diff --git "a/run-4/checkpoint-620/trainer_state.json" "b/run-4/checkpoint-620/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/run-4/checkpoint-620/trainer_state.json"
@@ -0,0 +1,4440 @@
+{
+  "best_metric": 1.9000149676084739,
+  "best_model_checkpoint": "./modernBERT-content-regression/run-4/checkpoint-620",
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 620,
+  "is_hyper_param_search": true,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008064516129032258,
+      "grad_norm": 312.8161315917969,
+      "learning_rate": 2.305260772966439e-06,
+      "loss": 21.3087,
+      "step": 1
+    },
+    {
+      "epoch": 0.016129032258064516,
+      "grad_norm": 88.59754180908203,
+      "learning_rate": 2.3015366037047808e-06,
+      "loss": 1.6705,
+      "step": 2
+    },
+    {
+      "epoch": 0.024193548387096774,
+      "grad_norm": 525.8513793945312,
+      "learning_rate": 2.2978124344431226e-06,
+      "loss": 151.777,
+      "step": 3
+    },
+    {
+      "epoch": 0.03225806451612903,
+      "grad_norm": 207.6021270751953,
+      "learning_rate": 2.2940882651814644e-06,
+      "loss": 16.9986,
+      "step": 4
+    },
+    {
+      "epoch": 0.04032258064516129,
+      "grad_norm": 95.76493072509766,
+      "learning_rate": 2.2903640959198062e-06,
+      "loss": 3.2381,
+      "step": 5
+    },
+    {
+      "epoch": 0.04838709677419355,
+      "grad_norm": 26.680326461791992,
+      "learning_rate": 2.286639926658148e-06,
+      "loss": 0.3182,
+      "step": 6
+    },
+    {
+      "epoch": 0.056451612903225805,
+      "grad_norm": 134.4029083251953,
+      "learning_rate": 2.28291575739649e-06,
+      "loss": 7.6237,
+      "step": 7
+    },
+    {
+      "epoch": 0.06451612903225806,
+      "grad_norm": 299.9959716796875,
+      "learning_rate": 2.2791915881348312e-06,
+      "loss": 16.5691,
+      "step": 8
+    },
+    {
+      "epoch": 0.07258064516129033,
+      "grad_norm": 8.101390838623047,
+      "learning_rate": 2.275467418873173e-06,
+      "loss": 0.0379,
+      "step": 9
+    },
+    {
+      "epoch": 0.08064516129032258,
+      "grad_norm": 109.74885559082031,
+      "learning_rate": 2.271743249611515e-06,
+      "loss": 3.1775,
+      "step": 10
+    },
+    {
+      "epoch": 0.08870967741935484,
+      "grad_norm": 204.84671020507812,
+      "learning_rate": 2.2680190803498567e-06,
+      "loss": 14.3084,
+      "step": 11
+    },
+    {
+      "epoch": 0.0967741935483871,
+      "grad_norm": 97.9755859375,
+      "learning_rate": 2.2642949110881985e-06,
+      "loss": 0.7043,
+      "step": 12
+    },
+    {
+      "epoch": 0.10483870967741936,
+      "grad_norm": 107.0622787475586,
+      "learning_rate": 2.2605707418265403e-06,
+      "loss": 1.1753,
+      "step": 13
+    },
+    {
+      "epoch": 0.11290322580645161,
+      "grad_norm": 61.44425964355469,
+      "learning_rate": 2.256846572564882e-06,
+      "loss": 1.8918,
+      "step": 14
+    },
+    {
+      "epoch": 0.12096774193548387,
+      "grad_norm": 99.69449615478516,
+      "learning_rate": 2.253122403303224e-06,
+      "loss": 0.9187,
+      "step": 15
+    },
+    {
+      "epoch": 0.12903225806451613,
+      "grad_norm": 160.03836059570312,
+      "learning_rate": 2.2493982340415658e-06,
+      "loss": 8.8633,
+      "step": 16
+    },
+    {
+      "epoch": 0.13709677419354838,
+      "grad_norm": 130.26107788085938,
+      "learning_rate": 2.245674064779907e-06,
+      "loss": 1.5865,
+      "step": 17
+    },
+    {
+      "epoch": 0.14516129032258066,
+      "grad_norm": 51.57841491699219,
+      "learning_rate": 2.241949895518249e-06,
+      "loss": 1.7055,
+      "step": 18
+    },
+    {
+      "epoch": 0.1532258064516129,
+      "grad_norm": 119.63735961914062,
+      "learning_rate": 2.238225726256591e-06,
+      "loss": 1.5964,
+      "step": 19
+    },
+    {
+      "epoch": 0.16129032258064516,
+      "grad_norm": 218.9719696044922,
+      "learning_rate": 2.2345015569949326e-06,
+      "loss": 8.2474,
+      "step": 20
+    },
+    {
+      "epoch": 0.1693548387096774,
+      "grad_norm": 213.2459716796875,
+      "learning_rate": 2.2307773877332744e-06,
+      "loss": 2.9405,
+      "step": 21
+    },
+    {
+      "epoch": 0.1774193548387097,
+      "grad_norm": 313.0417175292969,
+      "learning_rate": 2.2270532184716162e-06,
+      "loss": 4.7883,
+      "step": 22
+    },
+    {
+      "epoch": 0.18548387096774194,
+      "grad_norm": 178.1278839111328,
+      "learning_rate": 2.223329049209958e-06,
+      "loss": 16.9713,
+      "step": 23
+    },
+    {
+      "epoch": 0.1935483870967742,
+      "grad_norm": 340.25732421875,
+      "learning_rate": 2.2196048799483e-06,
+      "loss": 5.7798,
+      "step": 24
+    },
+    {
+      "epoch": 0.20161290322580644,
+      "grad_norm": 145.1334228515625,
+      "learning_rate": 2.2158807106866417e-06,
+      "loss": 1.827,
+      "step": 25
+    },
+    {
+      "epoch": 0.20967741935483872,
+      "grad_norm": 162.4819793701172,
+      "learning_rate": 2.212156541424983e-06,
+      "loss": 1.9449,
+      "step": 26
+    },
+    {
+      "epoch": 0.21774193548387097,
+      "grad_norm": 120.5811538696289,
+      "learning_rate": 2.208432372163325e-06,
+      "loss": 3.0724,
+      "step": 27
+    },
+    {
+      "epoch": 0.22580645161290322,
+      "grad_norm": 184.12374877929688,
+      "learning_rate": 2.2047082029016667e-06,
+      "loss": 2.8759,
+      "step": 28
+    },
+    {
+      "epoch": 0.23387096774193547,
+      "grad_norm": 85.72262573242188,
+      "learning_rate": 2.200984033640009e-06,
+      "loss": 1.3223,
+      "step": 29
+    },
+    {
+      "epoch": 0.24193548387096775,
+      "grad_norm": 45.07551193237305,
+      "learning_rate": 2.1972598643783508e-06,
+      "loss": 0.5763,
+      "step": 30
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 17.517133712768555,
+      "learning_rate": 2.193535695116692e-06,
+      "loss": 0.9204,
+      "step": 31
+    },
+    {
+      "epoch": 0.25806451612903225,
+      "grad_norm": 67.82421112060547,
+      "learning_rate": 2.189811525855034e-06,
+      "loss": 0.9659,
+      "step": 32
+    },
+    {
+      "epoch": 0.2661290322580645,
+      "grad_norm": 32.993446350097656,
+      "learning_rate": 2.186087356593376e-06,
+      "loss": 0.8629,
+      "step": 33
+    },
+    {
+      "epoch": 0.27419354838709675,
+      "grad_norm": 56.122249603271484,
+      "learning_rate": 2.1823631873317176e-06,
+      "loss": 1.5599,
+      "step": 34
+    },
+    {
+      "epoch": 0.28225806451612906,
+      "grad_norm": 348.0657958984375,
+      "learning_rate": 2.1786390180700594e-06,
+      "loss": 24.736,
+      "step": 35
+    },
+    {
+      "epoch": 0.2903225806451613,
+      "grad_norm": 26.41761016845703,
+      "learning_rate": 2.1749148488084013e-06,
+      "loss": 0.5115,
+      "step": 36
+    },
+    {
+      "epoch": 0.29838709677419356,
+      "grad_norm": 109.63630676269531,
+      "learning_rate": 2.171190679546743e-06,
+      "loss": 1.4894,
+      "step": 37
+    },
+    {
+      "epoch": 0.3064516129032258,
+      "grad_norm": 254.22276306152344,
+      "learning_rate": 2.167466510285085e-06,
+      "loss": 26.0623,
+      "step": 38
+    },
+    {
+      "epoch": 0.31451612903225806,
+      "grad_norm": 41.65399932861328,
+      "learning_rate": 2.1637423410234267e-06,
+      "loss": 0.3779,
+      "step": 39
+    },
+    {
+      "epoch": 0.3225806451612903,
+      "grad_norm": 450.0879821777344,
+      "learning_rate": 2.160018171761768e-06,
+      "loss": 34.1566,
+      "step": 40
+    },
+    {
+      "epoch": 0.33064516129032256,
+      "grad_norm": 57.8630256652832,
+      "learning_rate": 2.15629400250011e-06,
+      "loss": 1.1786,
+      "step": 41
+    },
+    {
+      "epoch": 0.3387096774193548,
+      "grad_norm": 253.47035217285156,
+      "learning_rate": 2.1525698332384517e-06,
+      "loss": 19.6974,
+      "step": 42
+    },
+    {
+      "epoch": 0.3467741935483871,
+      "grad_norm": 172.5160675048828,
+      "learning_rate": 2.1488456639767935e-06,
+      "loss": 27.7034,
+      "step": 43
+    },
+    {
+      "epoch": 0.3548387096774194,
+      "grad_norm": 25.08224868774414,
+      "learning_rate": 2.1451214947151354e-06,
+      "loss": 0.331,
+      "step": 44
+    },
+    {
+      "epoch": 0.3629032258064516,
+      "grad_norm": 40.45454788208008,
+      "learning_rate": 2.141397325453477e-06,
+      "loss": 1.2259,
+      "step": 45
+    },
+    {
+      "epoch": 0.3709677419354839,
+      "grad_norm": 62.238983154296875,
+      "learning_rate": 2.137673156191819e-06,
+      "loss": 2.3635,
+      "step": 46
+    },
+    {
+      "epoch": 0.3790322580645161,
+      "grad_norm": 40.89179229736328,
+      "learning_rate": 2.133948986930161e-06,
+      "loss": 2.3359,
+      "step": 47
+    },
+    {
+      "epoch": 0.3870967741935484,
+      "grad_norm": 143.2526397705078,
+      "learning_rate": 2.1302248176685026e-06,
+      "loss": 1.7717,
+      "step": 48
+    },
+    {
+      "epoch": 0.3951612903225806,
+      "grad_norm": 97.91305541992188,
+      "learning_rate": 2.1265006484068444e-06,
+      "loss": 0.9276,
+      "step": 49
+    },
+    {
+      "epoch": 0.4032258064516129,
+      "grad_norm": 146.67626953125,
+      "learning_rate": 2.122776479145186e-06,
+      "loss": 7.5945,
+      "step": 50
+    },
+    {
+      "epoch": 0.4112903225806452,
+      "grad_norm": 132.352294921875,
+      "learning_rate": 2.1190523098835277e-06,
+      "loss": 1.4938,
+      "step": 51
+    },
+    {
+      "epoch": 0.41935483870967744,
+      "grad_norm": 167.54727172851562,
+      "learning_rate": 2.1153281406218695e-06,
+      "loss": 0.3165,
+      "step": 52
+    },
+    {
+      "epoch": 0.4274193548387097,
+      "grad_norm": 300.7011413574219,
+      "learning_rate": 2.1116039713602113e-06,
+      "loss": 13.5344,
+      "step": 53
+    },
+    {
+      "epoch": 0.43548387096774194,
+      "grad_norm": 63.001609802246094,
+      "learning_rate": 2.1078798020985535e-06,
+      "loss": 1.0206,
+      "step": 54
+    },
+    {
+      "epoch": 0.4435483870967742,
+      "grad_norm": 37.191322326660156,
+      "learning_rate": 2.104155632836895e-06,
+      "loss": 4.3386,
+      "step": 55
+    },
+    {
+      "epoch": 0.45161290322580644,
+      "grad_norm": 42.49811935424805,
+      "learning_rate": 2.1004314635752367e-06,
+      "loss": 0.7406,
+      "step": 56
+    },
+    {
+      "epoch": 0.4596774193548387,
+      "grad_norm": 106.07703399658203,
+      "learning_rate": 2.0967072943135785e-06,
+      "loss": 1.2151,
+      "step": 57
+    },
+    {
+      "epoch": 0.46774193548387094,
+      "grad_norm": 204.26516723632812,
+      "learning_rate": 2.0929831250519204e-06,
+      "loss": 7.2585,
+      "step": 58
+    },
+    {
+      "epoch": 0.47580645161290325,
+      "grad_norm": 70.08997344970703,
+      "learning_rate": 2.089258955790262e-06,
+      "loss": 0.737,
+      "step": 59
+    },
+    {
+      "epoch": 0.4838709677419355,
+      "grad_norm": 213.41036987304688,
+      "learning_rate": 2.085534786528604e-06,
+      "loss": 14.7889,
+      "step": 60
+    },
+    {
+      "epoch": 0.49193548387096775,
+      "grad_norm": 98.5826644897461,
+      "learning_rate": 2.081810617266946e-06,
+      "loss": 0.8851,
+      "step": 61
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 118.79777526855469,
+      "learning_rate": 2.0780864480052876e-06,
+      "loss": 23.6807,
+      "step": 62
+    },
+    {
+      "epoch": 0.5080645161290323,
+      "grad_norm": 118.90921783447266,
+      "learning_rate": 2.0743622787436294e-06,
+      "loss": 1.4873,
+      "step": 63
+    },
+    {
+      "epoch": 0.5161290322580645,
+      "grad_norm": 45.01438903808594,
+      "learning_rate": 2.070638109481971e-06,
+      "loss": 2.5075,
+      "step": 64
+    },
+    {
+      "epoch": 0.5241935483870968,
+      "grad_norm": 143.5907440185547,
+      "learning_rate": 2.0669139402203127e-06,
+      "loss": 14.8131,
+      "step": 65
+    },
+    {
+      "epoch": 0.532258064516129,
+      "grad_norm": 58.62938690185547,
+      "learning_rate": 2.0631897709586545e-06,
+      "loss": 1.2206,
+      "step": 66
+    },
+    {
+      "epoch": 0.5403225806451613,
+      "grad_norm": 83.41098022460938,
+      "learning_rate": 2.0594656016969963e-06,
+      "loss": 1.4715,
+      "step": 67
+    },
+    {
+      "epoch": 0.5483870967741935,
+      "grad_norm": 72.7550048828125,
+      "learning_rate": 2.055741432435338e-06,
+      "loss": 0.8449,
+      "step": 68
+    },
+    {
+      "epoch": 0.5564516129032258,
+      "grad_norm": 99.98751831054688,
+      "learning_rate": 2.05201726317368e-06,
+      "loss": 1.5294,
+      "step": 69
+    },
+    {
+      "epoch": 0.5645161290322581,
+      "grad_norm": 32.452632904052734,
+      "learning_rate": 2.0482930939120217e-06,
+      "loss": 0.3351,
+      "step": 70
+    },
+    {
+      "epoch": 0.5725806451612904,
+      "grad_norm": 87.99835205078125,
+      "learning_rate": 2.0445689246503636e-06,
+      "loss": 2.0365,
+      "step": 71
+    },
+    {
+      "epoch": 0.5806451612903226,
+      "grad_norm": 263.8376770019531,
+      "learning_rate": 2.0408447553887054e-06,
+      "loss": 15.161,
+      "step": 72
+    },
+    {
+      "epoch": 0.5887096774193549,
+      "grad_norm": 39.59327697753906,
+      "learning_rate": 2.0371205861270468e-06,
+      "loss": 0.6412,
+      "step": 73
+    },
+    {
+      "epoch": 0.5967741935483871,
+      "grad_norm": 184.743896484375,
+      "learning_rate": 2.0333964168653886e-06,
+      "loss": 2.4842,
+      "step": 74
+    },
+    {
+      "epoch": 0.6048387096774194,
+      "grad_norm": 118.40076446533203,
+      "learning_rate": 2.0296722476037304e-06,
+      "loss": 15.6392,
+      "step": 75
+    },
+    {
+      "epoch": 0.6129032258064516,
+      "grad_norm": 42.980979919433594,
+      "learning_rate": 2.0259480783420722e-06,
+      "loss": 0.2714,
+      "step": 76
+    },
+    {
+      "epoch": 0.6209677419354839,
+      "grad_norm": 78.06953430175781,
+      "learning_rate": 2.022223909080414e-06,
+      "loss": 0.5073,
