.988733673546135e-09,
+      "loss": 0.9594,
+      "step": 27137
+    },
+    {
+      "epoch": 0.9869440302578463,
+      "grad_norm": 0.0,
+      "learning_rate": 8.938873052497521e-09,
+      "loss": 0.9026,
+      "step": 27138
+    },
+    {
+      "epoch": 0.9869803978615849,
+      "grad_norm": 0.0,
+      "learning_rate": 8.889151043245392e-09,
+      "loss": 0.8459,
+      "step": 27139
+    },
+    {
+      "epoch": 0.9870167654653235,
+      "grad_norm": 0.0,
+      "learning_rate": 8.839567646481417e-09,
+      "loss": 0.8869,
+      "step": 27140
+    },
+    {
+      "epoch": 0.9870531330690621,
+      "grad_norm": 0.0,
+      "learning_rate": 8.790122862892824e-09,
+      "loss": 0.9275,
+      "step": 27141
+    },
+    {
+      "epoch": 0.9870895006728007,
+      "grad_norm": 0.0,
+      "learning_rate": 8.740816693165733e-09,
+      "loss": 0.9544,
+      "step": 27142
+    },
+    {
+      "epoch": 0.9871258682765393,
+      "grad_norm": 0.0,
+      "learning_rate": 8.691649137984038e-09,
+      "loss": 0.9916,
+      "step": 27143
+    },
+    {
+      "epoch": 0.9871622358802778,
+      "grad_norm": 0.0,
+      "learning_rate": 8.642620198029417e-09,
+      "loss": 0.9648,
+      "step": 27144
+    },
+    {
+      "epoch": 0.9871986034840164,
+      "grad_norm": 0.0,
+      "learning_rate": 8.593729873982437e-09,
+      "loss": 0.8721,
+      "step": 27145
+    },
+    {
+      "epoch": 0.987234971087755,
+      "grad_norm": 0.0,
+      "learning_rate": 8.544978166522556e-09,
+      "loss": 0.9268,
+      "step": 27146
+    },
+    {
+      "epoch": 0.9872713386914936,
+      "grad_norm": 0.0,
+      "learning_rate": 8.496365076323677e-09,
+      "loss": 0.9227,
+      "step": 27147
+    },
+    {
+      "epoch": 0.9873077062952322,
+      "grad_norm": 0.0,
+      "learning_rate": 8.447890604063036e-09,
+      "loss": 0.9434,
+      "step": 27148
+    },
+    {
+      "epoch": 0.9873440738989708,
+      "grad_norm": 0.0,
+      "learning_rate": 8.39955475041121e-09,
+      "loss": 0.9982,
+      "step": 27149
+    },
+    {
+      "epoch": 0.9873804415027094,
+      "grad_norm": 0.0,
+      "learning_rate": 8.351357516038772e-09,
+      "loss": 0.8094,
+      "step": 27150
+    },
+    {
+      "epoch": 0.987416809106448,
+      "grad_norm": 0.0,
+      "learning_rate": 8.303298901616297e-09,
+      "loss": 0.7404,
+      "step": 27151
+    },
+    {
+      "epoch": 0.9874531767101866,
+      "grad_norm": 0.0,
+      "learning_rate": 8.255378907807698e-09,
+      "loss": 0.997,
+      "step": 27152
+    },
+    {
+      "epoch": 0.9874895443139251,
+      "grad_norm": 0.0,
+      "learning_rate": 8.207597535280221e-09,
+      "loss": 0.833,
+      "step": 27153
+    },
+    {
+      "epoch": 0.9875259119176637,
+      "grad_norm": 0.0,
+      "learning_rate": 8.159954784695556e-09,
+      "loss": 0.9138,
+      "step": 27154
+    },
+    {
+      "epoch": 0.9875622795214023,
+      "grad_norm": 0.0,
+      "learning_rate": 8.112450656714289e-09,
+      "loss": 0.8599,
+      "step": 27155
+    },
+    {
+      "epoch": 0.9875986471251409,
+      "grad_norm": 0.0,
+      "learning_rate": 8.065085151996999e-09,
+      "loss": 0.9455,
+      "step": 27156
+    },
+    {
+      "epoch": 0.9876350147288795,
+      "grad_norm": 0.0,
+      "learning_rate": 8.01785827119983e-09,
+      "loss": 0.843,
+      "step": 27157
+    },
+    {
+      "epoch": 0.9876713823326181,
+      "grad_norm": 0.0,
+      "learning_rate": 7.970770014977814e-09,
+      "loss": 0.7771,
+      "step": 27158
+    },
+    {
+      "epoch": 0.9877077499363567,
+      "grad_norm": 0.0,
+      "learning_rate": 7.923820383984871e-09,
+      "loss": 0.9555,
+      "step": 27159
+    },
+    {
+      "epoch": 0.9877441175400953,
+      "grad_norm": 0.0,
+      "learning_rate": 7.877009378871592e-09,
+      "loss": 0.9304,
+      "step": 27160
+    },
+    {
+      "epoch": 0.9877804851438339,
+      "grad_norm": 0.0,
+      "learning_rate": 7.830337000288568e-09,
+      "loss": 0.9462,
+      "step": 27161
+    },
+    {
+      "epoch": 0.9878168527475725,
+      "grad_norm": 0.0,
+      "learning_rate": 7.78380324888195e-09,
+      "loss": 0.7983,
+      "step": 27162
+    },
+    {
+      "epoch": 0.9878532203513111,
+      "grad_norm": 0.0,
+      "learning_rate": 7.737408125297884e-09,
+      "loss": 0.8635,
+      "step": 27163
+    },
+    {
+      "epoch": 0.9878895879550497,
+      "grad_norm": 0.0,
+      "learning_rate": 7.691151630180305e-09,
+      "loss": 0.8069,
+      "step": 27164
+    },
+    {
+      "epoch": 0.9879259555587883,
+      "grad_norm": 0.0,
+      "learning_rate": 7.645033764170917e-09,
+      "loss": 0.9338,
+      "step": 27165
+    },
+    {
+      "epoch": 0.9879623231625269,
+      "grad_norm": 0.0,
+      "learning_rate": 7.599054527909211e-09,
+      "loss": 0.9343,
+      "step": 27166
+    },
+    {
+      "epoch": 0.9879986907662655,
+      "grad_norm": 0.0,
+      "learning_rate": 7.553213922033564e-09,
+      "loss": 0.9033,
+      "step": 27167
+    },
+    {
+      "epoch": 0.988035058370004,
+      "grad_norm": 0.0,
+      "learning_rate": 7.507511947180136e-09,
+      "loss": 1.0347,
+      "step": 27168
+    },
+    {
+      "epoch": 0.9880714259737426,
+      "grad_norm": 0.0,
+      "learning_rate": 7.46194860398064e-09,
+      "loss": 0.879,
+      "step": 27169
+    },
+    {
+      "epoch": 0.9881077935774811,
+      "grad_norm": 0.0,
+      "learning_rate": 7.4165238930712365e-09,
+      "loss": 0.8608,
+      "step": 27170
+    },
+    {
+      "epoch": 0.9881441611812197,
+      "grad_norm": 0.0,
+      "learning_rate": 7.371237815078091e-09,
+      "loss": 0.9976,
+      "step": 27171
+    },
+    {
+      "epoch": 0.9881805287849583,
+      "grad_norm": 0.0,
+      "learning_rate": 7.3260903706329215e-09,
+      "loss": 0.9397,
+      "step": 27172
+    },
+    {
+      "epoch": 0.9882168963886969,
+      "grad_norm": 0.0,
+      "learning_rate": 7.281081560359671e-09,
+      "loss": 0.9138,
+      "step": 27173
+    },
+    {
+      "epoch": 0.9882532639924355,
+      "grad_norm": 0.0,
+      "learning_rate": 7.236211384883396e-09,
+      "loss": 0.9149,
+      "step": 27174
+    },
+    {
+      "epoch": 0.9882896315961741,
+      "grad_norm": 0.0,
+      "learning_rate": 7.191479844826932e-09,
+      "loss": 1.0038,
+      "step": 27175
+    },
+    {
+      "epoch": 0.9883259991999127,
+      "grad_norm": 0.0,
+      "learning_rate": 7.146886940810893e-09,
+      "loss": 0.8283,
+      "step": 27176
+    },
+    {
+      "epoch": 0.9883623668036513,
+      "grad_norm": 0.0,
+      "learning_rate": 7.102432673453674e-09,
+      "loss": 0.9883,
+      "step": 27177
+    },
+    {
+      "epoch": 0.9883987344073899,
+      "grad_norm": 0.0,
+      "learning_rate": 7.058117043372559e-09,
+      "loss": 1.0078,
+      "step": 27178
+    },
+    {
+      "epoch": 0.9884351020111285,
+      "grad_norm": 0.0,
+      "learning_rate": 7.0139400511815e-09,
+      "loss": 0.9097,
+      "step": 27179
+    },
+    {
+      "epoch": 0.9884714696148671,
+      "grad_norm": 0.0,
+      "learning_rate": 6.969901697494452e-09,
+      "loss": 0.8464,
+      "step": 27180
+    },
+    {
+      "epoch": 0.9885078372186057,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9260019829209265e-09,
+      "loss": 0.8466,
+      "step": 27181
+    },
+    {
+      "epoch": 0.9885442048223443,
+      "grad_norm": 0.0,
+      "learning_rate": 6.882240908070436e-09,
+      "loss": 0.8614,
+      "step": 27182
+    },
+    {
