diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31792 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 5294,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.886792452830189e-07,
+      "loss": 2.623,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.773584905660378e-07,
+      "loss": 2.8262,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.660377358490566e-07,
+      "loss": 2.9824,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.547169811320755e-07,
+      "loss": 2.6719,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.433962264150943e-07,
+      "loss": 2.5938,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1320754716981131e-06,
+      "loss": 2.748,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.320754716981132e-06,
+      "loss": 2.6074,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.509433962264151e-06,
+      "loss": 2.6816,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.69811320754717e-06,
+      "loss": 2.6562,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8867924528301887e-06,
+      "loss": 2.4512,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0754716981132075e-06,
+      "loss": 2.3457,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2641509433962262e-06,
+      "loss": 2.4609,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4528301886792453e-06,
+      "loss": 2.2979,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.641509433962264e-06,
+      "loss": 2.2578,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.830188679245283e-06,
+      "loss": 1.438,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.018867924528302e-06,
+      "loss": 1.3418,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.207547169811321e-06,
+      "loss": 1.2544,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.39622641509434e-06,
+      "loss": 1.2607,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5849056603773586e-06,
+      "loss": 1.2295,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7735849056603773e-06,
+      "loss": 1.0605,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.962264150943396e-06,
+      "loss": 1.1099,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.150943396226415e-06,
+      "loss": 1.104,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.339622641509434e-06,
+      "loss": 1.0688,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5283018867924524e-06,
+      "loss": 1.0791,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.716981132075472e-06,
+      "loss": 1.022,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.905660377358491e-06,
+      "loss": 0.9424,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.094339622641509e-06,
+      "loss": 0.8696,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.283018867924528e-06,
+      "loss": 0.8594,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.471698113207547e-06,
+      "loss": 0.9575,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 0.8467,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.849056603773586e-06,
+      "loss": 0.9233,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.037735849056604e-06,
+      "loss": 0.9834,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.226415094339623e-06,
+      "loss": 0.9043,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.415094339622642e-06,
+      "loss": 0.9033,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.60377358490566e-06,
+      "loss": 0.938,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.79245283018868e-06,
+      "loss": 0.8281,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.981132075471699e-06,
+      "loss": 0.7893,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.169811320754717e-06,
+      "loss": 0.7881,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.358490566037736e-06,
+      "loss": 0.8513,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.547169811320755e-06,
+      "loss": 0.771,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.735849056603773e-06,
+      "loss": 0.8271,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.924528301886793e-06,
+      "loss": 0.9448,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.11320754716981e-06,
+      "loss": 0.7722,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.30188679245283e-06,
+      "loss": 0.781,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.49056603773585e-06,
+      "loss": 0.7759,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.679245283018868e-06,
+      "loss": 0.8115,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.867924528301887e-06,
+      "loss": 0.8174,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.056603773584905e-06,
+      "loss": 0.8481,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.245283018867924e-06,
+      "loss": 0.9204,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.433962264150944e-06,
+      "loss": 0.8037,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.622641509433962e-06,
+      "loss": 0.73,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.811320754716981e-06,
+      "loss": 0.9517,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 0.9531,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0188679245283019e-05,
+      "loss": 0.7363,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0377358490566038e-05,
+      "loss": 1.0015,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0566037735849056e-05,
+      "loss": 0.9541,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0754716981132076e-05,
+      "loss": 0.6877,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0943396226415093e-05,
+      "loss": 0.8228,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1132075471698113e-05,
+      "loss": 0.9409,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 0.8716,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1509433962264152e-05,
+      "loss": 0.7451,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1698113207547171e-05,
+      "loss": 0.8638,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.188679245283019e-05,
+      "loss": 0.7529,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2075471698113209e-05,
+      "loss": 0.8003,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2264150943396227e-05,
+      "loss": 0.897,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2452830188679246e-05,
+      "loss": 0.7886,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2641509433962266e-05,
+      "loss": 0.7834,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2830188679245283e-05,
+      "loss": 0.8267,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3018867924528303e-05,
+      "loss": 0.8284,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.320754716981132e-05,
+      "loss": 0.6465,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.339622641509434e-05,
+      "loss": 0.9009,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.358490566037736e-05,
+      "loss": 0.9436,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3773584905660378e-05,
+      "loss": 0.7446,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3962264150943397e-05,
+      "loss": 0.7549,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4150943396226415e-05,
+      "loss": 0.8477,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4339622641509435e-05,
+      "loss": 0.8423,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4528301886792454e-05,
+      "loss": 0.7461,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4716981132075472e-05,
+      "loss": 0.8506,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4905660377358491e-05,
+      "loss": 0.7629,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.509433962264151e-05,
+      "loss": 0.7063,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.528301886792453e-05,
+      "loss": 0.8052,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5471698113207547e-05,
+      "loss": 0.7554,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5660377358490568e-05,
+      "loss": 0.7896,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5849056603773586e-05,
+      "loss": 0.7651,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6037735849056604e-05,
+      "loss": 0.6606,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.622641509433962e-05,
+      "loss": 0.813,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6415094339622643e-05,
+      "loss": 0.729,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.660377358490566e-05,
+      "loss": 0.7358,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6792452830188678e-05,
+      "loss": 0.7407,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.69811320754717e-05,
+      "loss": 0.834,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7169811320754717e-05,
+      "loss": 0.7607,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7358490566037735e-05,
+      "loss": 0.6758,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7547169811320753e-05,
+      "loss": 0.8472,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7735849056603774e-05,
+      "loss": 0.6694,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7924528301886792e-05,
+      "loss": 0.8237,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.811320754716981e-05,
+      "loss": 0.6929,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.830188679245283e-05,
+      "loss": 0.8442,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.849056603773585e-05,
+      "loss": 0.7068,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8679245283018867e-05,
+      "loss": 0.8203,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 0.7139,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9056603773584906e-05,
+      "loss": 0.7998,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9245283018867924e-05,
+      "loss": 0.7222,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.943396226415094e-05,
+      "loss": 0.7534,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9622641509433963e-05,
+      "loss": 0.8525,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.981132075471698e-05,
+      "loss": 0.6733,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 0.6902,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.018867924528302e-05,
+      "loss": 0.7295,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0377358490566037e-05,
+      "loss": 0.6677,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0566037735849055e-05,
+      "loss": 0.7827,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0754716981132076e-05,
+      "loss": 0.771,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0943396226415094e-05,
+      "loss": 0.668,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1132075471698112e-05,
+      "loss": 0.7434,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.132075471698113e-05,
+      "loss": 0.6292,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.150943396226415e-05,
+      "loss": 0.7954,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.169811320754717e-05,
+      "loss": 0.6765,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1886792452830187e-05,
+      "loss": 0.7102,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2075471698113208e-05,
+      "loss": 0.7417,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2264150943396226e-05,
+      "loss": 0.7104,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2452830188679244e-05,
+      "loss": 0.7004,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2641509433962265e-05,
+      "loss": 0.7314,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2830188679245286e-05,
+      "loss": 0.6741,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3018867924528304e-05,
+      "loss": 0.7124,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.320754716981132e-05,
+      "loss": 0.7021,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3396226415094343e-05,
+      "loss": 0.657,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.358490566037736e-05,
+      "loss": 0.6921,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.377358490566038e-05,
+      "loss": 0.7957,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.39622641509434e-05,
+      "loss": 0.6458,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4150943396226418e-05,
+      "loss": 0.7388,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4339622641509435e-05,
+      "loss": 0.7979,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4528301886792453e-05,
+      "loss": 0.7144,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4716981132075474e-05,
+      "loss": 0.8606,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4905660377358492e-05,
+      "loss": 0.7371,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.509433962264151e-05,
+      "loss": 0.6599,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.528301886792453e-05,
+      "loss": 0.6846,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.547169811320755e-05,
+      "loss": 0.5925,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5660377358490567e-05,
+      "loss": 0.698,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5849056603773585e-05,
+      "loss": 0.6755,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6037735849056606e-05,
+      "loss": 0.7681,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6226415094339624e-05,
+      "loss": 0.7351,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.641509433962264e-05,
+      "loss": 0.7871,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6603773584905663e-05,
+      "loss": 0.7654,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.679245283018868e-05,
+      "loss": 0.6035,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.69811320754717e-05,
+      "loss": 0.7273,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.716981132075472e-05,
+      "loss": 0.6704,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7358490566037738e-05,
+      "loss": 0.6777,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7547169811320755e-05,
+      "loss": 0.6846,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7735849056603773e-05,
+      "loss": 0.6841,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7924528301886794e-05,
+      "loss": 0.7275,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8113207547169812e-05,
+      "loss": 0.8164,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.830188679245283e-05,
+      "loss": 0.7949,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.849056603773585e-05,
+      "loss": 0.6221,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.867924528301887e-05,
+      "loss": 0.658,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8867924528301887e-05,
+      "loss": 0.7087,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9056603773584908e-05,
+      "loss": 0.8079,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9245283018867926e-05,
+      "loss": 0.5879,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9433962264150944e-05,
+      "loss": 0.7322,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9622641509433962e-05,
+      "loss": 0.832,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9811320754716983e-05,
+      "loss": 0.803,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3e-05,
+      "loss": 0.7861,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999999719275641e-05,
+      "loss": 0.6348,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999988771026676e-05,
+      "loss": 0.823,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999974734813963e-05,
+      "loss": 0.6875,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999955084123522e-05,
+      "loss": 0.6348,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999929818962706e-05,
+      "loss": 0.6797,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999989893934097e-05,
+      "loss": 0.6582,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999862445269875e-05,
+      "loss": 0.6465,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999982033676308e-05,
+      "loss": 0.793,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999772613836344e-05,
+      "loss": 0.7715,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999971927650753e-05,
+      "loss": 0.7061,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999966032479661e-05,
+      "loss": 0.6863,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999959575872564e-05,
+      "loss": 0.6882,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999952557831879e-05,
+      "loss": 0.6279,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999944978360233e-05,
+      "loss": 0.6528,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999936837460463e-05,
+      "loss": 0.7786,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999281351356162e-05,
+      "loss": 0.6572,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999188713889488e-05,
+      "loss": 0.8184,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.99990904622393e-05,
+      "loss": 0.6714,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999898659644236e-05,
+      "loss": 0.8179,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998877116537553e-05,
+      "loss": 0.8191,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999876202256585e-05,
+      "loss": 0.6501,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998641314570337e-05,
+      "loss": 0.688,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999851499259619e-05,
+      "loss": 0.6934,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998383056690696e-05,
+      "loss": 0.8242,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999824550690323e-05,
+      "loss": 0.6604,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998102343285288e-05,
+      "loss": 0.7407,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997953565890448e-05,
+      "loss": 0.7456,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.99977991747744e-05,
+      "loss": 0.658,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997639169994934e-05,
+      "loss": 0.6423,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997473551611936e-05,
+      "loss": 0.8242,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.99973023196874e-05,
+      "loss": 0.6514,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997125474285416e-05,
+      "loss": 0.6179,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996943015472177e-05,
+      "loss": 0.6378,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999675494331598e-05,
+      "loss": 0.6775,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996561257887218e-05,
+      "loss": 0.6069,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999636195925839e-05,
+      "loss": 0.7812,
+      "step": 195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996157047504083e-05,
+      "loss": 0.8076,
+      "step": 196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9995946522701007e-05,
+      "loss": 0.9609,
+      "step": 197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999573038492796e-05,
+      "loss": 0.6953,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999550863426584e-05,
+      "loss": 0.7329,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995281270797645e-05,
+      "loss": 0.772,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999504829460848e-05,
+      "loss": 0.7954,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999480970578555e-05,
+      "loss": 0.4868,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999456550441815e-05,
+      "loss": 0.6855,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994315690597694e-05,
+      "loss": 0.6792,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994060264417684e-05,
+      "loss": 0.7346,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993799225973724e-05,
+      "loss": 0.7073,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993532575363526e-05,
+      "loss": 0.6304,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999326031268689e-05,
+      "loss": 0.6907,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999298243804572e-05,
+      "loss": 0.6477,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999269895154404e-05,
+      "loss": 0.6672,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992409853287942e-05,
+      "loss": 0.7158,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992115143385647e-05,
+      "loss": 0.7,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999181482194746e-05,
+      "loss": 0.6174,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991508889085793e-05,
+      "loss": 0.5923,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999119734491515e-05,
+      "loss": 0.6548,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990880189552153e-05,
+      "loss": 0.5374,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.99905574231155e-05,
+      "loss": 0.646,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990229045726015e-05,
+      "loss": 0.7349,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.99898950575066e-05,
+      "loss": 0.6248,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998955545858227e-05,
+      "loss": 0.698,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989210249080133e-05,
+      "loss": 0.5923,
+      "step": 221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998885942912941e-05,
+      "loss": 0.7009,
+      "step": 222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988502998861406e-05,
+      "loss": 0.7878,
+      "step": 223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988140958409528e-05,
+      "loss": 0.7898,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9987773307909298e-05,
+      "loss": 0.77,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9987400047498318e-05,
+      "loss": 0.7615,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9987021177316302e-05,
+      "loss": 0.7253,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986636697505063e-05,
+      "loss": 0.7017,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986246608208508e-05,
+      "loss": 0.6184,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985850909572653e-05,
+      "loss": 0.7856,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985449601745602e-05,
+      "loss": 0.667,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985042684877566e-05,
+      "loss": 0.8391,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984630159120855e-05,
+      "loss": 0.6201,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984212024629868e-05,
+      "loss": 0.7563,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983788281561127e-05,
+      "loss": 0.5835,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983358930073232e-05,
+      "loss": 0.7515,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982923970326887e-05,
+      "loss": 0.7095,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99824834024849e-05,
+      "loss": 0.6387,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982037226712173e-05,
+      "loss": 0.6047,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998158544317571e-05,
+      "loss": 0.7007,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998112805204461e-05,
+      "loss": 0.5969,
+      "step": 241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9980665053490083e-05,
+      "loss": 0.5979,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9980196447685417e-05,
+      "loss": 0.5461,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9979722234806024e-05,
+      "loss": 0.7012,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9979242415029387e-05,
+      "loss": 0.5808,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9978756988535115e-05,
+      "loss": 0.7109,
+      "step": 246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9978265955504896e-05,
+      "loss": 0.6814,
+      "step": 247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9977769316122523e-05,
+      "loss": 0.5625,
+      "step": 248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9977267070573892e-05,
+      "loss": 0.6907,
+      "step": 249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9976759219046988e-05,
+      "loss": 0.8984,
+      "step": 250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99762457617319e-05,
+      "loss": 0.7354,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9975726698820822e-05,
+      "loss": 0.7495,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997520203050803e-05,
+      "loss": 0.6653,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997467175698991e-05,
+      "loss": 0.6804,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9974135878464945e-05,
+      "loss": 0.7197,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997359439513371e-05,
+      "loss": 0.697,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997304730719888e-05,
+      "loss": 0.5208,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9972494614865243e-05,
+      "loss": 0.7952,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9971936318339653e-05,
+      "loss": 0.7351,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9971372417831093e-05,
+      "loss": 0.7637,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9970802913550624e-05,
+      "loss": 0.7227,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997022780571141e-05,
+      "loss": 0.6465,
+      "step": 262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9969647094528718e-05,
+      "loss": 0.707,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9969060780219908e-05,
+      "loss": 0.7686,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9968468863004433e-05,
+      "loss": 0.6074,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9967871343103847e-05,
+      "loss": 0.7336,
+      "step": 266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9967268220741807e-05,
+      "loss": 0.6836,
+      "step": 267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9966659496144057e-05,
+      "loss": 0.656,
+      "step": 268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996604516953844e-05,
+      "loss": 0.7795,
+      "step": 269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9965425241154903e-05,
+      "loss": 0.6064,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996479971122548e-05,
+      "loss": 0.7681,
+      "step": 271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996416857998431e-05,
+      "loss": 0.6877,
+      "step": 272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9963531847667622e-05,
+      "loss": 0.6875,
+      "step": 273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9962889514513746e-05,
+      "loss": 0.7871,
+      "step": 274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9962241580763107e-05,
+      "loss": 0.7295,
+      "step": 275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9961588046658222e-05,
+      "loss": 0.6724,
+      "step": 276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9960928912443718e-05,
+      "loss": 0.6375,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9960264178366294e-05,
+      "loss": 0.6238,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995959384467477e-05,
+      "loss": 0.6416,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995891791162005e-05,
+      "loss": 0.7329,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9958236379455137e-05,
+      "loss": 0.6841,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995754924843512e-05,
+      "loss": 0.6589,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9956856518817198e-05,
+      "loss": 0.6533,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9956158190860658e-05,
+      "loss": 0.8232,
+      "step": 284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9955454264826883e-05,
+      "loss": 0.7883,
+      "step": 285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995474474097935e-05,
+      "loss": 0.6338,
+      "step": 286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995402961958364e-05,
+      "loss": 0.6685,
+      "step": 287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9953308900907413e-05,
+      "loss": 0.5989,
+      "step": 288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995258258522044e-05,
+      "loss": 0.6338,
+      "step": 289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9951850672794574e-05,
+      "loss": 0.6118,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9951113163903776e-05,
+      "loss": 0.74,
+      "step": 291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9950370058824095e-05,
+      "loss": 0.6177,
+      "step": 292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994962135783367e-05,
+      "loss": 0.6763,
+      "step": 293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994886706121274e-05,
+      "loss": 0.5908,
+      "step": 294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994810716924364e-05,
+      "loss": 0.7236,
+      "step": 295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994734168221079e-05,
+      "loss": 0.6865,
+      "step": 296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994657060040072e-05,
+      "loss": 0.6626,
+      "step": 297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9945793924102047e-05,
+      "loss": 0.7983,
+      "step": 298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9945011653605475e-05,
+      "loss": 0.6978,
+      "step": 299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9944223789203802e-05,
+      "loss": 0.6882,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994343033119193e-05,
+      "loss": 0.6355,
+      "step": 301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9942631279866853e-05,
+      "loss": 0.6426,
+      "step": 302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9941826635527648e-05,
+      "loss": 0.6892,
+      "step": 303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9941016398475504e-05,
+      "loss": 0.6135,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9940200569013683e-05,
+      "loss": 0.6279,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9939379147447544e-05,
+      "loss": 0.7629,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993855213408456e-05,
+      "loss": 0.6128,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9937719529234268e-05,
+      "loss": 0.6672,
+      "step": 308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993688133320832e-05,
+      "loss": 0.7209,
+      "step": 309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9936037546320446e-05,
+      "loss": 0.7502,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993518816888648e-05,
+      "loss": 0.687,
+      "step": 311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993433320122434e-05,
+      "loss": 0.5911,
+      "step": 312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9933472643654035e-05,
+      "loss": 0.7473,
+      "step": 313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9932606496497678e-05,
+      "loss": 0.6689,
+      "step": 314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993173476007947e-05,
+      "loss": 0.6399,
+      "step": 315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9930857434725688e-05,
+      "loss": 0.6619,
+      "step": 316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992997452076473e-05,
+      "loss": 0.7778,
+      "step": 317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992908601852706e-05,
+      "loss": 0.6306,
+      "step": 318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9928191928345245e-05,
+      "loss": 0.5918,
+      "step": 319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9927292250553944e-05,
+      "loss": 0.6814,
+      "step": 320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9926386985489902e-05,
+      "loss": 0.6899,
+      "step": 321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9925476133491967e-05,
+      "loss": 0.571,
+      "step": 322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9924559694901062e-05,
+      "loss": 0.5291,
+      "step": 323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992363767006021e-05,
+      "loss": 0.6753,
+      "step": 324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9922710059314527e-05,
+      "loss": 0.6008,
+      "step": 325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9921776863011217e-05,
+      "loss": 0.6514,
+      "step": 326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9920838081499576e-05,
+      "loss": 0.7148,
+      "step": 327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9919893715130983e-05,
+      "loss": 0.6628,
+      "step": 328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.991894376425892e-05,
+      "loss": 0.6084,
+      "step": 329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.991798822923895e-05,
+      "loss": 0.6086,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9917027110428724e-05,
+      "loss": 0.6387,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9916060408188003e-05,
+      "loss": 0.7305,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9915088122878604e-05,
+      "loss": 0.626,
+      "step": 333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991411025486446e-05,
+      "loss": 0.6467,
+      "step": 334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9913126804511596e-05,
+      "loss": 0.7102,
+      "step": 335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9912137772188102e-05,
+      "loss": 0.5754,
+      "step": 336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991114315826418e-05,
+      "loss": 0.6528,
+      "step": 337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9910142963112113e-05,
+      "loss": 0.7251,
+      "step": 338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990913718710627e-05,
+      "loss": 0.7178,
+      "step": 339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990812583062311e-05,
+      "loss": 0.6133,
+      "step": 340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990710889404119e-05,
+      "loss": 0.6294,
+      "step": 341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9906086377741143e-05,
+      "loss": 0.7332,
+      "step": 342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9905058282105703e-05,
+      "loss": 0.7856,
+      "step": 343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9904024607519676e-05,
+      "loss": 0.6704,
+      "step": 344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990298535436997e-05,
+      "loss": 0.7441,
+      "step": 345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9901940523045578e-05,
+      "loss": 0.6357,
+      "step": 346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9900890113937576e-05,
+      "loss": 0.5847,
+      "step": 347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9899834127439133e-05,
+      "loss": 0.6628,
+      "step": 348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9898772563945503e-05,
+      "loss": 0.5984,
+      "step": 349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9897705423854034e-05,
+      "loss": 0.741,
+      "step": 350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9896632707564146e-05,
+      "loss": 0.635,
+      "step": 351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9895554415477358e-05,
+      "loss": 0.604,
+      "step": 352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9894470547997284e-05,
+      "loss": 0.5771,
+      "step": 353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.98933811055296e-05,
+      "loss": 0.6216,
+      "step": 354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.98922860884821e-05,
+      "loss": 0.6772,
+      "step": 355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989118549726463e-05,
+      "loss": 0.8359,
+      "step": 356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9890079332289148e-05,
+      "loss": 0.6321,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9888967593969695e-05,
+      "loss": 0.6733,
+      "step": 358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9887850282722388e-05,
+      "loss": 0.6953,
+      "step": 359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9886727398965437e-05,
+      "loss": 0.6482,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988559894311914e-05,
+      "loss": 0.6628,
+      "step": 361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9884464915605875e-05,
+      "loss": 0.6934,
+      "step": 362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9883325316850103e-05,
+      "loss": 0.7603,
+      "step": 363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988218014727838e-05,
+      "loss": 0.5847,
+      "step": 364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988102940731934e-05,
+      "loss": 0.6301,
+      "step": 365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9879873097403705e-05,
+      "loss": 0.6404,
+      "step": 366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9878711217964284e-05,
+      "loss": 0.7563,
+      "step": 367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9877543769435958e-05,
+      "loss": 0.6121,
+      "step": 368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9876370752255712e-05,
+      "loss": 0.5564,
+      "step": 369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9875192166862597e-05,
+      "loss": 0.6172,
+      "step": 370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987400801369776e-05,
+      "loss": 0.6492,
+      "step": 371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9872818293204435e-05,
+      "loss": 0.6025,
+      "step": 372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9871623005827922e-05,
+      "loss": 0.7026,
+      "step": 373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9870422152015625e-05,
+      "loss": 0.613,
+      "step": 374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9869215732217018e-05,
+      "loss": 0.6689,
+      "step": 375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9868003746883665e-05,
+      "loss": 0.7236,
+      "step": 376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986678619646921e-05,
+      "loss": 0.7229,
+      "step": 377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986556308142938e-05,
+      "loss": 0.8655,
+      "step": 378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986433440222199e-05,
+      "loss": 0.6582,
+      "step": 379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9863100159306923e-05,
+      "loss": 0.6497,
+      "step": 380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986186035314617e-05,
+      "loss": 0.6582,
+      "step": 381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986061498420378e-05,
+      "loss": 0.7061,
+      "step": 382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9859364052945894e-05,
+      "loss": 0.5833,
+      "step": 383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9858107559840738e-05,
+      "loss": 0.6846,
+      "step": 384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9856845505358618e-05,
+      "loss": 0.686,
+      "step": 385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9855577889971913e-05,
+      "loss": 0.7461,
+      "step": 386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985430471415509e-05,
+      "loss": 0.5894,
+      "step": 387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9853025978384707e-05,
+      "loss": 0.6384,
+      "step": 388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9851741683139383e-05,
+      "loss": 0.6226,
+      "step": 389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985045182889984e-05,
+      "loss": 0.6072,
+      "step": 390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9849156416148857e-05,
+      "loss": 0.6108,
+      "step": 391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9847855445371312e-05,
+      "loss": 0.5762,
+      "step": 392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.984654891705416e-05,
+      "loss": 0.6802,
+      "step": 393
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.984523683168643e-05,
+      "loss": 0.6553,
+      "step": 394
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9843919189759235e-05,
+      "loss": 0.7764,
+      "step": 395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9842595991765766e-05,
+      "loss": 0.7007,
+      "step": 396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.98412672382013e-05,
+      "loss": 0.6179,
+      "step": 397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9839932929563182e-05,
+      "loss": 0.6267,
+      "step": 398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9838593066350842e-05,
+      "loss": 0.74,
+      "step": 399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9837247649065797e-05,
+      "loss": 0.6897,
+      "step": 400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9835896678211626e-05,
+      "loss": 0.6528,
+      "step": 401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9834540154294007e-05,
+      "loss": 0.5906,
+      "step": 402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983317807782068e-05,
+      "loss": 0.7393,
+      "step": 403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983181044930147e-05,
+      "loss": 0.6807,
+      "step": 404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9830437269248273e-05,
+      "loss": 0.6655,
+      "step": 405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9829058538175076e-05,
+      "loss": 0.6555,
+      "step": 406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9827674256597936e-05,
+      "loss": 0.6775,
+      "step": 407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9826284425034993e-05,
+      "loss": 0.687,
+      "step": 408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9824889044006448e-05,
+      "loss": 0.6084,
+      "step": 409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.98234881140346e-05,
+      "loss": 0.6587,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9822081635643812e-05,
+      "loss": 0.7307,
+      "step": 411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9820669609360533e-05,
+      "loss": 0.6992,
+      "step": 412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9819252035713277e-05,
+      "loss": 0.6794,