+      "step": 77
+    },
+    {
+      "epoch": 0.6290322580645161,
+      "grad_norm": 124.31591796875,
+      "learning_rate": 2.018499739818756e-06,
+      "loss": 12.7986,
+      "step": 78
+    },
+    {
+      "epoch": 0.6370967741935484,
+      "grad_norm": 179.42996215820312,
+      "learning_rate": 2.0147755705570977e-06,
+      "loss": 22.7291,
+      "step": 79
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 24.730609893798828,
+      "learning_rate": 2.0110514012954395e-06,
+      "loss": 1.038,
+      "step": 80
+    },
+    {
+      "epoch": 0.6532258064516129,
+      "grad_norm": 120.6009292602539,
+      "learning_rate": 2.0073272320337813e-06,
+      "loss": 1.043,
+      "step": 81
+    },
+    {
+      "epoch": 0.6612903225806451,
+      "grad_norm": 211.92623901367188,
+      "learning_rate": 2.003603062772123e-06,
+      "loss": 2.2342,
+      "step": 82
+    },
+    {
+      "epoch": 0.6693548387096774,
+      "grad_norm": 17.749357223510742,
+      "learning_rate": 1.9998788935104645e-06,
+      "loss": 0.2666,
+      "step": 83
+    },
+    {
+      "epoch": 0.6774193548387096,
+      "grad_norm": 23.90791893005371,
+      "learning_rate": 1.9961547242488063e-06,
+      "loss": 0.9203,
+      "step": 84
+    },
+    {
+      "epoch": 0.6854838709677419,
+      "grad_norm": 304.3386535644531,
+      "learning_rate": 1.9924305549871486e-06,
+      "loss": 18.5165,
+      "step": 85
+    },
+    {
+      "epoch": 0.6935483870967742,
+      "grad_norm": 115.4675521850586,
+      "learning_rate": 1.9887063857254904e-06,
+      "loss": 14.8705,
+      "step": 86
+    },
+    {
+      "epoch": 0.7016129032258065,
+      "grad_norm": 121.4536361694336,
+      "learning_rate": 1.9849822164638318e-06,
+      "loss": 1.6822,
+      "step": 87
+    },
+    {
+      "epoch": 0.7096774193548387,
+      "grad_norm": 75.23461151123047,
+      "learning_rate": 1.9812580472021736e-06,
+      "loss": 0.75,
+      "step": 88
+    },
+    {
+      "epoch": 0.717741935483871,
+      "grad_norm": 109.62806701660156,
+      "learning_rate": 1.9775338779405154e-06,
+      "loss": 1.6612,
+      "step": 89
+    },
+    {
+      "epoch": 0.7258064516129032,
+      "grad_norm": 133.42919921875,
+      "learning_rate": 1.9738097086788572e-06,
+      "loss": 1.6338,
+      "step": 90
+    },
+    {
+      "epoch": 0.7338709677419355,
+      "grad_norm": 318.0464172363281,
+      "learning_rate": 1.970085539417199e-06,
+      "loss": 19.9659,
+      "step": 91
+    },
+    {
+      "epoch": 0.7419354838709677,
+      "grad_norm": 143.49618530273438,
+      "learning_rate": 1.966361370155541e-06,
+      "loss": 6.968,
+      "step": 92
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 41.44537353515625,
+      "learning_rate": 1.9626372008938827e-06,
+      "loss": 0.273,
+      "step": 93
+    },
+    {
+      "epoch": 0.7580645161290323,
+      "grad_norm": 122.76979064941406,
+      "learning_rate": 1.9589130316322245e-06,
+      "loss": 1.6946,
+      "step": 94
+    },
+    {
+      "epoch": 0.7661290322580645,
+      "grad_norm": 233.6233673095703,
+      "learning_rate": 1.9551888623705663e-06,
+      "loss": 9.8987,
+      "step": 95
+    },
+    {
+      "epoch": 0.7741935483870968,
+      "grad_norm": 103.65068817138672,
+      "learning_rate": 1.951464693108908e-06,
+      "loss": 1.8381,
+      "step": 96
+    },
+    {
+      "epoch": 0.782258064516129,
+      "grad_norm": 79.15303802490234,
+      "learning_rate": 1.9477405238472495e-06,
+      "loss": 0.8288,
+      "step": 97
+    },
+    {
+      "epoch": 0.7903225806451613,
+      "grad_norm": 77.49793243408203,
+      "learning_rate": 1.9440163545855913e-06,
+      "loss": 0.9571,
+      "step": 98
+    },
+    {
+      "epoch": 0.7983870967741935,
+      "grad_norm": 89.57393646240234,
+      "learning_rate": 1.940292185323933e-06,
+      "loss": 1.1071,
+      "step": 99
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 274.893798828125,
+      "learning_rate": 1.936568016062275e-06,
+      "loss": 19.8978,
+      "step": 100
+    },
+    {
+      "epoch": 0.8145161290322581,
+      "grad_norm": 44.62432098388672,
+      "learning_rate": 1.9328438468006168e-06,
+      "loss": 0.8877,
+      "step": 101
+    },
+    {
+      "epoch": 0.8225806451612904,
+      "grad_norm": 103.63294219970703,
+      "learning_rate": 1.9291196775389586e-06,
+      "loss": 0.6422,
+      "step": 102
+    },
+    {
+      "epoch": 0.8306451612903226,
+      "grad_norm": 118.71654510498047,
+      "learning_rate": 1.9253955082773004e-06,
+      "loss": 11.9281,
+      "step": 103
+    },
+    {
+      "epoch": 0.8387096774193549,
+      "grad_norm": 22.929773330688477,
+      "learning_rate": 1.9216713390156422e-06,
+      "loss": 0.5554,
+      "step": 104
+    },
+    {
+      "epoch": 0.8467741935483871,
+      "grad_norm": 164.58189392089844,
+      "learning_rate": 1.917947169753984e-06,
+      "loss": 1.0632,
+      "step": 105
+    },
+    {
+      "epoch": 0.8548387096774194,
+      "grad_norm": 131.65518188476562,
+      "learning_rate": 1.9142230004923254e-06,
+      "loss": 3.3713,
+      "step": 106
+    },
+    {
+      "epoch": 0.8629032258064516,
+      "grad_norm": 103.23025512695312,
+      "learning_rate": 1.9104988312306673e-06,
+      "loss": 2.2805,
+      "step": 107
+    },
+    {
+      "epoch": 0.8709677419354839,
+      "grad_norm": 43.43564224243164,
+      "learning_rate": 1.9067746619690093e-06,
+      "loss": 0.5328,
+      "step": 108
+    },
+    {
+      "epoch": 0.8790322580645161,
+      "grad_norm": 109.90870666503906,
+      "learning_rate": 1.903050492707351e-06,
+      "loss": 1.4934,
+      "step": 109
+    },
+    {
+      "epoch": 0.8870967741935484,
+      "grad_norm": 12.80819034576416,
+      "learning_rate": 1.899326323445693e-06,
+      "loss": 0.8101,
+      "step": 110
+    },
+    {
+      "epoch": 0.8951612903225806,
+      "grad_norm": 223.2318115234375,
+      "learning_rate": 1.8956021541840345e-06,
+      "loss": 4.6677,
+      "step": 111
+    },
+    {
+      "epoch": 0.9032258064516129,
+      "grad_norm": 18.64436912536621,
+      "learning_rate": 1.8918779849223763e-06,
+      "loss": 1.0659,
+      "step": 112
+    },
+    {
+      "epoch": 0.9112903225806451,
+      "grad_norm": 24.009183883666992,
+      "learning_rate": 1.8881538156607181e-06,
+      "loss": 0.2875,
+      "step": 113
+    },
+    {
+      "epoch": 0.9193548387096774,
+      "grad_norm": 21.12612533569336,
+      "learning_rate": 1.88442964639906e-06,
+      "loss": 0.7339,
+      "step": 114
+    },
+    {
+      "epoch": 0.9274193548387096,
+      "grad_norm": 187.74147033691406,
+      "learning_rate": 1.8807054771374018e-06,
+      "loss": 6.5702,
+      "step": 115
+    },
+    {
+      "epoch": 0.9354838709677419,
+      "grad_norm": 62.3062744140625,
+      "learning_rate": 1.8769813078757434e-06,
+      "loss": 1.5116,
+      "step": 116
+    },
+    {
+      "epoch": 0.9435483870967742,
+      "grad_norm": 57.31589126586914,
+      "learning_rate": 1.8732571386140852e-06,
+      "loss": 0.5057,
+      "step": 117
+    },
+    {
+      "epoch": 0.9516129032258065,
+      "grad_norm": 136.85366821289062,
+      "learning_rate": 1.869532969352427e-06,
+      "loss": 14.2211,
+      "step": 118
+    },
+    {
+      "epoch": 0.9596774193548387,
+      "grad_norm": 144.6485137939453,
+      "learning_rate": 1.8658088000907688e-06,
+      "loss": 1.5384,
+      "step": 119
+    },
+    {
+      "epoch": 0.967741935483871,
+      "grad_norm": 25.256959915161133,
+      "learning_rate": 1.8620846308291104e-06,
+      "loss": 1.369,
+      "step": 120
+    },
+    {
+      "epoch": 0.9758064516129032,
+      "grad_norm": 82.7289047241211,
+      "learning_rate": 1.8583604615674523e-06,
+      "loss": 0.8961,
+      "step": 121
+    },
+    {
+      "epoch": 0.9838709677419355,
+      "grad_norm": 14.033979415893555,
+      "learning_rate": 1.854636292305794e-06,
+      "loss": 0.4005,
+      "step": 122
+    },
+    {
+      "epoch": 0.9919354838709677,
+      "grad_norm": 33.84238052368164,
+      "learning_rate": 1.8509121230441359e-06,
+      "loss": 0.3,
+      "step": 123
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 44.46285629272461,
+      "learning_rate": 1.8471879537824777e-06,
+      "loss": 0.0856,
+      "step": 124
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 3.761341094970703,
+      "eval_mae": 1.156431794166565,
+      "eval_mse": 3.761340856552124,
+      "eval_r2": -0.02978968620300293,
+      "eval_rmse": 1.9394176591317622,
+      "eval_runtime": 1.3457,
+      "eval_samples_per_second": 40.87,
+      "eval_smape": 46.60126864910126,
+      "eval_steps_per_second": 10.403,
+      "step": 124
+    },
+    {
+      "epoch": 1.0080645161290323,
+      "grad_norm": 31.391159057617188,
+      "learning_rate": 1.8434637845208193e-06,
+      "loss": 0.669,
+      "step": 125
+    },
+    {
+      "epoch": 1.0161290322580645,
+      "grad_norm": 68.69688415527344,
+      "learning_rate": 1.8397396152591611e-06,
+      "loss": 0.7357,
+      "step": 126
+    },
+    {
+      "epoch": 1.0241935483870968,
+      "grad_norm": 149.95779418945312,
+      "learning_rate": 1.8360154459975032e-06,
+      "loss": 6.2925,
+      "step": 127
+    },
+    {
+      "epoch": 1.032258064516129,
+      "grad_norm": 29.75308609008789,
+      "learning_rate": 1.832291276735845e-06,
+      "loss": 0.5026,
+      "step": 128
+    },
+    {
+      "epoch": 1.0403225806451613,
+      "grad_norm": 31.049043655395508,
+      "learning_rate": 1.8285671074741868e-06,
+      "loss": 0.8776,
+      "step": 129
+    },
+    {
+      "epoch": 1.0483870967741935,
+      "grad_norm": 50.398834228515625,
+      "learning_rate": 1.8248429382125284e-06,
+      "loss": 2.3401,
+      "step": 130
+    },
+    {
+      "epoch": 1.0564516129032258,
+      "grad_norm": 28.888776779174805,
+      "learning_rate": 1.8211187689508702e-06,
+      "loss": 0.1173,
+      "step": 131
+    },
+    {
+      "epoch": 1.064516129032258,
+      "grad_norm": 161.4318389892578,
+      "learning_rate": 1.817394599689212e-06,
+      "loss": 5.2758,
+      "step": 132
+    },
+    {
+      "epoch": 1.0725806451612903,
+      "grad_norm": 61.20700454711914,
+      "learning_rate": 1.8136704304275538e-06,
+      "loss": 0.5886,
+      "step": 133
+    },
+    {
+      "epoch": 1.0806451612903225,
+      "grad_norm": 61.833534240722656,
+      "learning_rate": 1.8099462611658954e-06,
+      "loss": 0.5642,
+      "step": 134
+    },
+    {
+      "epoch": 1.0887096774193548,
+      "grad_norm": 181.53607177734375,
+      "learning_rate": 1.8062220919042373e-06,
+      "loss": 7.6422,
+      "step": 135
+    },
+    {
+      "epoch": 1.096774193548387,
+      "grad_norm": 244.12063598632812,
+      "learning_rate": 1.802497922642579e-06,
+      "loss": 16.4538,
+      "step": 136
+    },
+    {
+      "epoch": 1.1048387096774193,
+      "grad_norm": 203.72213745117188,
+      "learning_rate": 1.798773753380921e-06,
+      "loss": 10.7823,
+      "step": 137
+    },
+    {
+      "epoch": 1.1129032258064515,
+      "grad_norm": 12.267001152038574,
+      "learning_rate": 1.7950495841192627e-06,
+      "loss": 0.3536,
+      "step": 138
+    },
+    {
+      "epoch": 1.120967741935484,
+      "grad_norm": 407.8166809082031,
+      "learning_rate": 1.7913254148576043e-06,
+      "loss": 45.116,
+      "step": 139
+    },
+    {
+      "epoch": 1.129032258064516,
+      "grad_norm": 27.063257217407227,
+      "learning_rate": 1.7876012455959461e-06,
+      "loss": 0.077,
+      "step": 140
+    },
+    {
+      "epoch": 1.1370967741935485,
+      "grad_norm": 44.98855972290039,
+      "learning_rate": 1.783877076334288e-06,
+      "loss": 1.1241,
+      "step": 141
+    },
+    {
+      "epoch": 1.1451612903225807,
+      "grad_norm": 40.457061767578125,
+      "learning_rate": 1.7801529070726298e-06,
+      "loss": 0.9006,
+      "step": 142
+    },
+    {
+      "epoch": 1.153225806451613,
+      "grad_norm": 34.76171112060547,
+      "learning_rate": 1.7764287378109716e-06,
+      "loss": 0.2299,
+      "step": 143
+    },
+    {
+      "epoch": 1.1612903225806452,
+      "grad_norm": 28.188255310058594,
+      "learning_rate": 1.7727045685493132e-06,
+      "loss": 0.2326,
+      "step": 144
+    },
+    {
+      "epoch": 1.1693548387096775,
+      "grad_norm": 37.98823547363281,
+      "learning_rate": 1.768980399287655e-06,
+      "loss": 0.5014,
+      "step": 145
+    },
+    {
+      "epoch": 1.1774193548387097,
+      "grad_norm": 40.45034408569336,
+      "learning_rate": 1.7652562300259968e-06,
+      "loss": 2.0889,
+      "step": 146
+    },
+    {
+      "epoch": 1.185483870967742,
+      "grad_norm": 44.3226432800293,
+      "learning_rate": 1.7615320607643386e-06,
+      "loss": 0.4977,
+      "step": 147
+    },
+    {
+      "epoch": 1.1935483870967742,
+      "grad_norm": 29.797224044799805,
+      "learning_rate": 1.7578078915026802e-06,
+      "loss": 1.1666,
+      "step": 148
+    },
+    {
+      "epoch": 1.2016129032258065,
+      "grad_norm": 291.9981689453125,
+      "learning_rate": 1.754083722241022e-06,
+      "loss": 17.1074,
+      "step": 149
+    },
+    {
+      "epoch": 1.2096774193548387,
+      "grad_norm": 344.50335693359375,
+      "learning_rate": 1.7503595529793639e-06,
+      "loss": 35.2006,
+      "step": 150
+    },
+    {
+      "epoch": 1.217741935483871,
+      "grad_norm": 186.88414001464844,
+      "learning_rate": 1.7466353837177057e-06,
+      "loss": 16.7039,
+      "step": 151
+    },
+    {
+      "epoch": 1.2258064516129032,
+      "grad_norm": 192.2663116455078,
+      "learning_rate": 1.7429112144560475e-06,
+      "loss": 1.0027,
+      "step": 152
+    },
+    {
+      "epoch": 1.2338709677419355,
+      "grad_norm": 125.88935089111328,
+      "learning_rate": 1.7391870451943891e-06,
+      "loss": 8.6522,
+      "step": 153
+    },
+    {
+      "epoch": 1.2419354838709677,
+      "grad_norm": 282.10894775390625,
+      "learning_rate": 1.735462875932731e-06,
+      "loss": 11.9202,