+      "epoch": 0.9885805724260829,
+      "grad_norm": 0.0,
+      "learning_rate": 6.838618473551384e-09,
+      "loss": 0.8729,
+      "step": 27183
+    },
+    {
+      "epoch": 0.9886169400298215,
+      "grad_norm": 0.0,
+      "learning_rate": 6.79513467996773e-09,
+      "loss": 0.8914,
+      "step": 27184
+    },
+    {
+      "epoch": 0.9886533076335601,
+      "grad_norm": 0.0,
+      "learning_rate": 6.751789527923436e-09,
+      "loss": 0.9116,
+      "step": 27185
+    },
+    {
+      "epoch": 0.9886896752372986,
+      "grad_norm": 0.0,
+      "learning_rate": 6.708583018019132e-09,
+      "loss": 0.8738,
+      "step": 27186
+    },
+    {
+      "epoch": 0.9887260428410372,
+      "grad_norm": 0.0,
+      "learning_rate": 6.66551515085545e-09,
+      "loss": 0.8669,
+      "step": 27187
+    },
+    {
+      "epoch": 0.9887624104447758,
+      "grad_norm": 0.0,
+      "learning_rate": 6.622585927027469e-09,
+      "loss": 0.8668,
+      "step": 27188
+    },
+    {
+      "epoch": 0.9887987780485143,
+      "grad_norm": 0.0,
+      "learning_rate": 6.579795347133599e-09,
+      "loss": 0.8818,
+      "step": 27189
+    },
+    {
+      "epoch": 0.988835145652253,
+      "grad_norm": 0.0,
+      "learning_rate": 6.53714341176559e-09,
+      "loss": 0.8048,
+      "step": 27190
+    },
+    {
+      "epoch": 0.9888715132559915,
+      "grad_norm": 0.0,
+      "learning_rate": 6.494630121516299e-09,
+      "loss": 0.979,
+      "step": 27191
+    },
+    {
+      "epoch": 0.9889078808597301,
+      "grad_norm": 0.0,
+      "learning_rate": 6.452255476975256e-09,
+      "loss": 0.814,
+      "step": 27192
+    },
+    {
+      "epoch": 0.9889442484634687,
+      "grad_norm": 0.0,
+      "learning_rate": 6.410019478729767e-09,
+      "loss": 0.9364,
+      "step": 27193
+    },
+    {
+      "epoch": 0.9889806160672073,
+      "grad_norm": 0.0,
+      "learning_rate": 6.367922127366033e-09,
+      "loss": 0.9444,
+      "step": 27194
+    },
+    {
+      "epoch": 0.9890169836709459,
+      "grad_norm": 0.0,
+      "learning_rate": 6.3259634234691395e-09,
+      "loss": 0.9595,
+      "step": 27195
+    },
+    {
+      "epoch": 0.9890533512746845,
+      "grad_norm": 0.0,
+      "learning_rate": 6.2841433676197325e-09,
+      "loss": 1.0081,
+      "step": 27196
+    },
+    {
+      "epoch": 0.9890897188784231,
+      "grad_norm": 0.0,
+      "learning_rate": 6.242461960398461e-09,
+      "loss": 0.9127,
+      "step": 27197
+    },
+    {
+      "epoch": 0.9891260864821617,
+      "grad_norm": 0.0,
+      "learning_rate": 6.200919202383749e-09,
+      "loss": 0.8562,
+      "step": 27198
+    },
+    {
+      "epoch": 0.9891624540859003,
+      "grad_norm": 0.0,
+      "learning_rate": 6.159515094151802e-09,
+      "loss": 0.9489,
+      "step": 27199
+    },
+    {
+      "epoch": 0.9891988216896389,
+      "grad_norm": 0.0,
+      "learning_rate": 6.118249636277718e-09,
+      "loss": 1.09,
+      "step": 27200
+    },
+    {
+      "epoch": 0.9892351892933775,
+      "grad_norm": 0.0,
+      "learning_rate": 6.0771228293332595e-09,
+      "loss": 0.9073,
+      "step": 27201
+    },
+    {
+      "epoch": 0.9892715568971161,
+      "grad_norm": 0.0,
+      "learning_rate": 6.036134673889082e-09,
+      "loss": 0.8321,
+      "step": 27202
+    },
+    {
+      "epoch": 0.9893079245008546,
+      "grad_norm": 0.0,
+      "learning_rate": 5.995285170513621e-09,
+      "loss": 0.9942,
+      "step": 27203
+    },
+    {
+      "epoch": 0.9893442921045932,
+      "grad_norm": 0.0,
+      "learning_rate": 5.954574319775308e-09,
+      "loss": 0.8321,
+      "step": 27204
+    },
+    {
+      "epoch": 0.9893806597083318,
+      "grad_norm": 0.0,
+      "learning_rate": 5.9140021222359176e-09,
+      "loss": 0.8981,
+      "step": 27205
+    },
+    {
+      "epoch": 0.9894170273120704,
+      "grad_norm": 0.0,
+      "learning_rate": 5.8735685784605535e-09,
+      "loss": 1.0077,
+      "step": 27206
+    },
+    {
+      "epoch": 0.989453394915809,
+      "grad_norm": 0.0,
+      "learning_rate": 5.833273689009877e-09,
+      "loss": 1.0192,
+      "step": 27207
+    },
+    {
+      "epoch": 0.9894897625195476,
+      "grad_norm": 0.0,
+      "learning_rate": 5.793117454442332e-09,
+      "loss": 0.9439,
+      "step": 27208
+    },
+    {
+      "epoch": 0.9895261301232862,
+      "grad_norm": 0.0,
+      "learning_rate": 5.753099875316359e-09,
+      "loss": 0.8127,
+      "step": 27209
+    },
+    {
+      "epoch": 0.9895624977270248,
+      "grad_norm": 0.0,
+      "learning_rate": 5.71322095218485e-09,
+      "loss": 0.8081,
+      "step": 27210
+    },
+    {
+      "epoch": 0.9895988653307634,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6734806856029164e-09,
+      "loss": 0.9121,
+      "step": 27211
+    },
+    {
+      "epoch": 0.989635232934502,
+      "grad_norm": 0.0,
+      "learning_rate": 5.633879076121229e-09,
+      "loss": 0.9962,
+      "step": 27212
+    },
+    {
+      "epoch": 0.9896716005382405,
+      "grad_norm": 0.0,
+      "learning_rate": 5.594416124289348e-09,
+      "loss": 0.9767,
+      "step": 27213
+    },
+    {
+      "epoch": 0.9897079681419791,
+      "grad_norm": 0.0,
+      "learning_rate": 5.555091830654613e-09,
+      "loss": 0.9576,
+      "step": 27214
+    },
+    {
+      "epoch": 0.9897443357457177,
+      "grad_norm": 0.0,
+      "learning_rate": 5.515906195763254e-09,
+      "loss": 0.926,
+      "step": 27215
+    },
+    {
+      "epoch": 0.9897807033494563,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4768592201581705e-09,
+      "loss": 0.9557,
+      "step": 27216
+    },
+    {
+      "epoch": 0.9898170709531949,
+      "grad_norm": 0.0,
+      "learning_rate": 5.437950904381151e-09,
+      "loss": 0.8914,
+      "step": 27217
+    },
+    {
+      "epoch": 0.9898534385569335,
+      "grad_norm": 0.0,
+      "learning_rate": 5.399181248971763e-09,
+      "loss": 0.8517,
+      "step": 27218
+    },
+    {
+      "epoch": 0.9898898061606721,
+      "grad_norm": 0.0,
+      "learning_rate": 5.360550254468467e-09,
+      "loss": 0.8312,
+      "step": 27219
+    },
+    {
+      "epoch": 0.9899261737644106,
+      "grad_norm": 0.0,
+      "learning_rate": 5.322057921406387e-09,
+      "loss": 0.8421,
+      "step": 27220
+    },
+    {
+      "epoch": 0.9899625413681492,
+      "grad_norm": 0.0,
+      "learning_rate": 5.283704250321764e-09,
+      "loss": 0.7954,
+      "step": 27221
+    },
+    {
+      "epoch": 0.9899989089718878,
+      "grad_norm": 0.0,
+      "learning_rate": 5.245489241744173e-09,
+      "loss": 0.9622,
+      "step": 27222
+    },
+    {
+      "epoch": 0.9900352765756264,
+      "grad_norm": 0.0,
+      "learning_rate": 5.207412896204301e-09,
+      "loss": 0.8856,
+      "step": 27223
+    },
+    {
+      "epoch": 0.990071644179365,
+      "grad_norm": 0.0,
+      "learning_rate": 5.169475214231723e-09,
+      "loss": 0.9119,
+      "step": 27224
+    },
+    {
+      "epoch": 0.9901080117831036,
+      "grad_norm": 0.0,
+      "learning_rate": 5.131676196351576e-09,
+      "loss": 0.7737,
+      "step": 27225
+    },
+    {
+      "epoch": 0.9901443793868422,
+      "grad_norm": 0.0,
+      "learning_rate": 5.094015843087885e-09,
+      "loss": 0.9085,
+      "step": 27226
+    },
+    {
+      "epoch": 0.9901807469905808,
+      "grad_norm": 0.0,
+      "learning_rate": 5.0564941549646754e-09,
+      "loss": 0.933,
+      "step": 27227
+    },
+    {
+      "epoch": 0.9902171145943194,
+      "grad_norm": 0.0,