+      "step": 413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.981782891523264e-05,
+      "loss": 0.8115,
+      "step": 414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9816400248451302e-05,
+      "loss": 0.769,
+      "step": 415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9814966035904004e-05,
+      "loss": 0.6667,
+      "step": 416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9813526278127577e-05,
+      "loss": 0.7148,
+      "step": 417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.981208097566092e-05,
+      "loss": 0.5576,
+      "step": 418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9810630129045003e-05,
+      "loss": 0.6562,
+      "step": 419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.980917373882288e-05,
+      "loss": 0.5322,
+      "step": 420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.980771180553968e-05,
+      "loss": 0.5874,
+      "step": 421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9806244329742594e-05,
+      "loss": 0.6924,
+      "step": 422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9804771311980907e-05,
+      "loss": 0.6941,
+      "step": 423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9803292752805968e-05,
+      "loss": 0.7549,
+      "step": 424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.980180865277119e-05,
+      "loss": 0.611,
+      "step": 425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.980031901243208e-05,
+      "loss": 0.7385,
+      "step": 426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9798823832346212e-05,
+      "loss": 0.6003,
+      "step": 427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9797323113073216e-05,
+      "loss": 0.625,
+      "step": 428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9795816855174825e-05,
+      "loss": 0.6306,
+      "step": 429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.979430505921482e-05,
+      "loss": 0.7217,
+      "step": 430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.979278772575907e-05,
+      "loss": 0.7173,
+      "step": 431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9791264855375516e-05,
+      "loss": 0.7769,
+      "step": 432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9789736448634154e-05,
+      "loss": 0.7595,
+      "step": 433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9788202506107078e-05,
+      "loss": 0.6433,
+      "step": 434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9786663028368434e-05,
+      "loss": 0.594,
+      "step": 435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.978511801599445e-05,
+      "loss": 0.5813,
+      "step": 436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9783567469563425e-05,
+      "loss": 0.5522,
+      "step": 437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9782011389655715e-05,
+      "loss": 0.7048,
+      "step": 438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.978044977685378e-05,
+      "loss": 0.6646,
+      "step": 439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9778882631742113e-05,
+      "loss": 0.6895,
+      "step": 440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9777309954907304e-05,
+      "loss": 0.6245,
+      "step": 441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9775731746938002e-05,
+      "loss": 0.6987,
+      "step": 442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.977414800842493e-05,
+      "loss": 0.7046,
+      "step": 443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9772558739960876e-05,
+      "loss": 0.7817,
+      "step": 444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.977096394214071e-05,
+      "loss": 0.6067,
+      "step": 445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9769363615561345e-05,
+      "loss": 0.6648,
+      "step": 446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.976775776082181e-05,
+      "loss": 0.6538,
+      "step": 447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.976614637852315e-05,
+      "loss": 0.6467,
+      "step": 448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.976452946926852e-05,
+      "loss": 0.7803,
+      "step": 449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9762907033663117e-05,
+      "loss": 0.5413,
+      "step": 450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9761279072314223e-05,
+      "loss": 0.7053,
+      "step": 451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9759645585831187e-05,
+      "loss": 0.7141,
+      "step": 452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9758006574825413e-05,
+      "loss": 0.7134,
+      "step": 453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9756362039910385e-05,
+      "loss": 0.7664,
+      "step": 454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.975471198170165e-05,
+      "loss": 0.5891,
+      "step": 455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9753056400816825e-05,
+      "loss": 0.6624,
+      "step": 456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9751395297875586e-05,
+      "loss": 0.7798,
+      "step": 457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9749728673499695e-05,
+      "loss": 0.7705,
+      "step": 458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.974805652831296e-05,
+      "loss": 0.7407,
+      "step": 459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9746378862941263e-05,
+      "loss": 0.7405,
+      "step": 460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9744695678012554e-05,
+      "loss": 0.6223,
+      "step": 461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9743006974156847e-05,
+      "loss": 0.5461,
+      "step": 462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9741312752006224e-05,
+      "loss": 0.7207,
+      "step": 463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973961301219483e-05,
+      "loss": 0.697,
+      "step": 464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9737907755358877e-05,
+      "loss": 0.6541,
+      "step": 465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9736196982136638e-05,
+      "loss": 0.5803,
+      "step": 466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973448069316846e-05,
+      "loss": 0.7463,
+      "step": 467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973275888909675e-05,
+      "loss": 0.614,
+      "step": 468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9731031570565962e-05,
+      "loss": 0.6741,
+      "step": 469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9729298738222642e-05,
+      "loss": 0.6831,
+      "step": 470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9727560392715395e-05,
+      "loss": 0.7734,
+      "step": 471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.972581653469487e-05,
+      "loss": 0.5911,
+      "step": 472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9724067164813792e-05,
+      "loss": 0.7058,
+      "step": 473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.972231228372695e-05,
+      "loss": 0.6658,
+      "step": 474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.97205518920912e-05,
+      "loss": 0.7158,
+      "step": 475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.971878599056545e-05,
+      "loss": 0.7292,
+      "step": 476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9717014579810683e-05,
+      "loss": 0.6394,
+      "step": 477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9715237660489923e-05,
+      "loss": 0.6943,
+      "step": 478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9713455233268282e-05,
+      "loss": 0.7205,
+      "step": 479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9711667298812914e-05,
+      "loss": 0.6367,
+      "step": 480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9709873857793038e-05,
+      "loss": 0.6997,
+      "step": 481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9708074910879946e-05,
+      "loss": 0.7109,
+      "step": 482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9706270458746978e-05,
+      "loss": 0.6628,
+      "step": 483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9704460502069544e-05,
+      "loss": 0.6389,
+      "step": 484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.97026450415251e-05,
+      "loss": 0.5684,
+      "step": 485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9700824077793177e-05,
+      "loss": 0.6143,
+      "step": 486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9698997611555362e-05,
+      "loss": 0.5945,
+      "step": 487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9697165643495296e-05,
+      "loss": 0.7407,
+      "step": 488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9695328174298682e-05,
+      "loss": 0.7639,
+      "step": 489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.969348520465329e-05,
+      "loss": 0.5977,
+      "step": 490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9691636735248936e-05,
+      "loss": 0.6389,
+      "step": 491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.96897827667775e-05,
+      "loss": 0.729,
+      "step": 492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968792329993292e-05,
+      "loss": 0.6758,
+      "step": 493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.96860583354112e-05,
+      "loss": 0.6404,
+      "step": 494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9684187873910387e-05,
+      "loss": 0.7153,
+      "step": 495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.96823119161306e-05,
+      "loss": 0.7158,
+      "step": 496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9680430462774e-05,
+      "loss": 0.6509,
+      "step": 497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9678543514544822e-05,
+      "loss": 0.6655,
+      "step": 498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9676651072149344e-05,
+      "loss": 0.6851,
+      "step": 499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.967475313629591e-05,
+      "loss": 0.6301,
+      "step": 500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9672849707694908e-05,
+      "loss": 0.6892,
+      "step": 501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9670940787058793e-05,
+      "loss": 0.698,
+      "step": 502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966902637510208e-05,
+      "loss": 0.6018,
+      "step": 503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966710647254132e-05,
+      "loss": 0.5498,
+      "step": 504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966518108009514e-05,
+      "loss": 0.6033,
+      "step": 505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9663250198484205e-05,
+      "loss": 0.7322,
+      "step": 506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966131382843125e-05,
+      "loss": 0.6714,
+      "step": 507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.965937197066105e-05,
+      "loss": 0.8208,
+      "step": 508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9657424625900445e-05,
+      "loss": 0.6965,
+      "step": 509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9655471794878325e-05,
+      "loss": 0.6455,
+      "step": 510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9653513478325626e-05,
+      "loss": 0.6626,
+      "step": 511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9651549676975355e-05,
+      "loss": 0.7305,
+      "step": 512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9649580391562552e-05,
+      "loss": 0.6311,
+      "step": 513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9647605622824323e-05,
+      "loss": 0.5208,
+      "step": 514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9645625371499826e-05,
+      "loss": 0.7222,
+      "step": 515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.964363963833026e-05,
+      "loss": 0.7273,
+      "step": 516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9641648424058885e-05,
+      "loss": 0.6021,
+      "step": 517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.963965172943101e-05,
+      "loss": 0.6323,
+      "step": 518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9637649555194003e-05,
+      "loss": 0.604,
+      "step": 519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9635641902097272e-05,
+      "loss": 0.635,
+      "step": 520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9633628770892275e-05,
+      "loss": 0.6843,
+      "step": 521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9631610162332526e-05,
+      "loss": 0.6968,
+      "step": 522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.96295860771736e-05,
+      "loss": 0.675,
+      "step": 523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.96275565161731e-05,
+      "loss": 0.6724,
+      "step": 524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.962552148009069e-05,
+      "loss": 0.6113,
+      "step": 525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.962348096968808e-05,
+      "loss": 0.7073,
+      "step": 526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.962143498572904e-05,
+      "loss": 0.6338,
+      "step": 527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961938352897937e-05,
+      "loss": 0.6929,
+      "step": 528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961732660020694e-05,
+      "loss": 0.5513,
+      "step": 529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9615264200181643e-05,
+      "loss": 0.7085,
+      "step": 530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9613196329675447e-05,
+      "loss": 0.6448,
+      "step": 531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9611122989462342e-05,
+      "loss": 0.5774,
+      "step": 532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9609044180318387e-05,
+      "loss": 0.5837,
+      "step": 533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9606959903021673e-05,
+      "loss": 0.6731,
+      "step": 534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9604870158352345e-05,
+      "loss": 0.7246,
+      "step": 535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.960277494709259e-05,
+      "loss": 0.6086,
+      "step": 536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.960067427002665e-05,
+      "loss": 0.6826,
+      "step": 537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.95985681279408e-05,
+      "loss": 0.6333,
+      "step": 538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9596456521623373e-05,
+      "loss": 0.6519,
+      "step": 539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9594339451864737e-05,
+      "loss": 0.7991,
+      "step": 540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9592216919457308e-05,
+      "loss": 0.7295,
+      "step": 541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9590088925195556e-05,
+      "loss": 0.6445,
+      "step": 542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.958795546987598e-05,
+      "loss": 0.6592,
+      "step": 543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9585816554297135e-05,
+      "loss": 0.6677,
+      "step": 544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.958367217925961e-05,
+      "loss": 0.6382,
+      "step": 545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.958152234556605e-05,
+      "loss": 0.5972,
+      "step": 546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.957936705402113e-05,
+      "loss": 0.6953,
+      "step": 547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.957720630543157e-05,
+      "loss": 0.7012,
+      "step": 548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.957504010060615e-05,
+      "loss": 0.7751,
+      "step": 549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.957286844035567e-05,
+      "loss": 0.658,
+      "step": 550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.957069132549298e-05,
+      "loss": 0.6255,
+      "step": 551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956850875683297e-05,
+      "loss": 0.6128,
+      "step": 552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956632073519258e-05,
+      "loss": 0.5732,
+      "step": 553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956412726139078e-05,
+      "loss": 0.7139,
+      "step": 554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956192833624859e-05,
+      "loss": 0.6116,
+      "step": 555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.955972396058906e-05,
+      "loss": 0.7119,
+      "step": 556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.955751413523729e-05,
+      "loss": 0.5806,
+      "step": 557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9555298861020415e-05,
+      "loss": 0.6384,
+      "step": 558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9553078138767605e-05,
+      "loss": 0.7229,
+      "step": 559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9550851969310086e-05,
+      "loss": 0.6726,
+      "step": 560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.95486203534811e-05,
+      "loss": 0.6965,
+      "step": 561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9546383292115947e-05,
+      "loss": 0.5818,
+      "step": 562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9544140786051955e-05,
+      "loss": 0.7393,
+      "step": 563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9541892836128483e-05,
+      "loss": 0.7051,
+      "step": 564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.953963944318695e-05,
+      "loss": 0.6633,
+      "step": 565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9537380608070792e-05,
+      "loss": 0.5398,
+      "step": 566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9535116331625493e-05,
+      "loss": 0.7539,
+      "step": 567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9532846614698564e-05,
+      "loss": 0.9111,
+      "step": 568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.953057145813956e-05,
+      "loss": 0.6057,
+      "step": 569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.952829086280008e-05,
+      "loss": 0.5603,
+      "step": 570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9526004829533734e-05,
+      "loss": 0.5366,
+      "step": 571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.952371335919619e-05,
+      "loss": 0.5305,
+      "step": 572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9521416452645136e-05,
+      "loss": 0.6514,
+      "step": 573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9519114110740313e-05,
+      "loss": 0.6091,
+      "step": 574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9516806334343482e-05,
+      "loss": 0.5896,
+      "step": 575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.951449312431844e-05,
+      "loss": 0.8103,
+      "step": 576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.951217448153102e-05,
+      "loss": 0.6426,
+      "step": 577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9509850406849088e-05,
+      "loss": 0.6135,
+      "step": 578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9507520901142543e-05,
+      "loss": 0.5693,
+      "step": 579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.950518596528332e-05,
+      "loss": 0.6094,
+      "step": 580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.950284560014538e-05,
+      "loss": 0.6616,
+      "step": 581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9500499806604716e-05,
+      "loss": 0.5359,
+      "step": 582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9498148585539364e-05,
+      "loss": 0.5935,
+      "step": 583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9495791937829385e-05,
+      "loss": 0.5796,
+      "step": 584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949342986435686e-05,
+      "loss": 0.7192,
+      "step": 585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9491062366005917e-05,
+      "loss": 0.6079,
+      "step": 586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9488689443662713e-05,
+      "loss": 0.7451,
+      "step": 587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9486311098215425e-05,
+      "loss": 0.6497,
+      "step": 588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.948392733055427e-05,
+      "loss": 0.7905,
+      "step": 589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9481538141571484e-05,
+      "loss": 0.5627,
+      "step": 590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9479143532161346e-05,
+      "loss": 0.6074,
+      "step": 591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.947674350322015e-05,
+      "loss": 0.5618,
+      "step": 592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9474338055646227e-05,
+      "loss": 0.7871,
+      "step": 593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.947192719033993e-05,
+      "loss": 0.6729,
+      "step": 594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9469510908203655e-05,
+      "loss": 0.6096,
+      "step": 595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.94670892101418e-05,
+      "loss": 0.6438,
+      "step": 596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9464662097060816e-05,
+      "loss": 0.502,
+      "step": 597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9462229569869167e-05,
+      "loss": 0.6348,
+      "step": 598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.945979162947734e-05,
+      "loss": 0.7092,
+      "step": 599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9457348276797863e-05,
+      "loss": 0.6433,
+      "step": 600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.945489951274527e-05,
+      "loss": 0.5916,
+      "step": 601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.945244533823614e-05,
+      "loss": 0.5979,
+      "step": 602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9449985754189063e-05,
+      "loss": 0.7068,
+      "step": 603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9447520761524663e-05,
+      "loss": 0.6831,
+      "step": 604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9445050361165586e-05,
+      "loss": 0.6487,
+      "step": 605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9442574554036496e-05,
+      "loss": 0.7146,
+      "step": 606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9440093341064088e-05,
+      "loss": 0.738,
+      "step": 607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9437606723177076e-05,
+      "loss": 0.5598,
+      "step": 608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9435114701306204e-05,
+      "loss": 0.6912,
+      "step": 609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9432617276384225e-05,
+      "loss": 0.6643,
+      "step": 610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.943011444934593e-05,
+      "loss": 0.575,
+      "step": 611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9427606221128124e-05,
+      "loss": 0.5261,
+      "step": 612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9425092592669636e-05,
+      "loss": 0.6514,
+      "step": 613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9422573564911305e-05,
+      "loss": 0.7283,
+      "step": 614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.942004913879602e-05,
+      "loss": 0.6677,
+      "step": 615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9417519315268654e-05,
+      "loss": 0.5657,
+      "step": 616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.941498409527612e-05,
+      "loss": 0.8,
+      "step": 617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9412443479767356e-05,
+      "loss": 0.5764,
+      "step": 618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.940989746969331e-05,
+      "loss": 0.6428,
+      "step": 619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.940734606600695e-05,
+      "loss": 0.5596,
+      "step": 620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9404789269663266e-05,
+      "loss": 0.6831,
+      "step": 621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9402227081619263e-05,
+      "loss": 0.7339,
+      "step": 622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9399659502833964e-05,
+      "loss": 0.6174,
+      "step": 623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9397086534268412e-05,
+      "loss": 0.6377,
+      "step": 624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9394508176885673e-05,
+      "loss": 0.5404,
+      "step": 625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9391924431650817e-05,
+      "loss": 0.5161,
+      "step": 626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.938933529953094e-05,
+      "loss": 0.6785,
+      "step": 627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9386740781495152e-05,
+      "loss": 0.626,
+      "step": 628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9384140878514578e-05,
+      "loss": 0.5908,
+      "step": 629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9381535591562365e-05,
+      "loss": 0.6902,
+      "step": 630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9378924921613663e-05,
+      "loss": 0.5779,
+      "step": 631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9376308869645644e-05,
+      "loss": 0.5808,
+      "step": 632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9373687436637492e-05,
+      "loss": 0.5894,
+      "step": 633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9371060623570417e-05,
+      "loss": 0.5759,
+      "step": 634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9368428431427624e-05,
+      "loss": 0.592,
+      "step": 635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9365790861194344e-05,
+      "loss": 0.6882,
+      "step": 636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9363147913857815e-05,
+      "loss": 0.5916,
+      "step": 637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9360499590407292e-05,
+      "loss": 0.7466,
+      "step": 638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.935784589183404e-05,
+      "loss": 0.6174,
+      "step": 639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9355186819131334e-05,
+      "loss": 0.6504,
+      "step": 640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9352522373294465e-05,
+      "loss": 0.7705,
+      "step": 641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.934985255532073e-05,
+      "loss": 0.6169,
+      "step": 642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9347177366209446e-05,
+      "loss": 0.7874,
+      "step": 643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.934449680696192e-05,
+      "loss": 0.6492,
+      "step": 644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.93418108785815e-05,
+      "loss": 0.5288,
+      "step": 645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.933911958207352e-05,
+      "loss": 0.6233,
+      "step": 646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.933642291844533e-05,
+      "loss": 0.5906,
+      "step": 647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9333720888706282e-05,
+      "loss": 0.7139,
+      "step": 648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9331013493867754e-05,
+      "loss": 0.5813,
+      "step": 649
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9328300734943117e-05,
+      "loss": 0.666,
+      "step": 650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9325582612947755e-05,
+      "loss": 0.5659,
+      "step": 651
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9322859128899056e-05,
+      "loss": 0.665,
+      "step": 652
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9320130283816417e-05,
+      "loss": 0.6699,
+      "step": 653
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.931739607872125e-05,
+      "loss": 0.7344,
+      "step": 654
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.931465651463696e-05,
+      "loss": 0.5396,
+      "step": 655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9311911592588966e-05,
+      "loss": 0.7446,
+      "step": 656
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9309161313604683e-05,
+      "loss": 0.6191,
+      "step": 657
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.930640567871355e-05,
+      "loss": 0.647,
+      "step": 658
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.930364468894699e-05,
+      "loss": 0.6392,
+      "step": 659
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.930087834533844e-05,
+      "loss": 0.6235,
+      "step": 660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.929810664892334e-05,
+      "loss": 0.7129,
+      "step": 661
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.929532960073914e-05,
+      "loss": 0.6704,
+      "step": 662
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9292547201825278e-05,
+      "loss": 0.582,
+      "step": 663
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.928975945322321e-05,
+      "loss": 0.8191,
+      "step": 664
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.928696635597638e-05,
+      "loss": 0.6743,
+      "step": 665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9284167911130254e-05,
+      "loss": 0.6431,
+      "step": 666
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9281364119732277e-05,
+      "loss": 0.5549,
+      "step": 667
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9278554982831907e-05,
+      "loss": 0.6077,
+      "step": 668
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.927574050148061e-05,
+      "loss": 0.6411,
+      "step": 669
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9272920676731834e-05,
+      "loss": 0.4717,
+      "step": 670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.927009550964104e-05,
+      "loss": 0.6858,
+      "step": 671
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9267265001265685e-05,
+      "loss": 0.5962,
+      "step": 672
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9264429152665234e-05,
+      "loss": 0.5571,
+      "step": 673
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.926158796490113e-05,
+      "loss": 0.6941,
+      "step": 674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9258741439036835e-05,
+      "loss": 0.6626,
+      "step": 675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9255889576137803e-05,
+      "loss": 0.636,
+      "step": 676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.925303237727148e-05,
+      "loss": 0.5679,
+      "step": 677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.925016984350731e-05,
+      "loss": 0.6296,
+      "step": 678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.924730197591674e-05,
+      "loss": 0.6089,
+      "step": 679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.924442877557321e-05,
+      "loss": 0.6028,
+      "step": 680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.924155024355216e-05,
+      "loss": 0.7212,
+      "step": 681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.923866638093102e-05,
+      "loss": 0.5144,
+      "step": 682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9235777188789217e-05,
+      "loss": 0.5603,
+      "step": 683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9232882668208168e-05,
+      "loss": 0.6069,
+      "step": 684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.92299828202713e-05,
+      "loss": 0.6885,
+      "step": 685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9227077646064013e-05,
+      "loss": 0.6211,
+      "step": 686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9224167146673723e-05,
+      "loss": 0.8086,
+      "step": 687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9221251323189812e-05,
+      "loss": 0.6072,
+      "step": 688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9218330176703683e-05,
+      "loss": 0.646,
+      "step": 689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9215403708308713e-05,
+      "loss": 0.5374,
+      "step": 690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9212471919100278e-05,
+      "loss": 0.6521,
+      "step": 691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9209534810175747e-05,
+      "loss": 0.5378,
+      "step": 692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9206592382634473e-05,
+      "loss": 0.6125,
+      "step": 693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9203644637577806e-05,
+      "loss": 0.5679,
+      "step": 694
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.920069157610908e-05,
+      "loss": 0.7021,
+      "step": 695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9197733199333636e-05,
+      "loss": 0.6953,
+      "step": 696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9194769508358776e-05,
+      "loss": 0.6763,
+      "step": 697
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9191800504293822e-05,
+      "loss": 0.6292,
+      "step": 698
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9188826188250056e-05,
+      "loss": 0.551,
+      "step": 699
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.918584656134077e-05,
+      "loss": 0.6746,
+      "step": 700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9182861624681233e-05,
+      "loss": 0.625,
+      "step": 701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9179871379388706e-05,
+      "loss": 0.6499,
+      "step": 702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9176875826582435e-05,
+      "loss": 0.6687,
+      "step": 703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.917387496738365e-05,
+      "loss": 0.5137,
+      "step": 704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9170868802915576e-05,
+      "loss": 0.6025,
+      "step": 705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.916785733430341e-05,
+      "loss": 0.6919,
+      "step": 706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9164840562674346e-05,
+      "loss": 0.8159,
+      "step": 707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9161818489157565e-05,
+      "loss": 0.616,
+      "step": 708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9158791114884215e-05,
+      "loss": 0.5955,
+      "step": 709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9155758440987448e-05,
+      "loss": 0.5947,
+      "step": 710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.915272046860239e-05,
+      "loss": 0.5605,
+      "step": 711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9149677198866145e-05,
+      "loss": 0.5562,
+      "step": 712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9146628632917817e-05,
+      "loss": 0.646,
+      "step": 713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9143574771898474e-05,
+      "loss": 0.7471,
+      "step": 714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9140515616951176e-05,
+      "loss": 0.5867,
+      "step": 715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.913745116922096e-05,
+      "loss": 0.5229,
+      "step": 716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.913438142985485e-05,
+      "loss": 0.6296,
+      "step": 717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9131306400001843e-05,
+      "loss": 0.6973,
+      "step": 718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.912822608081292e-05,
+      "loss": 0.6497,
+      "step": 719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.912514047344105e-05,
+      "loss": 0.5586,
+      "step": 720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9122049579041163e-05,
+      "loss": 0.625,
+      "step": 721
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.911895339877018e-05,
+      "loss": 0.6104,
+      "step": 722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9115851933787e-05,
+      "loss": 0.6729,
+      "step": 723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9112745185252502e-05,
+      "loss": 0.5959,
+      "step": 724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.910963315432954e-05,
+      "loss": 0.6592,
+      "step": 725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9106515842182936e-05,
+      "loss": 0.5522,
+      "step": 726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9103393249979498e-05,
+      "loss": 0.6106,
+      "step": 727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9100265378888026e-05,
+      "loss": 0.5891,
+      "step": 728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.909713223007926e-05,
+      "loss": 0.5461,
+      "step": 729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.909399380472594e-05,
+      "loss": 0.6443,
+      "step": 730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9090850104002784e-05,
+      "loss": 0.5393,
+      "step": 731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9087701129086468e-05,
+      "loss": 0.6245,
+      "step": 732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.908454688115565e-05,
+      "loss": 0.6155,
+      "step": 733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.908138736139097e-05,
+      "loss": 0.5679,
+      "step": 734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.907822257097503e-05,
+      "loss": 0.6541,
+      "step": 735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9075052511092404e-05,
+      "loss": 0.6626,
+      "step": 736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.907187718292965e-05,
+      "loss": 0.604,
+      "step": 737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9068696587675285e-05,
+      "loss": 0.6479,
+      "step": 738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9065510726519805e-05,
+      "loss": 0.5637,
+      "step": 739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9062319600655674e-05,
+      "loss": 0.7017,
+      "step": 740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9059123211277333e-05,
+      "loss": 0.8066,
+      "step": 741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9055921559581178e-05,
+      "loss": 0.585,
+      "step": 742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.905271464676559e-05,
+      "loss": 0.7144,
+      "step": 743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.904950247403092e-05,
+      "loss": 0.6361,
+      "step": 744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.904628504257947e-05,
+      "loss": 0.5583,
+      "step": 745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.904306235361553e-05,
+      "loss": 0.6018,
+      "step": 746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9039834408345343e-05,
+      "loss": 0.5955,
+      "step": 747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9036601207977134e-05,
+      "loss": 0.7056,
+      "step": 748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9033362753721082e-05,
+      "loss": 0.5864,
+      "step": 749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.903011904678934e-05,
+      "loss": 0.6187,
+      "step": 750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9026870088396022e-05,
+      "loss": 0.6689,
+      "step": 751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9023615879757214e-05,
+      "loss": 0.6948,
+      "step": 752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9020356422090962e-05,
+      "loss": 0.5771,
+      "step": 753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9017091716617272e-05,
+      "loss": 0.6558,
+      "step": 754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.901382176455813e-05,
+      "loss": 0.5903,
+      "step": 755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.901054656713747e-05,
+      "loss": 0.6917,
+      "step": 756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9007266125581203e-05,
+      "loss": 0.7285,
+      "step": 757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.900398044111719e-05,
+      "loss": 0.594,
+      "step": 758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9000689514975258e-05,
+      "loss": 0.5676,
+      "step": 759
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.89973933483872e-05,
+      "loss": 0.5544,
+      "step": 760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8994091942586767e-05,
+      "loss": 0.6953,
+      "step": 761
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.899078529880968e-05,
+      "loss": 0.6111,
+      "step": 762
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.89874734182936e-05,
+      "loss": 0.5657,
+      "step": 763
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.898415630227817e-05,
+      "loss": 0.7568,
+      "step": 764
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8980833952004984e-05,
+      "loss": 0.6171,
+      "step": 765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.897750636871759e-05,
+      "loss": 0.553,
+      "step": 766
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.89741735536615e-05,
+      "loss": 0.4788,
+      "step": 767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.897083550808419e-05,
+      "loss": 0.4927,
+      "step": 768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.896749223323508e-05,
+      "loss": 0.646,
+      "step": 769
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8964143730365564e-05,
+      "loss": 0.6023,
+      "step": 770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8960790000728972e-05,
+      "loss": 0.6296,
+      "step": 771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.895743104558061e-05,
+      "loss": 0.6042,
+      "step": 772
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.895406686617773e-05,
+      "loss": 0.636,
+      "step": 773
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8950697463779547e-05,
+      "loss": 0.6826,
+      "step": 774
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8947322839647216e-05,
+      "loss": 0.5974,
+      "step": 775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.894394299504386e-05,
+      "loss": 0.6082,
+      "step": 776
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.894055793123455e-05,
+      "loss": 0.5579,
+      "step": 777
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.893716764948632e-05,
+      "loss": 0.6982,
+      "step": 778
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8933772151068144e-05,
+      "loss": 0.5115,
+      "step": 779
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8930371437250948e-05,