+      "step": 154
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 87.92141723632812,
+      "learning_rate": 1.7317387066710727e-06,
+      "loss": 0.8521,
+      "step": 155
+    },
+    {
+      "epoch": 1.2580645161290323,
+      "grad_norm": 123.09964752197266,
+      "learning_rate": 1.7280145374094148e-06,
+      "loss": 1.2827,
+      "step": 156
+    },
+    {
+      "epoch": 1.2661290322580645,
+      "grad_norm": 137.37051391601562,
+      "learning_rate": 1.7242903681477566e-06,
+      "loss": 2.6597,
+      "step": 157
+    },
+    {
+      "epoch": 1.2741935483870968,
+      "grad_norm": 298.48602294921875,
+      "learning_rate": 1.7205661988860982e-06,
+      "loss": 12.6311,
+      "step": 158
+    },
+    {
+      "epoch": 1.282258064516129,
+      "grad_norm": 1922.1727294921875,
+      "learning_rate": 1.71684202962444e-06,
+      "loss": 22.0268,
+      "step": 159
+    },
+    {
+      "epoch": 1.2903225806451613,
+      "grad_norm": 130.57371520996094,
+      "learning_rate": 1.7131178603627818e-06,
+      "loss": 9.7016,
+      "step": 160
+    },
+    {
+      "epoch": 1.2983870967741935,
+      "grad_norm": 325.2424011230469,
+      "learning_rate": 1.7093936911011236e-06,
+      "loss": 37.0139,
+      "step": 161
+    },
+    {
+      "epoch": 1.3064516129032258,
+      "grad_norm": 180.130615234375,
+      "learning_rate": 1.7056695218394655e-06,
+      "loss": 1.6304,
+      "step": 162
+    },
+    {
+      "epoch": 1.314516129032258,
+      "grad_norm": 190.7817840576172,
+      "learning_rate": 1.701945352577807e-06,
+      "loss": 1.9045,
+      "step": 163
+    },
+    {
+      "epoch": 1.3225806451612903,
+      "grad_norm": 175.99176025390625,
+      "learning_rate": 1.6982211833161489e-06,
+      "loss": 2.5629,
+      "step": 164
+    },
+    {
+      "epoch": 1.3306451612903225,
+      "grad_norm": 416.7662353515625,
+      "learning_rate": 1.6944970140544907e-06,
+      "loss": 5.4869,
+      "step": 165
+    },
+    {
+      "epoch": 1.3387096774193548,
+      "grad_norm": 271.3612060546875,
+      "learning_rate": 1.6907728447928325e-06,
+      "loss": 3.779,
+      "step": 166
+    },
+    {
+      "epoch": 1.346774193548387,
+      "grad_norm": 48.5273323059082,
+      "learning_rate": 1.6870486755311741e-06,
+      "loss": 1.5463,
+      "step": 167
+    },
+    {
+      "epoch": 1.3548387096774195,
+      "grad_norm": 139.34823608398438,
+      "learning_rate": 1.683324506269516e-06,
+      "loss": 1.3512,
+      "step": 168
+    },
+    {
+      "epoch": 1.3629032258064515,
+      "grad_norm": 262.9454650878906,
+      "learning_rate": 1.6796003370078577e-06,
+      "loss": 4.0148,
+      "step": 169
+    },
+    {
+      "epoch": 1.370967741935484,
+      "grad_norm": 1780.4842529296875,
+      "learning_rate": 1.6758761677461996e-06,
+      "loss": 125.1284,
+      "step": 170
+    },
+    {
+      "epoch": 1.379032258064516,
+      "grad_norm": 82.82109069824219,
+      "learning_rate": 1.6721519984845414e-06,
+      "loss": 1.0954,
+      "step": 171
+    },
+    {
+      "epoch": 1.3870967741935485,
+      "grad_norm": 317.63446044921875,
+      "learning_rate": 1.668427829222883e-06,
+      "loss": 5.8565,
+      "step": 172
+    },
+    {
+      "epoch": 1.3951612903225805,
+      "grad_norm": 186.61788940429688,
+      "learning_rate": 1.6647036599612248e-06,
+      "loss": 2.093,
+      "step": 173
+    },
+    {
+      "epoch": 1.403225806451613,
+      "grad_norm": 192.48291015625,
+      "learning_rate": 1.6609794906995666e-06,
+      "loss": 8.7957,
+      "step": 174
+    },
+    {
+      "epoch": 1.4112903225806452,
+      "grad_norm": 83.75372314453125,
+      "learning_rate": 1.6572553214379084e-06,
+      "loss": 0.6524,
+      "step": 175
+    },
+    {
+      "epoch": 1.4193548387096775,
+      "grad_norm": 387.80767822265625,
+      "learning_rate": 1.6535311521762503e-06,
+      "loss": 14.9674,
+      "step": 176
+    },
+    {
+      "epoch": 1.4274193548387097,
+      "grad_norm": 83.91477966308594,
+      "learning_rate": 1.6498069829145919e-06,
+      "loss": 0.8564,
+      "step": 177
+    },
+    {
+      "epoch": 1.435483870967742,
+      "grad_norm": 138.35964965820312,
+      "learning_rate": 1.6460828136529337e-06,
+      "loss": 1.4829,
+      "step": 178
+    },
+    {
+      "epoch": 1.4435483870967742,
+      "grad_norm": 137.02264404296875,
+      "learning_rate": 1.6423586443912755e-06,
+      "loss": 2.6322,
+      "step": 179
+    },
+    {
+      "epoch": 1.4516129032258065,
+      "grad_norm": 164.8358154296875,
+      "learning_rate": 1.6386344751296173e-06,
+      "loss": 1.4194,
+      "step": 180
+    },
+    {
+      "epoch": 1.4596774193548387,
+      "grad_norm": 109.77494812011719,
+      "learning_rate": 1.634910305867959e-06,
+      "loss": 0.9658,
+      "step": 181
+    },
+    {
+      "epoch": 1.467741935483871,
+      "grad_norm": 44.13063430786133,
+      "learning_rate": 1.6311861366063007e-06,
+      "loss": 0.7388,
+      "step": 182
+    },
+    {
+      "epoch": 1.4758064516129032,
+      "grad_norm": 88.6048355102539,
+      "learning_rate": 1.6274619673446425e-06,
+      "loss": 1.3648,
+      "step": 183
+    },
+    {
+      "epoch": 1.4838709677419355,
+      "grad_norm": 123.55364990234375,
+      "learning_rate": 1.6237377980829846e-06,
+      "loss": 3.6786,
+      "step": 184
+    },
+    {
+      "epoch": 1.4919354838709677,
+      "grad_norm": 113.06607818603516,
+      "learning_rate": 1.6200136288213264e-06,
+      "loss": 1.6744,
+      "step": 185
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 32.589027404785156,
+      "learning_rate": 1.616289459559668e-06,
+      "loss": 0.4416,
+      "step": 186
+    },
+    {
+      "epoch": 1.5080645161290323,
+      "grad_norm": 161.1331329345703,
+      "learning_rate": 1.6125652902980098e-06,
+      "loss": 4.6922,
+      "step": 187
+    },
+    {
+      "epoch": 1.5161290322580645,
+      "grad_norm": 59.624202728271484,
+      "learning_rate": 1.6088411210363516e-06,
+      "loss": 1.7379,
+      "step": 188
+    },
+    {
+      "epoch": 1.5241935483870968,
+      "grad_norm": 125.68768310546875,
+      "learning_rate": 1.6051169517746934e-06,
+      "loss": 2.1186,
+      "step": 189
+    },
+    {
+      "epoch": 1.532258064516129,
+      "grad_norm": 47.66250991821289,
+      "learning_rate": 1.6013927825130353e-06,
+      "loss": 0.5983,
+      "step": 190
+    },
+    {
+      "epoch": 1.5403225806451613,
+      "grad_norm": 49.23503112792969,
+      "learning_rate": 1.5976686132513769e-06,
+      "loss": 0.6384,
+      "step": 191
+    },
+    {
+      "epoch": 1.5483870967741935,
+      "grad_norm": 30.719045639038086,
+      "learning_rate": 1.5939444439897187e-06,
+      "loss": 0.4656,
+      "step": 192
+    },
+    {
+      "epoch": 1.5564516129032258,
+      "grad_norm": 42.6551513671875,
+      "learning_rate": 1.5902202747280605e-06,
+      "loss": 0.8564,
+      "step": 193
+    },
+    {
+      "epoch": 1.564516129032258,
+      "grad_norm": 50.98031234741211,
+      "learning_rate": 1.5864961054664023e-06,
+      "loss": 0.6173,
+      "step": 194
+    },
+    {
+      "epoch": 1.5725806451612905,
+      "grad_norm": 45.17729568481445,
+      "learning_rate": 1.582771936204744e-06,
+      "loss": 0.368,
+      "step": 195
+    },
+    {
+      "epoch": 1.5806451612903225,
+      "grad_norm": 266.4801330566406,
+      "learning_rate": 1.5790477669430857e-06,
+      "loss": 7.7041,
+      "step": 196
+    },
+    {
+      "epoch": 1.588709677419355,
+      "grad_norm": 53.33144760131836,
+      "learning_rate": 1.5753235976814275e-06,
+      "loss": 1.2489,
+      "step": 197
+    },
+    {
+      "epoch": 1.596774193548387,
+      "grad_norm": 13.414701461791992,
+      "learning_rate": 1.5715994284197694e-06,
+      "loss": 0.5211,
+      "step": 198
+    },
+    {
+      "epoch": 1.6048387096774195,
+      "grad_norm": 29.101728439331055,
+      "learning_rate": 1.5678752591581112e-06,
+      "loss": 0.1087,
+      "step": 199
+    },
+    {
+      "epoch": 1.6129032258064515,
+      "grad_norm": 103.00382995605469,
+      "learning_rate": 1.5641510898964528e-06,
+      "loss": 0.6299,
+      "step": 200
+    },
+    {
+      "epoch": 1.620967741935484,
+      "grad_norm": 21.447538375854492,
+      "learning_rate": 1.5604269206347946e-06,
+      "loss": 0.1628,
+      "step": 201
+    },
+    {
+      "epoch": 1.629032258064516,
+      "grad_norm": 69.59345245361328,
+      "learning_rate": 1.5567027513731364e-06,
+      "loss": 1.4756,
+      "step": 202
+    },
+    {
+      "epoch": 1.6370967741935485,
+      "grad_norm": 25.250011444091797,
+      "learning_rate": 1.5529785821114782e-06,
+      "loss": 0.1545,
+      "step": 203
+    },
+    {
+      "epoch": 1.6451612903225805,
+      "grad_norm": 60.4410400390625,
+      "learning_rate": 1.54925441284982e-06,
+      "loss": 0.5283,
+      "step": 204
+    },
+    {
+      "epoch": 1.653225806451613,
+      "grad_norm": 234.9252471923828,
+      "learning_rate": 1.5455302435881617e-06,
+      "loss": 16.5494,
+      "step": 205
+    },
+    {
+      "epoch": 1.661290322580645,
+      "grad_norm": 79.00119018554688,
+      "learning_rate": 1.5418060743265035e-06,
+      "loss": 0.3539,
+      "step": 206
+    },
+    {
+      "epoch": 1.6693548387096775,
+      "grad_norm": 37.666534423828125,
+      "learning_rate": 1.5380819050648453e-06,
+      "loss": 0.5521,
+      "step": 207
+    },
+    {
+      "epoch": 1.6774193548387095,
+      "grad_norm": 32.151283264160156,
+      "learning_rate": 1.5343577358031871e-06,
+      "loss": 0.3343,
+      "step": 208
+    },
+    {
+      "epoch": 1.685483870967742,
+      "grad_norm": 23.681564331054688,
+      "learning_rate": 1.5306335665415291e-06,
+      "loss": 0.4023,
+      "step": 209
+    },
+    {
+      "epoch": 1.6935483870967742,
+      "grad_norm": 51.292144775390625,
+      "learning_rate": 1.5269093972798705e-06,
+      "loss": 0.3559,
+      "step": 210
+    },
+    {
+      "epoch": 1.7016129032258065,
+      "grad_norm": 126.73843383789062,
+      "learning_rate": 1.5231852280182123e-06,
+      "loss": 2.4061,
+      "step": 211
+    },
+    {
+      "epoch": 1.7096774193548387,
+      "grad_norm": 57.402000427246094,
+      "learning_rate": 1.5194610587565544e-06,
+      "loss": 1.1266,
+      "step": 212
+    },
+    {
+      "epoch": 1.717741935483871,
+      "grad_norm": 44.26993942260742,
+      "learning_rate": 1.5157368894948962e-06,
+      "loss": 0.1871,
+      "step": 213
+    },
+    {
+      "epoch": 1.7258064516129032,
+      "grad_norm": 331.17962646484375,
+      "learning_rate": 1.5120127202332378e-06,
+      "loss": 24.3463,
+      "step": 214
+    },
+    {
+      "epoch": 1.7338709677419355,
+      "grad_norm": 202.72023010253906,
+      "learning_rate": 1.5082885509715796e-06,
+      "loss": 16.1255,
+      "step": 215
+    },
+    {
+      "epoch": 1.7419354838709677,
+      "grad_norm": 140.6495819091797,
+      "learning_rate": 1.5045643817099214e-06,
+      "loss": 1.2096,
+      "step": 216
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 305.29217529296875,
+      "learning_rate": 1.5008402124482632e-06,
+      "loss": 6.5256,
+      "step": 217
+    },
+    {
+      "epoch": 1.7580645161290323,
+      "grad_norm": 100.94656372070312,
+      "learning_rate": 1.497116043186605e-06,
+      "loss": 0.1109,
+      "step": 218
+    },
+    {
+      "epoch": 1.7661290322580645,
+      "grad_norm": 426.759521484375,
+      "learning_rate": 1.4933918739249467e-06,
+      "loss": 12.6703,
+      "step": 219
+    },
+    {
+      "epoch": 1.7741935483870968,
+      "grad_norm": 24.68358039855957,
+      "learning_rate": 1.4896677046632885e-06,
+      "loss": 0.1721,
+      "step": 220
+    },
+    {
+      "epoch": 1.782258064516129,
+      "grad_norm": 49.39876174926758,
+      "learning_rate": 1.4859435354016303e-06,
+      "loss": 0.3496,
+      "step": 221
+    },
+    {
+      "epoch": 1.7903225806451613,
+      "grad_norm": 294.5503845214844,
+      "learning_rate": 1.4822193661399721e-06,
+      "loss": 0.629,
+      "step": 222
+    },
+    {
+      "epoch": 1.7983870967741935,
+      "grad_norm": 7.117827415466309,
+      "learning_rate": 1.478495196878314e-06,
+      "loss": 0.0899,
+      "step": 223
+    },
+    {
+      "epoch": 1.8064516129032258,
+      "grad_norm": 274.92755126953125,
+      "learning_rate": 1.4747710276166555e-06,
+      "loss": 17.6147,
+      "step": 224
+    },
+    {
+      "epoch": 1.814516129032258,
+      "grad_norm": 445.4465026855469,
+      "learning_rate": 1.4710468583549973e-06,
+      "loss": 5.6809,
+      "step": 225
+    },
+    {
+      "epoch": 1.8225806451612905,
+      "grad_norm": 42.991477966308594,
+      "learning_rate": 1.4673226890933392e-06,
+      "loss": 1.7588,
+      "step": 226
+    },
+    {
+      "epoch": 1.8306451612903225,
+      "grad_norm": 113.24928283691406,
+      "learning_rate": 1.463598519831681e-06,
+      "loss": 0.8243,
+      "step": 227
+    },
+    {
+      "epoch": 1.838709677419355,
+      "grad_norm": 121.32020568847656,
+      "learning_rate": 1.4598743505700226e-06,
+      "loss": 1.5972,
+      "step": 228
+    },
+    {
+      "epoch": 1.846774193548387,
+      "grad_norm": 52.4281005859375,
+      "learning_rate": 1.4561501813083644e-06,
+      "loss": 0.1293,
+      "step": 229
+    },
+    {
+      "epoch": 1.8548387096774195,
+      "grad_norm": 232.2115936279297,
+      "learning_rate": 1.4524260120467062e-06,
+      "loss": 6.8061,
+      "step": 230
+    },
+    {
+      "epoch": 1.8629032258064515,
+      "grad_norm": 178.65858459472656,
+      "learning_rate": 1.448701842785048e-06,
+      "loss": 1.2484,
+      "step": 231
+    },
+    {
+      "epoch": 1.870967741935484,
+      "grad_norm": 20.669795989990234,
+      "learning_rate": 1.4449776735233899e-06,
+      "loss": 0.6529,
+      "step": 232
+    },
+    {
+      "epoch": 1.879032258064516,
+      "grad_norm": 65.12823486328125,
+      "learning_rate": 1.4412535042617315e-06,
+      "loss": 0.7871,
+      "step": 233