+      "learning_rate": 5.019111132501531e-09,
+      "loss": 0.8981,
+      "step": 27228
+    },
+    {
+      "epoch": 0.990253482198058,
+      "grad_norm": 0.0,
+      "learning_rate": 4.981866776216926e-09,
+      "loss": 0.9364,
+      "step": 27229
+    },
+    {
+      "epoch": 0.9902898498017966,
+      "grad_norm": 0.0,
+      "learning_rate": 4.944761086627115e-09,
+      "loss": 0.9708,
+      "step": 27230
+    },
+    {
+      "epoch": 0.9903262174055352,
+      "grad_norm": 0.0,
+      "learning_rate": 4.90779406424835e-09,
+      "loss": 0.9379,
+      "step": 27231
+    },
+    {
+      "epoch": 0.9903625850092738,
+      "grad_norm": 0.0,
+      "learning_rate": 4.870965709593556e-09,
+      "loss": 0.9122,
+      "step": 27232
+    },
+    {
+      "epoch": 0.9903989526130124,
+      "grad_norm": 0.0,
+      "learning_rate": 4.834276023171214e-09,
+      "loss": 0.9599,
+      "step": 27233
+    },
+    {
+      "epoch": 0.990435320216751,
+      "grad_norm": 0.0,
+      "learning_rate": 4.797725005493137e-09,
+      "loss": 0.9831,
+      "step": 27234
+    },
+    {
+      "epoch": 0.9904716878204896,
+      "grad_norm": 0.0,
+      "learning_rate": 4.7613126570655865e-09,
+      "loss": 0.8868,
+      "step": 27235
+    },
+    {
+      "epoch": 0.9905080554242282,
+      "grad_norm": 0.0,
+      "learning_rate": 4.725038978392604e-09,
+      "loss": 0.8898,
+      "step": 27236
+    },
+    {
+      "epoch": 0.9905444230279666,
+      "grad_norm": 0.0,
+      "learning_rate": 4.688903969978231e-09,
+      "loss": 0.9095,
+      "step": 27237
+    },
+    {
+      "epoch": 0.9905807906317052,
+      "grad_norm": 0.0,
+      "learning_rate": 4.652907632324288e-09,
+      "loss": 0.8916,
+      "step": 27238
+    },
+    {
+      "epoch": 0.9906171582354438,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6170499659292655e-09,
+      "loss": 0.8658,
+      "step": 27239
+    },
+    {
+      "epoch": 0.9906535258391824,
+      "grad_norm": 0.0,
+      "learning_rate": 4.581330971291653e-09,
+      "loss": 0.9654,
+      "step": 27240
+    },
+    {
+      "epoch": 0.990689893442921,
+      "grad_norm": 0.0,
+      "learning_rate": 4.545750648906611e-09,
+      "loss": 0.7889,
+      "step": 27241
+    },
+    {
+      "epoch": 0.9907262610466596,
+      "grad_norm": 0.0,
+      "learning_rate": 4.510308999267077e-09,
+      "loss": 0.8828,
+      "step": 27242
+    },
+    {
+      "epoch": 0.9907626286503982,
+      "grad_norm": 0.0,
+      "learning_rate": 4.475006022864881e-09,
+      "loss": 0.8534,
+      "step": 27243
+    },
+    {
+      "epoch": 0.9907989962541368,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4398417201907405e-09,
+      "loss": 0.7343,
+      "step": 27244
+    },
+    {
+      "epoch": 0.9908353638578754,
+      "grad_norm": 0.0,
+      "learning_rate": 4.404816091732045e-09,
+      "loss": 0.9249,
+      "step": 27245
+    },
+    {
+      "epoch": 0.990871731461614,
+      "grad_norm": 0.0,
+      "learning_rate": 4.36992913797396e-09,
+      "loss": 0.8477,
+      "step": 27246
+    },
+    {
+      "epoch": 0.9909080990653526,
+      "grad_norm": 0.0,
+      "learning_rate": 4.335180859401656e-09,
+      "loss": 0.8713,
+      "step": 27247
+    },
+    {
+      "epoch": 0.9909444666690912,
+      "grad_norm": 0.0,
+      "learning_rate": 4.300571256496966e-09,
+      "loss": 1.0042,
+      "step": 27248
+    },
+    {
+      "epoch": 0.9909808342728298,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2661003297395084e-09,
+      "loss": 0.8433,
+      "step": 27249
+    },
+    {
+      "epoch": 0.9910172018765684,
+      "grad_norm": 0.0,
+      "learning_rate": 4.23176807960779e-09,
+      "loss": 0.8197,
+      "step": 27250
+    },
+    {
+      "epoch": 0.991053569480307,
+      "grad_norm": 0.0,
+      "learning_rate": 4.197574506578095e-09,
+      "loss": 0.9324,
+      "step": 27251
+    },
+    {
+      "epoch": 0.9910899370840456,
+      "grad_norm": 0.0,
+      "learning_rate": 4.163519611124489e-09,
+      "loss": 0.9349,
+      "step": 27252
+    },
+    {
+      "epoch": 0.9911263046877842,
+      "grad_norm": 0.0,
+      "learning_rate": 4.129603393721038e-09,
+      "loss": 1.1034,
+      "step": 27253
+    },
+    {
+      "epoch": 0.9911626722915227,
+      "grad_norm": 0.0,
+      "learning_rate": 4.095825854836255e-09,
+      "loss": 0.899,
+      "step": 27254
+    },
+    {
+      "epoch": 0.9911990398952613,
+      "grad_norm": 0.0,
+      "learning_rate": 4.062186994939765e-09,
+      "loss": 0.917,
+      "step": 27255
+    },
+    {
+      "epoch": 0.9912354074989999,
+      "grad_norm": 0.0,
+      "learning_rate": 4.028686814497862e-09,
+      "loss": 0.9265,
+      "step": 27256
+    },
+    {
+      "epoch": 0.9912717751027385,
+      "grad_norm": 0.0,
+      "learning_rate": 3.995325313975728e-09,
+      "loss": 0.9235,
+      "step": 27257
+    },
+    {
+      "epoch": 0.991308142706477,
+      "grad_norm": 0.0,
+      "learning_rate": 3.962102493835218e-09,
+      "loss": 0.8604,
+      "step": 27258
+    },
+    {
+      "epoch": 0.9913445103102156,
+      "grad_norm": 0.0,
+      "learning_rate": 3.929018354538183e-09,
+      "loss": 0.918,
+      "step": 27259
+    },
+    {
+      "epoch": 0.9913808779139542,
+      "grad_norm": 0.0,
+      "learning_rate": 3.896072896544256e-09,
+      "loss": 0.8253,
+      "step": 27260
+    },
+    {
+      "epoch": 0.9914172455176928,
+      "grad_norm": 0.0,
+      "learning_rate": 3.863266120309739e-09,
+      "loss": 0.8889,
+      "step": 27261
+    },
+    {
+      "epoch": 0.9914536131214314,
+      "grad_norm": 0.0,
+      "learning_rate": 3.830598026288712e-09,
+      "loss": 0.9919,
+      "step": 27262
+    },
+    {
+      "epoch": 0.99148998072517,
+      "grad_norm": 0.0,
+      "learning_rate": 3.798068614936368e-09,
+      "loss": 0.977,
+      "step": 27263
+    },
+    {
+      "epoch": 0.9915263483289086,
+      "grad_norm": 0.0,
+      "learning_rate": 3.765677886702346e-09,
+      "loss": 0.921,
+      "step": 27264
+    },
+    {
+      "epoch": 0.9915627159326472,
+      "grad_norm": 0.0,
+      "learning_rate": 3.733425842037397e-09,
+      "loss": 0.9003,
+      "step": 27265
+    },
+    {
+      "epoch": 0.9915990835363858,
+      "grad_norm": 0.0,
+      "learning_rate": 3.701312481387831e-09,
+      "loss": 0.7846,
+      "step": 27266
+    },
+    {
+      "epoch": 0.9916354511401244,
+      "grad_norm": 0.0,
+      "learning_rate": 3.6693378051999574e-09,
+      "loss": 0.9077,
+      "step": 27267
+    },
+    {
+      "epoch": 0.991671818743863,
+      "grad_norm": 0.0,
+      "learning_rate": 3.637501813916755e-09,
+      "loss": 0.9039,
+      "step": 27268
+    },
+    {
+      "epoch": 0.9917081863476016,
+      "grad_norm": 0.0,
+      "learning_rate": 3.605804507980093e-09,
+      "loss": 0.871,
+      "step": 27269
+    },
+    {
+      "epoch": 0.9917445539513401,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5742458878307294e-09,
+      "loss": 0.7986,
+      "step": 27270
+    },
+    {
+      "epoch": 0.9917809215550787,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5428259539049826e-09,
+      "loss": 0.9553,
+      "step": 27271
+    },
+    {
+      "epoch": 0.9918172891588173,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5115447066402795e-09,
+      "loss": 0.9478,
+      "step": 27272
+    },
+    {
+      "epoch": 0.9918536567625559,
+      "grad_norm": 0.0,
+      "learning_rate": 3.480402146468498e-09,
+      "loss": 0.9159,
+      "step": 27273
+    },