+      "loss": 0.6907,
+      "step": 780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.892696550930763e-05,
+      "loss": 0.5195,
+      "step": 781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8923554368513013e-05,
+      "loss": 0.626,
+      "step": 782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8920138016143894e-05,
+      "loss": 0.5612,
+      "step": 783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8916716453479005e-05,
+      "loss": 0.6167,
+      "step": 784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8913289681799028e-05,
+      "loss": 0.7285,
+      "step": 785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.890985770238662e-05,
+      "loss": 0.6704,
+      "step": 786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.890642051652635e-05,
+      "loss": 0.5471,
+      "step": 787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.890297812550476e-05,
+      "loss": 0.7075,
+      "step": 788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.889953053061033e-05,
+      "loss": 0.53,
+      "step": 789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8896077733133507e-05,
+      "loss": 0.6257,
+      "step": 790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.889261973436665e-05,
+      "loss": 0.6772,
+      "step": 791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8889156535604097e-05,
+      "loss": 0.6274,
+      "step": 792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8885688138142113e-05,
+      "loss": 0.6746,
+      "step": 793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8882214543278924e-05,
+      "loss": 0.6367,
+      "step": 794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8878735752314688e-05,
+      "loss": 0.6067,
+      "step": 795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.887525176655152e-05,
+      "loss": 0.6951,
+      "step": 796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.887176258729346e-05,
+      "loss": 0.7024,
+      "step": 797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8868268215846518e-05,
+      "loss": 0.6497,
+      "step": 798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.886476865351863e-05,
+      "loss": 0.698,
+      "step": 799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8861263901619677e-05,
+      "loss": 0.5972,
+      "step": 800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8857753961461486e-05,
+      "loss": 0.717,
+      "step": 801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8854238834357824e-05,
+      "loss": 0.4991,
+      "step": 802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8850718521624408e-05,
+      "loss": 0.6396,
+      "step": 803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.884719302457887e-05,
+      "loss": 0.6094,
+      "step": 804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8843662344540818e-05,
+      "loss": 0.5618,
+      "step": 805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8840126482831778e-05,
+      "loss": 0.6174,
+      "step": 806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8836585440775214e-05,
+      "loss": 0.5977,
+      "step": 807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.883303921969654e-05,
+      "loss": 0.5737,
+      "step": 808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.88294878209231e-05,
+      "loss": 0.6697,
+      "step": 809
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8825931245784183e-05,
+      "loss": 0.6711,
+      "step": 810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8822369495611014e-05,
+      "loss": 0.6655,
+      "step": 811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.881880257173675e-05,
+      "loss": 0.7351,
+      "step": 812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8815230475496487e-05,
+      "loss": 0.5935,
+      "step": 813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8811653208227257e-05,
+      "loss": 0.6648,
+      "step": 814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8808070771268033e-05,
+      "loss": 0.5376,
+      "step": 815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8804483165959706e-05,
+      "loss": 0.5649,
+      "step": 816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8800890393645128e-05,
+      "loss": 0.6099,
+      "step": 817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8797292455669066e-05,
+      "loss": 0.6135,
+      "step": 818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.879368935337822e-05,
+      "loss": 0.6506,
+      "step": 819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8790081088121234e-05,
+      "loss": 0.6221,
+      "step": 820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8786467661248674e-05,
+      "loss": 0.6399,
+      "step": 821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.878284907411305e-05,
+      "loss": 0.6338,
+      "step": 822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8779225328068785e-05,
+      "loss": 0.6301,
+      "step": 823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8775596424472257e-05,
+      "loss": 0.7644,
+      "step": 824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8771962364681753e-05,
+      "loss": 0.6265,
+      "step": 825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8768323150057497e-05,
+      "loss": 0.6235,
+      "step": 826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.876467878196165e-05,
+      "loss": 0.8726,
+      "step": 827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8761029261758287e-05,
+      "loss": 0.6311,
+      "step": 828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.875737459081343e-05,
+      "loss": 0.5793,
+      "step": 829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.875371477049502e-05,
+      "loss": 0.4973,
+      "step": 830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.875004980217292e-05,
+      "loss": 0.5332,
+      "step": 831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8746379687218917e-05,
+      "loss": 0.6094,
+      "step": 832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8742704427006747e-05,
+      "loss": 0.6013,
+      "step": 833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8739024022912044e-05,
+      "loss": 0.6082,
+      "step": 834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8735338476312386e-05,
+      "loss": 0.6426,
+      "step": 835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8731647788587267e-05,
+      "loss": 0.5376,
+      "step": 836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8727951961118117e-05,
+      "loss": 0.6729,
+      "step": 837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8724250995288272e-05,
+      "loss": 0.554,
+      "step": 838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8720544892483e-05,
+      "loss": 0.6064,
+      "step": 839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.87168336540895e-05,
+      "loss": 0.7244,
+      "step": 840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8713117281496876e-05,
+      "loss": 0.5938,
+      "step": 841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8709395776096174e-05,
+      "loss": 0.5789,
+      "step": 842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8705669139280335e-05,
+      "loss": 0.7029,
+      "step": 843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8701937372444248e-05,
+      "loss": 0.7869,
+      "step": 844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.869820047698471e-05,
+      "loss": 0.5886,
+      "step": 845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8694458454300436e-05,
+      "loss": 0.6335,
+      "step": 846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.869071130579206e-05,
+      "loss": 0.5916,
+      "step": 847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.868695903286213e-05,
+      "loss": 0.6169,
+      "step": 848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.868320163691513e-05,
+      "loss": 0.6543,
+      "step": 849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8679439119357447e-05,
+      "loss": 0.5007,
+      "step": 850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8675671481597385e-05,
+      "loss": 0.6768,
+      "step": 851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.867189872504517e-05,
+      "loss": 0.7815,
+      "step": 852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8668120851112945e-05,
+      "loss": 0.5326,
+      "step": 853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8664337861214756e-05,
+      "loss": 0.6826,
+      "step": 854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8660549756766583e-05,
+      "loss": 0.5514,
+      "step": 855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.86567565391863e-05,
+      "loss": 0.6278,
+      "step": 856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8652958209893714e-05,
+      "loss": 0.5901,
+      "step": 857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.864915477031053e-05,
+      "loss": 0.5552,
+      "step": 858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8645346221860376e-05,
+      "loss": 0.4963,
+      "step": 859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.864153256596878e-05,
+      "loss": 0.6045,
+      "step": 860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8637713804063207e-05,
+      "loss": 0.6331,
+      "step": 861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8633889937573004e-05,
+      "loss": 0.676,
+      "step": 862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8630060967929444e-05,
+      "loss": 0.5881,
+      "step": 863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8626226896565708e-05,
+      "loss": 0.5742,
+      "step": 864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8622387724916885e-05,
+      "loss": 0.7534,
+      "step": 865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8618543454419967e-05,
+      "loss": 0.6311,
+      "step": 866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.861469408651387e-05,
+      "loss": 0.6375,
+      "step": 867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.861083962263941e-05,
+      "loss": 0.6682,
+      "step": 868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8606980064239305e-05,
+      "loss": 0.519,
+      "step": 869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.860311541275818e-05,
+      "loss": 0.7058,
+      "step": 870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8599245669642583e-05,
+      "loss": 0.6836,
+      "step": 871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8595370836340944e-05,
+      "loss": 0.6343,
+      "step": 872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8591490914303618e-05,
+      "loss": 0.5557,
+      "step": 873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.858760590498285e-05,
+      "loss": 0.6506,
+      "step": 874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8583715809832798e-05,
+      "loss": 0.6039,
+      "step": 875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8579820630309524e-05,
+      "loss": 0.5359,
+      "step": 876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8575920367870986e-05,
+      "loss": 0.6711,
+      "step": 877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8572015023977054e-05,
+      "loss": 0.5361,
+      "step": 878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.856810460008949e-05,
+      "loss": 0.5314,
+      "step": 879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8564189097671962e-05,
+      "loss": 0.6108,
+      "step": 880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8560268518190045e-05,
+      "loss": 0.5623,
+      "step": 881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8556342863111206e-05,
+      "loss": 0.6033,
+      "step": 882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.855241213390481e-05,
+      "loss": 0.5947,
+      "step": 883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.854847633204213e-05,
+      "loss": 0.532,
+      "step": 884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8544535458996336e-05,
+      "loss": 0.7051,
+      "step": 885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8540589516242487e-05,
+      "loss": 0.5715,
+      "step": 886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.853663850525755e-05,
+      "loss": 0.7063,
+      "step": 887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.853268242752038e-05,
+      "loss": 0.5728,
+      "step": 888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.852872128451174e-05,
+      "loss": 0.6748,
+      "step": 889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.852475507771428e-05,
+      "loss": 0.6458,
+      "step": 890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8520783808612546e-05,
+      "loss": 0.5979,
+      "step": 891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8516807478692983e-05,
+      "loss": 0.6445,
+      "step": 892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.851282608944393e-05,
+      "loss": 0.6477,
+      "step": 893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8508839642355608e-05,
+      "loss": 0.7275,
+      "step": 894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8504848138920152e-05,
+      "loss": 0.572,
+      "step": 895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.850085158063157e-05,
+      "loss": 0.6465,
+      "step": 896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.849684996898578e-05,
+      "loss": 0.5857,
+      "step": 897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8492843305480573e-05,
+      "loss": 0.7073,
+      "step": 898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8488831591615643e-05,
+      "loss": 0.6292,
+      "step": 899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8484814828892568e-05,
+      "loss": 0.6123,
+      "step": 900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8480793018814822e-05,
+      "loss": 0.7222,
+      "step": 901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8476766162887768e-05,
+      "loss": 0.687,
+      "step": 902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.847273426261865e-05,
+      "loss": 0.5991,
+      "step": 903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8468697319516605e-05,
+      "loss": 0.6375,
+      "step": 904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.846465533509266e-05,
+      "loss": 0.6116,
+      "step": 905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.846060831085972e-05,
+      "loss": 0.636,
+      "step": 906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8456556248332595e-05,
+      "loss": 0.6086,
+      "step": 907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.845249914902796e-05,
+      "loss": 0.5918,
+      "step": 908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.844843701446438e-05,
+      "loss": 0.707,
+      "step": 909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8444369846162316e-05,
+      "loss": 0.5762,
+      "step": 910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8440297645644106e-05,
+      "loss": 0.5798,
+      "step": 911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8436220414433965e-05,
+      "loss": 0.6831,
+      "step": 912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8432138154058e-05,
+      "loss": 0.5654,
+      "step": 913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8428050866044198e-05,
+      "loss": 0.6621,
+      "step": 914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8423958551922426e-05,
+      "loss": 0.6509,
+      "step": 915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8419861213224436e-05,
+      "loss": 0.5728,
+      "step": 916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.841575885148386e-05,
+      "loss": 0.6804,
+      "step": 917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8411651468236205e-05,
+      "loss": 0.7837,
+      "step": 918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.840753906501886e-05,
+      "loss": 0.5288,
+      "step": 919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.84034216433711e-05,
+      "loss": 0.554,
+      "step": 920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8399299204834063e-05,
+      "loss": 0.5234,
+      "step": 921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.839517175095078e-05,
+      "loss": 0.6201,
+      "step": 922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8391039283266157e-05,
+      "loss": 0.6394,
+      "step": 923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8386901803326966e-05,
+      "loss": 0.6814,
+      "step": 924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.838275931268187e-05,
+      "loss": 0.5557,
+      "step": 925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.837861181288139e-05,
+      "loss": 0.583,
+      "step": 926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.837445930547794e-05,
+      "loss": 0.6487,
+      "step": 927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8370301792025792e-05,
+      "loss": 0.7593,
+      "step": 928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8366139274081108e-05,
+      "loss": 0.585,
+      "step": 929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8361971753201915e-05,
+      "loss": 0.707,
+      "step": 930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8357799230948102e-05,
+      "loss": 0.4905,
+      "step": 931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8353621708881453e-05,
+      "loss": 0.6089,
+      "step": 932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8349439188565597e-05,
+      "loss": 0.6409,
+      "step": 933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8345251671566065e-05,
+      "loss": 0.6528,
+      "step": 934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.834105915945023e-05,
+      "loss": 0.5264,
+      "step": 935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8336861653787344e-05,
+      "loss": 0.7153,
+      "step": 936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8332659156148533e-05,
+      "loss": 0.7458,
+      "step": 937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8328451668106793e-05,
+      "loss": 0.6489,
+      "step": 938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.832423919123698e-05,
+      "loss": 0.5442,
+      "step": 939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8320021727115817e-05,
+      "loss": 0.6636,
+      "step": 940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8315799277321902e-05,
+      "loss": 0.6484,
+      "step": 941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.831157184343569e-05,
+      "loss": 0.6084,
+      "step": 942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8307339427039506e-05,
+      "loss": 0.6282,
+      "step": 943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8303102029717543e-05,
+      "loss": 0.7817,
+      "step": 944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8298859653055855e-05,
+      "loss": 0.6199,
+      "step": 945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.829461229864236e-05,
+      "loss": 0.6189,
+      "step": 946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8290359968066838e-05,
+      "loss": 0.6018,
+      "step": 947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8286102662920933e-05,
+      "loss": 0.7161,
+      "step": 948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8281840384798147e-05,
+      "loss": 0.6978,
+      "step": 949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8277573135293853e-05,
+      "loss": 0.5164,
+      "step": 950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8273300916005277e-05,
+      "loss": 0.7112,
+      "step": 951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8269023728531513e-05,
+      "loss": 0.5796,
+      "step": 952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.826474157447349e-05,
+      "loss": 0.6692,
+      "step": 953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8260454455434033e-05,
+      "loss": 0.637,
+      "step": 954
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8256162373017797e-05,
+      "loss": 0.5564,
+      "step": 955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8251865328831313e-05,
+      "loss": 0.6484,
+      "step": 956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8247563324482956e-05,
+      "loss": 0.5621,
+      "step": 957
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.824325636158296e-05,
+      "loss": 0.6191,
+      "step": 958
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8238944441743422e-05,
+      "loss": 0.5488,
+      "step": 959
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8234627566578288e-05,
+      "loss": 0.6023,
+      "step": 960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8230305737703362e-05,
+      "loss": 0.5779,
+      "step": 961
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8225978956736298e-05,
+      "loss": 0.6519,
+      "step": 962
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8221647225296615e-05,
+      "loss": 0.709,
+      "step": 963
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8217310545005663e-05,
+      "loss": 0.7041,
+      "step": 964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8212968917486666e-05,
+      "loss": 0.614,
+      "step": 965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8208622344364687e-05,
+      "loss": 0.6375,
+      "step": 966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8204270827266652e-05,
+      "loss": 0.5715,
+      "step": 967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8199914367821326e-05,
+      "loss": 0.6035,
+      "step": 968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8195552967659326e-05,
+      "loss": 0.6614,
+      "step": 969
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.819118662841312e-05,
+      "loss": 0.5444,
+      "step": 970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.818681535171703e-05,
+      "loss": 0.533,
+      "step": 971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.818243913920722e-05,
+      "loss": 0.5933,
+      "step": 972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8178057992521697e-05,
+      "loss": 0.8257,
+      "step": 973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8173671913300328e-05,
+      "loss": 0.6111,
+      "step": 974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8169280903184808e-05,
+      "loss": 0.5757,
+      "step": 975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.81648849638187e-05,
+      "loss": 0.5996,
+      "step": 976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8160484096847398e-05,
+      "loss": 0.6606,
+      "step": 977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.815607830391814e-05,
+      "loss": 0.5571,
+      "step": 978
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8151667586680007e-05,
+      "loss": 0.5571,
+      "step": 979
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8147251946783935e-05,
+      "loss": 0.5596,
+      "step": 980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.814283138588269e-05,
+      "loss": 0.4993,
+      "step": 981
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8138405905630878e-05,
+      "loss": 0.4928,
+      "step": 982
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8133975507684966e-05,
+      "loss": 0.6077,
+      "step": 983
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8129540193703237e-05,
+      "loss": 0.5232,
+      "step": 984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8125099965345832e-05,
+      "loss": 0.5796,
+      "step": 985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8120654824274718e-05,
+      "loss": 0.5728,
+      "step": 986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8116204772153715e-05,
+      "loss": 0.5691,
+      "step": 987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8111749810648465e-05,
+      "loss": 0.707,
+      "step": 988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8107289941426463e-05,
+      "loss": 0.6584,
+      "step": 989
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8102825166157032e-05,
+      "loss": 0.5769,
+      "step": 990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.809835548651134e-05,
+      "loss": 0.5862,
+      "step": 991
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8093880904162373e-05,
+      "loss": 0.5562,
+      "step": 992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.808940142078497e-05,
+      "loss": 0.6113,
+      "step": 993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8084917038055794e-05,
+      "loss": 0.6387,
+      "step": 994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.808042775765335e-05,
+      "loss": 0.6528,
+      "step": 995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.807593358125797e-05,
+      "loss": 0.5287,
+      "step": 996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.807143451055182e-05,
+      "loss": 0.6384,
+      "step": 997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8066930547218896e-05,
+      "loss": 0.5869,
+      "step": 998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8062421692945035e-05,
+      "loss": 0.8325,
+      "step": 999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.805790794941789e-05,
+      "loss": 0.6207,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8053389318326953e-05,
+      "loss": 0.6492,
+      "step": 1001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8048865801363546e-05,
+      "loss": 0.6665,
+      "step": 1002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8044337400220814e-05,
+      "loss": 0.5957,
+      "step": 1003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8039804116593743e-05,
+      "loss": 0.5637,
+      "step": 1004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8035265952179125e-05,
+      "loss": 0.6716,
+      "step": 1005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8030722908675593e-05,
+      "loss": 0.7236,
+      "step": 1006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8026174987783607e-05,
+      "loss": 0.5935,
+      "step": 1007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8021622191205456e-05,
+      "loss": 0.5168,
+      "step": 1008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8017064520645238e-05,
+      "loss": 0.563,
+      "step": 1009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.801250197780889e-05,
+      "loss": 0.5718,
+      "step": 1010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8007934564404168e-05,
+      "loss": 0.5784,
+      "step": 1011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.800336228214065e-05,
+      "loss": 0.6707,
+      "step": 1012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7998785132729737e-05,
+      "loss": 0.7288,
+      "step": 1013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7994203117884654e-05,
+      "loss": 0.5151,
+      "step": 1014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7989616239320446e-05,
+      "loss": 0.6614,
+      "step": 1015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.798502449875397e-05,
+      "loss": 0.7195,
+      "step": 1016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.798042789790392e-05,
+      "loss": 0.5938,
+      "step": 1017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7975826438490795e-05,
+      "loss": 0.5872,
+      "step": 1018
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.797122012223692e-05,
+      "loss": 0.5457,
+      "step": 1019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.796660895086643e-05,
+      "loss": 0.4954,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7961992926105288e-05,
+      "loss": 0.7065,
+      "step": 1021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7957372049681265e-05,
+      "loss": 0.5769,
+      "step": 1022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7952746323323955e-05,
+      "loss": 0.6162,
+      "step": 1023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.794811574876476e-05,
+      "loss": 0.5698,
+      "step": 1024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7943480327736898e-05,
+      "loss": 0.5737,
+      "step": 1025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7938840061975405e-05,
+      "loss": 0.5935,
+      "step": 1026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7934194953217134e-05,
+      "loss": 0.6089,
+      "step": 1027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7929545003200738e-05,
+      "loss": 0.614,
+      "step": 1028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.792489021366669e-05,
+      "loss": 0.6182,
+      "step": 1029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7920230586357283e-05,
+      "loss": 0.6592,
+      "step": 1030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.79155661230166e-05,
+      "loss": 0.5186,
+      "step": 1031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7910896825390547e-05,
+      "loss": 0.5132,
+      "step": 1032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7906222695226843e-05,
+      "loss": 0.6951,
+      "step": 1033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.790154373427501e-05,
+      "loss": 0.6606,
+      "step": 1034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.789685994428638e-05,
+      "loss": 0.6393,
+      "step": 1035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7892171327014086e-05,
+      "loss": 0.6565,
+      "step": 1036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7887477884213076e-05,
+      "loss": 0.6084,
+      "step": 1037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7882779617640107e-05,
+      "loss": 0.6189,
+      "step": 1038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7878076529053727e-05,
+      "loss": 0.6262,
+      "step": 1039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7873368620214308e-05,
+      "loss": 0.6064,
+      "step": 1040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7868655892884006e-05,
+      "loss": 0.6169,
+      "step": 1041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7863938348826802e-05,
+      "loss": 0.6091,
+      "step": 1042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.785921598980846e-05,
+      "loss": 0.6189,
+      "step": 1043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7854488817596556e-05,
+      "loss": 0.5505,
+      "step": 1044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7849756833960466e-05,
+      "loss": 0.5681,
+      "step": 1045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7845020040671374e-05,
+      "loss": 0.5095,
+      "step": 1046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7840278439502254e-05,
+      "loss": 0.5703,
+      "step": 1047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7835532032227878e-05,
+      "loss": 0.6589,
+      "step": 1048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.783078082062483e-05,
+      "loss": 0.6648,
+      "step": 1049
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.782602480647148e-05,
+      "loss": 0.609,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7821263991548003e-05,
+      "loss": 0.7261,
+      "step": 1051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.781649837763636e-05,
+      "loss": 0.7263,
+      "step": 1052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7811727966520328e-05,
+      "loss": 0.8037,
+      "step": 1053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7806952759985464e-05,
+      "loss": 0.6985,
+      "step": 1054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.780217275981912e-05,
+      "loss": 0.5955,
+      "step": 1055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7797387967810444e-05,
+      "loss": 0.5454,
+      "step": 1056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7792598385750388e-05,
+      "loss": 0.5569,
+      "step": 1057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.778780401543168e-05,
+      "loss": 0.6121,
+      "step": 1058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7783004858648855e-05,
+      "loss": 0.6201,
+      "step": 1059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.777820091719823e-05,
+      "loss": 0.5598,
+      "step": 1060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.777339219287792e-05,
+      "loss": 0.6174,
+      "step": 1061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7768578687487824e-05,
+      "loss": 0.6873,
+      "step": 1062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7763760402829622e-05,
+      "loss": 0.5283,
+      "step": 1063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7758937340706813e-05,
+      "loss": 0.717,
+      "step": 1064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7754109502924657e-05,
+      "loss": 0.792,
+      "step": 1065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.774927689129021e-05,
+      "loss": 0.6638,
+      "step": 1066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.774443950761231e-05,
+      "loss": 0.4989,
+      "step": 1067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.773959735370159e-05,
+      "loss": 0.6257,
+      "step": 1068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7734750431370462e-05,
+      "loss": 0.6079,
+      "step": 1069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7729898742433125e-05,
+      "loss": 0.5525,
+      "step": 1070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7725042288705564e-05,
+      "loss": 0.5496,
+      "step": 1071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7720181072005543e-05,
+      "loss": 0.6489,
+      "step": 1072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.771531509415261e-05,
+      "loss": 0.5439,
+      "step": 1073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.771044435696811e-05,
+      "loss": 0.6511,
+      "step": 1074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7705568862275133e-05,
+      "loss": 0.5354,
+      "step": 1075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.770068861189859e-05,
+      "loss": 0.7087,
+      "step": 1076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7695803607665148e-05,
+      "loss": 0.5371,
+      "step": 1077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.769091385140326e-05,
+      "loss": 0.4751,
+      "step": 1078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7686019344943157e-05,
+      "loss": 0.5757,
+      "step": 1079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.768112009011685e-05,
+      "loss": 0.583,
+      "step": 1080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7676216088758128e-05,
+      "loss": 0.7288,
+      "step": 1081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.767130734270255e-05,
+      "loss": 0.5884,
+      "step": 1082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.766639385378746e-05,
+      "loss": 0.7358,
+      "step": 1083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7661475623851967e-05,
+      "loss": 0.6367,
+      "step": 1084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7656552654736967e-05,
+      "loss": 0.6289,
+      "step": 1085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7651624948285118e-05,
+      "loss": 0.5383,
+      "step": 1086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.764669250634086e-05,
+      "loss": 0.5891,
+      "step": 1087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7641755330750398e-05,
+      "loss": 0.5835,
+      "step": 1088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7636813423361716e-05,
+      "loss": 0.5911,
+      "step": 1089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7631866786024565e-05,
+      "loss": 0.5742,
+      "step": 1090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7626915420590462e-05,
+      "loss": 0.6248,
+      "step": 1091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7621959328912704e-05,
+      "loss": 0.6602,
+      "step": 1092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7616998512846355e-05,
+      "loss": 0.6875,
+      "step": 1093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.761203297424824e-05,
+      "loss": 0.5388,
+      "step": 1094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.760706271497695e-05,
+      "loss": 0.6931,
+      "step": 1095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7602087736892862e-05,
+      "loss": 0.5061,
+      "step": 1096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.75971080418581e-05,
+      "loss": 0.5928,
+      "step": 1097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7592123631736555e-05,
+      "loss": 0.5381,
+      "step": 1098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.758713450839389e-05,
+      "loss": 0.8206,
+      "step": 1099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7582140673697532e-05,
+      "loss": 0.6006,
+      "step": 1100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.757714212951667e-05,
+      "loss": 0.6163,
+      "step": 1101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.757213887772225e-05,
+      "loss": 0.6416,
+      "step": 1102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7567130920186986e-05,
+      "loss": 0.7334,
+      "step": 1103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7562118258785354e-05,
+      "loss": 0.532,
+      "step": 1104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.755710089539359e-05,
+      "loss": 0.5828,
+      "step": 1105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7552078831889685e-05,
+      "loss": 0.5945,
+      "step": 1106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.75470520701534e-05,
+      "loss": 0.574,
+      "step": 1107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.754202061206624e-05,
+      "loss": 0.6245,
+      "step": 1108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.753698445951148e-05,
+      "loss": 0.5601,
+      "step": 1109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.753194361437415e-05,
+      "loss": 0.6423,
+      "step": 1110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7526898078541025e-05,
+      "loss": 0.5942,
+      "step": 1111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.752184785390065e-05,
+      "loss": 0.6226,
+      "step": 1112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.751679294234332e-05,
+      "loss": 0.5718,
+      "step": 1113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7511733345761085e-05,
+      "loss": 0.7388,
+      "step": 1114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7506669066047747e-05,
+      "loss": 0.6477,
+      "step": 1115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7501600105098856e-05,
+      "loss": 0.6421,
+      "step": 1116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7496526464811733e-05,
+      "loss": 0.6594,
+      "step": 1117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7491448147085422e-05,
+      "loss": 0.657,
+      "step": 1118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7486365153820745e-05,
+      "loss": 0.6228,
+      "step": 1119
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7481277486920252e-05,
+      "loss": 0.6094,
+      "step": 1120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.747618514828826e-05,
+      "loss": 0.7117,
+      "step": 1121
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7471088139830827e-05,
+      "loss": 0.5779,
+      "step": 1122
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.746598646345575e-05,
+      "loss": 0.6809,
+      "step": 1123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.746088012107259e-05,
+      "loss": 0.5745,
+      "step": 1124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.745576911459265e-05,
+      "loss": 0.5247,
+      "step": 1125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7450653445928968e-05,
+      "loss": 0.5601,
+      "step": 1126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7445533116996338e-05,
+      "loss": 0.6726,
+      "step": 1127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.744040812971129e-05,
+      "loss": 0.5317,
+      "step": 1128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7435278485992107e-05,
+      "loss": 0.5291,
+      "step": 1129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.743014418775881e-05,
+      "loss": 0.6946,
+      "step": 1130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.742500523693316e-05,
+      "loss": 0.6455,
+      "step": 1131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7419861635438667e-05,
+      "loss": 0.8027,
+      "step": 1132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.741471338520057e-05,
+      "loss": 0.6008,
+      "step": 1133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7409560488145863e-05,
+      "loss": 0.5779,
+      "step": 1134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7404402946203257e-05,
+      "loss": 0.5149,
+      "step": 1135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.739924076130323e-05,
+      "loss": 0.6233,
+      "step": 1136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7394073935377974e-05,
+      "loss": 0.6194,
+      "step": 1137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7388902470361438e-05,
+      "loss": 0.5815,
+      "step": 1138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7383726368189286e-05,
+      "loss": 0.7551,
+      "step": 1139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7378545630798933e-05,
+      "loss": 0.7124,
+      "step": 1140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7373360260129526e-05,
+      "loss": 0.5898,
+      "step": 1141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.736817025812194e-05,
+      "loss": 0.5818,
+      "step": 1142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.736297562671879e-05,
+      "loss": 0.5964,
+      "step": 1143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.735777636786443e-05,
+      "loss": 0.635,
+      "step": 1144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.735257248350493e-05,
+      "loss": 0.5076,