+    },
+    {
+      "epoch": 1.8870967741935485,
+      "grad_norm": 19.540246963500977,
+      "learning_rate": 1.4375293350000733e-06,
+      "loss": 0.2503,
+      "step": 234
+    },
+    {
+      "epoch": 1.8951612903225805,
+      "grad_norm": 37.89284133911133,
+      "learning_rate": 1.433805165738415e-06,
+      "loss": 0.378,
+      "step": 235
+    },
+    {
+      "epoch": 1.903225806451613,
+      "grad_norm": 282.1019592285156,
+      "learning_rate": 1.430080996476757e-06,
+      "loss": 8.7697,
+      "step": 236
+    },
+    {
+      "epoch": 1.911290322580645,
+      "grad_norm": 80.56645965576172,
+      "learning_rate": 1.426356827215099e-06,
+      "loss": 6.0227,
+      "step": 237
+    },
+    {
+      "epoch": 1.9193548387096775,
+      "grad_norm": 38.571327209472656,
+      "learning_rate": 1.4226326579534403e-06,
+      "loss": 0.6088,
+      "step": 238
+    },
+    {
+      "epoch": 1.9274193548387095,
+      "grad_norm": 59.85017395019531,
+      "learning_rate": 1.4189084886917821e-06,
+      "loss": 0.6164,
+      "step": 239
+    },
+    {
+      "epoch": 1.935483870967742,
+      "grad_norm": 112.14440155029297,
+      "learning_rate": 1.4151843194301242e-06,
+      "loss": 1.8763,
+      "step": 240
+    },
+    {
+      "epoch": 1.9435483870967742,
+      "grad_norm": 99.5566177368164,
+      "learning_rate": 1.411460150168466e-06,
+      "loss": 0.8691,
+      "step": 241
+    },
+    {
+      "epoch": 1.9516129032258065,
+      "grad_norm": 49.559993743896484,
+      "learning_rate": 1.4077359809068074e-06,
+      "loss": 1.2076,
+      "step": 242
+    },
+    {
+      "epoch": 1.9596774193548387,
+      "grad_norm": 172.5569305419922,
+      "learning_rate": 1.4040118116451494e-06,
+      "loss": 1.7824,
+      "step": 243
+    },
+    {
+      "epoch": 1.967741935483871,
+      "grad_norm": 334.1643981933594,
+      "learning_rate": 1.4002876423834912e-06,
+      "loss": 4.1051,
+      "step": 244
+    },
+    {
+      "epoch": 1.9758064516129032,
+      "grad_norm": 71.8653564453125,
+      "learning_rate": 1.396563473121833e-06,
+      "loss": 0.7466,
+      "step": 245
+    },
+    {
+      "epoch": 1.9838709677419355,
+      "grad_norm": 295.7119445800781,
+      "learning_rate": 1.3928393038601749e-06,
+      "loss": 4.5561,
+      "step": 246
+    },
+    {
+      "epoch": 1.9919354838709677,
+      "grad_norm": 332.00482177734375,
+      "learning_rate": 1.3891151345985165e-06,
+      "loss": 1.4417,
+      "step": 247
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 193.66639709472656,
+      "learning_rate": 1.3853909653368583e-06,
+      "loss": 2.9134,
+      "step": 248
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 3.7568323612213135,
+      "eval_mae": 1.2384544610977173,
+      "eval_mse": 3.756831407546997,
+      "eval_r2": -0.02855515480041504,
+      "eval_rmse": 1.9382547323680128,
+      "eval_runtime": 1.4194,
+      "eval_samples_per_second": 38.75,
+      "eval_smape": 49.87496733665466,
+      "eval_steps_per_second": 9.864,
+      "step": 248
+    },
+    {
+      "epoch": 2.0080645161290325,
+      "grad_norm": 177.89308166503906,
+      "learning_rate": 1.3816667960752e-06,
+      "loss": 1.4892,
+      "step": 249
+    },
+    {
+      "epoch": 2.0161290322580645,
+      "grad_norm": 119.0855941772461,
+      "learning_rate": 1.377942626813542e-06,
+      "loss": 1.4447,
+      "step": 250
+    },
+    {
+      "epoch": 2.024193548387097,
+      "grad_norm": 174.97743225097656,
+      "learning_rate": 1.3742184575518837e-06,
+      "loss": 13.8914,
+      "step": 251
+    },
+    {
+      "epoch": 2.032258064516129,
+      "grad_norm": 231.81077575683594,
+      "learning_rate": 1.3704942882902253e-06,
+      "loss": 4.5327,
+      "step": 252
+    },
+    {
+      "epoch": 2.0403225806451615,
+      "grad_norm": 81.22002410888672,
+      "learning_rate": 1.3667701190285671e-06,
+      "loss": 0.3658,
+      "step": 253
+    },
+    {
+      "epoch": 2.0483870967741935,
+      "grad_norm": 226.5207061767578,
+      "learning_rate": 1.363045949766909e-06,
+      "loss": 21.8189,
+      "step": 254
+    },
+    {
+      "epoch": 2.056451612903226,
+      "grad_norm": 20.884220123291016,
+      "learning_rate": 1.3593217805052508e-06,
+      "loss": 0.1301,
+      "step": 255
+    },
+    {
+      "epoch": 2.064516129032258,
+      "grad_norm": 371.02484130859375,
+      "learning_rate": 1.3555976112435924e-06,
+      "loss": 21.3565,
+      "step": 256
+    },
+    {
+      "epoch": 2.0725806451612905,
+      "grad_norm": 92.34310150146484,
+      "learning_rate": 1.3518734419819342e-06,
+      "loss": 0.707,
+      "step": 257
+    },
+    {
+      "epoch": 2.0806451612903225,
+      "grad_norm": 173.9854736328125,
+      "learning_rate": 1.348149272720276e-06,
+      "loss": 1.7342,
+      "step": 258
+    },
+    {
+      "epoch": 2.088709677419355,
+      "grad_norm": 202.5366668701172,
+      "learning_rate": 1.3444251034586178e-06,
+      "loss": 25.2022,
+      "step": 259
+    },
+    {
+      "epoch": 2.096774193548387,
+      "grad_norm": 33.74102783203125,
+      "learning_rate": 1.3407009341969597e-06,
+      "loss": 0.4951,
+      "step": 260
+    },
+    {
+      "epoch": 2.1048387096774195,
+      "grad_norm": 548.375244140625,
+      "learning_rate": 1.3369767649353013e-06,
+      "loss": 23.6786,
+      "step": 261
+    },
+    {
+      "epoch": 2.1129032258064515,
+      "grad_norm": 203.58383178710938,
+      "learning_rate": 1.333252595673643e-06,
+      "loss": 2.1325,
+      "step": 262
+    },
+    {
+      "epoch": 2.120967741935484,
+      "grad_norm": 64.47084045410156,
+      "learning_rate": 1.3295284264119849e-06,
+      "loss": 0.5467,
+      "step": 263
+    },
+    {
+      "epoch": 2.129032258064516,
+      "grad_norm": 30.41888427734375,
+      "learning_rate": 1.3258042571503267e-06,
+      "loss": 0.6327,
+      "step": 264
+    },
+    {
+      "epoch": 2.1370967741935485,
+      "grad_norm": 217.5379180908203,
+      "learning_rate": 1.3220800878886685e-06,
+      "loss": 1.7505,
+      "step": 265
+    },
+    {
+      "epoch": 2.1451612903225805,
+      "grad_norm": 101.2957534790039,
+      "learning_rate": 1.3183559186270101e-06,
+      "loss": 2.8499,
+      "step": 266
+    },
+    {
+      "epoch": 2.153225806451613,
+      "grad_norm": 91.45832061767578,
+      "learning_rate": 1.314631749365352e-06,
+      "loss": 0.3068,
+      "step": 267
+    },
+    {
+      "epoch": 2.161290322580645,
+      "grad_norm": 99.60401153564453,
+      "learning_rate": 1.310907580103694e-06,
+      "loss": 1.2989,
+      "step": 268
+    },
+    {
+      "epoch": 2.1693548387096775,
+      "grad_norm": 112.3869857788086,
+      "learning_rate": 1.3071834108420358e-06,
+      "loss": 2.3015,
+      "step": 269
+    },
+    {
+      "epoch": 2.1774193548387095,
+      "grad_norm": 89.12266540527344,
+      "learning_rate": 1.3034592415803776e-06,
+      "loss": 1.0322,
+      "step": 270
+    },
+    {
+      "epoch": 2.185483870967742,
+      "grad_norm": 161.4855194091797,
+      "learning_rate": 1.2997350723187192e-06,
+      "loss": 11.7275,
+      "step": 271
+    },
+    {
+      "epoch": 2.193548387096774,
+      "grad_norm": 313.5152893066406,
+      "learning_rate": 1.296010903057061e-06,
+      "loss": 16.6577,
+      "step": 272
+    },
+    {
+      "epoch": 2.2016129032258065,
+      "grad_norm": 139.5433807373047,
+      "learning_rate": 1.2922867337954028e-06,
+      "loss": 14.2802,
+      "step": 273
+    },
+    {
+      "epoch": 2.2096774193548385,
+      "grad_norm": 81.8010482788086,
+      "learning_rate": 1.2885625645337447e-06,
+      "loss": 1.0866,
+      "step": 274
+    },
+    {
+      "epoch": 2.217741935483871,
+      "grad_norm": 36.10691833496094,
+      "learning_rate": 1.2848383952720863e-06,
+      "loss": 0.2724,
+      "step": 275
+    },
+    {
+      "epoch": 2.225806451612903,
+      "grad_norm": 44.03886032104492,
+      "learning_rate": 1.281114226010428e-06,
+      "loss": 0.1664,
+      "step": 276
+    },
+    {
+      "epoch": 2.2338709677419355,
+      "grad_norm": 62.077606201171875,
+      "learning_rate": 1.27739005674877e-06,
+      "loss": 0.3426,
+      "step": 277
+    },
+    {
+      "epoch": 2.241935483870968,
+      "grad_norm": 70.45355987548828,
+      "learning_rate": 1.2736658874871117e-06,
+      "loss": 0.9209,
+      "step": 278
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 61.781490325927734,
+      "learning_rate": 1.2699417182254535e-06,
+      "loss": 0.7348,
+      "step": 279
+    },
+    {
+      "epoch": 2.258064516129032,
+      "grad_norm": 209.58901977539062,
+      "learning_rate": 1.2662175489637951e-06,
+      "loss": 12.0011,
+      "step": 280
+    },
+    {
+      "epoch": 2.2661290322580645,
+      "grad_norm": 87.33316802978516,
+      "learning_rate": 1.262493379702137e-06,
+      "loss": 1.418,
+      "step": 281
+    },
+    {
+      "epoch": 2.274193548387097,
+      "grad_norm": 493.7389221191406,
+      "learning_rate": 1.2587692104404788e-06,
+      "loss": 14.2359,
+      "step": 282
+    },
+    {
+      "epoch": 2.282258064516129,
+      "grad_norm": 861.2841186523438,
+      "learning_rate": 1.2550450411788206e-06,
+      "loss": 3.9053,
+      "step": 283
+    },
+    {
+      "epoch": 2.2903225806451615,
+      "grad_norm": 558.491943359375,
+      "learning_rate": 1.2513208719171624e-06,
+      "loss": 6.4746,
+      "step": 284
+    },
+    {
+      "epoch": 2.2983870967741935,
+      "grad_norm": 89.52096557617188,
+      "learning_rate": 1.247596702655504e-06,
+      "loss": 0.7692,
+      "step": 285
+    },
+    {
+      "epoch": 2.306451612903226,
+      "grad_norm": 158.73118591308594,
+      "learning_rate": 1.2438725333938458e-06,
+      "loss": 1.7061,
+      "step": 286
+    },
+    {
+      "epoch": 2.314516129032258,
+      "grad_norm": 93.90080261230469,
+      "learning_rate": 1.2401483641321876e-06,
+      "loss": 0.8225,
+      "step": 287
+    },
+    {
+      "epoch": 2.3225806451612905,
+      "grad_norm": 27.100187301635742,
+      "learning_rate": 1.2364241948705295e-06,
+      "loss": 0.425,
+      "step": 288
+    },
+    {
+      "epoch": 2.3306451612903225,
+      "grad_norm": 123.69001770019531,
+      "learning_rate": 1.232700025608871e-06,
+      "loss": 1.7102,
+      "step": 289
+    },
+    {
+      "epoch": 2.338709677419355,
+      "grad_norm": 128.8670196533203,
+      "learning_rate": 1.2289758563472129e-06,
+      "loss": 12.3249,
+      "step": 290
+    },
+    {
+      "epoch": 2.346774193548387,
+      "grad_norm": 194.0598602294922,
+      "learning_rate": 1.2252516870855547e-06,
+      "loss": 0.415,
+      "step": 291
+    },
+    {
+      "epoch": 2.3548387096774195,
+      "grad_norm": 960.8411254882812,
+      "learning_rate": 1.2215275178238965e-06,
+      "loss": 5.3627,
+      "step": 292
+    },
+    {
+      "epoch": 2.3629032258064515,
+      "grad_norm": 50.68828201293945,
+      "learning_rate": 1.2178033485622383e-06,
+      "loss": 0.2609,
+      "step": 293
+    },
+    {
+      "epoch": 2.370967741935484,
+      "grad_norm": 1389.3763427734375,
+      "learning_rate": 1.21407917930058e-06,
+      "loss": 12.0135,
+      "step": 294
+    },
+    {
+      "epoch": 2.379032258064516,
+      "grad_norm": 157.8043212890625,
+      "learning_rate": 1.2103550100389217e-06,
+      "loss": 16.9635,
+      "step": 295
+    },
+    {
+      "epoch": 2.3870967741935485,
+      "grad_norm": 160.09109497070312,
+      "learning_rate": 1.2066308407772638e-06,
+      "loss": 1.2352,
+      "step": 296
+    },
+    {
+      "epoch": 2.3951612903225805,
+      "grad_norm": 27.7835750579834,
+      "learning_rate": 1.2029066715156056e-06,
+      "loss": 0.2556,
+      "step": 297
+    },
+    {
+      "epoch": 2.403225806451613,
+      "grad_norm": 177.45115661621094,
+      "learning_rate": 1.1991825022539474e-06,
+      "loss": 2.1015,
+      "step": 298
+    },
+    {
+      "epoch": 2.411290322580645,
+      "grad_norm": 44.05683517456055,
+      "learning_rate": 1.195458332992289e-06,
+      "loss": 0.976,
+      "step": 299
+    },
+    {
+      "epoch": 2.4193548387096775,
+      "grad_norm": 273.5810852050781,
+      "learning_rate": 1.1917341637306308e-06,
+      "loss": 2.6291,
+      "step": 300
+    },
+    {
+      "epoch": 2.4274193548387095,
+      "grad_norm": 151.48826599121094,
+      "learning_rate": 1.1880099944689726e-06,
+      "loss": 1.2252,
+      "step": 301
+    },
+    {
+      "epoch": 2.435483870967742,
+      "grad_norm": 101.94556427001953,
+      "learning_rate": 1.1842858252073145e-06,
+      "loss": 0.1799,
+      "step": 302
+    },
+    {
+      "epoch": 2.443548387096774,
+      "grad_norm": 254.22804260253906,
+      "learning_rate": 1.180561655945656e-06,
+      "loss": 6.3368,
+      "step": 303
+    },
+    {
+      "epoch": 2.4516129032258065,
+      "grad_norm": 137.3714141845703,
+      "learning_rate": 1.1768374866839979e-06,
+      "loss": 0.7803,
+      "step": 304
+    },
+    {
+      "epoch": 2.4596774193548385,
+      "grad_norm": 140.2755584716797,
+      "learning_rate": 1.1731133174223397e-06,
+      "loss": 14.2433,
+      "step": 305
+    },
+    {
+      "epoch": 2.467741935483871,
+      "grad_norm": 43.264320373535156,
+      "learning_rate": 1.1693891481606815e-06,
+      "loss": 0.3709,
+      "step": 306
+    },
+    {
+      "epoch": 2.475806451612903,
+      "grad_norm": 185.11683654785156,
+      "learning_rate": 1.1656649788990233e-06,
+      "loss": 1.7764,
+      "step": 307
+    },
+    {
+      "epoch": 2.4838709677419355,
+      "grad_norm": 303.35791015625,
+      "learning_rate": 1.161940809637365e-06,
+      "loss": 11.4762,
+      "step": 308
+    },
+    {
+      "epoch": 2.491935483870968,
+      "grad_norm": 91.71321868896484,
+      "learning_rate": 1.1582166403757067e-06,
+      "loss": 0.6848,
+      "step": 309
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 12.553592681884766,
+      "learning_rate": 1.1544924711140486e-06,
+      "loss": 0.3027,