+    {
+      "epoch": 0.9918900243662945,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4493982738237342e-09,
+      "loss": 0.8489,
+      "step": 27274
+    },
+    {
+      "epoch": 0.9919263919700331,
+      "grad_norm": 0.0,
+      "learning_rate": 3.4185330891345347e-09,
+      "loss": 0.8876,
+      "step": 27275
+    },
+    {
+      "epoch": 0.9919627595737717,
+      "grad_norm": 0.0,
+      "learning_rate": 3.387806592830556e-09,
+      "loss": 0.9657,
+      "step": 27276
+    },
+    {
+      "epoch": 0.9919991271775103,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3572187853370132e-09,
+      "loss": 0.8408,
+      "step": 27277
+    },
+    {
+      "epoch": 0.9920354947812489,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3267696670791215e-09,
+      "loss": 1.0617,
+      "step": 27278
+    },
+    {
+      "epoch": 0.9920718623849875,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2964592384787664e-09,
+      "loss": 0.8753,
+      "step": 27279
+    },
+    {
+      "epoch": 0.992108229988726,
+      "grad_norm": 0.0,
+      "learning_rate": 3.266287499956722e-09,
+      "loss": 0.9706,
+      "step": 27280
+    },
+    {
+      "epoch": 0.9921445975924646,
+      "grad_norm": 0.0,
+      "learning_rate": 3.2362544519304316e-09,
+      "loss": 0.8131,
+      "step": 27281
+    },
+    {
+      "epoch": 0.9921809651962032,
+      "grad_norm": 0.0,
+      "learning_rate": 3.20636009481845e-09,
+      "loss": 0.946,
+      "step": 27282
+    },
+    {
+      "epoch": 0.9922173327999418,
+      "grad_norm": 0.0,
+      "learning_rate": 3.176604429034891e-09,
+      "loss": 0.9701,
+      "step": 27283
+    },
+    {
+      "epoch": 0.9922537004036804,
+      "grad_norm": 0.0,
+      "learning_rate": 3.1469874549916455e-09,
+      "loss": 0.8982,
+      "step": 27284
+    },
+    {
+      "epoch": 0.992290068007419,
+      "grad_norm": 0.0,
+      "learning_rate": 3.117509173100608e-09,
+      "loss": 0.9039,
+      "step": 27285
+    },
+    {
+      "epoch": 0.9923264356111576,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0881695837703396e-09,
+      "loss": 0.8706,
+      "step": 27286
+    },
+    {
+      "epoch": 0.9923628032148961,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0589686874082926e-09,
+      "loss": 0.9334,
+      "step": 27287
+    },
+    {
+      "epoch": 0.9923991708186347,
+      "grad_norm": 0.0,
+      "learning_rate": 3.029906484418588e-09,
+      "loss": 0.8085,
+      "step": 27288
+    },
+    {
+      "epoch": 0.9924355384223733,
+      "grad_norm": 0.0,
+      "learning_rate": 3.0009829752053466e-09,
+      "loss": 0.8826,
+      "step": 27289
+    },
+    {
+      "epoch": 0.9924719060261119,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9721981601693597e-09,
+      "loss": 0.9955,
+      "step": 27290
+    },
+    {
+      "epoch": 0.9925082736298505,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9435520397103067e-09,
+      "loss": 0.9816,
+      "step": 27291
+    },
+    {
+      "epoch": 0.9925446412335891,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9150446142256484e-09,
+      "loss": 0.9525,
+      "step": 27292
+    },
+    {
+      "epoch": 0.9925810088373277,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8866758841106237e-09,
+      "loss": 0.9309,
+      "step": 27293
+    },
+    {
+      "epoch": 0.9926173764410663,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8584458497593613e-09,
+      "loss": 0.7957,
+      "step": 27294
+    },
+    {
+      "epoch": 0.9926537440448049,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8303545115626607e-09,
+      "loss": 0.9606,
+      "step": 27295
+    },
+    {
+      "epoch": 0.9926901116485435,
+      "grad_norm": 0.0,
+      "learning_rate": 2.8024018699113196e-09,
+      "loss": 0.9185,
+      "step": 27296
+    },
+    {
+      "epoch": 0.9927264792522821,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7745879251916964e-09,
+      "loss": 0.9517,
+      "step": 27297
+    },
+    {
+      "epoch": 0.9927628468560207,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7469126777901476e-09,
+      "loss": 0.9132,
+      "step": 27298
+    },
+    {
+      "epoch": 0.9927992144597593,
+      "grad_norm": 0.0,
+      "learning_rate": 2.719376128091922e-09,
+      "loss": 0.9096,
+      "step": 27299
+    },
+    {
+      "epoch": 0.9928355820634979,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6919782764778247e-09,
+      "loss": 0.9674,
+      "step": 27300
+    },
+    {
+      "epoch": 0.9928719496672365,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6647191233275525e-09,
+      "loss": 0.9157,
+      "step": 27301
+    },
+    {
+      "epoch": 0.9929083172709751,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6375986690208023e-09,
+      "loss": 0.9233,
+      "step": 27302
+    },
+    {
+      "epoch": 0.9929446848747137,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6106169139328285e-09,
+      "loss": 0.8683,
+      "step": 27303
+    },
+    {
+      "epoch": 0.9929810524784521,
+      "grad_norm": 0.0,
+      "learning_rate": 2.583773858437777e-09,
+      "loss": 0.9348,
+      "step": 27304
+    },
+    {
+      "epoch": 0.9930174200821907,
+      "grad_norm": 0.0,
+      "learning_rate": 2.557069502908682e-09,
+      "loss": 0.8213,
+      "step": 27305
+    },
+    {
+      "epoch": 0.9930537876859293,
+      "grad_norm": 0.0,
+      "learning_rate": 2.530503847715249e-09,
+      "loss": 0.9444,
+      "step": 27306
+    },
+    {
+      "epoch": 0.9930901552896679,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5040768932260705e-09,
+      "loss": 0.8117,
+      "step": 27307
+    },
+    {
+      "epoch": 0.9931265228934065,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4777886398097416e-09,
+      "loss": 0.9722,
+      "step": 27308
+    },
+    {
+      "epoch": 0.9931628904971451,
+      "grad_norm": 0.0,
+      "learning_rate": 2.451639087828195e-09,
+      "loss": 0.9846,
+      "step": 27309
+    },
+    {
+      "epoch": 0.9931992581008837,
+      "grad_norm": 0.0,
+      "learning_rate": 2.425628237645583e-09,
+      "loss": 0.8396,
+      "step": 27310
+    },
+    {
+      "epoch": 0.9932356257046223,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3997560896238394e-09,
+      "loss": 0.9513,
+      "step": 27311
+    },
+    {
+      "epoch": 0.9932719933083609,
+      "grad_norm": 0.0,
+      "learning_rate": 2.3740226441193448e-09,
+      "loss": 0.8131,
+      "step": 27312
+    },
+    {
+      "epoch": 0.9933083609120995,
+      "grad_norm": 0.0,
+      "learning_rate": 2.348427901490702e-09,
+      "loss": 0.8537,
+      "step": 27313
+    },
+    {
+      "epoch": 0.9933447285158381,
+      "grad_norm": 0.0,
+      "learning_rate": 2.322971862094292e-09,
+      "loss": 0.9207,
+      "step": 27314
+    },
+    {
+      "epoch": 0.9933810961195767,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2976545262798354e-09,
+      "loss": 0.9523,
+      "step": 27315
+    },
+    {
+      "epoch": 0.9934174637233153,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2724758944014935e-09,
+      "loss": 0.8437,
+      "step": 27316
+    },
+    {
+      "epoch": 0.9934538313270539,
+      "grad_norm": 0.0,
+      "learning_rate": 2.247435966807876e-09,
+      "loss": 0.7702,
+      "step": 27317
+    },
+    {
+      "epoch": 0.9934901989307925,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2225347438453726e-09,
+      "loss": 0.8409,
+      "step": 27318
+    },
+    {