+      "step": 1145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7347363975588097e-05,
+      "loss": 0.5244,
+      "step": 1146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7342150846063478e-05,
+      "loss": 0.52,
+      "step": 1147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.733693309688234e-05,
+      "loss": 0.7571,
+      "step": 1148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7331710729997682e-05,
+      "loss": 0.6292,
+      "step": 1149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.732648374736423e-05,
+      "loss": 0.6729,
+      "step": 1150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7321252150938446e-05,
+      "loss": 0.5562,
+      "step": 1151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7316015942678503e-05,
+      "loss": 0.6401,
+      "step": 1152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7310775124544314e-05,
+      "loss": 0.5947,
+      "step": 1153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7305529698497515e-05,
+      "loss": 0.7285,
+      "step": 1154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7300279666501457e-05,
+      "loss": 0.5024,
+      "step": 1155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7295025030521227e-05,
+      "loss": 0.5022,
+      "step": 1156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7289765792523635e-05,
+      "loss": 0.5234,
+      "step": 1157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7284501954477202e-05,
+      "loss": 0.6978,
+      "step": 1158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.727923351835218e-05,
+      "loss": 0.6489,
+      "step": 1159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7273960486120547e-05,
+      "loss": 0.6565,
+      "step": 1160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.726868285975598e-05,
+      "loss": 0.5525,
+      "step": 1161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7263400641233903e-05,
+      "loss": 0.5621,
+      "step": 1162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7258113832531443e-05,
+      "loss": 0.5466,
+      "step": 1163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7252822435627443e-05,
+      "loss": 0.5332,
+      "step": 1164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7247526452502477e-05,
+      "loss": 0.6206,
+      "step": 1165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7242225885138818e-05,
+      "loss": 0.6049,
+      "step": 1166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.723692073552047e-05,
+      "loss": 0.5782,
+      "step": 1167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7231611005633144e-05,
+      "loss": 0.665,
+      "step": 1168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7226296697464265e-05,
+      "loss": 0.6088,
+      "step": 1169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.722097781300298e-05,
+      "loss": 0.5681,
+      "step": 1170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7215654354240134e-05,
+      "loss": 0.6252,
+      "step": 1171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.72103263231683e-05,
+      "loss": 0.6895,
+      "step": 1172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7204993721781757e-05,
+      "loss": 0.542,
+      "step": 1173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.719965655207648e-05,
+      "loss": 0.6909,
+      "step": 1174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7194314816050186e-05,
+      "loss": 0.5662,
+      "step": 1175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7188968515702267e-05,
+      "loss": 0.6233,
+      "step": 1176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.718361765303385e-05,
+      "loss": 0.5784,
+      "step": 1177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.717826223004775e-05,
+      "loss": 0.7009,
+      "step": 1178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7172902248748502e-05,
+      "loss": 0.7939,
+      "step": 1179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.716753771114234e-05,
+      "loss": 0.562,
+      "step": 1180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7162168619237207e-05,
+      "loss": 0.5168,
+      "step": 1181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.715679497504275e-05,
+      "loss": 0.4839,
+      "step": 1182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7151416780570313e-05,
+      "loss": 0.6282,
+      "step": 1183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.714603403783296e-05,
+      "loss": 0.7183,
+      "step": 1184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7140646748845442e-05,
+      "loss": 0.5339,
+      "step": 1185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7135254915624213e-05,
+      "loss": 0.5876,
+      "step": 1186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7129858540187437e-05,
+      "loss": 0.6453,
+      "step": 1187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.712445762455497e-05,
+      "loss": 0.6084,
+      "step": 1188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7119052170748372e-05,
+      "loss": 0.5806,
+      "step": 1189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.71136421807909e-05,
+      "loss": 0.512,
+      "step": 1190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7108227656707507e-05,
+      "loss": 0.7451,
+      "step": 1191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7102808600524844e-05,
+      "loss": 0.793,
+      "step": 1192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7097385014271263e-05,
+      "loss": 0.551,
+      "step": 1193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.70919568999768e-05,
+      "loss": 0.5315,
+      "step": 1194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.70865242596732e-05,
+      "loss": 0.6455,
+      "step": 1195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.70810870953939e-05,
+      "loss": 0.53,
+      "step": 1196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7075645409174016e-05,
+      "loss": 0.5994,
+      "step": 1197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7070199203050373e-05,
+      "loss": 0.6423,
+      "step": 1198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7064748479061476e-05,
+      "loss": 0.5039,
+      "step": 1199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.705929323924753e-05,
+      "loss": 0.5464,
+      "step": 1200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.705383348565042e-05,
+      "loss": 0.6521,
+      "step": 1201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7048369220313743e-05,
+      "loss": 0.5312,
+      "step": 1202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.704290044528275e-05,
+      "loss": 0.5593,
+      "step": 1203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7037427162604414e-05,
+      "loss": 0.6418,
+      "step": 1204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.703194937432737e-05,
+      "loss": 0.4175,
+      "step": 1205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7026467082501957e-05,
+      "loss": 0.5234,
+      "step": 1206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7020980289180185e-05,
+      "loss": 0.6353,
+      "step": 1207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.701548899641576e-05,
+      "loss": 0.541,
+      "step": 1208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7009993206264067e-05,
+      "loss": 0.6428,
+      "step": 1209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7004492920782175e-05,
+      "loss": 0.6543,
+      "step": 1210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.699898814202884e-05,
+      "loss": 0.5339,
+      "step": 1211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6993478872064494e-05,
+      "loss": 0.6499,
+      "step": 1212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6987965112951248e-05,
+      "loss": 0.6074,
+      "step": 1213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69824468667529e-05,
+      "loss": 0.7085,
+      "step": 1214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.697692413553493e-05,
+      "loss": 0.5889,
+      "step": 1215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6971396921364482e-05,
+      "loss": 0.626,
+      "step": 1216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6965865226310398e-05,
+      "loss": 0.6057,
+      "step": 1217
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6960329052443177e-05,
+      "loss": 0.613,
+      "step": 1218
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.695478840183501e-05,
+      "loss": 0.6753,
+      "step": 1219
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6949243276559754e-05,
+      "loss": 0.6565,
+      "step": 1220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.694369367869295e-05,
+      "loss": 0.6082,
+      "step": 1221
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69381396103118e-05,
+      "loss": 0.6519,
+      "step": 1222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69325810734952e-05,
+      "loss": 0.5928,
+      "step": 1223
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.692701807032369e-05,
+      "loss": 0.6208,
+      "step": 1224
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6921450602879505e-05,
+      "loss": 0.5834,
+      "step": 1225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6915878673246548e-05,
+      "loss": 0.6289,
+      "step": 1226
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.691030228351038e-05,
+      "loss": 0.4783,
+      "step": 1227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.690472143575824e-05,
+      "loss": 0.6772,
+      "step": 1228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6899136132079033e-05,
+      "loss": 0.6597,
+      "step": 1229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.689354637456334e-05,
+      "loss": 0.4939,
+      "step": 1230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.68879521653034e-05,
+      "loss": 0.5063,
+      "step": 1231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6882353506393115e-05,
+      "loss": 0.6636,
+      "step": 1232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6876750399928066e-05,
+      "loss": 0.5813,
+      "step": 1233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.687114284800549e-05,
+      "loss": 0.5378,
+      "step": 1234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6865530852724278e-05,
+      "loss": 0.5466,
+      "step": 1235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6859914416185012e-05,
+      "loss": 0.814,
+      "step": 1236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.685429354048991e-05,
+      "loss": 0.6633,
+      "step": 1237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6848668227742858e-05,
+      "loss": 0.5803,
+      "step": 1238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.684303848004941e-05,
+      "loss": 0.6257,
+      "step": 1239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6837404299516782e-05,
+      "loss": 0.6895,
+      "step": 1240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.683176568825383e-05,
+      "loss": 0.54,
+      "step": 1241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.682612264837109e-05,
+      "loss": 0.6191,
+      "step": 1242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6820475181980743e-05,
+      "loss": 0.5764,
+      "step": 1243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6814823291196635e-05,
+      "loss": 0.4905,
+      "step": 1244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6809166978134258e-05,
+      "loss": 0.7112,
+      "step": 1245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6803506244910772e-05,
+      "loss": 0.6511,
+      "step": 1246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.679784109364498e-05,
+      "loss": 0.5583,
+      "step": 1247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6792171526457345e-05,
+      "loss": 0.5181,
+      "step": 1248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6786497545469984e-05,
+      "loss": 0.5662,
+      "step": 1249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6780819152806656e-05,
+      "loss": 0.6138,
+      "step": 1250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.677513635059278e-05,
+      "loss": 0.6926,
+      "step": 1251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.676944914095543e-05,
+      "loss": 0.5186,
+      "step": 1252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6763757526023322e-05,
+      "loss": 0.5471,
+      "step": 1253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.675806150792682e-05,
+      "loss": 0.657,
+      "step": 1254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6752361088797938e-05,
+      "loss": 0.6565,
+      "step": 1255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.674665627077034e-05,
+      "loss": 0.562,
+      "step": 1256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6740947055979332e-05,
+      "loss": 0.6738,
+      "step": 1257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6735233446561877e-05,
+      "loss": 0.6191,
+      "step": 1258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.672951544465656e-05,
+      "loss": 0.5383,
+      "step": 1259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6723793052403635e-05,
+      "loss": 0.5298,
+      "step": 1260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6718066271944984e-05,
+      "loss": 0.5859,
+      "step": 1261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.671233510542414e-05,
+      "loss": 0.5044,
+      "step": 1262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.670659955498627e-05,
+      "loss": 0.6724,
+      "step": 1263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6700859622778184e-05,
+      "loss": 0.5405,
+      "step": 1264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6695115310948335e-05,
+      "loss": 0.6484,
+      "step": 1265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.668936662164682e-05,
+      "loss": 0.5823,
+      "step": 1266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.668361355702536e-05,
+      "loss": 0.6016,
+      "step": 1267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.667785611923733e-05,
+      "loss": 0.6326,
+      "step": 1268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6672094310437726e-05,
+      "loss": 0.6116,
+      "step": 1269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6666328132783195e-05,
+      "loss": 0.5851,
+      "step": 1270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.666055758843201e-05,
+      "loss": 0.635,
+      "step": 1271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.665478267954408e-05,
+      "loss": 0.6531,
+      "step": 1272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6649003408280946e-05,
+      "loss": 0.5332,
+      "step": 1273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.664321977680579e-05,
+      "loss": 0.5967,
+      "step": 1274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6637431787283414e-05,
+      "loss": 0.6001,
+      "step": 1275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6631639441880264e-05,
+      "loss": 0.562,
+      "step": 1276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6625842742764404e-05,
+      "loss": 0.6356,
+      "step": 1277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.662004169210554e-05,
+      "loss": 0.4265,
+      "step": 1278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6614236292074994e-05,
+      "loss": 0.6108,
+      "step": 1279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6608426544845726e-05,
+      "loss": 0.5774,
+      "step": 1280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.660261245259232e-05,
+      "loss": 0.5889,
+      "step": 1281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6596794017490982e-05,
+      "loss": 0.6553,
+      "step": 1282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.659097124171955e-05,
+      "loss": 0.5605,
+      "step": 1283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6585144127457487e-05,
+      "loss": 0.7122,
+      "step": 1284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.657931267688587e-05,
+      "loss": 0.6274,
+      "step": 1285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6573476892187406e-05,
+      "loss": 0.626,
+      "step": 1286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.656763677554643e-05,
+      "loss": 0.6035,
+      "step": 1287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.656179232914889e-05,
+      "loss": 0.4546,
+      "step": 1288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.655594355518236e-05,
+      "loss": 0.5764,
+      "step": 1289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.655009045583603e-05,
+      "loss": 0.6392,
+      "step": 1290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6544233033300703e-05,
+      "loss": 0.7273,
+      "step": 1291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.653837128976882e-05,
+      "loss": 0.5391,
+      "step": 1292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6532505227434412e-05,
+      "loss": 0.5505,
+      "step": 1293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.652663484849315e-05,
+      "loss": 0.6318,
+      "step": 1294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6520760155142317e-05,
+      "loss": 0.478,
+      "step": 1295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.651488114958079e-05,
+      "loss": 0.6594,
+      "step": 1296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6508997834009092e-05,
+      "loss": 0.5527,
+      "step": 1297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6503110210629336e-05,
+      "loss": 0.6394,
+      "step": 1298
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.649721828164525e-05,
+      "loss": 0.6344,
+      "step": 1299
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6491322049262184e-05,
+      "loss": 0.6367,
+      "step": 1300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6485421515687092e-05,
+      "loss": 0.6357,
+      "step": 1301
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6479516683128538e-05,
+      "loss": 0.6331,
+      "step": 1302
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6473607553796696e-05,
+      "loss": 0.6917,
+      "step": 1303
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6467694129903345e-05,
+      "loss": 0.5356,
+      "step": 1304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6461776413661878e-05,
+      "loss": 0.5872,
+      "step": 1305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.645585440728729e-05,
+      "loss": 0.6594,
+      "step": 1306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6449928112996183e-05,
+      "loss": 0.6558,
+      "step": 1307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6443997533006766e-05,
+      "loss": 0.531,
+      "step": 1308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6438062669538846e-05,
+      "loss": 0.5784,
+      "step": 1309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.643212352481384e-05,
+      "loss": 0.551,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6426180101054764e-05,
+      "loss": 0.5623,
+      "step": 1311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6420232400486236e-05,
+      "loss": 0.5442,
+      "step": 1312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6414280425334478e-05,
+      "loss": 0.5659,
+      "step": 1313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6408324177827305e-05,
+      "loss": 0.5001,
+      "step": 1314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.640236366019414e-05,
+      "loss": 0.5383,
+      "step": 1315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6396398874665996e-05,
+      "loss": 0.4397,
+      "step": 1316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.639042982347549e-05,
+      "loss": 0.5781,
+      "step": 1317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.638445650885683e-05,
+      "loss": 0.4832,
+      "step": 1318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.637847893304582e-05,
+      "loss": 0.5393,
+      "step": 1319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6372497098279868e-05,
+      "loss": 0.6824,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6366511006797964e-05,
+      "loss": 0.5806,
+      "step": 1321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6360520660840703e-05,
+      "loss": 0.6082,
+      "step": 1322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6354526062650264e-05,
+      "loss": 0.5974,
+      "step": 1323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6348527214470417e-05,
+      "loss": 0.6306,
+      "step": 1324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6342524118546526e-05,
+      "loss": 0.7004,
+      "step": 1325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.633651677712555e-05,
+      "loss": 0.5566,
+      "step": 1326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6330505192456028e-05,
+      "loss": 0.5347,
+      "step": 1327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6324489366788087e-05,
+      "loss": 0.4675,
+      "step": 1328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6318469302373453e-05,
+      "loss": 0.5093,
+      "step": 1329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6312445001465427e-05,
+      "loss": 0.4434,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6306416466318898e-05,
+      "loss": 0.5825,
+      "step": 1331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.630038369919035e-05,
+      "loss": 0.6467,
+      "step": 1332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6294346702337828e-05,
+      "loss": 0.7754,
+      "step": 1333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6288305478020988e-05,
+      "loss": 0.551,
+      "step": 1334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.628226002850105e-05,
+      "loss": 0.5818,
+      "step": 1335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6276210356040824e-05,
+      "loss": 0.5754,
+      "step": 1336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6270156462904692e-05,
+      "loss": 0.5405,
+      "step": 1337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6264098351358617e-05,
+      "loss": 0.5713,
+      "step": 1338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.625803602367016e-05,
+      "loss": 0.5312,
+      "step": 1339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.625196948210843e-05,
+      "loss": 0.678,
+      "step": 1340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6245898728944137e-05,
+      "loss": 0.5989,
+      "step": 1341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6239823766449555e-05,
+      "loss": 0.5442,
+      "step": 1342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.623374459689854e-05,
+      "loss": 0.5662,
+      "step": 1343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6227661222566516e-05,
+      "loss": 0.6399,
+      "step": 1344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.622157364573049e-05,
+      "loss": 0.5859,
+      "step": 1345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6215481868669033e-05,
+      "loss": 0.5869,
+      "step": 1346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6209385893662293e-05,
+      "loss": 0.5525,
+      "step": 1347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.620328572299199e-05,
+      "loss": 0.6396,
+      "step": 1348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.619718135894141e-05,
+      "loss": 0.4757,
+      "step": 1349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6191072803795413e-05,
+      "loss": 0.6465,
+      "step": 1350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6184960059840423e-05,
+      "loss": 0.5227,
+      "step": 1351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6178843129364437e-05,
+      "loss": 0.5684,
+      "step": 1352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.617272201465702e-05,
+      "loss": 0.6338,
+      "step": 1353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.616659671800929e-05,
+      "loss": 0.6072,
+      "step": 1354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6160467241713954e-05,
+      "loss": 0.5862,
+      "step": 1355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6154333588065263e-05,
+      "loss": 0.3824,
+      "step": 1356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6148195759359035e-05,
+      "loss": 0.541,
+      "step": 1357
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6142053757892657e-05,
+      "loss": 0.5547,
+      "step": 1358
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6135907585965074e-05,
+      "loss": 0.6379,
+      "step": 1359
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6129757245876796e-05,
+      "loss": 0.5627,
+      "step": 1360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.612360273992989e-05,
+      "loss": 0.574,
+      "step": 1361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.611744407042798e-05,
+      "loss": 0.5537,
+      "step": 1362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.611128123967625e-05,
+      "loss": 0.6074,
+      "step": 1363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6105114249981446e-05,
+      "loss": 0.4731,
+      "step": 1364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.609894310365186e-05,
+      "loss": 0.5977,
+      "step": 1365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6092767802997353e-05,
+      "loss": 0.5767,
+      "step": 1366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6086588350329333e-05,
+      "loss": 0.7434,
+      "step": 1367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6080404747960768e-05,
+      "loss": 0.4106,
+      "step": 1368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6074216998206165e-05,
+      "loss": 0.4497,
+      "step": 1369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.60680251033816e-05,
+      "loss": 0.6035,
+      "step": 1370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6061829065804696e-05,
+      "loss": 0.5215,
+      "step": 1371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6055628887794624e-05,
+      "loss": 0.5979,
+      "step": 1372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6049424571672095e-05,
+      "loss": 0.5356,
+      "step": 1373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6043216119759395e-05,
+      "loss": 0.6187,
+      "step": 1374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6037003534380333e-05,
+      "loss": 0.4788,
+      "step": 1375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6030786817860275e-05,
+      "loss": 0.5408,
+      "step": 1376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6024565972526132e-05,
+      "loss": 0.5938,
+      "step": 1377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6018341000706365e-05,
+      "loss": 0.5481,
+      "step": 1378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6012111904730976e-05,
+      "loss": 0.7358,
+      "step": 1379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6005878686931504e-05,
+      "loss": 0.5347,
+      "step": 1380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5999641349641045e-05,
+      "loss": 0.5337,
+      "step": 1381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5993399895194222e-05,
+      "loss": 0.5013,
+      "step": 1382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5987154325927207e-05,
+      "loss": 0.6808,
+      "step": 1383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5980904644177716e-05,
+      "loss": 0.5557,
+      "step": 1384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5974650852285e-05,
+      "loss": 0.5842,
+      "step": 1385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5968392952589836e-05,
+      "loss": 0.5208,
+      "step": 1386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5962130947434563e-05,
+      "loss": 0.5024,
+      "step": 1387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5955864839163044e-05,
+      "loss": 0.6667,
+      "step": 1388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5949594630120674e-05,
+      "loss": 0.6396,
+      "step": 1389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5943320322654383e-05,
+      "loss": 0.4893,
+      "step": 1390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5937041919112647e-05,
+      "loss": 0.7102,
+      "step": 1391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.593075942184546e-05,
+      "loss": 0.5793,
+      "step": 1392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.592447283320436e-05,
+      "loss": 0.4219,
+      "step": 1393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5918182155542415e-05,
+      "loss": 0.6094,
+      "step": 1394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5911887391214208e-05,
+      "loss": 0.6033,
+      "step": 1395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5905588542575877e-05,
+      "loss": 0.4685,
+      "step": 1396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.589928561198507e-05,
+      "loss": 0.5967,
+      "step": 1397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.589297860180097e-05,
+      "loss": 0.6208,
+      "step": 1398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5886667514384278e-05,
+      "loss": 0.5874,
+      "step": 1399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.588035235209724e-05,
+      "loss": 0.5801,
+      "step": 1400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5874033117303608e-05,
+      "loss": 0.5166,
+      "step": 1401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.586770981236867e-05,
+      "loss": 0.5349,
+      "step": 1402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5861382439659235e-05,
+      "loss": 0.5525,
+      "step": 1403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5855051001543628e-05,
+      "loss": 0.634,
+      "step": 1404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5848715500391706e-05,
+      "loss": 0.5649,
+      "step": 1405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5842375938574837e-05,
+      "loss": 0.5161,
+      "step": 1406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5836032318465915e-05,
+      "loss": 0.6038,
+      "step": 1407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5829684642439356e-05,
+      "loss": 0.5815,
+      "step": 1408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5823332912871085e-05,
+      "loss": 0.571,
+      "step": 1409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.581697713213855e-05,
+      "loss": 0.5745,
+      "step": 1410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5810617302620714e-05,
+      "loss": 0.6006,
+      "step": 1411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5804253426698056e-05,
+      "loss": 0.5527,
+      "step": 1412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5797885506752573e-05,
+      "loss": 0.6445,
+      "step": 1413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.579151354516777e-05,
+      "loss": 0.6006,
+      "step": 1414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5785137544328657e-05,
+      "loss": 0.5513,
+      "step": 1415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5778757506621783e-05,
+      "loss": 0.6956,
+      "step": 1416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5772373434435176e-05,
+      "loss": 0.6548,
+      "step": 1417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5765985330158395e-05,
+      "loss": 0.5237,
+      "step": 1418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.57595931961825e-05,
+      "loss": 0.6748,
+      "step": 1419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.575319703490007e-05,
+      "loss": 0.5303,
+      "step": 1420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5746796848705168e-05,
+      "loss": 0.4911,
+      "step": 1421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.574039263999339e-05,
+      "loss": 0.5779,
+      "step": 1422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5733984411161818e-05,
+      "loss": 0.5715,
+      "step": 1423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.572757216460905e-05,
+      "loss": 0.6697,
+      "step": 1424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5721155902735186e-05,
+      "loss": 0.5684,
+      "step": 1425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5714735627941826e-05,
+      "loss": 0.4995,
+      "step": 1426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5708311342632077e-05,
+      "loss": 0.6147,
+      "step": 1427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5701883049210536e-05,
+      "loss": 0.5895,
+      "step": 1428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5695450750083306e-05,
+      "loss": 0.5662,
+      "step": 1429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5689014447658006e-05,
+      "loss": 0.585,
+      "step": 1430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5682574144343724e-05,
+      "loss": 0.6292,
+      "step": 1431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5676129842551065e-05,
+      "loss": 0.6255,
+      "step": 1432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5669681544692127e-05,
+      "loss": 0.6211,
+      "step": 1433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5663229253180497e-05,
+      "loss": 0.5496,
+      "step": 1434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5656772970431273e-05,
+      "loss": 0.6082,
+      "step": 1435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5650312698861023e-05,
+      "loss": 0.6956,
+      "step": 1436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5643848440887833e-05,
+      "loss": 0.4768,
+      "step": 1437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5637380198931263e-05,
+      "loss": 0.5525,
+      "step": 1438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5630907975412372e-05,
+      "loss": 0.5464,
+      "step": 1439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.56244317727537e-05,
+      "loss": 0.5476,
+      "step": 1440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.56179515933793e-05,
+      "loss": 0.4595,
+      "step": 1441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5611467439714685e-05,
+      "loss": 0.6082,
+      "step": 1442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.560497931418687e-05,
+      "loss": 0.6458,
+      "step": 1443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5598487219224358e-05,
+      "loss": 0.5632,
+      "step": 1444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5591991157257135e-05,
+      "loss": 0.5078,
+      "step": 1445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5585491130716662e-05,
+      "loss": 0.604,
+      "step": 1446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.557898714203591e-05,
+      "loss": 0.5576,
+      "step": 1447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.55724791936493e-05,
+      "loss": 0.4673,
+      "step": 1448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5565967287992766e-05,
+      "loss": 0.5295,
+      "step": 1449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5559451427503702e-05,
+      "loss": 0.5828,
+      "step": 1450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5552931614620983e-05,
+      "loss": 0.5859,
+      "step": 1451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5546407851784984e-05,
+      "loss": 0.5872,
+      "step": 1452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5539880141437525e-05,
+      "loss": 0.5588,
+      "step": 1453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5533348486021938e-05,
+      "loss": 0.5354,
+      "step": 1454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.552681288798301e-05,
+      "loss": 0.5708,
+      "step": 1455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.552027334976701e-05,
+      "loss": 0.6199,
+      "step": 1456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5513729873821684e-05,
+      "loss": 0.5798,
+      "step": 1457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5507182462596242e-05,
+      "loss": 0.6135,
+      "step": 1458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.550063111854138e-05,
+      "loss": 0.5234,
+      "step": 1459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5494075844109263e-05,
+      "loss": 0.6587,
+      "step": 1460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5487516641753523e-05,
+      "loss": 0.5728,
+      "step": 1461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.548095351392926e-05,
+      "loss": 0.5,
+      "step": 1462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5474386463093054e-05,
+      "loss": 0.6294,
+      "step": 1463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5467815491702936e-05,
+      "loss": 0.6348,
+      "step": 1464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5461240602218422e-05,
+      "loss": 0.7004,
+      "step": 1465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.545466179710049e-05,
+      "loss": 0.6282,
+      "step": 1466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5448079078811578e-05,
+      "loss": 0.5269,
+      "step": 1467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5441492449815588e-05,
+      "loss": 0.6941,
+      "step": 1468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.54349019125779e-05,
+      "loss": 0.5266,
+      "step": 1469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5428307469565333e-05,
+      "loss": 0.5739,
+      "step": 1470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5421709123246196e-05,
+      "loss": 0.5005,
+      "step": 1471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.541510687609023e-05,
+      "loss": 0.4716,
+      "step": 1472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5408500730568657e-05,
+      "loss": 0.4578,
+      "step": 1473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5401890689154152e-05,
+      "loss": 0.6951,
+      "step": 1474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.539527675432085e-05,
+      "loss": 0.6057,
+      "step": 1475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.538865892854434e-05,
+      "loss": 0.6082,
+      "step": 1476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5382037214301663e-05,
+      "loss": 0.4553,
+      "step": 1477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.537541161407133e-05,
+      "loss": 0.4719,
+      "step": 1478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5368782130333286e-05,
+      "loss": 0.5652,
+      "step": 1479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5362148765568954e-05,
+      "loss": 0.6182,
+      "step": 1480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5355511522261188e-05,
+      "loss": 0.6414,
+      "step": 1481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5348870402894306e-05,
+      "loss": 0.6001,
+      "step": 1482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.534222540995407e-05,
+      "loss": 0.6067,
+      "step": 1483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5335576545927703e-05,
+      "loss": 0.5159,
+      "step": 1484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.532892381330386e-05,
+      "loss": 0.541,
+      "step": 1485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5322267214572652e-05,
+      "loss": 0.5928,
+      "step": 1486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5315606752225652e-05,
+      "loss": 0.4984,
+      "step": 1487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5308942428755853e-05,
+      "loss": 0.5916,
+      "step": 1488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5302274246657707e-05,
+      "loss": 0.5249,
+      "step": 1489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5295602208427115e-05,
+      "loss": 0.4734,
+      "step": 1490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5288926316561405e-05,
+      "loss": 0.5021,
+      "step": 1491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.528224657355937e-05,
+      "loss": 0.5774,
+      "step": 1492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5275562981921225e-05,
+      "loss": 0.6313,
+      "step": 1493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5268875544148632e-05,
+      "loss": 0.6145,
+      "step": 1494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.526218426274469e-05,
+      "loss": 0.5386,
+      "step": 1495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.525548914021395e-05,
+      "loss": 0.4189,
+      "step": 1496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.524879017906238e-05,
+      "loss": 0.5796,
+      "step": 1497
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.524208738179741e-05,
+      "loss": 0.5608,
+      "step": 1498
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5235380750927875e-05,
+      "loss": 0.5874,
+      "step": 1499
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.522867028896407e-05,
+      "loss": 0.5312,
+      "step": 1500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5221955998417715e-05,
+      "loss": 0.6064,
+      "step": 1501
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.521523788180196e-05,
+      "loss": 0.483,
+      "step": 1502
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.520851594163139e-05,
+      "loss": 0.572,
+      "step": 1503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5201790180422025e-05,
+      "loss": 0.6248,
+      "step": 1504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.519506060069131e-05,
+      "loss": 0.5054,
+      "step": 1505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.518832720495812e-05,
+      "loss": 0.4353,
+      "step": 1506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5181589995742764e-05,
+      "loss": 0.6196,
+      "step": 1507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5174848975566968e-05,
+      "loss": 0.5781,
+      "step": 1508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.516810414695389e-05,