+      "step": 310
+    },
+    {
+      "epoch": 2.508064516129032,
+      "grad_norm": 66.78060150146484,
+      "learning_rate": 1.1507683018523904e-06,
+      "loss": 0.7009,
+      "step": 311
+    },
+    {
+      "epoch": 2.5161290322580645,
+      "grad_norm": 31.330020904541016,
+      "learning_rate": 1.1470441325907322e-06,
+      "loss": 0.2209,
+      "step": 312
+    },
+    {
+      "epoch": 2.524193548387097,
+      "grad_norm": 42.87683868408203,
+      "learning_rate": 1.143319963329074e-06,
+      "loss": 0.313,
+      "step": 313
+    },
+    {
+      "epoch": 2.532258064516129,
+      "grad_norm": 19.70355224609375,
+      "learning_rate": 1.1395957940674156e-06,
+      "loss": 0.5462,
+      "step": 314
+    },
+    {
+      "epoch": 2.540322580645161,
+      "grad_norm": 488.4886474609375,
+      "learning_rate": 1.1358716248057574e-06,
+      "loss": 8.1183,
+      "step": 315
+    },
+    {
+      "epoch": 2.5483870967741935,
+      "grad_norm": 14.261226654052734,
+      "learning_rate": 1.1321474555440993e-06,
+      "loss": 0.2522,
+      "step": 316
+    },
+    {
+      "epoch": 2.556451612903226,
+      "grad_norm": 31.870094299316406,
+      "learning_rate": 1.128423286282441e-06,
+      "loss": 1.1169,
+      "step": 317
+    },
+    {
+      "epoch": 2.564516129032258,
+      "grad_norm": 31.234668731689453,
+      "learning_rate": 1.1246991170207829e-06,
+      "loss": 0.3918,
+      "step": 318
+    },
+    {
+      "epoch": 2.5725806451612905,
+      "grad_norm": 17.941022872924805,
+      "learning_rate": 1.1209749477591245e-06,
+      "loss": 0.8012,
+      "step": 319
+    },
+    {
+      "epoch": 2.5806451612903225,
+      "grad_norm": 54.857208251953125,
+      "learning_rate": 1.1172507784974663e-06,
+      "loss": 1.1496,
+      "step": 320
+    },
+    {
+      "epoch": 2.588709677419355,
+      "grad_norm": 332.4779052734375,
+      "learning_rate": 1.1135266092358081e-06,
+      "loss": 8.3327,
+      "step": 321
+    },
+    {
+      "epoch": 2.596774193548387,
+      "grad_norm": 35.55912399291992,
+      "learning_rate": 1.10980243997415e-06,
+      "loss": 0.4927,
+      "step": 322
+    },
+    {
+      "epoch": 2.6048387096774195,
+      "grad_norm": 373.67242431640625,
+      "learning_rate": 1.1060782707124915e-06,
+      "loss": 18.8225,
+      "step": 323
+    },
+    {
+      "epoch": 2.6129032258064515,
+      "grad_norm": 25.760787963867188,
+      "learning_rate": 1.1023541014508334e-06,
+      "loss": 0.3753,
+      "step": 324
+    },
+    {
+      "epoch": 2.620967741935484,
+      "grad_norm": 58.30063247680664,
+      "learning_rate": 1.0986299321891754e-06,
+      "loss": 0.7956,
+      "step": 325
+    },
+    {
+      "epoch": 2.629032258064516,
+      "grad_norm": 31.588315963745117,
+      "learning_rate": 1.094905762927517e-06,
+      "loss": 0.6544,
+      "step": 326
+    },
+    {
+      "epoch": 2.6370967741935485,
+      "grad_norm": 79.70700073242188,
+      "learning_rate": 1.0911815936658588e-06,
+      "loss": 1.4362,
+      "step": 327
+    },
+    {
+      "epoch": 2.6451612903225805,
+      "grad_norm": 57.3786506652832,
+      "learning_rate": 1.0874574244042006e-06,
+      "loss": 0.2611,
+      "step": 328
+    },
+    {
+      "epoch": 2.653225806451613,
+      "grad_norm": 661.8717041015625,
+      "learning_rate": 1.0837332551425424e-06,
+      "loss": 137.7964,
+      "step": 329
+    },
+    {
+      "epoch": 2.661290322580645,
+      "grad_norm": 236.0816192626953,
+      "learning_rate": 1.080009085880884e-06,
+      "loss": 6.9625,
+      "step": 330
+    },
+    {
+      "epoch": 2.6693548387096775,
+      "grad_norm": 72.58594512939453,
+      "learning_rate": 1.0762849166192259e-06,
+      "loss": 0.8561,
+      "step": 331
+    },
+    {
+      "epoch": 2.6774193548387095,
+      "grad_norm": 65.97917175292969,
+      "learning_rate": 1.0725607473575677e-06,
+      "loss": 1.7101,
+      "step": 332
+    },
+    {
+      "epoch": 2.685483870967742,
+      "grad_norm": 55.23917770385742,
+      "learning_rate": 1.0688365780959095e-06,
+      "loss": 0.5448,
+      "step": 333
+    },
+    {
+      "epoch": 2.693548387096774,
+      "grad_norm": 501.1816711425781,
+      "learning_rate": 1.0651124088342513e-06,
+      "loss": 13.16,
+      "step": 334
+    },
+    {
+      "epoch": 2.7016129032258065,
+      "grad_norm": 55.06072235107422,
+      "learning_rate": 1.061388239572593e-06,
+      "loss": 0.9885,
+      "step": 335
+    },
+    {
+      "epoch": 2.709677419354839,
+      "grad_norm": 98.7667236328125,
+      "learning_rate": 1.0576640703109347e-06,
+      "loss": 1.4739,
+      "step": 336
+    },
+    {
+      "epoch": 2.717741935483871,
+      "grad_norm": 77.85026550292969,
+      "learning_rate": 1.0539399010492768e-06,
+      "loss": 0.9057,
+      "step": 337
+    },
+    {
+      "epoch": 2.725806451612903,
+      "grad_norm": 95.394287109375,
+      "learning_rate": 1.0502157317876184e-06,
+      "loss": 0.7198,
+      "step": 338
+    },
+    {
+      "epoch": 2.7338709677419355,
+      "grad_norm": 237.02577209472656,
+      "learning_rate": 1.0464915625259602e-06,
+      "loss": 13.3706,
+      "step": 339
+    },
+    {
+      "epoch": 2.741935483870968,
+      "grad_norm": 127.62625122070312,
+      "learning_rate": 1.042767393264302e-06,
+      "loss": 0.8813,
+      "step": 340
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 73.3454360961914,
+      "learning_rate": 1.0390432240026438e-06,
+      "loss": 0.5758,
+      "step": 341
+    },
+    {
+      "epoch": 2.758064516129032,
+      "grad_norm": 109.3171615600586,
+      "learning_rate": 1.0353190547409854e-06,
+      "loss": 9.7732,
+      "step": 342
+    },
+    {
+      "epoch": 2.7661290322580645,
+      "grad_norm": 163.61447143554688,
+      "learning_rate": 1.0315948854793272e-06,
+      "loss": 0.6441,
+      "step": 343
+    },
+    {
+      "epoch": 2.774193548387097,
+      "grad_norm": 36.76464080810547,
+      "learning_rate": 1.027870716217669e-06,
+      "loss": 0.5595,
+      "step": 344
+    },
+    {
+      "epoch": 2.782258064516129,
+      "grad_norm": 206.4422607421875,
+      "learning_rate": 1.0241465469560109e-06,
+      "loss": 3.4251,
+      "step": 345
+    },
+    {
+      "epoch": 2.790322580645161,
+      "grad_norm": 46.03797912597656,
+      "learning_rate": 1.0204223776943527e-06,
+      "loss": 0.1912,
+      "step": 346
+    },
+    {
+      "epoch": 2.7983870967741935,
+      "grad_norm": 196.0258026123047,
+      "learning_rate": 1.0166982084326943e-06,
+      "loss": 1.2507,
+      "step": 347
+    },
+    {
+      "epoch": 2.806451612903226,
+      "grad_norm": 171.45388793945312,
+      "learning_rate": 1.0129740391710361e-06,
+      "loss": 0.9022,
+      "step": 348
+    },
+    {
+      "epoch": 2.814516129032258,
+      "grad_norm": 128.89340209960938,
+      "learning_rate": 1.009249869909378e-06,
+      "loss": 5.8515,
+      "step": 349
+    },
+    {
+      "epoch": 2.8225806451612905,
+      "grad_norm": 44.62214660644531,
+      "learning_rate": 1.0055257006477197e-06,
+      "loss": 0.6465,
+      "step": 350
+    },
+    {
+      "epoch": 2.8306451612903225,
+      "grad_norm": 33.63302230834961,
+      "learning_rate": 1.0018015313860616e-06,
+      "loss": 0.1617,
+      "step": 351
+    },
+    {
+      "epoch": 2.838709677419355,
+      "grad_norm": 126.68191528320312,
+      "learning_rate": 9.980773621244032e-07,
+      "loss": 4.3174,
+      "step": 352
+    },
+    {
+      "epoch": 2.846774193548387,
+      "grad_norm": 18.371591567993164,
+      "learning_rate": 9.943531928627452e-07,
+      "loss": 0.4257,
+      "step": 353
+    },
+    {
+      "epoch": 2.8548387096774195,
+      "grad_norm": 324.5527648925781,
+      "learning_rate": 9.906290236010868e-07,
+      "loss": 7.0345,
+      "step": 354
+    },
+    {
+      "epoch": 2.8629032258064515,
+      "grad_norm": 295.57806396484375,
+      "learning_rate": 9.869048543394286e-07,
+      "loss": 4.2767,
+      "step": 355
+    },
+    {
+      "epoch": 2.870967741935484,
+      "grad_norm": 21.32406997680664,
+      "learning_rate": 9.831806850777704e-07,
+      "loss": 0.2288,
+      "step": 356
+    },
+    {
+      "epoch": 2.879032258064516,
+      "grad_norm": 163.4864959716797,
+      "learning_rate": 9.794565158161122e-07,
+      "loss": 19.382,
+      "step": 357
+    },
+    {
+      "epoch": 2.8870967741935485,
+      "grad_norm": 81.98368835449219,
+      "learning_rate": 9.75732346554454e-07,
+      "loss": 0.3589,
+      "step": 358
+    },
+    {
+      "epoch": 2.8951612903225805,
+      "grad_norm": 80.34703063964844,
+      "learning_rate": 9.720081772927957e-07,
+      "loss": 1.181,
+      "step": 359
+    },
+    {
+      "epoch": 2.903225806451613,
+      "grad_norm": 37.47127914428711,
+      "learning_rate": 9.682840080311375e-07,
+      "loss": 0.9547,
+      "step": 360
+    },
+    {
+      "epoch": 2.911290322580645,
+      "grad_norm": 212.42002868652344,
+      "learning_rate": 9.645598387694793e-07,
+      "loss": 17.7606,
+      "step": 361
+    },
+    {
+      "epoch": 2.9193548387096775,
+      "grad_norm": 25.195838928222656,
+      "learning_rate": 9.608356695078211e-07,
+      "loss": 0.3995,
+      "step": 362
+    },
+    {
+      "epoch": 2.9274193548387095,
+      "grad_norm": 39.5676383972168,
+      "learning_rate": 9.571115002461627e-07,
+      "loss": 0.3185,
+      "step": 363
+    },
+    {
+      "epoch": 2.935483870967742,
+      "grad_norm": 96.67916870117188,
+      "learning_rate": 9.533873309845046e-07,
+      "loss": 0.4761,
+      "step": 364
+    },
+    {
+      "epoch": 2.943548387096774,
+      "grad_norm": 67.7193374633789,
+      "learning_rate": 9.496631617228465e-07,
+      "loss": 0.6273,
+      "step": 365
+    },
+    {
+      "epoch": 2.9516129032258065,
+      "grad_norm": 78.17704010009766,
+      "learning_rate": 9.459389924611882e-07,
+      "loss": 0.9327,
+      "step": 366
+    },
+    {
+      "epoch": 2.959677419354839,
+      "grad_norm": 94.27801513671875,
+      "learning_rate": 9.4221482319953e-07,
+      "loss": 0.8413,
+      "step": 367
+    },
+    {
+      "epoch": 2.967741935483871,
+      "grad_norm": 195.948486328125,
+      "learning_rate": 9.384906539378717e-07,
+      "loss": 1.9969,
+      "step": 368
+    },
+    {
+      "epoch": 2.975806451612903,
+      "grad_norm": 78.65729522705078,
+      "learning_rate": 9.347664846762135e-07,
+      "loss": 0.8227,
+      "step": 369
+    },
+    {
+      "epoch": 2.9838709677419355,
+      "grad_norm": 51.40840530395508,
+      "learning_rate": 9.310423154145552e-07,
+      "loss": 0.2983,
+      "step": 370
+    },
+    {
+      "epoch": 2.991935483870968,
+      "grad_norm": 107.39090728759766,
+      "learning_rate": 9.27318146152897e-07,
+      "loss": 2.3893,
+      "step": 371
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 592.5924072265625,
+      "learning_rate": 9.235939768912389e-07,
+      "loss": 32.2766,
+      "step": 372
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 3.6544721126556396,
+      "eval_mae": 1.135090947151184,
+      "eval_mse": 3.654472827911377,
+      "eval_r2": -0.0005310773849487305,
+      "eval_rmse": 1.9116675516185802,
+      "eval_runtime": 1.4093,
+      "eval_samples_per_second": 39.026,
+      "eval_smape": 48.73234033584595,
+      "eval_steps_per_second": 9.934,
+      "step": 372
+    },
+    {
+      "epoch": 3.0080645161290325,
+      "grad_norm": 167.24183654785156,
+      "learning_rate": 9.198698076295806e-07,
+      "loss": 2.4298,
+      "step": 373
+    },
+    {
+      "epoch": 3.0161290322580645,
+      "grad_norm": 346.9148254394531,
+      "learning_rate": 9.161456383679225e-07,
+      "loss": 6.5046,
+      "step": 374
+    },
+    {
+      "epoch": 3.024193548387097,
+      "grad_norm": 42.78517532348633,
+      "learning_rate": 9.124214691062642e-07,
+      "loss": 0.3438,
+      "step": 375
+    },
+    {
+      "epoch": 3.032258064516129,
+      "grad_norm": 33.160282135009766,
+      "learning_rate": 9.08697299844606e-07,
+      "loss": 1.2097,
+      "step": 376
+    },
+    {
+      "epoch": 3.0403225806451615,
+      "grad_norm": 38.76713562011719,
+      "learning_rate": 9.049731305829477e-07,
+      "loss": 0.3487,
+      "step": 377
+    },
+    {
+      "epoch": 3.0483870967741935,
+      "grad_norm": 167.7181396484375,
+      "learning_rate": 9.012489613212895e-07,
+      "loss": 2.2915,
+      "step": 378
+    },
+    {
+      "epoch": 3.056451612903226,
+      "grad_norm": 91.34284973144531,
+      "learning_rate": 8.975247920596314e-07,
+      "loss": 1.243,
+      "step": 379
+    },
+    {
+      "epoch": 3.064516129032258,
+      "grad_norm": 86.71770477294922,
+      "learning_rate": 8.938006227979731e-07,
+      "loss": 1.3177,
+      "step": 380
+    },
+    {
+      "epoch": 3.0725806451612905,
+      "grad_norm": 674.9639892578125,
+      "learning_rate": 8.900764535363149e-07,
+      "loss": 16.8542,
+      "step": 381
+    },
+    {
+      "epoch": 3.0806451612903225,
+      "grad_norm": 88.91838836669922,
+      "learning_rate": 8.863522842746566e-07,
+      "loss": 2.1356,
+      "step": 382
+    },
+    {
+      "epoch": 3.088709677419355,
+      "grad_norm": 63.240787506103516,
+      "learning_rate": 8.826281150129984e-07,
+      "loss": 0.7736,
+      "step": 383
+    },
+    {
+      "epoch": 3.096774193548387,
+      "grad_norm": 82.86074829101562,
+      "learning_rate": 8.789039457513401e-07,
+      "loss": 2.2132,
+      "step": 384
+    },
+    {
+      "epoch": 3.1048387096774195,
+      "grad_norm": 34.56795883178711,
+      "learning_rate": 8.751797764896819e-07,
+      "loss": 1.0128,
+      "step": 385
+    },
+    {
+      "epoch": 3.1129032258064515,
+      "grad_norm": 24.039697647094727,
+      "learning_rate": 8.714556072280238e-07,
+      "loss": 0.4341,
+      "step": 386
+    },
+    {
+      "epoch": 3.120967741935484,
+      "grad_norm": 96.96089935302734,
+      "learning_rate": 8.677314379663655e-07,
+      "loss": 1.5847,
+      "step": 387
+    },