+      "epoch": 0.9935265665345311,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1977722258614832e-09,
+      "loss": 0.8831,
+      "step": 27319
+    },
+    {
+      "epoch": 0.9935629341382697,
+      "grad_norm": 0.0,
+      "learning_rate": 2.173148413197046e-09,
+      "loss": 0.9619,
+      "step": 27320
+    },
+    {
+      "epoch": 0.9935993017420082,
+      "grad_norm": 0.0,
+      "learning_rate": 2.14866330619623e-09,
+      "loss": 0.7842,
+      "step": 27321
+    },
+    {
+      "epoch": 0.9936356693457468,
+      "grad_norm": 0.0,
+      "learning_rate": 2.124316905196544e-09,
+      "loss": 0.9456,
+      "step": 27322
+    },
+    {
+      "epoch": 0.9936720369494854,
+      "grad_norm": 0.0,
+      "learning_rate": 2.100109210537715e-09,
+      "loss": 0.9388,
+      "step": 27323
+    },
+    {
+      "epoch": 0.993708404553224,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0760402225539213e-09,
+      "loss": 0.8966,
+      "step": 27324
+    },
+    {
+      "epoch": 0.9937447721569626,
+      "grad_norm": 0.0,
+      "learning_rate": 2.052109941580449e-09,
+      "loss": 0.7859,
+      "step": 27325
+    },
+    {
+      "epoch": 0.9937811397607011,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0283183679492557e-09,
+      "loss": 0.9801,
+      "step": 27326
+    },
+    {
+      "epoch": 0.9938175073644397,
+      "grad_norm": 0.0,
+      "learning_rate": 2.004665501988967e-09,
+      "loss": 0.8833,
+      "step": 27327
+    },
+    {
+      "epoch": 0.9938538749681783,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9811513440282092e-09,
+      "loss": 0.8511,
+      "step": 27328
+    },
+    {
+      "epoch": 0.9938902425719169,
+      "grad_norm": 0.0,
+      "learning_rate": 1.957775894394498e-09,
+      "loss": 0.8996,
+      "step": 27329
+    },
+    {
+      "epoch": 0.9939266101756555,
+      "grad_norm": 0.0,
+      "learning_rate": 1.934539153410908e-09,
+      "loss": 0.9754,
+      "step": 27330
+    },
+    {
+      "epoch": 0.9939629777793941,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9114411214005146e-09,
+      "loss": 0.9153,
+      "step": 27331
+    },
+    {
+      "epoch": 0.9939993453831327,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8884817986830617e-09,
+      "loss": 0.9024,
+      "step": 27332
+    },
+    {
+      "epoch": 0.9940357129868713,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8656611855782937e-09,
+      "loss": 0.8996,
+      "step": 27333
+    },
+    {
+      "epoch": 0.9940720805906099,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8429792824015136e-09,
+      "loss": 1.0499,
+      "step": 27334
+    },
+    {
+      "epoch": 0.9941084481943485,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8204360894680252e-09,
+      "loss": 0.9062,
+      "step": 27335
+    },
+    {
+      "epoch": 0.9941448157980871,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7980316070898007e-09,
+      "loss": 0.7631,
+      "step": 27336
+    },
+    {
+      "epoch": 0.9941811834018257,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7757658355788131e-09,
+      "loss": 0.8611,
+      "step": 27337
+    },
+    {
+      "epoch": 0.9942175510055642,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7536387752425943e-09,
+      "loss": 0.8292,
+      "step": 27338
+    },
+    {
+      "epoch": 0.9942539186093028,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7316504263897859e-09,
+      "loss": 0.8206,
+      "step": 27339
+    },
+    {
+      "epoch": 0.9942902862130414,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7098007893245893e-09,
+      "loss": 1.0493,
+      "step": 27340
+    },
+    {
+      "epoch": 0.99432665381678,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6880898643500954e-09,
+      "loss": 0.9389,
+      "step": 27341
+    },
+    {
+      "epoch": 0.9943630214205186,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6665176517682847e-09,
+      "loss": 0.8596,
+      "step": 27342
+    },
+    {
+      "epoch": 0.9943993890242572,
+      "grad_norm": 0.0,
+      "learning_rate": 1.645084151876697e-09,
+      "loss": 0.968,
+      "step": 27343
+    },
+    {
+      "epoch": 0.9944357566279958,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6237893649739823e-09,
+      "loss": 0.9599,
+      "step": 27344
+    },
+    {
+      "epoch": 0.9944721242317344,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6026332913554598e-09,
+      "loss": 0.9485,
+      "step": 27345
+    },
+    {
+      "epoch": 0.994508491835473,
+      "grad_norm": 0.0,
+      "learning_rate": 1.581615931315339e-09,
+      "loss": 0.9092,
+      "step": 27346
+    },
+    {
+      "epoch": 0.9945448594392116,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5607372851433878e-09,
+      "loss": 0.7914,
+      "step": 27347
+    },
+    {
+      "epoch": 0.9945812270429502,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5399973531304845e-09,
+      "loss": 0.8885,
+      "step": 27348
+    },
+    {
+      "epoch": 0.9946175946466888,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5193961355641773e-09,
+      "loss": 0.9868,
+      "step": 27349
+    },
+    {
+      "epoch": 0.9946539622504273,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4989336327297932e-09,
+      "loss": 0.9673,
+      "step": 27350
+    },
+    {
+      "epoch": 0.9946903298541659,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4786098449126596e-09,
+      "loss": 0.9778,
+      "step": 27351
+    },
+    {
+      "epoch": 0.9947266974579045,
+      "grad_norm": 0.0,
+      "learning_rate": 1.458424772393663e-09,
+      "loss": 0.8522,
+      "step": 27352
+    },
+    {
+      "epoch": 0.9947630650616431,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4383784154525793e-09,
+      "loss": 0.8103,
+      "step": 27353
+    },
+    {
+      "epoch": 0.9947994326653817,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4184707743680748e-09,
+      "loss": 0.9454,
+      "step": 27354
+    },
+    {
+      "epoch": 0.9948358002691202,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3987018494154847e-09,
+      "loss": 0.841,
+      "step": 27355
+    },
+    {
+      "epoch": 0.9948721678728588,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3790716408701443e-09,
+      "loss": 0.8672,
+      "step": 27356
+    },
+    {
+      "epoch": 0.9949085354765974,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3595801490040582e-09,
+      "loss": 0.8928,
+      "step": 27357
+    },
+    {
+      "epoch": 0.994944903080336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.340227374088121e-09,
+      "loss": 0.8919,
+      "step": 27358
+    },
+    {
+      "epoch": 0.9949812706840746,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3210133163898964e-09,
+      "loss": 0.9169,
+      "step": 27359
+    },
+    {
+      "epoch": 0.9950176382878132,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3019379761758378e-09,
+      "loss": 0.8725,
+      "step": 27360
+    },
+    {
+      "epoch": 0.9950540058915518,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2830013537112883e-09,
+      "loss": 0.8733,
+      "step": 27361
+    },
+    {
+      "epoch": 0.9950903734952904,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2642034492593713e-09,
+      "loss": 0.8515,
+      "step": 27362
+    },
+    {
+      "epoch": 0.995126741099029,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2455442630798786e-09,
+      "loss": 0.9611,
+      "step": 27363
+    },
+    {