+      "loss": 0.5251,
+      "step": 1509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5161355512428116e-05,
+      "loss": 0.4744,
+      "step": 1510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5154603074515653e-05,
+      "loss": 0.6179,
+      "step": 1511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.514784683574394e-05,
+      "loss": 0.5205,
+      "step": 1512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5141086798641817e-05,
+      "loss": 0.467,
+      "step": 1513
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.513432296573957e-05,
+      "loss": 0.5728,
+      "step": 1514
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5127555339568895e-05,
+      "loss": 0.5439,
+      "step": 1515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.512078392266291e-05,
+      "loss": 0.6682,
+      "step": 1516
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5114008717556147e-05,
+      "loss": 0.5823,
+      "step": 1517
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5107229726784554e-05,
+      "loss": 0.7307,
+      "step": 1518
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5100446952885517e-05,
+      "loss": 0.5381,
+      "step": 1519
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5093660398397807e-05,
+      "loss": 0.5366,
+      "step": 1520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5086870065861636e-05,
+      "loss": 0.5813,
+      "step": 1521
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5080075957818615e-05,
+      "loss": 0.4558,
+      "step": 1522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5073278076811776e-05,
+      "loss": 0.4985,
+      "step": 1523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5066476425385552e-05,
+      "loss": 0.5342,
+      "step": 1524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5059671006085804e-05,
+      "loss": 0.5183,
+      "step": 1525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.505286182145979e-05,
+      "loss": 0.5623,
+      "step": 1526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.504604887405619e-05,
+      "loss": 0.6611,
+      "step": 1527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.503923216642507e-05,
+      "loss": 0.5623,
+      "step": 1528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.503241170111793e-05,
+      "loss": 0.5427,
+      "step": 1529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5025587480687658e-05,
+      "loss": 0.5918,
+      "step": 1530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5018759507688556e-05,
+      "loss": 0.5613,
+      "step": 1531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.501192778467633e-05,
+      "loss": 0.6855,
+      "step": 1532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5005092314208085e-05,
+      "loss": 0.5015,
+      "step": 1533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4998253098842337e-05,
+      "loss": 0.615,
+      "step": 1534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.499141014113899e-05,
+      "loss": 0.5662,
+      "step": 1535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4984563443659364e-05,
+      "loss": 0.5171,
+      "step": 1536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.497771300896617e-05,
+      "loss": 0.6104,
+      "step": 1537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.497085883962352e-05,
+      "loss": 0.5505,
+      "step": 1538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.496400093819692e-05,
+      "loss": 0.5474,
+      "step": 1539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.495713930725329e-05,
+      "loss": 0.5774,
+      "step": 1540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4950273949360914e-05,
+      "loss": 0.4988,
+      "step": 1541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4943404867089505e-05,
+      "loss": 0.4985,
+      "step": 1542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4936532063010142e-05,
+      "loss": 0.5771,
+      "step": 1543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.492965553969532e-05,
+      "loss": 0.4585,
+      "step": 1544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4922775299718908e-05,
+      "loss": 0.5366,
+      "step": 1545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.491589134565618e-05,
+      "loss": 0.6008,
+      "step": 1546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4909003680083793e-05,
+      "loss": 0.6926,
+      "step": 1547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4902112305579793e-05,
+      "loss": 0.6187,
+      "step": 1548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4895217224723613e-05,
+      "loss": 0.6011,
+      "step": 1549
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4888318440096076e-05,
+      "loss": 0.5083,
+      "step": 1550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4881415954279397e-05,
+      "loss": 0.5654,
+      "step": 1551
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4874509769857165e-05,
+      "loss": 0.5854,
+      "step": 1552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.486759988941437e-05,
+      "loss": 0.5957,
+      "step": 1553
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4860686315537356e-05,
+      "loss": 0.5015,
+      "step": 1554
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.485376905081388e-05,
+      "loss": 0.5708,
+      "step": 1555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4846848097833063e-05,
+      "loss": 0.5564,
+      "step": 1556
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.483992345918542e-05,
+      "loss": 0.4905,
+      "step": 1557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4832995137462823e-05,
+      "loss": 0.5264,
+      "step": 1558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.482606313525856e-05,
+      "loss": 0.457,
+      "step": 1559
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4819127455167247e-05,
+      "loss": 0.6062,
+      "step": 1560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4812188099784916e-05,
+      "loss": 0.6082,
+      "step": 1561
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4805245071708966e-05,
+      "loss": 0.51,
+      "step": 1562
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4798298373538156e-05,
+      "loss": 0.4541,
+      "step": 1563
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.479134800787264e-05,
+      "loss": 0.606,
+      "step": 1564
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4784393977313926e-05,
+      "loss": 0.4768,
+      "step": 1565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4777436284464905e-05,
+      "loss": 0.6372,
+      "step": 1566
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4770474931929835e-05,
+      "loss": 0.5173,
+      "step": 1567
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4763509922314344e-05,
+      "loss": 0.6511,
+      "step": 1568
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4756541258225432e-05,
+      "loss": 0.5481,
+      "step": 1569
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4749568942271455e-05,
+      "loss": 0.4883,
+      "step": 1570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.474259297706215e-05,
+      "loss": 0.5618,
+      "step": 1571
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4735613365208626e-05,
+      "loss": 0.5798,
+      "step": 1572
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4728630109323328e-05,
+      "loss": 0.563,
+      "step": 1573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4721643212020094e-05,
+      "loss": 0.6182,
+      "step": 1574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.471465267591411e-05,
+      "loss": 0.4929,
+      "step": 1575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.470765850362192e-05,
+      "loss": 0.5151,
+      "step": 1576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4700660697761455e-05,
+      "loss": 0.6055,
+      "step": 1577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4693659260951974e-05,
+      "loss": 0.4822,
+      "step": 1578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4686654195814115e-05,
+      "loss": 0.5242,
+      "step": 1579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4679645504969855e-05,
+      "loss": 0.6422,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.467263319104256e-05,
+      "loss": 0.5654,
+      "step": 1581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4665617256656922e-05,
+      "loss": 0.465,
+      "step": 1582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4658597704439e-05,
+      "loss": 0.4915,
+      "step": 1583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4651574537016214e-05,
+      "loss": 0.551,
+      "step": 1584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4644547757017315e-05,
+      "loss": 0.5271,
+      "step": 1585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4637517367072433e-05,
+      "loss": 0.4851,
+      "step": 1586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.463048336981303e-05,
+      "loss": 0.4971,
+      "step": 1587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.462344576787193e-05,
+      "loss": 0.6541,
+      "step": 1588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4616404563883302e-05,
+      "loss": 0.6165,
+      "step": 1589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.460935976048266e-05,
+      "loss": 0.6538,
+      "step": 1590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4602311360306867e-05,
+      "loss": 0.5544,
+      "step": 1591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.459525936599413e-05,
+      "loss": 0.5991,
+      "step": 1592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4588203780184017e-05,
+      "loss": 0.6587,
+      "step": 1593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4581144605517414e-05,
+      "loss": 0.5967,
+      "step": 1594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4574081844636573e-05,
+      "loss": 0.5277,
+      "step": 1595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4567015500185076e-05,
+      "loss": 0.4983,
+      "step": 1596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4559945574807845e-05,
+      "loss": 0.5457,
+      "step": 1597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4552872071151154e-05,
+      "loss": 0.6294,
+      "step": 1598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4545794991862606e-05,
+      "loss": 0.5071,
+      "step": 1599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.453871433959115e-05,
+      "loss": 0.5212,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4531630116987057e-05,
+      "loss": 0.7915,
+      "step": 1601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4524542326701963e-05,
+      "loss": 0.5525,
+      "step": 1602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4517450971388812e-05,
+      "loss": 0.7568,
+      "step": 1603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4510356053701883e-05,
+      "loss": 0.553,
+      "step": 1604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4503257576296815e-05,
+      "loss": 0.5317,
+      "step": 1605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.449615554183055e-05,
+      "loss": 0.5336,
+      "step": 1606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4489049952961383e-05,
+      "loss": 0.551,
+      "step": 1607
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4481940812348916e-05,
+      "loss": 0.4928,
+      "step": 1608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4474828122654106e-05,
+      "loss": 0.5261,
+      "step": 1609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4467711886539228e-05,
+      "loss": 0.7485,
+      "step": 1610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4460592106667874e-05,
+      "loss": 0.5732,
+      "step": 1611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4453468785704978e-05,
+      "loss": 0.6072,
+      "step": 1612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4446341926316785e-05,
+      "loss": 0.5461,
+      "step": 1613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4439211531170887e-05,
+      "loss": 0.4934,
+      "step": 1614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.443207760293617e-05,
+      "loss": 0.6389,
+      "step": 1615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.442494014428287e-05,
+      "loss": 0.4749,
+      "step": 1616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.441779915788252e-05,
+      "loss": 0.53,
+      "step": 1617
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4410654646407987e-05,
+      "loss": 0.5952,
+      "step": 1618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4403506612533462e-05,
+      "loss": 0.4639,
+      "step": 1619
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4396355058934444e-05,
+      "loss": 0.5762,
+      "step": 1620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4389199988287757e-05,
+      "loss": 0.5828,
+      "step": 1621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4382041403271533e-05,
+      "loss": 0.5962,
+      "step": 1622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4374879306565224e-05,
+      "loss": 0.6594,
+      "step": 1623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.43677137008496e-05,
+      "loss": 0.6165,
+      "step": 1624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.436054458880674e-05,
+      "loss": 0.4312,
+      "step": 1625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4353371973120036e-05,
+      "loss": 0.5959,
+      "step": 1626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4346195856474194e-05,
+      "loss": 0.6255,
+      "step": 1627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4339016241555222e-05,
+      "loss": 0.5732,
+      "step": 1628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4331833131050454e-05,
+      "loss": 0.5613,
+      "step": 1629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4324646527648515e-05,
+      "loss": 0.5742,
+      "step": 1630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4317456434039345e-05,
+      "loss": 0.7346,
+      "step": 1631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4310262852914188e-05,
+      "loss": 0.5525,
+      "step": 1632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4303065786965604e-05,
+      "loss": 0.4647,
+      "step": 1633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.429586523888744e-05,
+      "loss": 0.6655,
+      "step": 1634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4288661211374856e-05,
+      "loss": 0.708,
+      "step": 1635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4281453707124313e-05,
+      "loss": 0.5459,
+      "step": 1636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4274242728833583e-05,
+      "loss": 0.5903,
+      "step": 1637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.426702827920172e-05,
+      "loss": 0.5134,
+      "step": 1638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4259810360929086e-05,
+      "loss": 0.5122,
+      "step": 1639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4252588976717343e-05,
+      "loss": 0.6514,
+      "step": 1640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.424536412926945e-05,
+      "loss": 0.4697,
+      "step": 1641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4238135821289664e-05,
+      "loss": 0.4202,
+      "step": 1642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.423090405548353e-05,
+      "loss": 0.6528,
+      "step": 1643
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4223668834557896e-05,
+      "loss": 0.7744,
+      "step": 1644
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.421643016122089e-05,
+      "loss": 0.438,
+      "step": 1645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.420918803818195e-05,
+      "loss": 0.5129,
+      "step": 1646
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4201942468151794e-05,
+      "loss": 0.6903,
+      "step": 1647
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4194693453842437e-05,
+      "loss": 0.5293,
+      "step": 1648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.418744099796717e-05,
+      "loss": 0.5833,
+      "step": 1649
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4180185103240585e-05,
+      "loss": 0.5215,
+      "step": 1650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.417292577237856e-05,
+      "loss": 0.5564,
+      "step": 1651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4165663008098252e-05,
+      "loss": 0.562,
+      "step": 1652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4158396813118112e-05,
+      "loss": 0.4646,
+      "step": 1653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4151127190157864e-05,
+      "loss": 0.5764,
+      "step": 1654
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.414385414193853e-05,
+      "loss": 0.5481,
+      "step": 1655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.41365776711824e-05,
+      "loss": 0.4524,
+      "step": 1656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4129297780613057e-05,
+      "loss": 0.6492,
+      "step": 1657
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4122014472955346e-05,
+      "loss": 0.5798,
+      "step": 1658
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4114727750935415e-05,
+      "loss": 0.4829,
+      "step": 1659
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.410743761728067e-05,
+      "loss": 0.5908,
+      "step": 1660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4100144074719806e-05,
+      "loss": 0.665,
+      "step": 1661
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4092847125982787e-05,
+      "loss": 0.6641,
+      "step": 1662
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4085546773800855e-05,
+      "loss": 0.6326,
+      "step": 1663
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.407824302090653e-05,
+      "loss": 0.5884,
+      "step": 1664
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4070935870033594e-05,
+      "loss": 0.5591,
+      "step": 1665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.406362532391711e-05,
+      "loss": 0.6194,
+      "step": 1666
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4056311385293413e-05,
+      "loss": 0.5918,
+      "step": 1667
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.40489940569001e-05,
+      "loss": 0.6194,
+      "step": 1668
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.404167334147604e-05,
+      "loss": 0.5757,
+      "step": 1669
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.403434924176137e-05,
+      "loss": 0.5061,
+      "step": 1670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4027021760497496e-05,
+      "loss": 0.5344,
+      "step": 1671
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.401969090042709e-05,
+      "loss": 0.4905,
+      "step": 1672
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.401235666429409e-05,
+      "loss": 0.6035,
+      "step": 1673
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4005019054843686e-05,
+      "loss": 0.5293,
+      "step": 1674
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.399767807482234e-05,
+      "loss": 0.5994,
+      "step": 1675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3990333726977777e-05,
+      "loss": 0.6697,
+      "step": 1676
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3982986014058977e-05,
+      "loss": 0.6262,
+      "step": 1677
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3975634938816185e-05,
+      "loss": 0.5188,
+      "step": 1678
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3968280504000905e-05,
+      "loss": 0.5491,
+      "step": 1679
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3960922712365893e-05,
+      "loss": 0.5256,
+      "step": 1680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.395356156666517e-05,
+      "loss": 0.5913,
+      "step": 1681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3946197069653996e-05,
+      "loss": 0.5439,
+      "step": 1682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3938829224088896e-05,
+      "loss": 0.5186,
+      "step": 1683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3931458032727664e-05,
+      "loss": 0.5957,
+      "step": 1684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3924083498329307e-05,
+      "loss": 0.5166,
+      "step": 1685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.391670562365413e-05,
+      "loss": 0.5337,
+      "step": 1686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3909324411463653e-05,
+      "loss": 0.542,
+      "step": 1687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3901939864520656e-05,
+      "loss": 0.5183,
+      "step": 1688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.389455198558917e-05,
+      "loss": 0.4734,
+      "step": 1689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.388716077743448e-05,
+      "loss": 0.5493,
+      "step": 1690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.38797662428231e-05,
+      "loss": 0.6519,
+      "step": 1691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.38723683845228e-05,
+      "loss": 0.7236,
+      "step": 1692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3864967205302595e-05,
+      "loss": 0.5854,
+      "step": 1693
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3857562707932735e-05,
+      "loss": 0.5017,
+      "step": 1694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3850154895184722e-05,
+      "loss": 0.6179,
+      "step": 1695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3842743769831287e-05,
+      "loss": 0.6448,
+      "step": 1696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.383532933464642e-05,
+      "loss": 0.5881,
+      "step": 1697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3827911592405322e-05,
+      "loss": 0.5413,
+      "step": 1698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.382049054588446e-05,
+      "loss": 0.6509,
+      "step": 1699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.381306619786152e-05,
+      "loss": 0.6084,
+      "step": 1700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3805638551115433e-05,
+      "loss": 0.576,
+      "step": 1701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3798207608426362e-05,
+      "loss": 0.5469,
+      "step": 1702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.379077337257569e-05,
+      "loss": 0.5562,
+      "step": 1703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.378333584634606e-05,
+      "loss": 0.5977,
+      "step": 1704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3775895032521325e-05,
+      "loss": 0.543,
+      "step": 1705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3768450933886572e-05,
+      "loss": 0.5513,
+      "step": 1706
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3761003553228128e-05,
+      "loss": 0.5254,
+      "step": 1707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3753552893333534e-05,
+      "loss": 0.614,
+      "step": 1708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3746098956991574e-05,
+      "loss": 0.4832,
+      "step": 1709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3738641746992246e-05,
+      "loss": 0.6487,
+      "step": 1710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3731181266126773e-05,
+      "loss": 0.6099,
+      "step": 1711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3723717517187612e-05,
+      "loss": 0.6321,
+      "step": 1712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3716250502968434e-05,
+      "loss": 0.5532,
+      "step": 1713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.370878022626414e-05,
+      "loss": 0.5237,
+      "step": 1714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.370130668987085e-05,
+      "loss": 0.5205,
+      "step": 1715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3693829896585895e-05,
+      "loss": 0.6914,
+      "step": 1716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3686349849207834e-05,
+      "loss": 0.6995,
+      "step": 1717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3678866550536452e-05,
+      "loss": 0.5312,
+      "step": 1718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3671380003372726e-05,
+      "loss": 0.5103,
+      "step": 1719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3663890210518877e-05,
+      "loss": 0.6572,
+      "step": 1720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3656397174778318e-05,
+      "loss": 0.5322,
+      "step": 1721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3648900898955697e-05,
+      "loss": 0.4951,
+      "step": 1722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3641401385856853e-05,
+      "loss": 0.4443,
+      "step": 1723
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.363389863828886e-05,
+      "loss": 0.5626,
+      "step": 1724
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.362639265905997e-05,
+      "loss": 0.6917,
+      "step": 1725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3618883450979683e-05,
+      "loss": 0.5527,
+      "step": 1726
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3611371016858683e-05,
+      "loss": 0.5359,
+      "step": 1727
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3603855359508867e-05,
+      "loss": 0.5923,
+      "step": 1728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.359633648174334e-05,
+      "loss": 0.572,
+      "step": 1729
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.358881438637641e-05,
+      "loss": 0.6921,
+      "step": 1730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.358128907622359e-05,
+      "loss": 0.5344,
+      "step": 1731
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.35737605541016e-05,
+      "loss": 0.4792,
+      "step": 1732
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3566228822828362e-05,
+      "loss": 0.5073,
+      "step": 1733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.355869388522299e-05,
+      "loss": 0.5593,
+      "step": 1734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3551155744105806e-05,
+      "loss": 0.446,
+      "step": 1735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3543614402298337e-05,
+      "loss": 0.7231,
+      "step": 1736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.353606986262329e-05,
+      "loss": 0.5317,
+      "step": 1737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3528522127904588e-05,
+      "loss": 0.5422,
+      "step": 1738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3520971200967337e-05,
+      "loss": 0.5315,
+      "step": 1739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3513417084637847e-05,
+      "loss": 0.5955,
+      "step": 1740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.350585978174361e-05,
+      "loss": 0.5508,
+      "step": 1741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.349829929511333e-05,
+      "loss": 0.5679,
+      "step": 1742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.349073562757688e-05,
+      "loss": 0.5442,
+      "step": 1743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3483168781965328e-05,
+      "loss": 0.5952,
+      "step": 1744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3475598761110962e-05,
+      "loss": 0.5425,
+      "step": 1745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3468025567847213e-05,
+      "loss": 0.5762,
+      "step": 1746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.346044920500873e-05,
+      "loss": 0.426,
+      "step": 1747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3452869675431332e-05,
+      "loss": 0.6707,
+      "step": 1748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.344528698195204e-05,
+      "loss": 0.4536,
+      "step": 1749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.343770112740904e-05,
+      "loss": 0.6326,
+      "step": 1750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3430112114641716e-05,
+      "loss": 0.6257,
+      "step": 1751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3422519946490634e-05,
+      "loss": 0.6089,
+      "step": 1752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3414924625797527e-05,
+      "loss": 0.5957,
+      "step": 1753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.340732615540532e-05,
+      "loss": 0.5913,
+      "step": 1754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3399724538158115e-05,
+      "loss": 0.6379,
+      "step": 1755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3392119776901192e-05,
+      "loss": 0.4967,
+      "step": 1756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3384511874481005e-05,
+      "loss": 0.5242,
+      "step": 1757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3376900833745185e-05,
+      "loss": 0.5115,
+      "step": 1758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3369286657542538e-05,
+      "loss": 0.5286,
+      "step": 1759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3361669348723046e-05,
+      "loss": 0.5378,
+      "step": 1760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.335404891013786e-05,
+      "loss": 0.5698,
+      "step": 1761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3346425344639304e-05,
+      "loss": 0.5352,
+      "step": 1762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3338798655080867e-05,
+      "loss": 0.4685,
+      "step": 1763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3331168844317218e-05,
+      "loss": 0.6099,
+      "step": 1764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3323535915204192e-05,
+      "loss": 0.5282,
+      "step": 1765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3315899870598775e-05,
+      "loss": 0.6318,
+      "step": 1766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3308260713359146e-05,
+      "loss": 0.5146,
+      "step": 1767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3300618446344622e-05,
+      "loss": 0.6104,
+      "step": 1768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.329297307241571e-05,
+      "loss": 0.689,
+      "step": 1769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3285324594434066e-05,
+      "loss": 0.5596,
+      "step": 1770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3277673015262495e-05,
+      "loss": 0.6792,
+      "step": 1771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3270018337764988e-05,
+      "loss": 0.4749,
+      "step": 1772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3262360564806682e-05,
+      "loss": 0.5986,
+      "step": 1773
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3254699699253874e-05,
+      "loss": 0.5225,
+      "step": 1774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.324703574397402e-05,
+      "loss": 0.501,
+      "step": 1775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.323936870183573e-05,
+      "loss": 0.4915,
+      "step": 1776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3231698575708776e-05,
+      "loss": 0.6438,
+      "step": 1777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3224025368464077e-05,
+      "loss": 0.6208,
+      "step": 1778
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.321634908297371e-05,
+      "loss": 0.4756,
+      "step": 1779
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.320866972211089e-05,
+      "loss": 0.4846,
+      "step": 1780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3200987288750016e-05,
+      "loss": 0.625,
+      "step": 1781
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.31933017857666e-05,
+      "loss": 0.4758,
+      "step": 1782
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3185613216037328e-05,
+      "loss": 0.5798,
+      "step": 1783
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3177921582440015e-05,
+      "loss": 0.647,
+      "step": 1784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3170226887853645e-05,
+      "loss": 0.5547,
+      "step": 1785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3162529135158327e-05,
+      "loss": 0.6423,
+      "step": 1786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3154828327235327e-05,
+      "loss": 0.6509,
+      "step": 1787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.314712446696705e-05,
+      "loss": 0.4697,
+      "step": 1788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3139417557237044e-05,
+      "loss": 0.6196,
+      "step": 1789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.313170760093e-05,
+      "loss": 0.5609,
+      "step": 1790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.312399460093175e-05,
+      "loss": 0.6584,
+      "step": 1791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3116278560129255e-05,
+      "loss": 0.5493,
+      "step": 1792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3108559481410632e-05,
+      "loss": 0.4915,
+      "step": 1793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3100837367665122e-05,
+      "loss": 0.5054,
+      "step": 1794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.309311222178311e-05,
+      "loss": 0.5811,
+      "step": 1795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3085384046656096e-05,
+      "loss": 0.6262,
+      "step": 1796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3077652845176745e-05,
+      "loss": 0.5056,
+      "step": 1797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3069918620238836e-05,
+      "loss": 0.6318,
+      "step": 1798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3062181374737278e-05,
+      "loss": 0.4749,
+      "step": 1799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.305444111156812e-05,
+      "loss": 0.5518,
+      "step": 1800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3046697833628528e-05,
+      "loss": 0.53,
+      "step": 1801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3038951543816816e-05,
+      "loss": 0.3949,
+      "step": 1802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.303120224503241e-05,
+      "loss": 0.5459,
+      "step": 1803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.302344994017586e-05,
+      "loss": 0.666,
+      "step": 1804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3015694632148855e-05,
+      "loss": 0.4849,
+      "step": 1805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3007936323854196e-05,
+      "loss": 0.5201,
+      "step": 1806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3000175018195808e-05,
+      "loss": 0.4993,
+      "step": 1807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2992410718078748e-05,
+      "loss": 0.4696,
+      "step": 1808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.298464342640918e-05,
+      "loss": 0.5664,
+      "step": 1809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2976873146094403e-05,
+      "loss": 0.6506,
+      "step": 1810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2969099880042818e-05,
+      "loss": 0.5569,
+      "step": 1811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2961323631163957e-05,
+      "loss": 0.5708,
+      "step": 1812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2953544402368463e-05,
+      "loss": 0.5342,
+      "step": 1813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2945762196568086e-05,
+      "loss": 0.5859,
+      "step": 1814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.293797701667571e-05,
+      "loss": 0.6091,
+      "step": 1815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.293018886560532e-05,
+      "loss": 0.5569,
+      "step": 1816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2922397746272013e-05,
+      "loss": 0.4929,
+      "step": 1817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.291460366159199e-05,
+      "loss": 0.5374,
+      "step": 1818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.290680661448258e-05,
+      "loss": 0.5093,
+      "step": 1819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2899006607862216e-05,
+      "loss": 0.5088,
+      "step": 1820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.289120364465042e-05,
+      "loss": 0.5544,
+      "step": 1821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.288339772776784e-05,
+      "loss": 0.609,
+      "step": 1822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2875588860136224e-05,
+      "loss": 0.6023,
+      "step": 1823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.286777704467843e-05,
+      "loss": 0.5327,
+      "step": 1824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2859962284318404e-05,
+      "loss": 0.6465,
+      "step": 1825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2852144581981208e-05,
+      "loss": 0.5144,
+      "step": 1826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2844323940593005e-05,
+      "loss": 0.5891,
+      "step": 1827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2836500363081053e-05,
+      "loss": 0.5842,
+      "step": 1828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2828673852373703e-05,
+      "loss": 0.5366,
+      "step": 1829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2820844411400427e-05,
+      "loss": 0.6057,
+      "step": 1830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2813012043091764e-05,
+      "loss": 0.5764,
+      "step": 1831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2805176750379367e-05,
+      "loss": 0.627,
+      "step": 1832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2797338536195976e-05,
+      "loss": 0.5269,
+      "step": 1833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2789497403475438e-05,
+      "loss": 0.5249,
+      "step": 1834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2781653355152675e-05,
+      "loss": 0.5127,
+      "step": 1835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2773806394163707e-05,
+      "loss": 0.5813,
+      "step": 1836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2765956523445645e-05,
+      "loss": 0.5027,
+      "step": 1837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.275810374593669e-05,
+      "loss": 0.5393,
+      "step": 1838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2750248064576134e-05,
+      "loss": 0.5005,
+      "step": 1839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2742389482304346e-05,
+      "loss": 0.7002,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2734528002062783e-05,
+      "loss": 0.5225,
+      "step": 1841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2726663626794002e-05,
+      "loss": 0.5168,
+      "step": 1842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2718796359441625e-05,
+      "loss": 0.4622,
+      "step": 1843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2710926202950366e-05,
+      "loss": 0.5725,
+      "step": 1844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2703053160266007e-05,
+      "loss": 0.6846,
+      "step": 1845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.269517723433544e-05,
+      "loss": 0.5229,
+      "step": 1846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2687298428106603e-05,
+      "loss": 0.4613,
+      "step": 1847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.267941674452853e-05,
+      "loss": 0.5614,
+      "step": 1848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2671532186551335e-05,
+      "loss": 0.5171,
+      "step": 1849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.266364475712619e-05,
+      "loss": 0.4421,
+      "step": 1850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2655754459205365e-05,
+      "loss": 0.5537,
+      "step": 1851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2647861295742177e-05,
+      "loss": 0.5662,
+      "step": 1852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2639965269691046e-05,
+      "loss": 0.5957,
+      "step": 1853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.263206638400744e-05,
+      "loss": 0.4683,
+      "step": 1854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2624164641647902e-05,
+      "loss": 0.5071,
+      "step": 1855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.261626004557005e-05,
+      "loss": 0.5215,
+      "step": 1856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2608352598732578e-05,
+      "loss": 0.5896,
+      "step": 1857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2600442304095222e-05,
+      "loss": 0.4821,
+      "step": 1858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2592529164618804e-05,
+      "loss": 0.615,
+      "step": 1859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2584613183265203e-05,
+      "loss": 0.4722,
+      "step": 1860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.257669436299737e-05,
+      "loss": 0.6226,
+      "step": 1861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2568772706779307e-05,
+      "loss": 0.4832,
+      "step": 1862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2560848217576083e-05,
+      "loss": 0.573,
+      "step": 1863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2552920898353833e-05,
+      "loss": 0.5806,
+      "step": 1864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.254499075207975e-05,
+      "loss": 0.6233,
+      "step": 1865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2537057781722062e-05,
+      "loss": 0.519,
+      "step": 1866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2529121990250093e-05,
+      "loss": 0.53,
+      "step": 1867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2521183380634184e-05,
+      "loss": 0.6472,
+      "step": 1868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.251324195584577e-05,
+      "loss": 0.5647,
+      "step": 1869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2505297718857307e-05,
+      "loss": 0.4628,
+      "step": 1870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2497350672642317e-05,
+      "loss": 0.4663,
+      "step": 1871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2489400820175374e-05,
+      "loss": 0.5052,