+    {
+      "epoch": 3.129032258064516,
+      "grad_norm": 91.98773956298828,
+      "learning_rate": 8.640072687047074e-07,
+      "loss": 0.9096,
+      "step": 388
+    },
+    {
+      "epoch": 3.1370967741935485,
+      "grad_norm": 66.54564666748047,
+      "learning_rate": 8.602830994430491e-07,
+      "loss": 0.3645,
+      "step": 389
+    },
+    {
+      "epoch": 3.1451612903225805,
+      "grad_norm": 64.6353759765625,
+      "learning_rate": 8.565589301813909e-07,
+      "loss": 0.5853,
+      "step": 390
+    },
+    {
+      "epoch": 3.153225806451613,
+      "grad_norm": 445.8139343261719,
+      "learning_rate": 8.528347609197327e-07,
+      "loss": 14.5169,
+      "step": 391
+    },
+    {
+      "epoch": 3.161290322580645,
+      "grad_norm": 43.0020637512207,
+      "learning_rate": 8.491105916580744e-07,
+      "loss": 0.8973,
+      "step": 392
+    },
+    {
+      "epoch": 3.1693548387096775,
+      "grad_norm": 74.25178527832031,
+      "learning_rate": 8.453864223964163e-07,
+      "loss": 0.2461,
+      "step": 393
+    },
+    {
+      "epoch": 3.1774193548387095,
+      "grad_norm": 24.665449142456055,
+      "learning_rate": 8.41662253134758e-07,
+      "loss": 0.4055,
+      "step": 394
+    },
+    {
+      "epoch": 3.185483870967742,
+      "grad_norm": 66.4891357421875,
+      "learning_rate": 8.379380838730998e-07,
+      "loss": 0.612,
+      "step": 395
+    },
+    {
+      "epoch": 3.193548387096774,
+      "grad_norm": 55.56448745727539,
+      "learning_rate": 8.342139146114415e-07,
+      "loss": 0.2977,
+      "step": 396
+    },
+    {
+      "epoch": 3.2016129032258065,
+      "grad_norm": 40.89321517944336,
+      "learning_rate": 8.304897453497833e-07,
+      "loss": 0.7316,
+      "step": 397
+    },
+    {
+      "epoch": 3.2096774193548385,
+      "grad_norm": 72.87995910644531,
+      "learning_rate": 8.267655760881251e-07,
+      "loss": 1.1076,
+      "step": 398
+    },
+    {
+      "epoch": 3.217741935483871,
+      "grad_norm": 199.8931427001953,
+      "learning_rate": 8.230414068264668e-07,
+      "loss": 22.663,
+      "step": 399
+    },
+    {
+      "epoch": 3.225806451612903,
+      "grad_norm": 380.7558898925781,
+      "learning_rate": 8.193172375648087e-07,
+      "loss": 12.9819,
+      "step": 400
+    },
+    {
+      "epoch": 3.2338709677419355,
+      "grad_norm": 94.97148895263672,
+      "learning_rate": 8.155930683031504e-07,
+      "loss": 2.038,
+      "step": 401
+    },
+    {
+      "epoch": 3.241935483870968,
+      "grad_norm": 770.9739990234375,
+      "learning_rate": 8.118688990414923e-07,
+      "loss": 6.0835,
+      "step": 402
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 276.0569152832031,
+      "learning_rate": 8.08144729779834e-07,
+      "loss": 5.7228,
+      "step": 403
+    },
+    {
+      "epoch": 3.258064516129032,
+      "grad_norm": 64.70281219482422,
+      "learning_rate": 8.044205605181758e-07,
+      "loss": 0.4039,
+      "step": 404
+    },
+    {
+      "epoch": 3.2661290322580645,
+      "grad_norm": 244.07774353027344,
+      "learning_rate": 8.006963912565176e-07,
+      "loss": 0.4348,
+      "step": 405
+    },
+    {
+      "epoch": 3.274193548387097,
+      "grad_norm": 516.8790283203125,
+      "learning_rate": 7.969722219948593e-07,
+      "loss": 20.4644,
+      "step": 406
+    },
+    {
+      "epoch": 3.282258064516129,
+      "grad_norm": 431.8797912597656,
+      "learning_rate": 7.932480527332012e-07,
+      "loss": 5.4049,
+      "step": 407
+    },
+    {
+      "epoch": 3.2903225806451615,
+      "grad_norm": 92.80670928955078,
+      "learning_rate": 7.895238834715429e-07,
+      "loss": 8.2338,
+      "step": 408
+    },
+    {
+      "epoch": 3.2983870967741935,
+      "grad_norm": 208.46914672851562,
+      "learning_rate": 7.857997142098847e-07,
+      "loss": 2.0624,
+      "step": 409
+    },
+    {
+      "epoch": 3.306451612903226,
+      "grad_norm": 66.8282241821289,
+      "learning_rate": 7.820755449482264e-07,
+      "loss": 1.6893,
+      "step": 410
+    },
+    {
+      "epoch": 3.314516129032258,
+      "grad_norm": 58.88878631591797,
+      "learning_rate": 7.783513756865682e-07,
+      "loss": 0.5589,
+      "step": 411
+    },
+    {
+      "epoch": 3.3225806451612905,
+      "grad_norm": 72.84752655029297,
+      "learning_rate": 7.7462720642491e-07,
+      "loss": 2.3565,
+      "step": 412
+    },
+    {
+      "epoch": 3.3306451612903225,
+      "grad_norm": 57.83047866821289,
+      "learning_rate": 7.709030371632517e-07,
+      "loss": 0.3691,
+      "step": 413
+    },
+    {
+      "epoch": 3.338709677419355,
+      "grad_norm": 38.37238311767578,
+      "learning_rate": 7.671788679015936e-07,
+      "loss": 0.2926,
+      "step": 414
+    },
+    {
+      "epoch": 3.346774193548387,
+      "grad_norm": 19.532394409179688,
+      "learning_rate": 7.634546986399353e-07,
+      "loss": 0.6592,
+      "step": 415
+    },
+    {
+      "epoch": 3.3548387096774195,
+      "grad_norm": 97.23928833007812,
+      "learning_rate": 7.597305293782772e-07,
+      "loss": 0.271,
+      "step": 416
+    },
+    {
+      "epoch": 3.3629032258064515,
+      "grad_norm": 107.05040740966797,
+      "learning_rate": 7.560063601166189e-07,
+      "loss": 0.3408,
+      "step": 417
+    },
+    {
+      "epoch": 3.370967741935484,
+      "grad_norm": 38.197364807128906,
+      "learning_rate": 7.522821908549607e-07,
+      "loss": 1.5913,
+      "step": 418
+    },
+    {
+      "epoch": 3.379032258064516,
+      "grad_norm": 74.39269256591797,
+      "learning_rate": 7.485580215933025e-07,
+      "loss": 0.534,
+      "step": 419
+    },
+    {
+      "epoch": 3.3870967741935485,
+      "grad_norm": 361.89697265625,
+      "learning_rate": 7.448338523316442e-07,
+      "loss": 11.104,
+      "step": 420
+    },
+    {
+      "epoch": 3.3951612903225805,
+      "grad_norm": 30.99224090576172,
+      "learning_rate": 7.411096830699861e-07,
+      "loss": 0.5005,
+      "step": 421
+    },
+    {
+      "epoch": 3.403225806451613,
+      "grad_norm": 187.10740661621094,
+      "learning_rate": 7.373855138083278e-07,
+      "loss": 1.9135,
+      "step": 422
+    },
+    {
+      "epoch": 3.411290322580645,
+      "grad_norm": 418.1996154785156,
+      "learning_rate": 7.336613445466696e-07,
+      "loss": 14.1697,
+      "step": 423
+    },
+    {
+      "epoch": 3.4193548387096775,
+      "grad_norm": 331.3507080078125,
+      "learning_rate": 7.299371752850113e-07,
+      "loss": 18.4025,
+      "step": 424
+    },
+    {
+      "epoch": 3.4274193548387095,
+      "grad_norm": 96.46886444091797,
+      "learning_rate": 7.262130060233531e-07,
+      "loss": 0.506,
+      "step": 425
+    },
+    {
+      "epoch": 3.435483870967742,
+      "grad_norm": 93.3166732788086,
+      "learning_rate": 7.224888367616949e-07,
+      "loss": 1.3464,
+      "step": 426
+    },
+    {
+      "epoch": 3.443548387096774,
+      "grad_norm": 60.4049072265625,
+      "learning_rate": 7.187646675000366e-07,
+      "loss": 0.2097,
+      "step": 427
+    },
+    {
+      "epoch": 3.4516129032258065,
+      "grad_norm": 43.35322570800781,
+      "learning_rate": 7.150404982383785e-07,
+      "loss": 0.4467,
+      "step": 428
+    },
+    {
+      "epoch": 3.4596774193548385,
+      "grad_norm": 45.652435302734375,
+      "learning_rate": 7.113163289767202e-07,
+      "loss": 0.4335,
+      "step": 429
+    },
+    {
+      "epoch": 3.467741935483871,
+      "grad_norm": 80.38420867919922,
+      "learning_rate": 7.075921597150621e-07,
+      "loss": 0.4553,
+      "step": 430
+    },
+    {
+      "epoch": 3.475806451612903,
+      "grad_norm": 332.3706970214844,
+      "learning_rate": 7.038679904534037e-07,
+      "loss": 22.9866,
+      "step": 431
+    },
+    {
+      "epoch": 3.4838709677419355,
+      "grad_norm": 263.8254699707031,
+      "learning_rate": 7.001438211917456e-07,
+      "loss": 5.5233,
+      "step": 432
+    },
+    {
+      "epoch": 3.491935483870968,
+      "grad_norm": 73.21070098876953,
+      "learning_rate": 6.964196519300874e-07,
+      "loss": 0.7063,
+      "step": 433
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 169.56521606445312,
+      "learning_rate": 6.926954826684291e-07,
+      "loss": 1.6993,
+      "step": 434
+    },
+    {
+      "epoch": 3.508064516129032,
+      "grad_norm": 29.43960189819336,
+      "learning_rate": 6.88971313406771e-07,
+      "loss": 0.5154,
+      "step": 435
+    },
+    {
+      "epoch": 3.5161290322580645,
+      "grad_norm": 29.282577514648438,
+      "learning_rate": 6.852471441451127e-07,
+      "loss": 0.3941,
+      "step": 436
+    },
+    {
+      "epoch": 3.524193548387097,
+      "grad_norm": 24.749528884887695,
+      "learning_rate": 6.815229748834545e-07,
+      "loss": 0.0609,
+      "step": 437
+    },
+    {
+      "epoch": 3.532258064516129,
+      "grad_norm": 268.2690734863281,
+      "learning_rate": 6.777988056217962e-07,
+      "loss": 2.3492,
+      "step": 438
+    },
+    {
+      "epoch": 3.540322580645161,
+      "grad_norm": 61.73430252075195,
+      "learning_rate": 6.74074636360138e-07,
+      "loss": 0.468,
+      "step": 439
+    },
+    {
+      "epoch": 3.5483870967741935,
+      "grad_norm": 148.60581970214844,
+      "learning_rate": 6.703504670984798e-07,
+      "loss": 4.6193,
+      "step": 440
+    },
+    {
+      "epoch": 3.556451612903226,
+      "grad_norm": 135.78884887695312,
+      "learning_rate": 6.666262978368215e-07,
+      "loss": 1.1676,
+      "step": 441
+    },
+    {
+      "epoch": 3.564516129032258,
+      "grad_norm": 49.250648498535156,
+      "learning_rate": 6.629021285751634e-07,
+      "loss": 0.2826,
+      "step": 442
+    },
+    {
+      "epoch": 3.5725806451612905,
+      "grad_norm": 304.4104919433594,
+      "learning_rate": 6.591779593135051e-07,
+      "loss": 5.2831,
+      "step": 443
+    },
+    {
+      "epoch": 3.5806451612903225,
+      "grad_norm": 46.258243560791016,
+      "learning_rate": 6.55453790051847e-07,
+      "loss": 0.741,
+      "step": 444
+    },
+    {
+      "epoch": 3.588709677419355,
+      "grad_norm": 24.82040786743164,
+      "learning_rate": 6.517296207901888e-07,
+      "loss": 0.9014,
+      "step": 445
+    },
+    {
+      "epoch": 3.596774193548387,
+      "grad_norm": 308.60443115234375,
+      "learning_rate": 6.480054515285305e-07,
+      "loss": 19.5394,
+      "step": 446
+    },
+    {
+      "epoch": 3.6048387096774195,
+      "grad_norm": 117.33186340332031,
+      "learning_rate": 6.442812822668723e-07,
+      "loss": 10.2945,
+      "step": 447
+    },
+    {
+      "epoch": 3.6129032258064515,
+      "grad_norm": 21.672733306884766,
+      "learning_rate": 6.40557113005214e-07,
+      "loss": 0.0768,
+      "step": 448
+    },
+    {
+      "epoch": 3.620967741935484,
+      "grad_norm": 68.24752807617188,
+      "learning_rate": 6.368329437435559e-07,
+      "loss": 0.3628,
+      "step": 449
+    },
+    {
+      "epoch": 3.629032258064516,
+      "grad_norm": 49.00190734863281,
+      "learning_rate": 6.331087744818976e-07,
+      "loss": 1.1004,
+      "step": 450
+    },
+    {
+      "epoch": 3.6370967741935485,
+      "grad_norm": 5.552606582641602,
+      "learning_rate": 6.293846052202394e-07,
+      "loss": 0.0698,
+      "step": 451
+    },
+    {
+      "epoch": 3.6451612903225805,
+      "grad_norm": 28.109525680541992,
+      "learning_rate": 6.256604359585812e-07,
+      "loss": 0.0813,
+      "step": 452
+    },
+    {
+      "epoch": 3.653225806451613,
+      "grad_norm": 12.806706428527832,
+      "learning_rate": 6.219362666969229e-07,
+      "loss": 0.2015,
+      "step": 453
+    },
+    {
+      "epoch": 3.661290322580645,
+      "grad_norm": 320.7244873046875,
+      "learning_rate": 6.182120974352647e-07,
+      "loss": 13.654,
+      "step": 454
+    },
+    {
+      "epoch": 3.6693548387096775,
+      "grad_norm": 267.62042236328125,
+      "learning_rate": 6.144879281736064e-07,
+      "loss": 21.5447,
+      "step": 455
+    },
+    {
+      "epoch": 3.6774193548387095,
+      "grad_norm": 177.5928192138672,
+      "learning_rate": 6.107637589119483e-07,
+      "loss": 1.9395,
+      "step": 456
+    },
+    {
+      "epoch": 3.685483870967742,
+      "grad_norm": 64.51294708251953,
+      "learning_rate": 6.0703958965029e-07,
+      "loss": 0.2744,
+      "step": 457
+    },
+    {
+      "epoch": 3.693548387096774,
+      "grad_norm": 225.3753662109375,
+      "learning_rate": 6.033154203886319e-07,
+      "loss": 11.8973,
+      "step": 458
+    },
+    {
+      "epoch": 3.7016129032258065,
+      "grad_norm": 138.8841094970703,
+      "learning_rate": 5.995912511269737e-07,
+      "loss": 0.8054,
+      "step": 459
+    },
+    {
+      "epoch": 3.709677419354839,
+      "grad_norm": 194.67889404296875,
+      "learning_rate": 5.958670818653154e-07,
+      "loss": 1.5434,
+      "step": 460
+    },
+    {
+      "epoch": 3.717741935483871,
+      "grad_norm": 84.84955596923828,
+      "learning_rate": 5.921429126036572e-07,
+      "loss": 0.7915,
+      "step": 461
+    },
+    {
+      "epoch": 3.725806451612903,
+      "grad_norm": 130.38189697265625,
+      "learning_rate": 5.884187433419989e-07,
+      "loss": 2.1479,
+      "step": 462
+    },
+    {
+      "epoch": 3.7338709677419355,
+      "grad_norm": 213.24395751953125,
+      "learning_rate": 5.846945740803408e-07,
+      "loss": 3.5013,
+      "step": 463
+    },
+    {
+      "epoch": 3.741935483870968,
+      "grad_norm": 207.3053436279297,
+      "learning_rate": 5.809704048186825e-07,
+      "loss": 17.0066,
+      "step": 464
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 41.873130798339844,
+      "learning_rate": 5.772462355570243e-07,
+      "loss": 0.6512,
+      "step": 465
+    },
+    {
+      "epoch": 3.758064516129032,
+      "grad_norm": 1999.728515625,
+      "learning_rate": 5.735220662953661e-07,
+      "loss": 134.0836,
+      "step": 466
+    },
+    {
+      "epoch": 3.7661290322580645,
+      "grad_norm": 66.1317367553711,
+      "learning_rate": 5.697978970337078e-07,