+      "epoch": 0.9951631087027676,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2270237954314924e-09,
+      "loss": 0.8993,
+      "step": 27364
+    },
+    {
+      "epoch": 0.9951994763065062,
+      "grad_norm": 0.0,
+      "learning_rate": 1.2086420465728943e-09,
+      "loss": 0.8583,
+      "step": 27365
+    },
+    {
+      "epoch": 0.9952358439102448,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1903990167572156e-09,
+      "loss": 0.9487,
+      "step": 27366
+    },
+    {
+      "epoch": 0.9952722115139834,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1722947062386968e-09,
+      "loss": 0.937,
+      "step": 27367
+    },
+    {
+      "epoch": 0.995308579117722,
+      "grad_norm": 0.0,
+      "learning_rate": 1.154329115268249e-09,
+      "loss": 0.7491,
+      "step": 27368
+    },
+    {
+      "epoch": 0.9953449467214606,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1365022440945616e-09,
+      "loss": 0.9106,
+      "step": 27369
+    },
+    {
+      "epoch": 0.9953813143251992,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1188140929652146e-09,
+      "loss": 0.9039,
+      "step": 27370
+    },
+    {
+      "epoch": 0.9954176819289376,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1012646621255674e-09,
+      "loss": 1.021,
+      "step": 27371
+    },
+    {
+      "epoch": 0.9954540495326762,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0838539518198686e-09,
+      "loss": 0.8373,
+      "step": 27372
+    },
+    {
+      "epoch": 0.9954904171364148,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0665819622890372e-09,
+      "loss": 0.9669,
+      "step": 27373
+    },
+    {
+      "epoch": 0.9955267847401534,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0494486937728809e-09,
+      "loss": 1.0885,
+      "step": 27374
+    },
+    {
+      "epoch": 0.995563152343892,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0324541465089876e-09,
+      "loss": 0.8265,
+      "step": 27375
+    },
+    {
+      "epoch": 0.9955995199476306,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0155983207327247e-09,
+      "loss": 0.9609,
+      "step": 27376
+    },
+    {
+      "epoch": 0.9956358875513692,
+      "grad_norm": 0.0,
+      "learning_rate": 9.98881216678349e-10,
+      "loss": 0.9011,
+      "step": 27377
+    },
+    {
+      "epoch": 0.9956722551551078,
+      "grad_norm": 0.0,
+      "learning_rate": 9.823028345778973e-10,
+      "loss": 0.8329,
+      "step": 27378
+    },
+    {
+      "epoch": 0.9957086227588464,
+      "grad_norm": 0.0,
+      "learning_rate": 9.658631746611858e-10,
+      "loss": 0.9275,
+      "step": 27379
+    },
+    {
+      "epoch": 0.995744990362585,
+      "grad_norm": 0.0,
+      "learning_rate": 9.4956223715581e-10,
+      "loss": 0.8573,
+      "step": 27380
+    },
+    {
+      "epoch": 0.9957813579663236,
+      "grad_norm": 0.0,
+      "learning_rate": 9.334000222893658e-10,
+      "loss": 0.9565,
+      "step": 27381
+    },
+    {
+      "epoch": 0.9958177255700622,
+      "grad_norm": 0.0,
+      "learning_rate": 9.173765302850079e-10,
+      "loss": 0.8533,
+      "step": 27382
+    },
+    {
+      "epoch": 0.9958540931738008,
+      "grad_norm": 0.0,
+      "learning_rate": 9.014917613647812e-10,
+      "loss": 0.872,
+      "step": 27383
+    },
+    {
+      "epoch": 0.9958904607775394,
+      "grad_norm": 0.0,
+      "learning_rate": 8.857457157496196e-10,
+      "loss": 0.9193,
+      "step": 27384
+    },
+    {
+      "epoch": 0.995926828381278,
+      "grad_norm": 0.0,
+      "learning_rate": 8.701383936582375e-10,
+      "loss": 0.8955,
+      "step": 27385
+    },
+    {
+      "epoch": 0.9959631959850166,
+      "grad_norm": 0.0,
+      "learning_rate": 8.546697953071282e-10,
+      "loss": 0.8611,
+      "step": 27386
+    },
+    {
+      "epoch": 0.9959995635887552,
+      "grad_norm": 0.0,
+      "learning_rate": 8.393399209094543e-10,
+      "loss": 0.96,
+      "step": 27387
+    },
+    {
+      "epoch": 0.9960359311924937,
+      "grad_norm": 0.0,
+      "learning_rate": 8.241487706794893e-10,
+      "loss": 1.015,
+      "step": 27388
+    },
+    {
+      "epoch": 0.9960722987962323,
+      "grad_norm": 0.0,
+      "learning_rate": 8.090963448270651e-10,
+      "loss": 0.9182,
+      "step": 27389
+    },
+    {
+      "epoch": 0.9961086663999709,
+      "grad_norm": 0.0,
+      "learning_rate": 7.941826435620136e-10,
+      "loss": 0.9848,
+      "step": 27390
+    },
+    {
+      "epoch": 0.9961450340037095,
+      "grad_norm": 0.0,
+      "learning_rate": 7.794076670908369e-10,
+      "loss": 0.9531,
+      "step": 27391
+    },
+    {
+      "epoch": 0.9961814016074481,
+      "grad_norm": 0.0,
+      "learning_rate": 7.647714156178154e-10,
+      "loss": 0.8734,
+      "step": 27392
+    },
+    {
+      "epoch": 0.9962177692111867,
+      "grad_norm": 0.0,
+      "learning_rate": 7.502738893472305e-10,
+      "loss": 0.9532,
+      "step": 27393
+    },
+    {
+      "epoch": 0.9962541368149253,
+      "grad_norm": 0.0,
+      "learning_rate": 7.359150884789223e-10,
+      "loss": 0.9062,
+      "step": 27394
+    },
+    {
+      "epoch": 0.9962905044186638,
+      "grad_norm": 0.0,
+      "learning_rate": 7.216950132127309e-10,
+      "loss": 0.9849,
+      "step": 27395
+    },
+    {
+      "epoch": 0.9963268720224024,
+      "grad_norm": 0.0,
+      "learning_rate": 7.076136637451659e-10,
+      "loss": 0.9347,
+      "step": 27396
+    },
+    {
+      "epoch": 0.996363239626141,
+      "grad_norm": 0.0,
+      "learning_rate": 6.936710402738466e-10,
+      "loss": 0.9684,
+      "step": 27397
+    },
+    {
+      "epoch": 0.9963996072298796,
+      "grad_norm": 0.0,
+      "learning_rate": 6.798671429897319e-10,
+      "loss": 0.8952,
+      "step": 27398
+    },
+    {
+      "epoch": 0.9964359748336182,
+      "grad_norm": 0.0,
+      "learning_rate": 6.6620197208489e-10,
+      "loss": 0.9236,
+      "step": 27399
+    },
+    {
+      "epoch": 0.9964723424373568,
+      "grad_norm": 0.0,
+      "learning_rate": 6.526755277502795e-10,
+      "loss": 1.0256,
+      "step": 27400
+    },
+    {
+      "epoch": 0.9965087100410954,
+      "grad_norm": 0.0,
+      "learning_rate": 6.392878101724176e-10,
+      "loss": 0.8916,
+      "step": 27401
+    },
+    {
+      "epoch": 0.996545077644834,
+      "grad_norm": 0.0,
+      "learning_rate": 6.260388195367117e-10,
+      "loss": 0.8283,
+      "step": 27402
+    },
+    {
+      "epoch": 0.9965814452485726,
+      "grad_norm": 0.0,
+      "learning_rate": 6.12928556027459e-10,
+      "loss": 0.7574,
+      "step": 27403
+    },
+    {
+      "epoch": 0.9966178128523112,
+      "grad_norm": 0.0,
+      "learning_rate": 5.999570198267357e-10,
+      "loss": 0.9379,
+      "step": 27404
+    },
+    {
+      "epoch": 0.9966541804560497,
+      "grad_norm": 0.0,
+      "learning_rate": 5.871242111143982e-10,
+      "loss": 0.8943,
+      "step": 27405
+    },
+    {
+      "epoch": 0.9966905480597883,
+      "grad_norm": 0.0,
+      "learning_rate": 5.744301300691923e-10,
+      "loss": 0.8941,
+      "step": 27406
+    },
+    {
+      "epoch": 0.9967269156635269,
+      "grad_norm": 0.0,
+      "learning_rate": 5.618747768654231e-10,
+      "loss": 0.9082,
+      "step": 27407
+    },
+    {
+      "epoch": 0.9967632832672655,
+      "grad_norm": 0.0,
+      "learning_rate": 5.494581516785058e-10,
+      "loss": 0.9959,
+      "step": 27408
+    },
+    {
+      "epoch": 0.9967996508710041,
+      "grad_norm": 0.0,
+      "learning_rate": 5.37180254680525e-10,