+      "step": 1872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2481448164432106e-05,
+      "loss": 0.5212,
+      "step": 1873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2473492708389177e-05,
+      "loss": 0.501,
+      "step": 1874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2465534455024314e-05,
+      "loss": 0.5156,
+      "step": 1875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2457573407316284e-05,
+      "loss": 0.4878,
+      "step": 1876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2449609568244894e-05,
+      "loss": 0.4563,
+      "step": 1877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2441642940791008e-05,
+      "loss": 0.4099,
+      "step": 1878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.243367352793652e-05,
+      "loss": 0.5408,
+      "step": 1879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2425701332664388e-05,
+      "loss": 0.5374,
+      "step": 1880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2417726357958592e-05,
+      "loss": 0.515,
+      "step": 1881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.240974860680415e-05,
+      "loss": 0.4719,
+      "step": 1882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2401768082187138e-05,
+      "loss": 0.5098,
+      "step": 1883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.239378478709465e-05,
+      "loss": 0.4731,
+      "step": 1884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.238579872451484e-05,
+      "loss": 0.6233,
+      "step": 1885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.237780989743687e-05,
+      "loss": 0.5366,
+      "step": 1886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2369818308850964e-05,
+      "loss": 0.5312,
+      "step": 1887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2361823961748354e-05,
+      "loss": 0.4202,
+      "step": 1888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2353826859121324e-05,
+      "loss": 0.4724,
+      "step": 1889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2345827003963185e-05,
+      "loss": 0.5264,
+      "step": 1890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2337824399268276e-05,
+      "loss": 0.686,
+      "step": 1891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2329819048031958e-05,
+      "loss": 0.5793,
+      "step": 1892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2321810953250633e-05,
+      "loss": 0.6681,
+      "step": 1893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2313800117921724e-05,
+      "loss": 0.6169,
+      "step": 1894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.230578654504368e-05,
+      "loss": 0.54,
+      "step": 1895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2297770237615967e-05,
+      "loss": 0.5762,
+      "step": 1896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.228975119863909e-05,
+      "loss": 0.4783,
+      "step": 1897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2281729431114572e-05,
+      "loss": 0.541,
+      "step": 1898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2273704938044944e-05,
+      "loss": 0.5706,
+      "step": 1899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2265677722433767e-05,
+      "loss": 0.4912,
+      "step": 1900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2257647787285632e-05,
+      "loss": 0.5674,
+      "step": 1901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2249615135606128e-05,
+      "loss": 0.4836,
+      "step": 1902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.224157977040187e-05,
+      "loss": 0.5088,
+      "step": 1903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2233541694680484e-05,
+      "loss": 0.6079,
+      "step": 1904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.222550091145063e-05,
+      "loss": 0.5945,
+      "step": 1905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.221745742372195e-05,
+      "loss": 0.6421,
+      "step": 1906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.220941123450513e-05,
+      "loss": 0.5837,
+      "step": 1907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2201362346811836e-05,
+      "loss": 0.6033,
+      "step": 1908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.219331076365477e-05,
+      "loss": 0.4753,
+      "step": 1909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2185256488047633e-05,
+      "loss": 0.5579,
+      "step": 1910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2177199523005124e-05,
+      "loss": 0.5283,
+      "step": 1911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.216913987154297e-05,
+      "loss": 0.5785,
+      "step": 1912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2161077536677884e-05,
+      "loss": 0.5293,
+      "step": 1913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2153012521427593e-05,
+      "loss": 0.5327,
+      "step": 1914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.214494482881082e-05,
+      "loss": 0.5334,
+      "step": 1915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.213687446184731e-05,
+      "loss": 0.6067,
+      "step": 1916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.212880142355778e-05,
+      "loss": 0.644,
+      "step": 1917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2120725716963972e-05,
+      "loss": 0.5889,
+      "step": 1918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2112647345088603e-05,
+      "loss": 0.5347,
+      "step": 1919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2104566310955413e-05,
+      "loss": 0.5352,
+      "step": 1920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2096482617589125e-05,
+      "loss": 0.4493,
+      "step": 1921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2088396268015447e-05,
+      "loss": 0.4834,
+      "step": 1922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.208030726526111e-05,
+      "loss": 0.543,
+      "step": 1923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2072215612353802e-05,
+      "loss": 0.5515,
+      "step": 1924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.206412131232224e-05,
+      "loss": 0.6362,
+      "step": 1925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2056024368196097e-05,
+      "loss": 0.5986,
+      "step": 1926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2047924783006067e-05,
+      "loss": 0.4677,
+      "step": 1927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2039822559783813e-05,
+      "loss": 0.5813,
+      "step": 1928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2031717701561982e-05,
+      "loss": 0.5598,
+      "step": 1929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2023610211374224e-05,
+      "loss": 0.4706,
+      "step": 1930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2015500092255166e-05,
+      "loss": 0.5278,
+      "step": 1931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2007387347240415e-05,
+      "loss": 0.5112,
+      "step": 1932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1999271979366563e-05,
+      "loss": 0.5287,
+      "step": 1933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.199115399167119e-05,
+      "loss": 0.5483,
+      "step": 1934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1983033387192855e-05,
+      "loss": 0.5485,
+      "step": 1935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1974910168971078e-05,
+      "loss": 0.4756,
+      "step": 1936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1966784340046392e-05,
+      "loss": 0.551,
+      "step": 1937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.195865590346027e-05,
+      "loss": 0.5139,
+      "step": 1938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1950524862255195e-05,
+      "loss": 0.5366,
+      "step": 1939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.19423912194746e-05,
+      "loss": 0.6694,
+      "step": 1940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.19342549781629e-05,
+      "loss": 0.5303,
+      "step": 1941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1926116141365486e-05,
+      "loss": 0.5361,
+      "step": 1942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.191797471212872e-05,
+      "loss": 0.4482,
+      "step": 1943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.190983069349992e-05,
+      "loss": 0.5469,
+      "step": 1944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1901684088527398e-05,
+      "loss": 0.421,
+      "step": 1945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1893534900260422e-05,
+      "loss": 0.488,
+      "step": 1946
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1885383131749218e-05,
+      "loss": 0.5916,
+      "step": 1947
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1877228786044987e-05,
+      "loss": 0.6162,
+      "step": 1948
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1869071866199898e-05,
+      "loss": 0.4204,
+      "step": 1949
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.186091237526708e-05,
+      "loss": 0.4333,
+      "step": 1950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1852750316300617e-05,
+      "loss": 0.4976,
+      "step": 1951
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1844585692355563e-05,
+      "loss": 0.5349,
+      "step": 1952
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1836418506487937e-05,
+      "loss": 0.4636,
+      "step": 1953
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.18282487617547e-05,
+      "loss": 0.623,
+      "step": 1954
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1820076461213782e-05,
+      "loss": 0.5457,
+      "step": 1955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1811901607924074e-05,
+      "loss": 0.604,
+      "step": 1956
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1803724204945414e-05,
+      "loss": 0.5708,
+      "step": 1957
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1795544255338596e-05,
+      "loss": 0.5845,
+      "step": 1958
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1787361762165363e-05,
+      "loss": 0.5332,
+      "step": 1959
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1779176728488424e-05,
+      "loss": 0.6743,
+      "step": 1960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1770989157371426e-05,
+      "loss": 0.5618,
+      "step": 1961
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1762799051878967e-05,
+      "loss": 0.5693,
+      "step": 1962
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1754606415076603e-05,
+      "loss": 0.4639,
+      "step": 1963
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1746411250030833e-05,
+      "loss": 0.6323,
+      "step": 1964
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.173821355980909e-05,
+      "loss": 0.6177,
+      "step": 1965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1730013347479765e-05,
+      "loss": 0.5082,
+      "step": 1966
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1721810616112196e-05,
+      "loss": 0.6138,
+      "step": 1967
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1713605368776653e-05,
+      "loss": 0.5479,
+      "step": 1968
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.170539760854436e-05,
+      "loss": 0.5139,
+      "step": 1969
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1697187338487465e-05,
+      "loss": 0.5442,
+      "step": 1970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1688974561679075e-05,
+      "loss": 0.4792,
+      "step": 1971
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1680759281193216e-05,
+      "loss": 0.3821,
+      "step": 1972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.167254150010487e-05,
+      "loss": 0.6494,
+      "step": 1973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.166432122148993e-05,
+      "loss": 0.5527,
+      "step": 1974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.165609844842526e-05,
+      "loss": 0.5779,
+      "step": 1975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1647873183988624e-05,
+      "loss": 0.4753,
+      "step": 1976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1639645431258733e-05,
+      "loss": 0.4946,
+      "step": 1977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1631415193315225e-05,
+      "loss": 0.5911,
+      "step": 1978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.162318247323868e-05,
+      "loss": 0.5503,
+      "step": 1979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.161494727411059e-05,
+      "loss": 0.543,
+      "step": 1980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1606709599013385e-05,
+      "loss": 0.4785,
+      "step": 1981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1598469451030426e-05,
+      "loss": 0.6025,
+      "step": 1982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1590226833245985e-05,
+      "loss": 0.626,
+      "step": 1983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.158198174874527e-05,
+      "loss": 0.5322,
+      "step": 1984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1573734200614403e-05,
+      "loss": 0.5305,
+      "step": 1985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1565484191940453e-05,
+      "loss": 0.469,
+      "step": 1986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1557231725811368e-05,
+      "loss": 0.6221,
+      "step": 1987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1548976805316054e-05,
+      "loss": 0.5427,
+      "step": 1988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1540719433544314e-05,
+      "loss": 0.4722,
+      "step": 1989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1532459613586878e-05,
+      "loss": 0.4993,
+      "step": 1990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1524197348535387e-05,
+      "loss": 0.5593,
+      "step": 1991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1515932641482404e-05,
+      "loss": 0.5583,
+      "step": 1992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1507665495521398e-05,
+      "loss": 0.5056,
+      "step": 1993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1499395913746756e-05,
+      "loss": 0.4822,
+      "step": 1994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.149112389925378e-05,
+      "loss": 0.5217,
+      "step": 1995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.148284945513866e-05,
+      "loss": 0.5481,
+      "step": 1996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.147457258449854e-05,
+      "loss": 0.5186,
+      "step": 1997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.146629329043143e-05,
+      "loss": 0.48,
+      "step": 1998
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1458011576036266e-05,
+      "loss": 0.5159,
+      "step": 1999
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.144972744441288e-05,
+      "loss": 0.4912,
+      "step": 2000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1441440898662025e-05,
+      "loss": 0.5586,
+      "step": 2001
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1433151941885343e-05,
+      "loss": 0.5735,
+      "step": 2002
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1424860577185383e-05,
+      "loss": 0.5684,
+      "step": 2003
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1416566807665595e-05,
+      "loss": 0.4539,
+      "step": 2004
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.140827063643034e-05,
+      "loss": 0.5388,
+      "step": 2005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1399972066584855e-05,
+      "loss": 0.5491,
+      "step": 2006
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1391671101235292e-05,
+      "loss": 0.5133,
+      "step": 2007
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1383367743488698e-05,
+      "loss": 0.4438,
+      "step": 2008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.137506199645301e-05,
+      "loss": 0.6477,
+      "step": 2009
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1366753863237062e-05,
+      "loss": 0.5243,
+      "step": 2010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1358443346950577e-05,
+      "loss": 0.5762,
+      "step": 2011
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1350130450704184e-05,
+      "loss": 0.5566,
+      "step": 2012
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.134181517760939e-05,
+      "loss": 0.6387,
+      "step": 2013
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1333497530778583e-05,
+      "loss": 0.5046,
+      "step": 2014
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.132517751332507e-05,
+      "loss": 0.5344,
+      "step": 2015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1316855128363014e-05,
+      "loss": 0.5356,
+      "step": 2016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1308530379007477e-05,
+      "loss": 0.5042,
+      "step": 2017
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1300203268374408e-05,
+      "loss": 0.4884,
+      "step": 2018
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1291873799580638e-05,
+      "loss": 0.6086,
+      "step": 2019
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1283541975743885e-05,
+      "loss": 0.4929,
+      "step": 2020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.127520779998273e-05,
+      "loss": 0.4768,
+      "step": 2021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.126687127541666e-05,
+      "loss": 0.5667,
+      "step": 2022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1258532405166024e-05,
+      "loss": 0.6296,
+      "step": 2023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1250191192352055e-05,
+      "loss": 0.5542,
+      "step": 2024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.124184764009686e-05,
+      "loss": 0.5515,
+      "step": 2025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.123350175152343e-05,
+      "loss": 0.5989,
+      "step": 2026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1225153529755618e-05,
+      "loss": 0.6235,
+      "step": 2027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1216802977918154e-05,
+      "loss": 0.4236,
+      "step": 2028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1208450099136647e-05,
+      "loss": 0.6421,
+      "step": 2029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1200094896537572e-05,
+      "loss": 0.5471,
+      "step": 2030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1191737373248277e-05,
+      "loss": 0.6469,
+      "step": 2031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.118337753239697e-05,
+      "loss": 0.5833,
+      "step": 2032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.117501537711273e-05,
+      "loss": 0.6367,
+      "step": 2033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1166650910525512e-05,
+      "loss": 0.5488,
+      "step": 2034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1158284135766127e-05,
+      "loss": 0.4966,
+      "step": 2035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1149915055966252e-05,
+      "loss": 0.6331,
+      "step": 2036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1141543674258418e-05,
+      "loss": 0.5176,
+      "step": 2037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.113316999377604e-05,
+      "loss": 0.502,
+      "step": 2038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1124794017653368e-05,
+      "loss": 0.5854,
+      "step": 2039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.111641574902553e-05,
+      "loss": 0.5952,
+      "step": 2040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1108035191028502e-05,
+      "loss": 0.4893,
+      "step": 2041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1099652346799117e-05,
+      "loss": 0.575,
+      "step": 2042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1091267219475075e-05,
+      "loss": 0.5735,
+      "step": 2043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.108287981219491e-05,
+      "loss": 0.4889,
+      "step": 2044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1074490128098028e-05,
+      "loss": 0.5481,
+      "step": 2045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.106609817032468e-05,
+      "loss": 0.5764,
+      "step": 2046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1057703942015974e-05,
+      "loss": 0.6057,
+      "step": 2047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1049307446313848e-05,
+      "loss": 0.6357,
+      "step": 2048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.104090868636112e-05,
+      "loss": 0.5035,
+      "step": 2049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.103250766530143e-05,
+      "loss": 0.4944,
+      "step": 2050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.102410438627927e-05,
+      "loss": 0.4587,
+      "step": 2051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.101569885243999e-05,
+      "loss": 0.5173,
+      "step": 2052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1007291066929762e-05,
+      "loss": 0.533,
+      "step": 2053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.099888103289562e-05,
+      "loss": 0.542,
+      "step": 2054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.099046875348543e-05,
+      "loss": 0.4146,
+      "step": 2055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.098205423184791e-05,
+      "loss": 0.5886,
+      "step": 2056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0973637471132597e-05,
+      "loss": 0.5891,
+      "step": 2057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0965218474489883e-05,
+      "loss": 0.4578,
+      "step": 2058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0956797245070986e-05,
+      "loss": 0.4475,
+      "step": 2059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.094837378602797e-05,
+      "loss": 0.6267,
+      "step": 2060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.093994810051373e-05,
+      "loss": 0.4924,
+      "step": 2061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.093152019168199e-05,
+      "loss": 0.5808,
+      "step": 2062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0923090062687305e-05,
+      "loss": 0.5601,
+      "step": 2063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.091465771668507e-05,
+      "loss": 0.5183,
+      "step": 2064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.090622315683151e-05,
+      "loss": 0.5481,
+      "step": 2065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.089778638628366e-05,
+      "loss": 0.4744,
+      "step": 2066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0889347408199406e-05,
+      "loss": 0.4551,
+      "step": 2067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.088090622573745e-05,
+      "loss": 0.4792,
+      "step": 2068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0872462842057313e-05,
+      "loss": 0.5437,
+      "step": 2069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0864017260319356e-05,
+      "loss": 0.6294,
+      "step": 2070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.085556948368474e-05,
+      "loss": 0.5427,
+      "step": 2071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0847119515315476e-05,
+      "loss": 0.4429,
+      "step": 2072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.083866735837437e-05,
+      "loss": 0.5474,
+      "step": 2073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.083021301602506e-05,
+      "loss": 0.4504,
+      "step": 2074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0821756491431996e-05,
+      "loss": 0.4595,
+      "step": 2075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.081329778776045e-05,
+      "loss": 0.5397,
+      "step": 2076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0804836908176508e-05,
+      "loss": 0.5403,
+      "step": 2077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0796373855847073e-05,
+      "loss": 0.5698,
+      "step": 2078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.078790863393985e-05,
+      "loss": 0.517,
+      "step": 2079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0779441245623375e-05,
+      "loss": 0.5024,
+      "step": 2080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0770971694066975e-05,
+      "loss": 0.5042,
+      "step": 2081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.07624999824408e-05,
+      "loss": 0.4917,
+      "step": 2082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0754026113915807e-05,
+      "loss": 0.5225,
+      "step": 2083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.074555009166375e-05,
+      "loss": 0.5339,
+      "step": 2084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0737071918857203e-05,
+      "loss": 0.6064,
+      "step": 2085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0728591598669543e-05,
+      "loss": 0.4419,
+      "step": 2086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0720109134274935e-05,
+      "loss": 0.5183,
+      "step": 2087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0711624528848367e-05,
+      "loss": 0.5249,
+      "step": 2088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0703137785565616e-05,
+      "loss": 0.5651,
+      "step": 2089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0694648907603265e-05,
+      "loss": 0.5542,
+      "step": 2090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0686157898138688e-05,
+      "loss": 0.447,
+      "step": 2091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0677664760350067e-05,
+      "loss": 0.5413,
+      "step": 2092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0669169497416385e-05,
+      "loss": 0.4919,
+      "step": 2093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0660672112517394e-05,
+      "loss": 0.4829,
+      "step": 2094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.065217260883367e-05,
+      "loss": 0.4597,
+      "step": 2095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0643670989546564e-05,
+      "loss": 0.5977,
+      "step": 2096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0635167257838233e-05,
+      "loss": 0.4851,
+      "step": 2097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0626661416891606e-05,
+      "loss": 0.5542,
+      "step": 2098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0618153469890416e-05,
+      "loss": 0.5037,
+      "step": 2099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0609643420019187e-05,
+      "loss": 0.5269,
+      "step": 2100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0601131270463217e-05,
+      "loss": 0.4912,
+      "step": 2101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.05926170244086e-05,
+      "loss": 0.5894,
+      "step": 2102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0584100685042194e-05,
+      "loss": 0.4709,
+      "step": 2103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0575582255551687e-05,
+      "loss": 0.4418,
+      "step": 2104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0567061739125495e-05,
+      "loss": 0.4922,
+      "step": 2105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0558539138952858e-05,
+      "loss": 0.5833,
+      "step": 2106
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0550014458223757e-05,
+      "loss": 0.4771,
+      "step": 2107
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0541487700128995e-05,
+      "loss": 0.6157,
+      "step": 2108
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0532958867860115e-05,
+      "loss": 0.4741,
+      "step": 2109
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0524427964609465e-05,
+      "loss": 0.4531,
+      "step": 2110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.051589499357014e-05,
+      "loss": 0.4639,
+      "step": 2111
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0507359957936036e-05,
+      "loss": 0.5837,
+      "step": 2112
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0498822860901802e-05,
+      "loss": 0.345,
+      "step": 2113
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0490283705662863e-05,
+      "loss": 0.509,
+      "step": 2114
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0481742495415428e-05,
+      "loss": 0.4695,
+      "step": 2115
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047319923335646e-05,
+      "loss": 0.5522,
+      "step": 2116
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0464653922683695e-05,
+      "loss": 0.5432,
+      "step": 2117
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0456106566595628e-05,
+      "loss": 0.582,
+      "step": 2118
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.044755716829154e-05,
+      "loss": 0.5726,
+      "step": 2119
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043900573097146e-05,
+      "loss": 0.5591,
+      "step": 2120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0430452257836178e-05,
+      "loss": 0.54,
+      "step": 2121
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.042189675208725e-05,
+      "loss": 0.5381,
+      "step": 2122
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0413339216927005e-05,
+      "loss": 0.5173,
+      "step": 2123
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040477965555851e-05,
+      "loss": 0.564,
+      "step": 2124
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.039621807118561e-05,
+      "loss": 0.4514,
+      "step": 2125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03876544670129e-05,
+      "loss": 0.6218,
+      "step": 2126
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0379088846245718e-05,
+      "loss": 0.5254,
+      "step": 2127
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037052121209018e-05,
+      "loss": 0.488,
+      "step": 2128
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0361951567753135e-05,
+      "loss": 0.4176,
+      "step": 2129
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03533799164422e-05,
+      "loss": 0.511,
+      "step": 2130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0344806261365743e-05,
+      "loss": 0.5593,
+      "step": 2131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0336230605732863e-05,
+      "loss": 0.6216,
+      "step": 2132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0327652952753424e-05,
+      "loss": 0.521,
+      "step": 2133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0319073305638035e-05,
+      "loss": 0.5144,
+      "step": 2134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0310491667598054e-05,
+      "loss": 0.5784,
+      "step": 2135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0301908041845577e-05,
+      "loss": 0.4949,
+      "step": 2136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.029332243159345e-05,
+      "loss": 0.5193,
+      "step": 2137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0284734840055258e-05,
+      "loss": 0.6399,
+      "step": 2138
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0276145270445328e-05,
+      "loss": 0.6582,
+      "step": 2139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.026755372597873e-05,
+      "loss": 0.5144,
+      "step": 2140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0258960209871272e-05,
+      "loss": 0.6436,
+      "step": 2141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0250364725339507e-05,
+      "loss": 0.4551,
+      "step": 2142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0241767275600702e-05,
+      "loss": 0.4976,
+      "step": 2143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0233167863872885e-05,
+      "loss": 0.5186,
+      "step": 2144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0224566493374808e-05,
+      "loss": 0.5974,
+      "step": 2145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0215963167325954e-05,
+      "loss": 0.5989,
+      "step": 2146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0207357888946544e-05,
+      "loss": 0.6045,
+      "step": 2147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0198750661457522e-05,
+      "loss": 0.4668,
+      "step": 2148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0190141488080575e-05,
+      "loss": 0.5159,
+      "step": 2149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0181530372038096e-05,
+      "loss": 0.5278,
+      "step": 2150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0172917316553225e-05,
+      "loss": 0.4224,
+      "step": 2151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.016430232484982e-05,
+      "loss": 0.5342,
+      "step": 2152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0155685400152472e-05,
+      "loss": 0.5178,
+      "step": 2153
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0147066545686478e-05,
+      "loss": 0.4575,
+      "step": 2154
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0138445764677875e-05,
+      "loss": 0.5364,
+      "step": 2155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0129823060353413e-05,
+      "loss": 0.3582,
+      "step": 2156
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0121198435940557e-05,
+      "loss": 0.3835,
+      "step": 2157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0112571894667504e-05,
+      "loss": 0.5996,
+      "step": 2158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.010394343976316e-05,
+      "loss": 0.4583,
+      "step": 2159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.009531307445714e-05,
+      "loss": 0.5977,
+      "step": 2160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0086680801979794e-05,
+      "loss": 0.4358,
+      "step": 2161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0078046625562166e-05,
+      "loss": 0.4736,
+      "step": 2162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.006941054843603e-05,
+      "loss": 0.4858,
+      "step": 2163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.006077257383385e-05,
+      "loss": 0.5881,
+      "step": 2164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.005213270498882e-05,
+      "loss": 0.521,
+      "step": 2165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0043490945134834e-05,
+      "loss": 0.5197,
+      "step": 2166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0034847297506494e-05,
+      "loss": 0.6426,
+      "step": 2167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0026201765339117e-05,
+      "loss": 0.5486,
+      "step": 2168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0017554351868708e-05,
+      "loss": 0.593,
+      "step": 2169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0008905060331986e-05,
+      "loss": 0.4554,
+      "step": 2170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.000025389396639e-05,
+      "loss": 0.5771,
+      "step": 2171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999160085601003e-05,
+      "loss": 0.5256,
+      "step": 2172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9982945949701727e-05,
+      "loss": 0.5183,
+      "step": 2173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.997428917828102e-05,
+      "loss": 0.4233,
+      "step": 2174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9965630544988116e-05,
+      "loss": 0.5281,
+      "step": 2175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9956970053063948e-05,
+      "loss": 0.4026,
+      "step": 2176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.994830770575011e-05,
+      "loss": 0.4841,
+      "step": 2177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.993964350628894e-05,
+      "loss": 0.6045,
+      "step": 2178
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9930977457923416e-05,
+      "loss": 0.5237,
+      "step": 2179
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.992230956389725e-05,
+      "loss": 0.5686,
+      "step": 2180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9913639827454806e-05,
+      "loss": 0.49,
+      "step": 2181
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.990496825184118e-05,
+      "loss": 0.5225,
+      "step": 2182
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9896294840302127e-05,
+      "loss": 0.6057,
+      "step": 2183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9887619596084096e-05,
+      "loss": 0.5518,
+      "step": 2184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.987894252243423e-05,
+      "loss": 0.5039,
+      "step": 2185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.987026362260034e-05,
+      "loss": 0.4957,
+      "step": 2186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9861582899830937e-05,
+      "loss": 0.6034,
+      "step": 2187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.985290035737521e-05,
+      "loss": 0.4678,
+      "step": 2188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9844215998483024e-05,
+      "loss": 0.6018,
+      "step": 2189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9835529826404926e-05,
+      "loss": 0.515,
+      "step": 2190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9826841844392146e-05,
+      "loss": 0.4022,
+      "step": 2191
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9818152055696585e-05,
+      "loss": 0.5308,
+      "step": 2192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9809460463570827e-05,
+      "loss": 0.4175,
+      "step": 2193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9800767071268127e-05,
+      "loss": 0.5168,
+      "step": 2194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9792071882042412e-05,
+      "loss": 0.5115,
+      "step": 2195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9783374899148286e-05,
+      "loss": 0.4658,
+      "step": 2196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9774676125841015e-05,
+      "loss": 0.5717,
+      "step": 2197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9765975565376557e-05,
+      "loss": 0.4236,
+      "step": 2198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.975727322101151e-05,
+      "loss": 0.5497,
+      "step": 2199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9748569096003166e-05,
+      "loss": 0.5146,
+      "step": 2200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9739863193609464e-05,
+      "loss": 0.5457,
+      "step": 2201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9731155517089015e-05,
+      "loss": 0.5813,
+      "step": 2202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9722446069701096e-05,
+      "loss": 0.5242,
+      "step": 2203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.971373485470565e-05,
+      "loss": 0.4475,
+      "step": 2204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9705021875363275e-05,
+      "loss": 0.5135,
+      "step": 2205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9696307134935234e-05,
+      "loss": 0.4644,
+      "step": 2206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.968759063668344e-05,
+      "loss": 0.46,
+      "step": 2207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9678872383870475e-05,
+      "loss": 0.5287,
+      "step": 2208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9670152379759574e-05,
+      "loss": 0.5037,
+      "step": 2209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.966143062761463e-05,
+      "loss": 0.4805,
+      "step": 2210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9652707130700177e-05,
+      "loss": 0.7625,
+      "step": 2211
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9643981892281425e-05,
+      "loss": 0.4742,
+      "step": 2212
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.963525491562421e-05,
+      "loss": 0.4453,
+      "step": 2213
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9626526203995048e-05,
+      "loss": 0.4951,
+      "step": 2214
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9617795760661076e-05,
+      "loss": 0.4873,
+      "step": 2215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.96090635888901e-05,
+      "loss": 0.5293,
+      "step": 2216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9600329691950554e-05,
+      "loss": 0.5383,
+      "step": 2217
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9591594073111535e-05,
+      "loss": 0.4985,
+      "step": 2218
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.958285673564278e-05,
+      "loss": 0.5261,
+      "step": 2219
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.957411768281466e-05,
+      "loss": 0.5267,
+      "step": 2220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.95653769178982e-05,
+      "loss": 0.5564,
+      "step": 2221
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9556634444165058e-05,
+      "loss": 0.4031,
+      "step": 2222
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9547890264887544e-05,
+      "loss": 0.55,
+      "step": 2223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.953914438333858e-05,
+      "loss": 0.5518,
+      "step": 2224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9530396802791753e-05,
+      "loss": 0.5146,
+      "step": 2225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9521647526521272e-05,
+      "loss": 0.5359,
+      "step": 2226
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9512896557801987e-05,
+      "loss": 0.5479,
+      "step": 2227
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.950414389990937e-05,
+      "loss": 0.5425,
+      "step": 2228
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9495389556119542e-05,
+      "loss": 0.48,
+      "step": 2229
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9486633529709246e-05,
+      "loss": 0.5042,
+      "step": 2230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.947787582395585e-05,
+      "loss": 0.426,
+      "step": 2231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9469116442137358e-05,
+      "loss": 0.4878,
+      "step": 2232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9460355387532398e-05,
+      "loss": 0.5149,
+      "step": 2233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.945159266342023e-05,
+      "loss": 0.5645,
+      "step": 2234
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9442828273080728e-05,