+      "loss": 0.7145,
+      "step": 467
+    },
+    {
+      "epoch": 3.774193548387097,
+      "grad_norm": 211.88003540039062,
+      "learning_rate": 5.660737277720496e-07,
+      "loss": 14.5329,
+      "step": 468
+    },
+    {
+      "epoch": 3.782258064516129,
+      "grad_norm": 81.95831298828125,
+      "learning_rate": 5.623495585103914e-07,
+      "loss": 0.6901,
+      "step": 469
+    },
+    {
+      "epoch": 3.790322580645161,
+      "grad_norm": 54.79182815551758,
+      "learning_rate": 5.586253892487332e-07,
+      "loss": 0.6836,
+      "step": 470
+    },
+    {
+      "epoch": 3.7983870967741935,
+      "grad_norm": 65.97078704833984,
+      "learning_rate": 5.54901219987075e-07,
+      "loss": 5.0991,
+      "step": 471
+    },
+    {
+      "epoch": 3.806451612903226,
+      "grad_norm": 85.9794921875,
+      "learning_rate": 5.511770507254167e-07,
+      "loss": 0.2896,
+      "step": 472
+    },
+    {
+      "epoch": 3.814516129032258,
+      "grad_norm": 71.18566131591797,
+      "learning_rate": 5.474528814637585e-07,
+      "loss": 0.4102,
+      "step": 473
+    },
+    {
+      "epoch": 3.8225806451612905,
+      "grad_norm": 62.09203338623047,
+      "learning_rate": 5.437287122021003e-07,
+      "loss": 0.3776,
+      "step": 474
+    },
+    {
+      "epoch": 3.8306451612903225,
+      "grad_norm": 128.81930541992188,
+      "learning_rate": 5.40004542940442e-07,
+      "loss": 1.9145,
+      "step": 475
+    },
+    {
+      "epoch": 3.838709677419355,
+      "grad_norm": 312.0795593261719,
+      "learning_rate": 5.362803736787838e-07,
+      "loss": 2.3146,
+      "step": 476
+    },
+    {
+      "epoch": 3.846774193548387,
+      "grad_norm": 265.22119140625,
+      "learning_rate": 5.325562044171257e-07,
+      "loss": 4.3763,
+      "step": 477
+    },
+    {
+      "epoch": 3.8548387096774195,
+      "grad_norm": 207.3704376220703,
+      "learning_rate": 5.288320351554674e-07,
+      "loss": 20.7729,
+      "step": 478
+    },
+    {
+      "epoch": 3.8629032258064515,
+      "grad_norm": 418.6147766113281,
+      "learning_rate": 5.251078658938092e-07,
+      "loss": 0.3353,
+      "step": 479
+    },
+    {
+      "epoch": 3.870967741935484,
+      "grad_norm": 69.67323303222656,
+      "learning_rate": 5.21383696632151e-07,
+      "loss": 1.5328,
+      "step": 480
+    },
+    {
+      "epoch": 3.879032258064516,
+      "grad_norm": 18.951583862304688,
+      "learning_rate": 5.176595273704927e-07,
+      "loss": 0.1113,
+      "step": 481
+    },
+    {
+      "epoch": 3.8870967741935485,
+      "grad_norm": 271.2349853515625,
+      "learning_rate": 5.139353581088345e-07,
+      "loss": 2.2297,
+      "step": 482
+    },
+    {
+      "epoch": 3.8951612903225805,
+      "grad_norm": 78.0976791381836,
+      "learning_rate": 5.102111888471763e-07,
+      "loss": 0.9436,
+      "step": 483
+    },
+    {
+      "epoch": 3.903225806451613,
+      "grad_norm": 124.9957046508789,
+      "learning_rate": 5.064870195855181e-07,
+      "loss": 11.5244,
+      "step": 484
+    },
+    {
+      "epoch": 3.911290322580645,
+      "grad_norm": 119.20372009277344,
+      "learning_rate": 5.027628503238599e-07,
+      "loss": 1.2656,
+      "step": 485
+    },
+    {
+      "epoch": 3.9193548387096775,
+      "grad_norm": 111.48886108398438,
+      "learning_rate": 4.990386810622016e-07,
+      "loss": 0.9424,
+      "step": 486
+    },
+    {
+      "epoch": 3.9274193548387095,
+      "grad_norm": 44.763328552246094,
+      "learning_rate": 4.953145118005434e-07,
+      "loss": 0.406,
+      "step": 487
+    },
+    {
+      "epoch": 3.935483870967742,
+      "grad_norm": 33.71810531616211,
+      "learning_rate": 4.915903425388852e-07,
+      "loss": 0.7287,
+      "step": 488
+    },
+    {
+      "epoch": 3.943548387096774,
+      "grad_norm": 314.32232666015625,
+      "learning_rate": 4.87866173277227e-07,
+      "loss": 14.9761,
+      "step": 489
+    },
+    {
+      "epoch": 3.9516129032258065,
+      "grad_norm": 161.79278564453125,
+      "learning_rate": 4.841420040155687e-07,
+      "loss": 1.6247,
+      "step": 490
+    },
+    {
+      "epoch": 3.959677419354839,
+      "grad_norm": 19.090133666992188,
+      "learning_rate": 4.804178347539106e-07,
+      "loss": 0.5247,
+      "step": 491
+    },
+    {
+      "epoch": 3.967741935483871,
+      "grad_norm": 227.18374633789062,
+      "learning_rate": 4.766936654922523e-07,
+      "loss": 1.9452,
+      "step": 492
+    },
+    {
+      "epoch": 3.975806451612903,
+      "grad_norm": 42.912105560302734,
+      "learning_rate": 4.729694962305941e-07,
+      "loss": 0.6671,
+      "step": 493
+    },
+    {
+      "epoch": 3.9838709677419355,
+      "grad_norm": 229.20773315429688,
+      "learning_rate": 4.6924532696893585e-07,
+      "loss": 23.1584,
+      "step": 494
+    },
+    {
+      "epoch": 3.991935483870968,
+      "grad_norm": 16.33729362487793,
+      "learning_rate": 4.655211577072776e-07,
+      "loss": 0.6326,
+      "step": 495
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 31.148290634155273,
+      "learning_rate": 4.6179698844561943e-07,
+      "loss": 0.083,
+      "step": 496
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 3.6658711433410645,
+      "eval_mae": 1.1627670526504517,
+      "eval_mse": 3.6658709049224854,
+      "eval_r2": -0.0036516189575195312,
+      "eval_rmse": 1.9146464177289981,
+      "eval_runtime": 1.3825,
+      "eval_samples_per_second": 39.783,
+      "eval_smape": 49.43971037864685,
+      "eval_steps_per_second": 10.127,
+      "step": 496
+    },
+    {
+      "epoch": 4.008064516129032,
+      "grad_norm": 40.192630767822266,
+      "learning_rate": 4.5807281918396124e-07,
+      "loss": 0.4606,
+      "step": 497
+    },
+    {
+      "epoch": 4.016129032258065,
+      "grad_norm": 34.81431579589844,
+      "learning_rate": 4.54348649922303e-07,
+      "loss": 0.6634,
+      "step": 498
+    },
+    {
+      "epoch": 4.024193548387097,
+      "grad_norm": 44.28828048706055,
+      "learning_rate": 4.5062448066064477e-07,
+      "loss": 0.4906,
+      "step": 499
+    },
+    {
+      "epoch": 4.032258064516129,
+      "grad_norm": 144.46249389648438,
+      "learning_rate": 4.4690031139898653e-07,
+      "loss": 1.7835,
+      "step": 500
+    },
+    {
+      "epoch": 4.040322580645161,
+      "grad_norm": 46.331146240234375,
+      "learning_rate": 4.431761421373283e-07,
+      "loss": 1.3651,
+      "step": 501
+    },
+    {
+      "epoch": 4.048387096774194,
+      "grad_norm": 217.1621856689453,
+      "learning_rate": 4.3945197287567006e-07,
+      "loss": 2.7421,
+      "step": 502
+    },
+    {
+      "epoch": 4.056451612903226,
+      "grad_norm": 90.83784484863281,
+      "learning_rate": 4.357278036140119e-07,
+      "loss": 0.5116,
+      "step": 503
+    },
+    {
+      "epoch": 4.064516129032258,
+      "grad_norm": 176.08767700195312,
+      "learning_rate": 4.320036343523537e-07,
+      "loss": 1.0464,
+      "step": 504
+    },
+    {
+      "epoch": 4.07258064516129,
+      "grad_norm": 27.863637924194336,
+      "learning_rate": 4.2827946509069546e-07,
+      "loss": 0.4235,
+      "step": 505
+    },
+    {
+      "epoch": 4.080645161290323,
+      "grad_norm": 33.19697189331055,
+      "learning_rate": 4.245552958290372e-07,
+      "loss": 0.261,
+      "step": 506
+    },
+    {
+      "epoch": 4.088709677419355,
+      "grad_norm": 124.42127227783203,
+      "learning_rate": 4.20831126567379e-07,
+      "loss": 9.2822,
+      "step": 507
+    },
+    {
+      "epoch": 4.096774193548387,
+      "grad_norm": 21.310752868652344,
+      "learning_rate": 4.1710695730572075e-07,
+      "loss": 0.0854,
+      "step": 508
+    },
+    {
+      "epoch": 4.104838709677419,
+      "grad_norm": 163.42176818847656,
+      "learning_rate": 4.1338278804406256e-07,
+      "loss": 0.4038,
+      "step": 509
+    },
+    {
+      "epoch": 4.112903225806452,
+      "grad_norm": 68.20549774169922,
+      "learning_rate": 4.0965861878240433e-07,
+      "loss": 0.4068,
+      "step": 510
+    },
+    {
+      "epoch": 4.120967741935484,
+      "grad_norm": 96.01313781738281,
+      "learning_rate": 4.0593444952074614e-07,
+      "loss": 1.2494,
+      "step": 511
+    },
+    {
+      "epoch": 4.129032258064516,
+      "grad_norm": 153.09539794921875,
+      "learning_rate": 4.022102802590879e-07,
+      "loss": 0.2314,
+      "step": 512
+    },
+    {
+      "epoch": 4.137096774193548,
+      "grad_norm": 231.76513671875,
+      "learning_rate": 3.9848611099742967e-07,
+      "loss": 12.2955,
+      "step": 513
+    },
+    {
+      "epoch": 4.145161290322581,
+      "grad_norm": 497.2821044921875,
+      "learning_rate": 3.9476194173577143e-07,
+      "loss": 14.7263,
+      "step": 514
+    },
+    {
+      "epoch": 4.153225806451613,
+      "grad_norm": 100.38178253173828,
+      "learning_rate": 3.910377724741132e-07,
+      "loss": 0.6703,
+      "step": 515
+    },
+    {
+      "epoch": 4.161290322580645,
+      "grad_norm": 20.106956481933594,
+      "learning_rate": 3.87313603212455e-07,
+      "loss": 0.2879,
+      "step": 516
+    },
+    {
+      "epoch": 4.169354838709677,
+      "grad_norm": 421.3139343261719,
+      "learning_rate": 3.835894339507968e-07,
+      "loss": 4.0681,
+      "step": 517
+    },
+    {
+      "epoch": 4.17741935483871,
+      "grad_norm": 224.01904296875,
+      "learning_rate": 3.798652646891386e-07,
+      "loss": 5.5452,
+      "step": 518
+    },
+    {
+      "epoch": 4.185483870967742,
+      "grad_norm": 75.63765716552734,
+      "learning_rate": 3.7614109542748036e-07,
+      "loss": 1.4427,
+      "step": 519
+    },
+    {
+      "epoch": 4.193548387096774,
+      "grad_norm": 164.29302978515625,
+      "learning_rate": 3.724169261658221e-07,
+      "loss": 18.1201,
+      "step": 520
+    },
+    {
+      "epoch": 4.201612903225806,
+      "grad_norm": 23.143348693847656,
+      "learning_rate": 3.686927569041639e-07,
+      "loss": 0.5357,
+      "step": 521
+    },
+    {
+      "epoch": 4.209677419354839,
+      "grad_norm": 26.469867706298828,
+      "learning_rate": 3.6496858764250565e-07,
+      "loss": 0.6538,
+      "step": 522
+    },
+    {
+      "epoch": 4.217741935483871,
+      "grad_norm": 184.11761474609375,
+      "learning_rate": 3.6124441838084746e-07,
+      "loss": 20.6509,
+      "step": 523
+    },
+    {
+      "epoch": 4.225806451612903,
+      "grad_norm": 124.52848815917969,
+      "learning_rate": 3.5752024911918923e-07,
+      "loss": 1.5531,
+      "step": 524
+    },
+    {
+      "epoch": 4.233870967741935,
+      "grad_norm": 147.81829833984375,
+      "learning_rate": 3.5379607985753104e-07,
+      "loss": 4.8944,
+      "step": 525
+    },
+    {
+      "epoch": 4.241935483870968,
+      "grad_norm": 54.121559143066406,
+      "learning_rate": 3.500719105958728e-07,
+      "loss": 0.5221,
+      "step": 526
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 83.79217529296875,
+      "learning_rate": 3.4634774133421457e-07,
+      "loss": 1.6032,
+      "step": 527
+    },
+    {
+      "epoch": 4.258064516129032,
+      "grad_norm": 27.414520263671875,
+      "learning_rate": 3.4262357207255633e-07,
+      "loss": 0.4982,
+      "step": 528
+    },
+    {
+      "epoch": 4.266129032258064,
+      "grad_norm": 176.4253692626953,
+      "learning_rate": 3.388994028108981e-07,
+      "loss": 3.1434,
+      "step": 529
+    },
+    {
+      "epoch": 4.274193548387097,
+      "grad_norm": 63.877079010009766,
+      "learning_rate": 3.351752335492399e-07,
+      "loss": 0.5721,
+      "step": 530
+    },
+    {
+      "epoch": 4.282258064516129,
+      "grad_norm": 105.18173217773438,
+      "learning_rate": 3.314510642875817e-07,
+      "loss": 0.4412,
+      "step": 531
+    },
+    {
+      "epoch": 4.290322580645161,
+      "grad_norm": 18.525733947753906,
+      "learning_rate": 3.277268950259235e-07,
+      "loss": 0.1196,
+      "step": 532
+    },
+    {
+      "epoch": 4.298387096774194,
+      "grad_norm": 126.90892028808594,
+      "learning_rate": 3.2400272576426526e-07,
+      "loss": 1.7735,
+      "step": 533
+    },
+    {
+      "epoch": 4.306451612903226,
+      "grad_norm": 13.654480934143066,
+      "learning_rate": 3.20278556502607e-07,
+      "loss": 0.8871,
+      "step": 534
+    },
+    {
+      "epoch": 4.314516129032258,
+      "grad_norm": 183.49330139160156,
+      "learning_rate": 3.165543872409488e-07,
+      "loss": 1.7886,
+      "step": 535
+    },
+    {
+      "epoch": 4.32258064516129,
+      "grad_norm": 651.6168823242188,
+      "learning_rate": 3.128302179792906e-07,
+      "loss": 2.3902,
+      "step": 536
+    },
+    {
+      "epoch": 4.330645161290323,
+      "grad_norm": 76.452880859375,
+      "learning_rate": 3.0910604871763236e-07,
+      "loss": 0.6681,
+      "step": 537
+    },
+    {
+      "epoch": 4.338709677419355,
+      "grad_norm": 70.49311065673828,
+      "learning_rate": 3.0538187945597413e-07,
+      "loss": 0.3893,
+      "step": 538
+    },
+    {
+      "epoch": 4.346774193548387,
+      "grad_norm": 24.761642456054688,
+      "learning_rate": 3.0165771019431594e-07,
+      "loss": 0.0661,
+      "step": 539
+    },
+    {
+      "epoch": 4.354838709677419,
+      "grad_norm": 54.143436431884766,
+      "learning_rate": 2.979335409326577e-07,
+      "loss": 0.6459,
+      "step": 540
+    },
+    {
+      "epoch": 4.362903225806452,
+      "grad_norm": 70.6890640258789,
+      "learning_rate": 2.9420937167099947e-07,
+      "loss": 0.2079,
+      "step": 541
+    },
+    {
+      "epoch": 4.370967741935484,
+      "grad_norm": 44.36699295043945,
+      "learning_rate": 2.9048520240934123e-07,
+      "loss": 0.2182,
+      "step": 542
+    },
+    {
+      "epoch": 4.379032258064516,
+      "grad_norm": 24.9780330657959,
+      "learning_rate": 2.8676103314768305e-07,
+      "loss": 0.1024,
+      "step": 543
+    },
+    {
+      "epoch": 4.387096774193548,
+      "grad_norm": 600.659423828125,
+      "learning_rate": 2.830368638860248e-07,
+      "loss": 20.363,