+      "loss": 0.9355,
+      "step": 27409
+    },
+    {
+      "epoch": 0.9968360184747427,
+      "grad_norm": 0.0,
+      "learning_rate": 5.250410860413447e-10,
+      "loss": 0.8946,
+      "step": 27410
+    },
+    {
+      "epoch": 0.9968723860784813,
+      "grad_norm": 0.0,
+      "learning_rate": 5.13040645930829e-10,
+      "loss": 0.8692,
+      "step": 27411
+    },
+    {
+      "epoch": 0.9969087536822199,
+      "grad_norm": 0.0,
+      "learning_rate": 5.011789345132911e-10,
+      "loss": 0.9437,
+      "step": 27412
+    },
+    {
+      "epoch": 0.9969451212859585,
+      "grad_norm": 0.0,
+      "learning_rate": 4.894559519552644e-10,
+      "loss": 0.9252,
+      "step": 27413
+    },
+    {
+      "epoch": 0.9969814888896971,
+      "grad_norm": 0.0,
+      "learning_rate": 4.778716984177311e-10,
+      "loss": 0.8099,
+      "step": 27414
+    },
+    {
+      "epoch": 0.9970178564934357,
+      "grad_norm": 0.0,
+      "learning_rate": 4.664261740638942e-10,
+      "loss": 0.8756,
+      "step": 27415
+    },
+    {
+      "epoch": 0.9970542240971743,
+      "grad_norm": 0.0,
+      "learning_rate": 4.551193790491848e-10,
+      "loss": 0.895,
+      "step": 27416
+    },
+    {
+      "epoch": 0.9970905917009129,
+      "grad_norm": 0.0,
+      "learning_rate": 4.4395131353347497e-10,
+      "loss": 0.9523,
+      "step": 27417
+    },
+    {
+      "epoch": 0.9971269593046515,
+      "grad_norm": 0.0,
+      "learning_rate": 4.3292197766997554e-10,
+      "loss": 0.8848,
+      "step": 27418
+    },
+    {
+      "epoch": 0.99716332690839,
+      "grad_norm": 0.0,
+      "learning_rate": 4.2203137161189733e-10,
+      "loss": 0.998,
+      "step": 27419
+    },
+    {
+      "epoch": 0.9971996945121286,
+      "grad_norm": 0.0,
+      "learning_rate": 4.112794955102306e-10,
+      "loss": 0.9358,
+      "step": 27420
+    },
+    {
+      "epoch": 0.9972360621158672,
+      "grad_norm": 0.0,
+      "learning_rate": 4.006663495148555e-10,
+      "loss": 1.0469,
+      "step": 27421
+    },
+    {
+      "epoch": 0.9972724297196057,
+      "grad_norm": 0.0,
+      "learning_rate": 3.901919337734317e-10,
+      "loss": 0.8778,
+      "step": 27422
+    },
+    {
+      "epoch": 0.9973087973233443,
+      "grad_norm": 0.0,
+      "learning_rate": 3.798562484291779e-10,
+      "loss": 0.9936,
+      "step": 27423
+    },
+    {
+      "epoch": 0.9973451649270829,
+      "grad_norm": 0.0,
+      "learning_rate": 3.696592936275334e-10,
+      "loss": 0.809,
+      "step": 27424
+    },
+    {
+      "epoch": 0.9973815325308215,
+      "grad_norm": 0.0,
+      "learning_rate": 3.5960106950949646e-10,
+      "loss": 1.0352,
+      "step": 27425
+    },
+    {
+      "epoch": 0.9974179001345601,
+      "grad_norm": 0.0,
+      "learning_rate": 3.49681576213845e-10,
+      "loss": 0.968,
+      "step": 27426
+    },
+    {
+      "epoch": 0.9974542677382987,
+      "grad_norm": 0.0,
+      "learning_rate": 3.399008138793569e-10,
+      "loss": 0.7843,
+      "step": 27427
+    },
+    {
+      "epoch": 0.9974906353420373,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3025878264036915e-10,
+      "loss": 0.7874,
+      "step": 27428
+    },
+    {
+      "epoch": 0.9975270029457759,
+      "grad_norm": 0.0,
+      "learning_rate": 3.207554826323289e-10,
+      "loss": 0.8674,
+      "step": 27429
+    },
+    {
+      "epoch": 0.9975633705495145,
+      "grad_norm": 0.0,
+      "learning_rate": 3.113909139851323e-10,
+      "loss": 0.8635,
+      "step": 27430
+    },
+    {
+      "epoch": 0.9975997381532531,
+      "grad_norm": 0.0,
+      "learning_rate": 3.021650768297857e-10,
+      "loss": 0.9334,
+      "step": 27431
+    },
+    {
+      "epoch": 0.9976361057569917,
+      "grad_norm": 0.0,
+      "learning_rate": 2.9307797129507485e-10,
+      "loss": 0.978,
+      "step": 27432
+    },
+    {
+      "epoch": 0.9976724733607303,
+      "grad_norm": 0.0,
+      "learning_rate": 2.841295975053449e-10,
+      "loss": 0.7867,
+      "step": 27433
+    },
+    {
+      "epoch": 0.9977088409644689,
+      "grad_norm": 0.0,
+      "learning_rate": 2.753199555849406e-10,
+      "loss": 0.8673,
+      "step": 27434
+    },
+    {
+      "epoch": 0.9977452085682075,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6664904565820715e-10,
+      "loss": 0.9623,
+      "step": 27435
+    },
+    {
+      "epoch": 0.9977815761719461,
+      "grad_norm": 0.0,
+      "learning_rate": 2.581168678428281e-10,
+      "loss": 0.8368,
+      "step": 27436
+    },
+    {
+      "epoch": 0.9978179437756847,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4972342225870747e-10,
+      "loss": 0.8305,
+      "step": 27437
+    },
+    {
+      "epoch": 0.9978543113794233,
+      "grad_norm": 0.0,
+      "learning_rate": 2.414687090224188e-10,
+      "loss": 0.8422,
+      "step": 27438
+    },
+    {
+      "epoch": 0.9978906789831617,
+      "grad_norm": 0.0,
+      "learning_rate": 2.333527282472048e-10,
+      "loss": 0.9788,
+      "step": 27439
+    },
+    {
+      "epoch": 0.9979270465869003,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2537548004630816e-10,
+      "loss": 0.8673,
+      "step": 27440
+    },
+    {
+      "epoch": 0.9979634141906389,
+      "grad_norm": 0.0,
+      "learning_rate": 2.1753696453075125e-10,
+      "loss": 0.9479,
+      "step": 27441
+    },
+    {
+      "epoch": 0.9979997817943775,
+      "grad_norm": 0.0,
+      "learning_rate": 2.098371818093359e-10,
+      "loss": 0.8472,
+      "step": 27442
+    },
+    {
+      "epoch": 0.9980361493981161,
+      "grad_norm": 0.0,
+      "learning_rate": 2.022761319886435e-10,
+      "loss": 0.9606,
+      "step": 27443
+    },
+    {
+      "epoch": 0.9980725170018547,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9485381517303503e-10,
+      "loss": 0.9062,
+      "step": 27444
+    },
+    {
+      "epoch": 0.9981088846055933,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8757023146576125e-10,
+      "loss": 0.9348,
+      "step": 27445
+    },
+    {
+      "epoch": 0.9981452522093319,
+      "grad_norm": 0.0,
+      "learning_rate": 1.8042538096896266e-10,
+      "loss": 0.9472,
+      "step": 27446
+    },
+    {
+      "epoch": 0.9981816198130705,
+      "grad_norm": 0.0,
+      "learning_rate": 1.734192637803389e-10,
+      "loss": 0.9379,
+      "step": 27447
+    },
+    {
+      "epoch": 0.9982179874168091,
+      "grad_norm": 0.0,
+      "learning_rate": 1.665518799975896e-10,
+      "loss": 0.9762,
+      "step": 27448
+    },
+    {
+      "epoch": 0.9982543550205477,
+      "grad_norm": 0.0,
+      "learning_rate": 1.598232297161939e-10,
+      "loss": 0.7963,
+      "step": 27449
+    },
+    {
+      "epoch": 0.9982907226242863,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5323331302941058e-10,
+      "loss": 0.9323,
+      "step": 27450
+    },
+    {
+      "epoch": 0.9983270902280249,
+      "grad_norm": 0.0,
+      "learning_rate": 1.467821300282779e-10,
+      "loss": 0.892,
+      "step": 27451
+    },
+    {
+      "epoch": 0.9983634578317635,
+      "grad_norm": 0.0,
+      "learning_rate": 1.4046968080272394e-10,
+      "loss": 0.9335,
+      "step": 27452
+    },
+    {
+      "epoch": 0.9983998254355021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3429596544045631e-10,
+      "loss": 1.0096,
+      "step": 27453
+    },
+    {
+      "epoch": 0.9984361930392407,
+      "grad_norm": 0.0,
+      "learning_rate": 1.282609840269622e-10,
+      "loss": 0.7919,
+      "step": 27454
+    },
+    {