+      "loss": 0.5122,
+      "step": 2235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9434062219794394e-05,
+      "loss": 0.5654,
+      "step": 2236
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.942529450684237e-05,
+      "loss": 0.6235,
+      "step": 2237
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9416525137506385e-05,
+      "loss": 0.5303,
+      "step": 2238
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9407754115068814e-05,
+      "loss": 0.5376,
+      "step": 2239
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9398981442812643e-05,
+      "loss": 0.5148,
+      "step": 2240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.939020712402148e-05,
+      "loss": 0.4381,
+      "step": 2241
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9381431161979535e-05,
+      "loss": 0.6147,
+      "step": 2242
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.937265355997165e-05,
+      "loss": 0.4976,
+      "step": 2243
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9363874321283275e-05,
+      "loss": 0.5261,
+      "step": 2244
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.935509344920047e-05,
+      "loss": 0.5601,
+      "step": 2245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9346310947009905e-05,
+      "loss": 0.6245,
+      "step": 2246
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.933752681799886e-05,
+      "loss": 0.6216,
+      "step": 2247
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9328741065455242e-05,
+      "loss": 0.4844,
+      "step": 2248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9319953692667535e-05,
+      "loss": 0.5848,
+      "step": 2249
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9311164702924854e-05,
+      "loss": 0.5461,
+      "step": 2250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9302374099516904e-05,
+      "loss": 0.5021,
+      "step": 2251
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9293581885734007e-05,
+      "loss": 0.5742,
+      "step": 2252
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9284788064867077e-05,
+      "loss": 0.6099,
+      "step": 2253
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9275992640207635e-05,
+      "loss": 0.5449,
+      "step": 2254
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9267195615047798e-05,
+      "loss": 0.7058,
+      "step": 2255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9258396992680287e-05,
+      "loss": 0.6138,
+      "step": 2256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9249596776398424e-05,
+      "loss": 0.4702,
+      "step": 2257
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9240794969496114e-05,
+      "loss": 0.4753,
+      "step": 2258
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9231991575267868e-05,
+      "loss": 0.5078,
+      "step": 2259
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9223186597008795e-05,
+      "loss": 0.424,
+      "step": 2260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9214380038014583e-05,
+      "loss": 0.4812,
+      "step": 2261
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9205571901581522e-05,
+      "loss": 0.6313,
+      "step": 2262
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9196762191006494e-05,
+      "loss": 0.5093,
+      "step": 2263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9187950909586964e-05,
+      "loss": 0.475,
+      "step": 2264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.917913806062098e-05,
+      "loss": 0.4768,
+      "step": 2265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9170323647407195e-05,
+      "loss": 0.5916,
+      "step": 2266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9161507673244824e-05,
+      "loss": 0.5139,
+      "step": 2267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.91526901414337e-05,
+      "loss": 0.5647,
+      "step": 2268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.914387105527419e-05,
+      "loss": 0.3865,
+      "step": 2269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9135050418067286e-05,
+      "loss": 0.5051,
+      "step": 2270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9126228233114546e-05,
+      "loss": 0.417,
+      "step": 2271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9117404503718102e-05,
+      "loss": 0.5381,
+      "step": 2272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9108579233180667e-05,
+      "loss": 0.5366,
+      "step": 2273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9099752424805538e-05,
+      "loss": 0.5439,
+      "step": 2274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.909092408189658e-05,
+      "loss": 0.4539,
+      "step": 2275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9082094207758227e-05,
+      "loss": 0.4795,
+      "step": 2276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9073262805695505e-05,
+      "loss": 0.499,
+      "step": 2277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.906442987901399e-05,
+      "loss": 0.3397,
+      "step": 2278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9055595431019848e-05,
+      "loss": 0.4719,
+      "step": 2279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9046759465019794e-05,
+      "loss": 0.4849,
+      "step": 2280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9037921984321132e-05,
+      "loss": 0.4468,
+      "step": 2281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9029082992231723e-05,
+      "loss": 0.5879,
+      "step": 2282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.902024249205999e-05,
+      "loss": 0.5703,
+      "step": 2283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.901140048711492e-05,
+      "loss": 0.5081,
+      "step": 2284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9002556980706086e-05,
+      "loss": 0.5244,
+      "step": 2285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.899371197614359e-05,
+      "loss": 0.4377,
+      "step": 2286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8984865476738104e-05,
+      "loss": 0.5244,
+      "step": 2287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8976017485800882e-05,
+      "loss": 0.511,
+      "step": 2288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896716800664371e-05,
+      "loss": 0.4885,
+      "step": 2289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.895831704257894e-05,
+      "loss": 0.6301,
+      "step": 2290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8949464596919483e-05,
+      "loss": 0.4182,
+      "step": 2291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8940610672978803e-05,
+      "loss": 0.5056,
+      "step": 2292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8931755274070914e-05,
+      "loss": 0.5554,
+      "step": 2293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8922898403510388e-05,
+      "loss": 0.4175,
+      "step": 2294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.891404006461233e-05,
+      "loss": 0.4492,
+      "step": 2295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.890518026069243e-05,
+      "loss": 0.4253,
+      "step": 2296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.889631899506689e-05,
+      "loss": 0.469,
+      "step": 2297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.888745627105248e-05,
+      "loss": 0.5891,
+      "step": 2298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.88785920919665e-05,
+      "loss": 0.5986,
+      "step": 2299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8869726461126825e-05,
+      "loss": 0.5439,
+      "step": 2300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8860859381851837e-05,
+      "loss": 0.5605,
+      "step": 2301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8851990857460476e-05,
+      "loss": 0.5916,
+      "step": 2302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8843120891272232e-05,
+      "loss": 0.4827,
+      "step": 2303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.883424948660712e-05,
+      "loss": 0.4595,
+      "step": 2304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8825376646785707e-05,
+      "loss": 0.4907,
+      "step": 2305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.881650237512907e-05,
+      "loss": 0.5627,
+      "step": 2306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8807626674958862e-05,
+      "loss": 0.459,
+      "step": 2307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.879874954959724e-05,
+      "loss": 0.4846,
+      "step": 2308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.878987100236691e-05,
+      "loss": 0.3438,
+      "step": 2309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8780991036591096e-05,
+      "loss": 0.573,
+      "step": 2310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.877210965559357e-05,
+      "loss": 0.472,
+      "step": 2311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8763226862698626e-05,
+      "loss": 0.4701,
+      "step": 2312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8754342661231075e-05,
+      "loss": 0.4844,
+      "step": 2313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8745457054516275e-05,
+      "loss": 0.5723,
+      "step": 2314
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.87365700458801e-05,
+      "loss": 0.4604,
+      "step": 2315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.872768163864895e-05,
+      "loss": 0.4885,
+      "step": 2316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8718791836149743e-05,
+      "loss": 0.5254,
+      "step": 2317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.870990064170993e-05,
+      "loss": 0.501,
+      "step": 2318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8701008058657473e-05,
+      "loss": 0.4651,
+      "step": 2319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8692114090320866e-05,
+      "loss": 0.5125,
+      "step": 2320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8683218740029108e-05,
+      "loss": 0.4089,
+      "step": 2321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8674322011111722e-05,
+      "loss": 0.4949,
+      "step": 2322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8665423906898748e-05,
+      "loss": 0.406,
+      "step": 2323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8656524430720732e-05,
+      "loss": 0.5645,
+      "step": 2324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8647623585908746e-05,
+      "loss": 0.4233,
+      "step": 2325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8638721375794365e-05,
+      "loss": 0.6599,
+      "step": 2326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8629817803709688e-05,
+      "loss": 0.5278,
+      "step": 2327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8620912872987303e-05,
+      "loss": 0.4946,
+      "step": 2328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8612006586960324e-05,
+      "loss": 0.5005,
+      "step": 2329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8603098948962367e-05,
+      "loss": 0.3738,
+      "step": 2330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.859418996232755e-05,
+      "loss": 0.4833,
+      "step": 2331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8585279630390497e-05,
+      "loss": 0.6519,
+      "step": 2332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8576367956486342e-05,
+      "loss": 0.5984,
+      "step": 2333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.856745494395072e-05,
+      "loss": 0.4592,
+      "step": 2334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8558540596119757e-05,
+      "loss": 0.4659,
+      "step": 2335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.854962491633009e-05,
+      "loss": 0.5383,
+      "step": 2336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8540707907918844e-05,
+      "loss": 0.5193,
+      "step": 2337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.853178957422366e-05,
+      "loss": 0.4834,
+      "step": 2338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.852286991858265e-05,
+      "loss": 0.5503,
+      "step": 2339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.851394894433444e-05,
+      "loss": 0.4325,
+      "step": 2340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8505026654818147e-05,
+      "loss": 0.5603,
+      "step": 2341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8496103053373364e-05,
+      "loss": 0.5374,
+      "step": 2342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8487178143340195e-05,
+      "loss": 0.4829,
+      "step": 2343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.847825192805923e-05,
+      "loss": 0.5083,
+      "step": 2344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8469324410871537e-05,
+      "loss": 0.5654,
+      "step": 2345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.846039559511868e-05,
+      "loss": 0.4375,
+      "step": 2346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.845146548414271e-05,
+      "loss": 0.5865,
+      "step": 2347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.844253408128616e-05,
+      "loss": 0.5957,
+      "step": 2348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.843360138989204e-05,
+      "loss": 0.5425,
+      "step": 2349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8424667413303848e-05,
+      "loss": 0.4897,
+      "step": 2350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8415732154865576e-05,
+      "loss": 0.542,
+      "step": 2351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8406795617921676e-05,
+      "loss": 0.5129,
+      "step": 2352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8397857805817084e-05,
+      "loss": 0.54,
+      "step": 2353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8388918721897218e-05,
+      "loss": 0.6299,
+      "step": 2354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.837997836950797e-05,
+      "loss": 0.4116,
+      "step": 2355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8371036751995704e-05,
+      "loss": 0.4763,
+      "step": 2356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8362093872707267e-05,
+      "loss": 0.5679,
+      "step": 2357
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8353149734989956e-05,
+      "loss": 0.4899,
+      "step": 2358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.834420434219157e-05,
+      "loss": 0.6274,
+      "step": 2359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8335257697660353e-05,
+      "loss": 0.5208,
+      "step": 2360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.832630980474503e-05,
+      "loss": 0.6782,
+      "step": 2361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8317360666794784e-05,
+      "loss": 0.533,
+      "step": 2362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8308410287159283e-05,
+      "loss": 0.5754,
+      "step": 2363
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8299458669188634e-05,
+      "loss": 0.6858,
+      "step": 2364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.829050581623342e-05,
+      "loss": 0.2985,
+      "step": 2365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8281551731644692e-05,
+      "loss": 0.447,
+      "step": 2366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8272596418773954e-05,
+      "loss": 0.5596,
+      "step": 2367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8263639880973174e-05,
+      "loss": 0.4253,
+      "step": 2368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.825468212159477e-05,
+      "loss": 0.6052,
+      "step": 2369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8245723143991634e-05,
+      "loss": 0.5908,
+      "step": 2370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8236762951517095e-05,
+      "loss": 0.5347,
+      "step": 2371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8227801547524948e-05,
+      "loss": 0.4927,
+      "step": 2372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8218838935369437e-05,
+      "loss": 0.593,
+      "step": 2373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.820987511840526e-05,
+      "loss": 0.5911,
+      "step": 2374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8200910099987575e-05,
+      "loss": 0.5432,
+      "step": 2375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8191943883471968e-05,
+      "loss": 0.5581,
+      "step": 2376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8182976472214498e-05,
+      "loss": 0.4927,
+      "step": 2377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.817400786957165e-05,
+      "loss": 0.4275,
+      "step": 2378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.816503807890037e-05,
+      "loss": 0.4055,
+      "step": 2379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.815606710355803e-05,
+      "loss": 0.4185,
+      "step": 2380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8147094946902483e-05,
+      "loss": 0.5558,
+      "step": 2381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8138121612291985e-05,
+      "loss": 0.6178,
+      "step": 2382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8129147103085242e-05,
+      "loss": 0.5752,
+      "step": 2383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.812017142264141e-05,
+      "loss": 0.6162,
+      "step": 2384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8111194574320082e-05,
+      "loss": 0.5474,
+      "step": 2385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8102216561481286e-05,
+      "loss": 0.5059,
+      "step": 2386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8093237387485477e-05,
+      "loss": 0.5649,
+      "step": 2387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.808425705569356e-05,
+      "loss": 0.4233,
+      "step": 2388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8075275569466855e-05,
+      "loss": 0.5439,
+      "step": 2389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.806629293216714e-05,
+      "loss": 0.5046,
+      "step": 2390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.805730914715659e-05,
+      "loss": 0.424,
+      "step": 2391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8048324217797843e-05,
+      "loss": 0.4827,
+      "step": 2392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.803933814745395e-05,
+      "loss": 0.514,
+      "step": 2393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8030350939488375e-05,
+      "loss": 0.4207,
+      "step": 2394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8021362597265038e-05,
+      "loss": 0.5103,
+      "step": 2395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.801237312414826e-05,
+      "loss": 0.6265,
+      "step": 2396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8003382523502796e-05,
+      "loss": 0.4668,
+      "step": 2397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7994390798693818e-05,
+      "loss": 0.6177,
+      "step": 2398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.798539795308692e-05,
+      "loss": 0.4302,
+      "step": 2399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7976403990048126e-05,
+      "loss": 0.5171,
+      "step": 2400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7967408912943856e-05,
+      "loss": 0.4922,
+      "step": 2401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7958412725140963e-05,
+      "loss": 0.4854,
+      "step": 2402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7949415430006713e-05,
+      "loss": 0.5332,
+      "step": 2403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7940417030908793e-05,
+      "loss": 0.5483,
+      "step": 2404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7931417531215285e-05,
+      "loss": 0.5967,
+      "step": 2405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7922416934294702e-05,
+      "loss": 0.5598,
+      "step": 2406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7913415243515953e-05,
+      "loss": 0.4609,
+      "step": 2407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.790441246224837e-05,
+      "loss": 0.49,
+      "step": 2408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7895408593861684e-05,
+      "loss": 0.5723,
+      "step": 2409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7886403641726032e-05,
+      "loss": 0.4889,
+      "step": 2410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7877397609211965e-05,
+      "loss": 0.5374,
+      "step": 2411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.786839049969043e-05,
+      "loss": 0.5464,
+      "step": 2412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.785938231653278e-05,
+      "loss": 0.501,
+      "step": 2413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7850373063110772e-05,
+      "loss": 0.5173,
+      "step": 2414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7841362742796566e-05,
+      "loss": 0.47,
+      "step": 2415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.783235135896271e-05,
+      "loss": 0.4851,
+      "step": 2416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7823338914982154e-05,
+      "loss": 0.5808,
+      "step": 2417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.781432541422826e-05,
+      "loss": 0.4631,
+      "step": 2418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7805310860074766e-05,
+      "loss": 0.5315,
+      "step": 2419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.779629525589581e-05,
+      "loss": 0.4556,
+      "step": 2420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7787278605065932e-05,
+      "loss": 0.4106,
+      "step": 2421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7778260910960055e-05,
+      "loss": 0.4873,
+      "step": 2422
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.776924217695349e-05,
+      "loss": 0.457,
+      "step": 2423
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7760222406421943e-05,
+      "loss": 0.489,
+      "step": 2424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.775120160274151e-05,
+      "loss": 0.4775,
+      "step": 2425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7742179769288662e-05,
+      "loss": 0.5935,
+      "step": 2426
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7733156909440273e-05,
+      "loss": 0.4832,
+      "step": 2427
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7724133026573584e-05,
+      "loss": 0.5728,
+      "step": 2428
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7715108124066235e-05,
+      "loss": 0.4695,
+      "step": 2429
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.770608220529623e-05,
+      "loss": 0.5206,
+      "step": 2430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.769705527364197e-05,
+      "loss": 0.5017,
+      "step": 2431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7688027332482222e-05,
+      "loss": 0.5117,
+      "step": 2432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.767899838519614e-05,
+      "loss": 0.4667,
+      "step": 2433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.766996843516326e-05,
+      "loss": 0.4486,
+      "step": 2434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.766093748576346e-05,
+      "loss": 0.5566,
+      "step": 2435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7651905540377045e-05,
+      "loss": 0.5333,
+      "step": 2436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7642872602384645e-05,
+      "loss": 0.5242,
+      "step": 2437
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.763383867516729e-05,
+      "loss": 0.4521,
+      "step": 2438
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7624803762106364e-05,
+      "loss": 0.5586,
+      "step": 2439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7615767866583638e-05,
+      "loss": 0.552,
+      "step": 2440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7606730991981234e-05,
+      "loss": 0.4973,
+      "step": 2441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7597693141681643e-05,
+      "loss": 0.6274,
+      "step": 2442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.758865431906773e-05,
+      "loss": 0.5764,
+      "step": 2443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7579614527522716e-05,
+      "loss": 0.6384,
+      "step": 2444
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7570573770430188e-05,
+      "loss": 0.4614,
+      "step": 2445
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7561532051174094e-05,
+      "loss": 0.4658,
+      "step": 2446
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7552489373138737e-05,
+      "loss": 0.3853,
+      "step": 2447
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.754344573970879e-05,
+      "loss": 0.4685,
+      "step": 2448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.753440115426928e-05,
+      "loss": 0.4937,
+      "step": 2449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7525355620205574e-05,
+      "loss": 0.5037,
+      "step": 2450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7516309140903428e-05,
+      "loss": 0.4423,
+      "step": 2451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7507261719748918e-05,
+      "loss": 0.4731,
+      "step": 2452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7498213360128482e-05,
+      "loss": 0.6614,
+      "step": 2453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.748916406542892e-05,
+      "loss": 0.4976,
+      "step": 2454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7480113839037378e-05,
+      "loss": 0.4706,
+      "step": 2455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7471062684341346e-05,
+      "loss": 0.4526,
+      "step": 2456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7462010604728656e-05,
+      "loss": 0.5303,
+      "step": 2457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7452957603587502e-05,
+      "loss": 0.5293,
+      "step": 2458
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7443903684306414e-05,
+      "loss": 0.5217,
+      "step": 2459
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7434848850274267e-05,
+      "loss": 0.5022,
+      "step": 2460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.742579310488027e-05,
+      "loss": 0.5098,
+      "step": 2461
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7416736451513986e-05,
+      "loss": 0.5391,
+      "step": 2462
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7407678893565318e-05,
+      "loss": 0.5066,
+      "step": 2463
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7398620434424494e-05,
+      "loss": 0.4548,
+      "step": 2464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7389561077482085e-05,
+      "loss": 0.5066,
+      "step": 2465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7380500826129012e-05,
+      "loss": 0.5276,
+      "step": 2466
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7371439683756516e-05,
+      "loss": 0.6245,
+      "step": 2467
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.736237765375617e-05,
+      "loss": 0.5146,
+      "step": 2468
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7353314739519882e-05,
+      "loss": 0.53,
+      "step": 2469
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.73442509444399e-05,
+      "loss": 0.5059,
+      "step": 2470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7335186271908792e-05,
+      "loss": 0.377,
+      "step": 2471
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.732612072531945e-05,
+      "loss": 0.5685,
+      "step": 2472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7317054308065115e-05,
+      "loss": 0.4507,
+      "step": 2473
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7307987023539336e-05,
+      "loss": 0.5701,
+      "step": 2474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.729891887513598e-05,
+      "loss": 0.443,
+      "step": 2475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7289849866249254e-05,
+      "loss": 0.4893,
+      "step": 2476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7280780000273677e-05,
+      "loss": 0.4756,
+      "step": 2477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7271709280604096e-05,
+      "loss": 0.5404,
+      "step": 2478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7262637710635673e-05,
+      "loss": 0.5632,
+      "step": 2479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7253565293763885e-05,
+      "loss": 0.6279,
+      "step": 2480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7244492033384535e-05,
+      "loss": 0.4368,
+      "step": 2481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7235417932893734e-05,
+      "loss": 0.439,
+      "step": 2482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7226342995687907e-05,
+      "loss": 0.5205,
+      "step": 2483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.72172672251638e-05,
+      "loss": 0.5238,
+      "step": 2484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7208190624718466e-05,
+      "loss": 0.4424,
+      "step": 2485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7199113197749272e-05,
+      "loss": 0.5237,
+      "step": 2486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7190034947653878e-05,
+      "loss": 0.498,
+      "step": 2487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.718095587783028e-05,
+      "loss": 0.5266,
+      "step": 2488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7171875991676754e-05,
+      "loss": 0.4592,
+      "step": 2489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7162795292591903e-05,
+      "loss": 0.5249,
+      "step": 2490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.715371378397462e-05,
+      "loss": 0.517,
+      "step": 2491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7144631469224113e-05,
+      "loss": 0.4507,
+      "step": 2492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7135548351739874e-05,
+      "loss": 0.5747,
+      "step": 2493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.712646443492171e-05,
+      "loss": 0.4916,
+      "step": 2494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.711737972216973e-05,
+      "loss": 0.4707,
+      "step": 2495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7108294216884325e-05,
+      "loss": 0.3717,
+      "step": 2496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7099207922466198e-05,
+      "loss": 0.5437,
+      "step": 2497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.709012084231633e-05,
+      "loss": 0.4502,
+      "step": 2498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7081032979836027e-05,
+      "loss": 0.5156,
+      "step": 2499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7071944338426853e-05,
+      "loss": 0.5066,
+      "step": 2500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7062854921490684e-05,
+      "loss": 0.5825,
+      "step": 2501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7053764732429675e-05,
+      "loss": 0.6077,
+      "step": 2502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7044673774646283e-05,
+      "loss": 0.4736,
+      "step": 2503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7035582051543253e-05,
+      "loss": 0.4502,
+      "step": 2504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.702648956652359e-05,
+      "loss": 0.5437,
+      "step": 2505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7017396322990617e-05,
+      "loss": 0.4215,
+      "step": 2506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7008302324347923e-05,
+      "loss": 0.5469,
+      "step": 2507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6999207573999382e-05,
+      "loss": 0.5454,
+      "step": 2508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6990112075349152e-05,
+      "loss": 0.489,
+      "step": 2509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6981015831801682e-05,
+      "loss": 0.4504,
+      "step": 2510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.697191884676167e-05,
+      "loss": 0.5603,
+      "step": 2511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6962821123634125e-05,
+      "loss": 0.5256,
+      "step": 2512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6953722665824302e-05,
+      "loss": 0.5803,
+      "step": 2513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6944623476737756e-05,
+      "loss": 0.5537,
+      "step": 2514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.693552355978031e-05,
+      "loss": 0.52,
+      "step": 2515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6926422918358046e-05,
+      "loss": 0.593,
+      "step": 2516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6917321555877324e-05,
+      "loss": 0.4365,
+      "step": 2517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6908219475744787e-05,
+      "loss": 0.5227,
+      "step": 2518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.689911668136733e-05,
+      "loss": 0.468,
+      "step": 2519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6890013176152117e-05,
+      "loss": 0.4591,
+      "step": 2520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.688090896350659e-05,
+      "loss": 0.5303,
+      "step": 2521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6871804046838446e-05,
+      "loss": 0.5466,
+      "step": 2522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6862698429555644e-05,
+      "loss": 0.478,
+      "step": 2523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.685359211506641e-05,
+      "loss": 0.4531,
+      "step": 2524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6844485106779233e-05,
+      "loss": 0.4843,
+      "step": 2525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6835377408102858e-05,
+      "loss": 0.5386,
+      "step": 2526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6826269022446282e-05,
+      "loss": 0.4712,
+      "step": 2527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6817159953218773e-05,
+      "loss": 0.5039,
+      "step": 2528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6808050203829845e-05,
+      "loss": 0.4766,
+      "step": 2529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6798939777689273e-05,
+      "loss": 0.5601,
+      "step": 2530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6789828678207076e-05,
+      "loss": 0.5859,
+      "step": 2531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6780716908793535e-05,
+      "loss": 0.5176,
+      "step": 2532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.677160447285918e-05,
+      "loss": 0.4325,
+      "step": 2533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6762491373814775e-05,
+      "loss": 0.4841,
+      "step": 2534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6753377615071364e-05,
+      "loss": 0.473,
+      "step": 2535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.67442632000402e-05,
+      "loss": 0.5146,
+      "step": 2536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.673514813213282e-05,
+      "loss": 0.5659,
+      "step": 2537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6726032414760968e-05,
+      "loss": 0.5142,
+      "step": 2538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.671691605133666e-05,
+      "loss": 0.5671,
+      "step": 2539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6707799045272143e-05,
+      "loss": 0.5615,
+      "step": 2540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.66986813999799e-05,
+      "loss": 0.4749,
+      "step": 2541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6689563118872664e-05,
+      "loss": 0.6025,
+      "step": 2542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.668044420536339e-05,
+      "loss": 0.5156,
+      "step": 2543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.667132466286529e-05,
+      "loss": 0.5,
+      "step": 2544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6662204494791794e-05,
+      "loss": 0.5112,
+      "step": 2545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6653083704556573e-05,
+      "loss": 0.4067,
+      "step": 2546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6643962295573537e-05,
+      "loss": 0.5474,
+      "step": 2547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.663484027125682e-05,
+      "loss": 0.5774,
+      "step": 2548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6625717635020782e-05,
+      "loss": 0.5073,
+      "step": 2549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6616594390280025e-05,
+      "loss": 0.4912,
+      "step": 2550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6607470540449366e-05,
+      "loss": 0.5723,
+      "step": 2551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.659834608894386e-05,
+      "loss": 0.6694,
+      "step": 2552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.658922103917877e-05,
+      "loss": 0.5432,
+      "step": 2553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6580095394569614e-05,
+      "loss": 0.3705,
+      "step": 2554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.65709691585321e-05,
+      "loss": 0.5342,
+      "step": 2555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6561842334482167e-05,
+      "loss": 0.4761,
+      "step": 2556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.655271492583598e-05,
+      "loss": 0.4241,
+      "step": 2557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6543586936009927e-05,
+      "loss": 0.3838,
+      "step": 2558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.65344583684206e-05,
+      "loss": 0.5291,
+      "step": 2559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6525329226484825e-05,
+      "loss": 0.354,
+      "step": 2560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.651619951361961e-05,
+      "loss": 0.4482,
+      "step": 2561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.650706923324222e-05,
+      "loss": 0.5236,
+      "step": 2562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6497938388770102e-05,
+      "loss": 0.584,
+      "step": 2563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6488806983620927e-05,
+      "loss": 0.5339,
+      "step": 2564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6479675021212567e-05,
+      "loss": 0.5671,
+      "step": 2565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6470542504963112e-05,
+      "loss": 0.437,
+      "step": 2566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6461409438290855e-05,
+      "loss": 0.5217,
+      "step": 2567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6452275824614286e-05,
+      "loss": 0.5701,
+      "step": 2568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6443141667352123e-05,
+      "loss": 0.4861,
+      "step": 2569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6434006969923264e-05,
+      "loss": 0.5427,
+      "step": 2570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6424871735746824e-05,
+      "loss": 0.4932,
+      "step": 2571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6415735968242107e-05,
+      "loss": 0.5017,
+      "step": 2572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.640659967082863e-05,
+      "loss": 0.5212,
+      "step": 2573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.63974628469261e-05,
+      "loss": 0.4641,
+      "step": 2574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6388325499954413e-05,
+      "loss": 0.4722,
+      "step": 2575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.637918763333368e-05,
+      "loss": 0.5664,
+      "step": 2576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6370049250484195e-05,
+      "loss": 0.4304,
+      "step": 2577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6360910354826447e-05,
+      "loss": 0.4713,
+      "step": 2578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6351770949781113e-05,
+      "loss": 0.4534,
+      "step": 2579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6342631038769067e-05,
+      "loss": 0.4521,
+      "step": 2580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.633349062521137e-05,
+      "loss": 0.5171,
+      "step": 2581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6324349712529275e-05,
+      "loss": 0.4946,
+      "step": 2582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6315208304144213e-05,
+      "loss": 0.4426,
+      "step": 2583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6306066403477802e-05,
+      "loss": 0.6052,
+      "step": 2584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6296924013951858e-05,
+      "loss": 0.4719,
+      "step": 2585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6287781138988357e-05,
+      "loss": 0.48,
+      "step": 2586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.627863778200948e-05,
+      "loss": 0.6562,
+      "step": 2587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6269493946437568e-05,
+      "loss": 0.4603,
+      "step": 2588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6260349635695166e-05,
+      "loss": 0.562,
+      "step": 2589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6251204853204963e-05,
+      "loss": 0.467,
+      "step": 2590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6242059602389855e-05,
+      "loss": 0.5122,
+      "step": 2591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.62329138866729e-05,
+      "loss": 0.5327,
+      "step": 2592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6223767709477333e-05,
+      "loss": 0.4412,
+      "step": 2593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.621462107422655e-05,
+      "loss": 0.4526,
+      "step": 2594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6205473984344142e-05,
+      "loss": 0.5435,
+      "step": 2595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6196326443253853e-05,
+      "loss": 0.5654,
+      "step": 2596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6187178454379593e-05,
+      "loss": 0.5094,
+      "step": 2597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6178030021145455e-05,