+      "step": 544
+    },
+    {
+      "epoch": 4.395161290322581,
+      "grad_norm": 11.253837585449219,
+      "learning_rate": 2.793126946243666e-07,
+      "loss": 0.1866,
+      "step": 545
+    },
+    {
+      "epoch": 4.403225806451613,
+      "grad_norm": 28.648752212524414,
+      "learning_rate": 2.7558852536270834e-07,
+      "loss": 0.3628,
+      "step": 546
+    },
+    {
+      "epoch": 4.411290322580645,
+      "grad_norm": 63.38084030151367,
+      "learning_rate": 2.7186435610105016e-07,
+      "loss": 0.7742,
+      "step": 547
+    },
+    {
+      "epoch": 4.419354838709677,
+      "grad_norm": 81.5975112915039,
+      "learning_rate": 2.681401868393919e-07,
+      "loss": 2.4472,
+      "step": 548
+    },
+    {
+      "epoch": 4.42741935483871,
+      "grad_norm": 96.23023223876953,
+      "learning_rate": 2.644160175777337e-07,
+      "loss": 0.5348,
+      "step": 549
+    },
+    {
+      "epoch": 4.435483870967742,
+      "grad_norm": 698.8837280273438,
+      "learning_rate": 2.606918483160755e-07,
+      "loss": 6.1945,
+      "step": 550
+    },
+    {
+      "epoch": 4.443548387096774,
+      "grad_norm": 105.67257690429688,
+      "learning_rate": 2.5696767905441726e-07,
+      "loss": 1.0181,
+      "step": 551
+    },
+    {
+      "epoch": 4.451612903225806,
+      "grad_norm": 369.3524475097656,
+      "learning_rate": 2.5324350979275903e-07,
+      "loss": 10.3212,
+      "step": 552
+    },
+    {
+      "epoch": 4.459677419354839,
+      "grad_norm": 68.52635955810547,
+      "learning_rate": 2.495193405311008e-07,
+      "loss": 0.8354,
+      "step": 553
+    },
+    {
+      "epoch": 4.467741935483871,
+      "grad_norm": 34.14206314086914,
+      "learning_rate": 2.457951712694426e-07,
+      "loss": 0.5059,
+      "step": 554
+    },
+    {
+      "epoch": 4.475806451612903,
+      "grad_norm": 160.20071411132812,
+      "learning_rate": 2.4207100200778437e-07,
+      "loss": 0.5112,
+      "step": 555
+    },
+    {
+      "epoch": 4.483870967741936,
+      "grad_norm": 38.58085632324219,
+      "learning_rate": 2.3834683274612616e-07,
+      "loss": 0.3525,
+      "step": 556
+    },
+    {
+      "epoch": 4.491935483870968,
+      "grad_norm": 65.05552673339844,
+      "learning_rate": 2.3462266348446792e-07,
+      "loss": 0.3484,
+      "step": 557
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 394.2702941894531,
+      "learning_rate": 2.3089849422280971e-07,
+      "loss": 20.5987,
+      "step": 558
+    },
+    {
+      "epoch": 4.508064516129032,
+      "grad_norm": 11.85052490234375,
+      "learning_rate": 2.271743249611515e-07,
+      "loss": 0.0769,
+      "step": 559
+    },
+    {
+      "epoch": 4.516129032258064,
+      "grad_norm": 186.5146484375,
+      "learning_rate": 2.2345015569949327e-07,
+      "loss": 11.3414,
+      "step": 560
+    },
+    {
+      "epoch": 4.524193548387097,
+      "grad_norm": 35.12519454956055,
+      "learning_rate": 2.1972598643783503e-07,
+      "loss": 0.5073,
+      "step": 561
+    },
+    {
+      "epoch": 4.532258064516129,
+      "grad_norm": 159.3316650390625,
+      "learning_rate": 2.1600181717617685e-07,
+      "loss": 1.5338,
+      "step": 562
+    },
+    {
+      "epoch": 4.540322580645161,
+      "grad_norm": 376.8158264160156,
+      "learning_rate": 2.122776479145186e-07,
+      "loss": 16.0564,
+      "step": 563
+    },
+    {
+      "epoch": 4.548387096774194,
+      "grad_norm": 38.60029220581055,
+      "learning_rate": 2.0855347865286037e-07,
+      "loss": 0.3901,
+      "step": 564
+    },
+    {
+      "epoch": 4.556451612903226,
+      "grad_norm": 14.886783599853516,
+      "learning_rate": 2.0482930939120216e-07,
+      "loss": 0.2495,
+      "step": 565
+    },
+    {
+      "epoch": 4.564516129032258,
+      "grad_norm": 274.8218688964844,
+      "learning_rate": 2.0110514012954395e-07,
+      "loss": 3.6786,
+      "step": 566
+    },
+    {
+      "epoch": 4.57258064516129,
+      "grad_norm": 865.7882080078125,
+      "learning_rate": 1.9738097086788572e-07,
+      "loss": 5.4778,
+      "step": 567
+    },
+    {
+      "epoch": 4.580645161290323,
+      "grad_norm": 167.94483947753906,
+      "learning_rate": 1.936568016062275e-07,
+      "loss": 13.7189,
+      "step": 568
+    },
+    {
+      "epoch": 4.588709677419355,
+      "grad_norm": 28.00385856628418,
+      "learning_rate": 1.899326323445693e-07,
+      "loss": 0.9164,
+      "step": 569
+    },
+    {
+      "epoch": 4.596774193548387,
+      "grad_norm": 492.3658752441406,
+      "learning_rate": 1.8620846308291106e-07,
+      "loss": 10.5385,
+      "step": 570
+    },
+    {
+      "epoch": 4.604838709677419,
+      "grad_norm": 91.81761169433594,
+      "learning_rate": 1.8248429382125282e-07,
+      "loss": 1.3811,
+      "step": 571
+    },
+    {
+      "epoch": 4.612903225806452,
+      "grad_norm": 53.9609489440918,
+      "learning_rate": 1.7876012455959461e-07,
+      "loss": 0.4604,
+      "step": 572
+    },
+    {
+      "epoch": 4.620967741935484,
+      "grad_norm": 91.66685485839844,
+      "learning_rate": 1.750359552979364e-07,
+      "loss": 3.1813,
+      "step": 573
+    },
+    {
+      "epoch": 4.629032258064516,
+      "grad_norm": 61.261680603027344,
+      "learning_rate": 1.7131178603627817e-07,
+      "loss": 0.2714,
+      "step": 574
+    },
+    {
+      "epoch": 4.637096774193548,
+      "grad_norm": 86.26275634765625,
+      "learning_rate": 1.6758761677461996e-07,
+      "loss": 1.7492,
+      "step": 575
+    },
+    {
+      "epoch": 4.645161290322581,
+      "grad_norm": 35.08537673950195,
+      "learning_rate": 1.6386344751296175e-07,
+      "loss": 1.0437,
+      "step": 576
+    },
+    {
+      "epoch": 4.653225806451613,
+      "grad_norm": 71.87916564941406,
+      "learning_rate": 1.601392782513035e-07,
+      "loss": 0.5856,
+      "step": 577
+    },
+    {
+      "epoch": 4.661290322580645,
+      "grad_norm": 362.8409118652344,
+      "learning_rate": 1.564151089896453e-07,
+      "loss": 23.2117,
+      "step": 578
+    },
+    {
+      "epoch": 4.669354838709677,
+      "grad_norm": 37.816200256347656,
+      "learning_rate": 1.5269093972798706e-07,
+      "loss": 0.5448,
+      "step": 579
+    },
+    {
+      "epoch": 4.67741935483871,
+      "grad_norm": 152.78346252441406,
+      "learning_rate": 1.4896677046632885e-07,
+      "loss": 7.6629,
+      "step": 580
+    },
+    {
+      "epoch": 4.685483870967742,
+      "grad_norm": 51.52225875854492,
+      "learning_rate": 1.4524260120467062e-07,
+      "loss": 0.3167,
+      "step": 581
+    },
+    {
+      "epoch": 4.693548387096774,
+      "grad_norm": 34.78017044067383,
+      "learning_rate": 1.415184319430124e-07,
+      "loss": 0.1883,
+      "step": 582
+    },
+    {
+      "epoch": 4.701612903225806,
+      "grad_norm": 33.89111328125,
+      "learning_rate": 1.3779426268135417e-07,
+      "loss": 0.5733,
+      "step": 583
+    },
+    {
+      "epoch": 4.709677419354839,
+      "grad_norm": 165.656494140625,
+      "learning_rate": 1.3407009341969596e-07,
+      "loss": 15.5164,
+      "step": 584
+    },
+    {
+      "epoch": 4.717741935483871,
+      "grad_norm": 187.8523712158203,
+      "learning_rate": 1.3034592415803775e-07,
+      "loss": 2.6475,
+      "step": 585
+    },
+    {
+      "epoch": 4.725806451612903,
+      "grad_norm": 226.56434631347656,
+      "learning_rate": 1.2662175489637951e-07,
+      "loss": 19.2786,
+      "step": 586
+    },
+    {
+      "epoch": 4.733870967741936,
+      "grad_norm": 93.69617462158203,
+      "learning_rate": 1.228975856347213e-07,
+      "loss": 0.3997,
+      "step": 587
+    },
+    {
+      "epoch": 4.741935483870968,
+      "grad_norm": 70.5687255859375,
+      "learning_rate": 1.1917341637306308e-07,
+      "loss": 0.8395,
+      "step": 588
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 139.98455810546875,
+      "learning_rate": 1.1544924711140486e-07,
+      "loss": 1.5936,
+      "step": 589
+    },
+    {
+      "epoch": 4.758064516129032,
+      "grad_norm": 61.73574447631836,
+      "learning_rate": 1.1172507784974663e-07,
+      "loss": 1.6269,
+      "step": 590
+    },
+    {
+      "epoch": 4.766129032258064,
+      "grad_norm": 267.8656921386719,
+      "learning_rate": 1.0800090858808842e-07,
+      "loss": 10.6706,
+      "step": 591
+    },
+    {
+      "epoch": 4.774193548387097,
+      "grad_norm": 103.93309783935547,
+      "learning_rate": 1.0427673932643019e-07,
+      "loss": 0.7362,
+      "step": 592
+    },
+    {
+      "epoch": 4.782258064516129,
+      "grad_norm": 33.18098831176758,
+      "learning_rate": 1.0055257006477198e-07,
+      "loss": 0.5736,
+      "step": 593
+    },
+    {
+      "epoch": 4.790322580645161,
+      "grad_norm": 67.95414733886719,
+      "learning_rate": 9.682840080311375e-08,
+      "loss": 0.3606,
+      "step": 594
+    },
+    {
+      "epoch": 4.798387096774194,
+      "grad_norm": 374.37744140625,
+      "learning_rate": 9.310423154145553e-08,
+      "loss": 4.9431,
+      "step": 595
+    },
+    {
+      "epoch": 4.806451612903226,
+      "grad_norm": 27.907346725463867,
+      "learning_rate": 8.938006227979731e-08,
+      "loss": 0.1885,
+      "step": 596
+    },
+    {
+      "epoch": 4.814516129032258,
+      "grad_norm": 54.07706069946289,
+      "learning_rate": 8.565589301813908e-08,
+      "loss": 1.0378,
+      "step": 597
+    },
+    {
+      "epoch": 4.82258064516129,
+      "grad_norm": 192.1484832763672,
+      "learning_rate": 8.193172375648087e-08,
+      "loss": 10.6858,
+      "step": 598
+    },
+    {
+      "epoch": 4.830645161290323,
+      "grad_norm": 415.5929870605469,
+      "learning_rate": 7.820755449482265e-08,
+      "loss": 4.9497,
+      "step": 599
+    },
+    {
+      "epoch": 4.838709677419355,
+      "grad_norm": 79.47122192382812,
+      "learning_rate": 7.448338523316443e-08,
+      "loss": 4.9277,
+      "step": 600
+    },
+    {
+      "epoch": 4.846774193548387,
+      "grad_norm": 214.2676544189453,
+      "learning_rate": 7.07592159715062e-08,
+      "loss": 5.4366,
+      "step": 601
+    },
+    {
+      "epoch": 4.854838709677419,
+      "grad_norm": 85.91707611083984,
+      "learning_rate": 6.703504670984798e-08,
+      "loss": 0.9629,
+      "step": 602
+    },
+    {
+      "epoch": 4.862903225806452,
+      "grad_norm": 59.75841522216797,
+      "learning_rate": 6.331087744818976e-08,
+      "loss": 0.2416,
+      "step": 603
+    },
+    {
+      "epoch": 4.870967741935484,
+      "grad_norm": 9.71762752532959,
+      "learning_rate": 5.958670818653154e-08,
+      "loss": 0.1288,
+      "step": 604
+    },
+    {
+      "epoch": 4.879032258064516,
+      "grad_norm": 139.64285278320312,
+      "learning_rate": 5.5862538924873317e-08,
+      "loss": 1.0523,
+      "step": 605
+    },
+    {
+      "epoch": 4.887096774193548,
+      "grad_norm": 222.7238006591797,
+      "learning_rate": 5.2138369663215093e-08,
+      "loss": 17.453,
+      "step": 606
+    },
+    {
+      "epoch": 4.895161290322581,
+      "grad_norm": 79.89140319824219,
+      "learning_rate": 4.8414200401556877e-08,
+      "loss": 0.4757,
+      "step": 607
+    },
+    {
+      "epoch": 4.903225806451613,
+      "grad_norm": 65.18061065673828,
+      "learning_rate": 4.4690031139898653e-08,
+      "loss": 0.239,
+      "step": 608
+    },
+    {
+      "epoch": 4.911290322580645,
+      "grad_norm": 2055.86767578125,
+      "learning_rate": 4.0965861878240437e-08,
+      "loss": 133.0742,
+      "step": 609
+    },
+    {
+      "epoch": 4.919354838709677,
+      "grad_norm": 53.200016021728516,
+      "learning_rate": 3.724169261658221e-08,
+      "loss": 0.4046,
+      "step": 610
+    },
+    {
+      "epoch": 4.92741935483871,
+      "grad_norm": 46.2671012878418,
+      "learning_rate": 3.351752335492399e-08,
+      "loss": 1.0537,
+      "step": 611
+    },
+    {
+      "epoch": 4.935483870967742,
+      "grad_norm": 121.25851440429688,
+      "learning_rate": 2.979335409326577e-08,
+      "loss": 11.5313,
+      "step": 612
+    },
+    {
+      "epoch": 4.943548387096774,
+      "grad_norm": 313.4679870605469,
+      "learning_rate": 2.6069184831607547e-08,
+      "loss": 14.306,
+      "step": 613
+    },
+    {
+      "epoch": 4.951612903225806,
+      "grad_norm": 26.671438217163086,
+      "learning_rate": 2.2345015569949327e-08,
+      "loss": 0.8477,
+      "step": 614
+    },
+    {
+      "epoch": 4.959677419354839,
+      "grad_norm": 42.424842834472656,
+      "learning_rate": 1.8620846308291107e-08,
+      "loss": 0.3301,
+      "step": 615
+    },
+    {
+      "epoch": 4.967741935483871,
+      "grad_norm": 152.9750213623047,
+      "learning_rate": 1.4896677046632885e-08,
+      "loss": 1.3814,
+      "step": 616
+    },
+    {
+      "epoch": 4.975806451612903,
+      "grad_norm": 17.42218017578125,
+      "learning_rate": 1.1172507784974663e-08,
+      "loss": 0.0749,
+      "step": 617
+    },
+    {
+      "epoch": 4.983870967741936,
+      "grad_norm": 88.0694808959961,
+      "learning_rate": 7.4483385233164425e-09,
+      "loss": 0.6241,
+      "step": 618
+    },
+    {
+      "epoch": 4.991935483870968,
+      "grad_norm": 240.30140686035156,
+      "learning_rate": 3.7241692616582212e-09,
+      "loss": 22.4261,
+      "step": 619
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 26.565181732177734,
+      "learning_rate": 0.0,
+      "loss": 0.0558,
+      "step": 620
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 3.6100568771362305,
+      "eval_mae": 1.1832215785980225,
+      "eval_mse": 3.6100568771362305,
+      "eval_r2": 0.011629223823547363,
+      "eval_rmse": 1.9000149676084739,
+      "eval_runtime": 1.3508,
+      "eval_samples_per_second": 40.717,
+      "eval_smape": 49.474382400512695,
+      "eval_steps_per_second": 10.364,
+      "step": 620
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 620,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3359849068769280.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": {
+    "learning_rate": 2.308984942228097e-06
+  }
+}