+      "epoch": 0.9984725606429792,
+      "grad_norm": 0.0,
+      "learning_rate": 1.223647366455083e-10,
+      "loss": 0.7917,
+      "step": 27455
+    },
+    {
+      "epoch": 0.9985089282467178,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1660722337825114e-10,
+      "loss": 0.929,
+      "step": 27456
+    },
+    {
+      "epoch": 0.9985452958504564,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1098844430512678e-10,
+      "loss": 1.0261,
+      "step": 27457
+    },
+    {
+      "epoch": 0.998581663454195,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0550839950385084e-10,
+      "loss": 0.9132,
+      "step": 27458
+    },
+    {
+      "epoch": 0.9986180310579336,
+      "grad_norm": 0.0,
+      "learning_rate": 1.0016708905102868e-10,
+      "loss": 0.7967,
+      "step": 27459
+    },
+    {
+      "epoch": 0.9986543986616722,
+      "grad_norm": 0.0,
+      "learning_rate": 9.496451301993503e-11,
+      "loss": 0.7986,
+      "step": 27460
+    },
+    {
+      "epoch": 0.9986907662654108,
+      "grad_norm": 0.0,
+      "learning_rate": 8.990067148384463e-11,
+      "loss": 0.9256,
+      "step": 27461
+    },
+    {
+      "epoch": 0.9987271338691494,
+      "grad_norm": 0.0,
+      "learning_rate": 8.497556451159128e-11,
+      "loss": 1.0734,
+      "step": 27462
+    },
+    {
+      "epoch": 0.998763501472888,
+      "grad_norm": 0.0,
+      "learning_rate": 8.018919217311905e-11,
+      "loss": 1.1017,
+      "step": 27463
+    },
+    {
+      "epoch": 0.9987998690766265,
+      "grad_norm": 0.0,
+      "learning_rate": 7.554155453393109e-11,
+      "loss": 0.9713,
+      "step": 27464
+    },
+    {
+      "epoch": 0.9988362366803651,
+      "grad_norm": 0.0,
+      "learning_rate": 7.103265165842032e-11,
+      "loss": 1.0049,
+      "step": 27465
+    },
+    {
+      "epoch": 0.9988726042841037,
+      "grad_norm": 0.0,
+      "learning_rate": 6.666248360875927e-11,
+      "loss": 0.9483,
+      "step": 27466
+    },
+    {
+      "epoch": 0.9989089718878423,
+      "grad_norm": 0.0,
+      "learning_rate": 6.243105044601016e-11,
+      "loss": 0.8193,
+      "step": 27467
+    },
+    {
+      "epoch": 0.9989453394915809,
+      "grad_norm": 0.0,
+      "learning_rate": 5.833835223012508e-11,
+      "loss": 0.9624,
+      "step": 27468
+    },
+    {
+      "epoch": 0.9989817070953195,
+      "grad_norm": 0.0,
+      "learning_rate": 5.438438901550491e-11,
+      "loss": 0.8292,
+      "step": 27469
+    },
+    {
+      "epoch": 0.9990180746990581,
+      "grad_norm": 0.0,
+      "learning_rate": 5.056916085877106e-11,
+      "loss": 0.785,
+      "step": 27470
+    },
+    {
+      "epoch": 0.9990544423027967,
+      "grad_norm": 0.0,
+      "learning_rate": 4.689266781321422e-11,
+      "loss": 0.939,
+      "step": 27471
+    },
+    {
+      "epoch": 0.9990908099065352,
+      "grad_norm": 0.0,
+      "learning_rate": 4.335490992768421e-11,
+      "loss": 0.9615,
+      "step": 27472
+    },
+    {
+      "epoch": 0.9991271775102738,
+      "grad_norm": 0.0,
+      "learning_rate": 3.995588725325128e-11,
+      "loss": 1.0252,
+      "step": 27473
+    },
+    {
+      "epoch": 0.9991635451140124,
+      "grad_norm": 0.0,
+      "learning_rate": 3.669559983543458e-11,
+      "loss": 0.8977,
+      "step": 27474
+    },
+    {
+      "epoch": 0.999199912717751,
+      "grad_norm": 0.0,
+      "learning_rate": 3.357404772086348e-11,
+      "loss": 1.063,
+      "step": 27475
+    },
+    {
+      "epoch": 0.9992362803214896,
+      "grad_norm": 0.0,
+      "learning_rate": 3.059123095283667e-11,
+      "loss": 0.8502,
+      "step": 27476
+    },
+    {
+      "epoch": 0.9992726479252282,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7747149571322185e-11,
+      "loss": 0.9259,
+      "step": 27477
+    },
+    {
+      "epoch": 0.9993090155289668,
+      "grad_norm": 0.0,
+      "learning_rate": 2.5041803617398276e-11,
+      "loss": 0.9555,
+      "step": 27478
+    },
+    {
+      "epoch": 0.9993453831327054,
+      "grad_norm": 0.0,
+      "learning_rate": 2.2475193127702298e-11,
+      "loss": 0.9517,
+      "step": 27479
+    },
+    {
+      "epoch": 0.999381750736444,
+      "grad_norm": 0.0,
+      "learning_rate": 2.0047318137761397e-11,
+      "loss": 0.8338,
+      "step": 27480
+    },
+    {
+      "epoch": 0.9994181183401826,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7758178681992476e-11,
+      "loss": 0.9263,
+      "step": 27481
+    },
+    {
+      "epoch": 0.9994544859439212,
+      "grad_norm": 0.0,
+      "learning_rate": 1.5607774791481788e-11,
+      "loss": 0.9316,
+      "step": 27482
+    },
+    {
+      "epoch": 0.9994908535476598,
+      "grad_norm": 0.0,
+      "learning_rate": 1.3596106496205352e-11,
+      "loss": 0.8595,
+      "step": 27483
+    },
+    {
+      "epoch": 0.9995272211513984,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1723173825028966e-11,
+      "loss": 0.8812,
+      "step": 27484
+    },
+    {
+      "epoch": 0.999563588755137,
+      "grad_norm": 0.0,
+      "learning_rate": 9.988976802377537e-12,
+      "loss": 0.8648,
+      "step": 27485
+    },
+    {
+      "epoch": 0.9995999563588756,
+      "grad_norm": 0.0,
+      "learning_rate": 8.393515452675971e-12,
+      "loss": 0.9586,
+      "step": 27486
+    },
+    {
+      "epoch": 0.9996363239626141,
+      "grad_norm": 0.0,
+      "learning_rate": 6.9367897992389525e-12,
+      "loss": 0.9505,
+      "step": 27487
+    },
+    {
+      "epoch": 0.9996726915663527,
+      "grad_norm": 0.0,
+      "learning_rate": 5.6187998609402716e-12,
+      "loss": 0.902,
+      "step": 27488
+    },
+    {
+      "epoch": 0.9997090591700912,
+      "grad_norm": 0.0,
+      "learning_rate": 4.439545655543498e-12,
+      "loss": 0.9914,
+      "step": 27489
+    },
+    {
+      "epoch": 0.9997454267738298,
+      "grad_norm": 0.0,
+      "learning_rate": 3.3990272019224226e-12,
+      "loss": 0.8977,
+      "step": 27490
+    },
+    {
+      "epoch": 0.9997817943775684,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4972445122894984e-12,
+      "loss": 0.9756,
+      "step": 27491
+    },
+    {
+      "epoch": 0.999818161981307,
+      "grad_norm": 0.0,
+      "learning_rate": 1.7341975999674021e-12,
+      "loss": 0.895,
+      "step": 27492
+    },
+    {
+      "epoch": 0.9998545295850456,
+      "grad_norm": 0.0,
+      "learning_rate": 1.1098864760583639e-12,
+      "loss": 0.9635,
+      "step": 27493
+    },
+    {
+      "epoch": 0.9998908971887842,
+      "grad_norm": 0.0,
+      "learning_rate": 6.243111472237218e-13,
+      "loss": 0.9478,
+      "step": 27494
+    },
+    {
+      "epoch": 0.9999272647925228,
+      "grad_norm": 0.0,
+      "learning_rate": 2.7747162234526006e-13,
+      "loss": 1.0227,
+      "step": 27495
+    },
+    {
+      "epoch": 0.9999636323962614,
+      "grad_norm": 0.0,
+      "learning_rate": 6.936790586387076e-14,
+      "loss": 0.924,
+      "step": 27496
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.8743,
+      "step": 27497
+    },
+    {
+      "epoch": 1.0,
+      "step": 27497,
+      "total_flos": 5.023469194434734e+19,
+      "train_loss": 1.027738457658696,
+      "train_runtime": 185450.2168,
+      "train_samples_per_second": 18.978,
+      "train_steps_per_second": 0.148
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 27497,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 3000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.023469194434734e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}