+      "loss": 0.5908,
+      "step": 2598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.616888114697568e-05,
+      "loss": 0.3904,
+      "step": 2599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6159731835294702e-05,
+      "loss": 0.4434,
+      "step": 2600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.615058208952708e-05,
+      "loss": 0.5293,
+      "step": 2601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.614143191309757e-05,
+      "loss": 0.4814,
+      "step": 2602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.613228130943107e-05,
+      "loss": 0.5525,
+      "step": 2603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6123130281952637e-05,
+      "loss": 0.4937,
+      "step": 2604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.61139788340875e-05,
+      "loss": 0.4937,
+      "step": 2605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6104826969261038e-05,
+      "loss": 0.564,
+      "step": 2606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.609567469089879e-05,
+      "loss": 0.5042,
+      "step": 2607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6086522002426437e-05,
+      "loss": 0.5918,
+      "step": 2608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6077368907269826e-05,
+      "loss": 0.5264,
+      "step": 2609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6068215408854958e-05,
+      "loss": 0.5835,
+      "step": 2610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6059061510607968e-05,
+      "loss": 0.5793,
+      "step": 2611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6049907215955163e-05,
+      "loss": 0.5095,
+      "step": 2612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.604075252832298e-05,
+      "loss": 0.4688,
+      "step": 2613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6031597451138024e-05,
+      "loss": 0.448,
+      "step": 2614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6022441987827014e-05,
+      "loss": 0.4937,
+      "step": 2615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6013286141816846e-05,
+      "loss": 0.6027,
+      "step": 2616
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6004129916534536e-05,
+      "loss": 0.5071,
+      "step": 2617
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.599497331540726e-05,
+      "loss": 0.5051,
+      "step": 2618
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5985816341862327e-05,
+      "loss": 0.5879,
+      "step": 2619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.597665899932717e-05,
+      "loss": 0.4827,
+      "step": 2620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.596750129122939e-05,
+      "loss": 0.4514,
+      "step": 2621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5958343220996697e-05,
+      "loss": 0.5996,
+      "step": 2622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.594918479205696e-05,
+      "loss": 0.4719,
+      "step": 2623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5940026007838163e-05,
+      "loss": 0.4485,
+      "step": 2624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5930866871768438e-05,
+      "loss": 0.405,
+      "step": 2625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.592170738727604e-05,
+      "loss": 0.4216,
+      "step": 2626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5912547557789345e-05,
+      "loss": 0.5396,
+      "step": 2627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.590338738673689e-05,
+      "loss": 0.5183,
+      "step": 2628
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5894226877547298e-05,
+      "loss": 0.5063,
+      "step": 2629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5885066033649358e-05,
+      "loss": 0.6418,
+      "step": 2630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5875904858471947e-05,
+      "loss": 0.5791,
+      "step": 2631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5866743355444104e-05,
+      "loss": 0.4939,
+      "step": 2632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5857581527994962e-05,
+      "loss": 0.5122,
+      "step": 2633
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5848419379553787e-05,
+      "loss": 0.4404,
+      "step": 2634
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5839256913549962e-05,
+      "loss": 0.4115,
+      "step": 2635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.583009413341299e-05,
+      "loss": 0.4915,
+      "step": 2636
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5820931042572497e-05,
+      "loss": 0.4668,
+      "step": 2637
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5811767644458214e-05,
+      "loss": 0.49,
+      "step": 2638
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5802603942499993e-05,
+      "loss": 0.5444,
+      "step": 2639
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.57934399401278e-05,
+      "loss": 0.4379,
+      "step": 2640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5784275640771723e-05,
+      "loss": 0.3812,
+      "step": 2641
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5775111047861933e-05,
+      "loss": 0.5178,
+      "step": 2642
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.576594616482875e-05,
+      "loss": 0.4359,
+      "step": 2643
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.575678099510257e-05,
+      "loss": 0.4927,
+      "step": 2644
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5747615542113906e-05,
+      "loss": 0.4856,
+      "step": 2645
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5738449809293383e-05,
+      "loss": 0.4595,
+      "step": 2646
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5729283800071725e-05,
+      "loss": 0.54,
+      "step": 2647
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.572011751787977e-05,
+      "loss": 0.2877,
+      "step": 2648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5710950966148442e-05,
+      "loss": 0.3829,
+      "step": 2649
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5701784148308773e-05,
+      "loss": 0.349,
+      "step": 2650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.56926170677919e-05,
+      "loss": 0.2703,
+      "step": 2651
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5683449728029047e-05,
+      "loss": 0.2881,
+      "step": 2652
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5674282132451546e-05,
+      "loss": 0.2078,
+      "step": 2653
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5665114284490813e-05,
+      "loss": 0.3035,
+      "step": 2654
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.565594618757838e-05,
+      "loss": 0.3257,
+      "step": 2655
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5646777845145847e-05,
+      "loss": 0.2402,
+      "step": 2656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5637609260624915e-05,
+      "loss": 0.3114,
+      "step": 2657
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5628440437447384e-05,
+      "loss": 0.2287,
+      "step": 2658
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5619271379045135e-05,
+      "loss": 0.2792,
+      "step": 2659
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5610102088850138e-05,
+      "loss": 0.3395,
+      "step": 2660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.560093257029445e-05,
+      "loss": 0.2914,
+      "step": 2661
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5591762826810213e-05,
+      "loss": 0.2117,
+      "step": 2662
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5582592861829654e-05,
+      "loss": 0.292,
+      "step": 2663
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5573422678785088e-05,
+      "loss": 0.313,
+      "step": 2664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5564252281108898e-05,
+      "loss": 0.3077,
+      "step": 2665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5555081672233568e-05,
+      "loss": 0.2985,
+      "step": 2666
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.554591085559164e-05,
+      "loss": 0.2484,
+      "step": 2667
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5536739834615742e-05,
+      "loss": 0.3408,
+      "step": 2668
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.552756861273858e-05,
+      "loss": 0.2604,
+      "step": 2669
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5518397193392945e-05,
+      "loss": 0.2538,
+      "step": 2670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550922558001168e-05,
+      "loss": 0.2496,
+      "step": 2671
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5500053776027716e-05,
+      "loss": 0.2769,
+      "step": 2672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549088178487405e-05,
+      "loss": 0.3563,
+      "step": 2673
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548170960998375e-05,
+      "loss": 0.2378,
+      "step": 2674
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547253725478996e-05,
+      "loss": 0.2703,
+      "step": 2675
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546336472272587e-05,
+      "loss": 0.224,
+      "step": 2676
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5454192017224767e-05,
+      "loss": 0.246,
+      "step": 2677
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5445019141719977e-05,
+      "loss": 0.2621,
+      "step": 2678
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5435846099644903e-05,
+      "loss": 0.2513,
+      "step": 2679
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542667289443301e-05,
+      "loss": 0.2532,
+      "step": 2680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5417499529517812e-05,
+      "loss": 0.2593,
+      "step": 2681
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5408326008332897e-05,
+      "loss": 0.254,
+      "step": 2682
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5399152334311905e-05,
+      "loss": 0.275,
+      "step": 2683
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5389978510888542e-05,
+      "loss": 0.2336,
+      "step": 2684
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.538080454149655e-05,
+      "loss": 0.2269,
+      "step": 2685
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.537163042956974e-05,
+      "loss": 0.2247,
+      "step": 2686
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5362456178541986e-05,
+      "loss": 0.2179,
+      "step": 2687
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535328179184719e-05,
+      "loss": 0.2555,
+      "step": 2688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534410727291932e-05,
+      "loss": 0.2179,
+      "step": 2689
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533493262519239e-05,
+      "loss": 0.2922,
+      "step": 2690
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532575785210047e-05,
+      "loss": 0.2789,
+      "step": 2691
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5316582957077664e-05,
+      "loss": 0.2975,
+      "step": 2692
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5307407943558123e-05,
+      "loss": 0.2927,
+      "step": 2693
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5298232814976053e-05,
+      "loss": 0.2999,
+      "step": 2694
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5289057574765697e-05,
+      "loss": 0.2704,
+      "step": 2695
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.527988222636134e-05,
+      "loss": 0.2535,
+      "step": 2696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5270706773197298e-05,
+      "loss": 0.3085,
+      "step": 2697
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5261531218707943e-05,
+      "loss": 0.3138,
+      "step": 2698
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5252355566327677e-05,
+      "loss": 0.2787,
+      "step": 2699
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5243179819490945e-05,
+      "loss": 0.2207,
+      "step": 2700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5234003981632202e-05,
+      "loss": 0.2815,
+      "step": 2701
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.522482805618598e-05,
+      "loss": 0.2446,
+      "step": 2702
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5215652046586801e-05,
+      "loss": 0.2415,
+      "step": 2703
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5206475956269251e-05,
+      "loss": 0.2628,
+      "step": 2704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5197299788667922e-05,
+      "loss": 0.3044,
+      "step": 2705
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5188123547217452e-05,
+      "loss": 0.271,
+      "step": 2706
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5178947235352502e-05,
+      "loss": 0.2629,
+      "step": 2707
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.516977085650775e-05,
+      "loss": 0.3361,
+      "step": 2708
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5160594414117913e-05,
+      "loss": 0.3047,
+      "step": 2709
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5151417911617726e-05,
+      "loss": 0.3765,
+      "step": 2710
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5142241352441943e-05,
+      "loss": 0.2834,
+      "step": 2711
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5133064740025343e-05,
+      "loss": 0.2589,
+      "step": 2712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5123888077802724e-05,
+      "loss": 0.3193,
+      "step": 2713
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5114711369208907e-05,
+      "loss": 0.2804,
+      "step": 2714
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5105534617678715e-05,
+      "loss": 0.3523,
+      "step": 2715
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.509635782664701e-05,
+      "loss": 0.2897,
+      "step": 2716
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5087180999548653e-05,
+      "loss": 0.291,
+      "step": 2717
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5078004139818524e-05,
+      "loss": 0.3036,
+      "step": 2718
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5068827250891509e-05,
+      "loss": 0.2383,
+      "step": 2719
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5059650336202514e-05,
+      "loss": 0.3223,
+      "step": 2720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5050473399186447e-05,
+      "loss": 0.2814,
+      "step": 2721
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5041296443278234e-05,
+      "loss": 0.317,
+      "step": 2722
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5032119471912797e-05,
+      "loss": 0.3099,
+      "step": 2723
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5022942488525066e-05,
+      "loss": 0.3273,
+      "step": 2724
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5013765496549983e-05,
+      "loss": 0.1783,
+      "step": 2725
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5004588499422487e-05,
+      "loss": 0.2472,
+      "step": 2726
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4995411500577517e-05,
+      "loss": 0.3279,
+      "step": 2727
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.498623450345002e-05,
+      "loss": 0.2758,
+      "step": 2728
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4977057511474938e-05,
+      "loss": 0.2812,
+      "step": 2729
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4967880528087207e-05,
+      "loss": 0.28,
+      "step": 2730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4958703556721772e-05,
+      "loss": 0.2805,
+      "step": 2731
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4949526600813554e-05,
+      "loss": 0.2495,
+      "step": 2732
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.494034966379749e-05,
+      "loss": 0.2057,
+      "step": 2733
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4931172749108493e-05,
+      "loss": 0.2377,
+      "step": 2734
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4921995860181482e-05,
+      "loss": 0.2588,
+      "step": 2735
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4912819000451348e-05,
+      "loss": 0.2628,
+      "step": 2736
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4903642173352991e-05,
+      "loss": 0.2853,
+      "step": 2737
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4894465382321288e-05,
+      "loss": 0.2178,
+      "step": 2738
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4885288630791101e-05,
+      "loss": 0.2805,
+      "step": 2739
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4876111922197282e-05,
+      "loss": 0.2871,
+      "step": 2740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.486693525997466e-05,
+      "loss": 0.2683,
+      "step": 2741
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4857758647558059e-05,
+      "loss": 0.2396,
+      "step": 2742
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4848582088382279e-05,
+      "loss": 0.24,
+      "step": 2743
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4839405585882092e-05,
+      "loss": 0.2419,
+      "step": 2744
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4830229143492252e-05,
+      "loss": 0.218,
+      "step": 2745
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4821052764647502e-05,
+      "loss": 0.2716,
+      "step": 2746
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4811876452782552e-05,
+      "loss": 0.2827,
+      "step": 2747
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4802700211332084e-05,
+      "loss": 0.3588,
+      "step": 2748
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4793524043730756e-05,
+      "loss": 0.2711,
+      "step": 2749
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4784347953413198e-05,
+      "loss": 0.2648,
+      "step": 2750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4775171943814025e-05,
+      "loss": 0.2769,
+      "step": 2751
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.47659960183678e-05,
+      "loss": 0.2806,
+      "step": 2752
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4756820180509062e-05,
+      "loss": 0.2257,
+      "step": 2753
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.474764443367232e-05,
+      "loss": 0.2931,
+      "step": 2754
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.473846878129206e-05,
+      "loss": 0.4055,
+      "step": 2755
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4729293226802705e-05,
+      "loss": 0.2854,
+      "step": 2756
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4720117773638667e-05,
+      "loss": 0.3069,
+      "step": 2757
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4710942425234309e-05,
+      "loss": 0.2021,
+      "step": 2758
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4701767185023948e-05,
+      "loss": 0.3497,
+      "step": 2759
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4692592056441876e-05,
+      "loss": 0.2411,
+      "step": 2760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4683417042922341e-05,
+      "loss": 0.3361,
+      "step": 2761
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4674242147899534e-05,
+      "loss": 0.2816,
+      "step": 2762
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4665067374807606e-05,
+      "loss": 0.2968,
+      "step": 2763
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.465589272708068e-05,
+      "loss": 0.2808,
+      "step": 2764
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4646718208152813e-05,
+      "loss": 0.2123,
+      "step": 2765
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.463754382145802e-05,
+      "loss": 0.2404,
+      "step": 2766
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.462836957043026e-05,
+      "loss": 0.244,
+      "step": 2767
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.461919545850345e-05,
+      "loss": 0.2919,
+      "step": 2768
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4610021489111462e-05,
+      "loss": 0.3522,
+      "step": 2769
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4600847665688097e-05,
+      "loss": 0.2217,
+      "step": 2770
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4591673991667107e-05,
+      "loss": 0.3086,
+      "step": 2771
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4582500470482189e-05,
+      "loss": 0.2135,
+      "step": 2772
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4573327105566994e-05,
+      "loss": 0.2517,
+      "step": 2773
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.45641539003551e-05,
+      "loss": 0.2851,
+      "step": 2774
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4554980858280026e-05,
+      "loss": 0.2813,
+      "step": 2775
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4545807982775239e-05,
+      "loss": 0.3162,
+      "step": 2776
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.453663527727413e-05,
+      "loss": 0.2491,
+      "step": 2777
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4527462745210046e-05,
+      "loss": 0.2737,
+      "step": 2778
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4518290390016252e-05,
+      "loss": 0.267,
+      "step": 2779
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4509118215125954e-05,
+      "loss": 0.196,
+      "step": 2780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4499946223972285e-05,
+      "loss": 0.207,
+      "step": 2781
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4490774419988321e-05,
+      "loss": 0.2771,
+      "step": 2782
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4481602806607056e-05,
+      "loss": 0.3383,
+      "step": 2783
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4472431387261419e-05,
+      "loss": 0.3041,
+      "step": 2784
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.446326016538426e-05,
+      "loss": 0.2532,
+      "step": 2785
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4454089144408365e-05,
+      "loss": 0.2965,
+      "step": 2786
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4444918327766433e-05,
+      "loss": 0.2704,
+      "step": 2787
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4435747718891104e-05,
+      "loss": 0.2383,
+      "step": 2788
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4426577321214916e-05,
+      "loss": 0.3171,
+      "step": 2789
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4417407138170343e-05,
+      "loss": 0.2848,
+      "step": 2790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.440823717318979e-05,
+      "loss": 0.2616,
+      "step": 2791
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4399067429705556e-05,
+      "loss": 0.3011,
+      "step": 2792
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4389897911149866e-05,
+      "loss": 0.2329,
+      "step": 2793
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.438072862095487e-05,
+      "loss": 0.287,
+      "step": 2794
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4371559562552616e-05,
+      "loss": 0.2627,
+      "step": 2795
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4362390739375088e-05,
+      "loss": 0.2657,
+      "step": 2796
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4353222154854157e-05,
+      "loss": 0.3281,
+      "step": 2797
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4344053812421624e-05,
+      "loss": 0.2426,
+      "step": 2798
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4334885715509184e-05,
+      "loss": 0.2618,
+      "step": 2799
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4325717867548459e-05,
+      "loss": 0.2979,
+      "step": 2800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4316550271970956e-05,
+      "loss": 0.2582,
+      "step": 2801
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4307382932208106e-05,
+      "loss": 0.3027,
+      "step": 2802
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4298215851691233e-05,
+      "loss": 0.3269,
+      "step": 2803
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4289049033851559e-05,
+      "loss": 0.3247,
+      "step": 2804
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.427988248212023e-05,
+      "loss": 0.2016,
+      "step": 2805
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4270716199928274e-05,
+      "loss": 0.2524,
+      "step": 2806
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4261550190706623e-05,
+      "loss": 0.2593,
+      "step": 2807
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4252384457886098e-05,
+      "loss": 0.329,
+      "step": 2808
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4243219004897434e-05,
+      "loss": 0.2806,
+      "step": 2809
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4234053835171253e-05,
+      "loss": 0.2789,
+      "step": 2810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.422488895213807e-05,
+      "loss": 0.2781,
+      "step": 2811
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4215724359228285e-05,
+      "loss": 0.2588,
+      "step": 2812
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4206560059872198e-05,
+      "loss": 0.29,
+      "step": 2813
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4197396057500011e-05,
+      "loss": 0.2599,
+      "step": 2814
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.418823235554179e-05,
+      "loss": 0.2714,
+      "step": 2815
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4179068957427509e-05,
+      "loss": 0.2443,
+      "step": 2816
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.416990586658701e-05,
+      "loss": 0.3203,
+      "step": 2817
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.416074308645004e-05,
+      "loss": 0.217,
+      "step": 2818
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4151580620446215e-05,
+      "loss": 0.3026,
+      "step": 2819
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.414241847200504e-05,
+      "loss": 0.3212,
+      "step": 2820
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.41332566445559e-05,
+      "loss": 0.2997,
+      "step": 2821
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.412409514152805e-05,
+      "loss": 0.293,
+      "step": 2822
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4114933966350646e-05,
+      "loss": 0.2836,
+      "step": 2823
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4105773122452703e-05,
+      "loss": 0.2789,
+      "step": 2824
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4096612613263119e-05,
+      "loss": 0.2837,
+      "step": 2825
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4087452442210652e-05,
+      "loss": 0.2776,
+      "step": 2826
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4078292612723961e-05,
+      "loss": 0.2731,
+      "step": 2827
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4069133128231563e-05,
+      "loss": 0.3473,
+      "step": 2828
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4059973992161838e-05,
+      "loss": 0.2911,
+      "step": 2829
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4050815207943042e-05,
+      "loss": 0.2842,
+      "step": 2830
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4041656779003299e-05,
+      "loss": 0.2894,
+      "step": 2831
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4032498708770614e-05,
+      "loss": 0.2871,
+      "step": 2832
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4023341000672833e-05,
+      "loss": 0.2955,
+      "step": 2833
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4014183658137679e-05,
+      "loss": 0.2576,
+      "step": 2834
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4005026684592736e-05,
+      "loss": 0.246,
+      "step": 2835
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3995870083465462e-05,
+      "loss": 0.2742,
+      "step": 2836
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3986713858183159e-05,
+      "loss": 0.2609,
+      "step": 2837
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3977558012172988e-05,
+      "loss": 0.2499,
+      "step": 2838
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3968402548861983e-05,
+      "loss": 0.3009,
+      "step": 2839
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3959247471677017e-05,
+      "loss": 0.373,
+      "step": 2840
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.395009278404484e-05,
+      "loss": 0.2916,
+      "step": 2841
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3940938489392033e-05,
+      "loss": 0.2554,
+      "step": 2842
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.393178459114505e-05,
+      "loss": 0.2302,
+      "step": 2843
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3922631092730173e-05,
+      "loss": 0.2889,
+      "step": 2844
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3913477997573565e-05,
+      "loss": 0.2245,
+      "step": 2845
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3904325309101213e-05,
+      "loss": 0.28,
+      "step": 2846
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3895173030738963e-05,
+      "loss": 0.2389,
+      "step": 2847
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.38860211659125e-05,
+      "loss": 0.2849,
+      "step": 2848
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3876869718047362e-05,
+      "loss": 0.2671,
+      "step": 2849
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3867718690568935e-05,
+      "loss": 0.2528,
+      "step": 2850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3858568086902433e-05,
+      "loss": 0.2985,
+      "step": 2851
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3849417910472921e-05,
+      "loss": 0.3123,
+      "step": 2852
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3840268164705297e-05,
+      "loss": 0.2677,
+      "step": 2853
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3831118853024314e-05,
+      "loss": 0.291,
+      "step": 2854
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.382196997885455e-05,
+      "loss": 0.2538,
+      "step": 2855
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3812821545620408e-05,
+      "loss": 0.2983,
+      "step": 2856
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3803673556746153e-05,
+      "loss": 0.2491,
+      "step": 2857
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3794526015655858e-05,
+      "loss": 0.2928,
+      "step": 2858
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3785378925773452e-05,
+      "loss": 0.2825,
+      "step": 2859
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3776232290522673e-05,
+      "loss": 0.2949,
+      "step": 2860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3767086113327104e-05,
+      "loss": 0.2144,
+      "step": 2861
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3757940397610144e-05,
+      "loss": 0.2328,
+      "step": 2862
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3748795146795039e-05,
+      "loss": 0.2893,
+      "step": 2863
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3739650364304839e-05,
+      "loss": 0.2854,
+      "step": 2864
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3730506053562433e-05,
+      "loss": 0.2579,
+      "step": 2865
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3721362217990527e-05,
+      "loss": 0.2451,
+      "step": 2866
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3712218861011644e-05,
+      "loss": 0.301,
+      "step": 2867
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3703075986048145e-05,
+      "loss": 0.3051,
+      "step": 2868
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.36939335965222e-05,
+      "loss": 0.3164,
+      "step": 2869
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3684791695855795e-05,
+      "loss": 0.3329,
+      "step": 2870
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.367565028747073e-05,
+      "loss": 0.2532,
+      "step": 2871
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3666509374788628e-05,
+      "loss": 0.2926,
+      "step": 2872
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3657368961230935e-05,
+      "loss": 0.2712,
+      "step": 2873
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.364822905021889e-05,
+      "loss": 0.3262,
+      "step": 2874
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.363908964517356e-05,
+      "loss": 0.2677,
+      "step": 2875
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3629950749515806e-05,
+      "loss": 0.2511,
+      "step": 2876
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3620812366666322e-05,
+      "loss": 0.2955,
+      "step": 2877
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3611674500045588e-05,
+      "loss": 0.3456,
+      "step": 2878
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3602537153073907e-05,
+      "loss": 0.2676,
+      "step": 2879
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3593400329171377e-05,
+      "loss": 0.2148,
+      "step": 2880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3584264031757894e-05,
+      "loss": 0.2809,
+      "step": 2881
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3575128264253178e-05,
+      "loss": 0.3134,
+      "step": 2882
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3565993030076738e-05,
+      "loss": 0.3068,
+      "step": 2883
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3556858332647884e-05,
+      "loss": 0.2722,
+      "step": 2884
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3547724175385715e-05,
+      "loss": 0.2576,
+      "step": 2885
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.353859056170915e-05,
+      "loss": 0.2487,
+      "step": 2886
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3529457495036894e-05,
+      "loss": 0.3201,
+      "step": 2887
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3520324978787439e-05,
+      "loss": 0.3052,
+      "step": 2888
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3511193016379079e-05,
+      "loss": 0.222,
+      "step": 2889
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3502061611229897e-05,
+      "loss": 0.2469,
+      "step": 2890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3492930766757783e-05,
+      "loss": 0.3262,
+      "step": 2891
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3483800486380394e-05,
+      "loss": 0.3359,
+      "step": 2892
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3474670773515185e-05,
+      "loss": 0.2963,
+      "step": 2893
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3465541631579398e-05,
+      "loss": 0.2603,
+      "step": 2894
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3456413063990074e-05,
+      "loss": 0.2614,
+      "step": 2895
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3447285074164023e-05,
+      "loss": 0.2623,
+      "step": 2896
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3438157665517839e-05,
+      "loss": 0.2328,
+      "step": 2897
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.342903084146791e-05,
+      "loss": 0.2562,
+      "step": 2898
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3419904605430388e-05,
+      "loss": 0.2867,
+      "step": 2899
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3410778960821228e-05,
+      "loss": 0.288,
+      "step": 2900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3401653911056143e-05,
+      "loss": 0.2723,
+      "step": 2901
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3392529459550638e-05,
+      "loss": 0.2852,
+      "step": 2902
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3383405609719977e-05,
+      "loss": 0.288,
+      "step": 2903
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.337428236497922e-05,
+      "loss": 0.2987,
+      "step": 2904
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3365159728743183e-05,
+      "loss": 0.2988,
+      "step": 2905
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3356037704426465e-05,
+      "loss": 0.2346,
+      "step": 2906
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3346916295443428e-05,
+      "loss": 0.2785,
+      "step": 2907
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3337795505208205e-05,
+      "loss": 0.2761,
+      "step": 2908
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3328675337134712e-05,
+      "loss": 0.266,
+      "step": 2909
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3319555794636613e-05,
+      "loss": 0.2545,
+      "step": 2910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3310436881127342e-05,
+      "loss": 0.2758,
+      "step": 2911
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3301318600020096e-05,
+      "loss": 0.3169,
+      "step": 2912
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3292200954727857e-05,
+      "loss": 0.2428,
+      "step": 2913
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.328308394866334e-05,
+      "loss": 0.2603,
+      "step": 2914
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3273967585239034e-05,
+      "loss": 0.3079,
+      "step": 2915
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3264851867867186e-05,
+      "loss": 0.2531,
+      "step": 2916
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3255736799959799e-05,
+      "loss": 0.2762,
+      "step": 2917
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3246622384928642e-05,
+      "loss": 0.2526,
+      "step": 2918
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3237508626185224e-05,
+      "loss": 0.2415,
+      "step": 2919
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3228395527140828e-05,
+      "loss": 0.267,
+      "step": 2920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3219283091206464e-05,
+      "loss": 0.2452,
+      "step": 2921
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3210171321792924e-05,
+      "loss": 0.2935,
+      "step": 2922
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3201060222310728e-05,
+      "loss": 0.2905,
+      "step": 2923
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3191949796170156e-05,
+      "loss": 0.2538,
+      "step": 2924
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3182840046781233e-05,
+      "loss": 0.374,
+      "step": 2925
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.317373097755372e-05,
+      "loss": 0.2862,
+      "step": 2926
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3164622591897145e-05,
+      "loss": 0.2834,
+      "step": 2927
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3155514893220771e-05,
+      "loss": 0.2522,
+      "step": 2928
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3146407884933595e-05,
+      "loss": 0.2822,
+      "step": 2929
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3137301570444357e-05,
+      "loss": 0.2665,
+      "step": 2930
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3128195953161555e-05,
+      "loss": 0.2871,
+      "step": 2931
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3119091036493411e-05,
+      "loss": 0.2799,
+      "step": 2932
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3109986823847885e-05,
+      "loss": 0.3551,
+      "step": 2933
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3100883318632677e-05,
+      "loss": 0.2898,
+      "step": 2934
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3091780524255214e-05,
+      "loss": 0.2648,
+      "step": 2935
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3082678444122677e-05,
+      "loss": 0.2922,
+      "step": 2936
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3073577081641958e-05,
+      "loss": 0.3223,
+      "step": 2937
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3064476440219695e-05,
+      "loss": 0.2714,
+      "step": 2938
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.305537652326224e-05,
+      "loss": 0.2755,
+      "step": 2939
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.30462773341757e-05,
+      "loss": 0.2744,
+      "step": 2940
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3037178876365879e-05,
+      "loss": 0.2656,
+      "step": 2941
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3028081153238333e-05,
+      "loss": 0.2988,
+      "step": 2942
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3018984168198326e-05,
+      "loss": 0.2936,
+      "step": 2943
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3009887924650845e-05,
+      "loss": 0.2152,
+      "step": 2944
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3000792426000619e-05,
+      "loss": 0.3053,
+      "step": 2945
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2991697675652083e-05,
+      "loss": 0.2344,
+      "step": 2946
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.298260367700939e-05,
+      "loss": 0.2747,
+      "step": 2947
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2973510433476413e-05,
+      "loss": 0.3009,
+      "step": 2948
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2964417948456751e-05,
+      "loss": 0.2208,
+      "step": 2949
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2955326225353716e-05,
+      "loss": 0.3042,
+      "step": 2950
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.294623526757033e-05,
+      "loss": 0.2844,
+      "step": 2951
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2937145078509324e-05,
+      "loss": 0.3138,
+      "step": 2952
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.292805566157315e-05,
+      "loss": 0.3113,
+      "step": 2953
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2918967020163978e-05,
+      "loss": 0.3311,
+      "step": 2954
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2909879157683673e-05,
+      "loss": 0.2546,
+      "step": 2955
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2900792077533811e-05,
+      "loss": 0.2616,
+      "step": 2956
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2891705783115678e-05,
+      "loss": 0.2442,
+      "step": 2957
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2882620277830275e-05,
+      "loss": 0.2614,
+      "step": 2958
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2873535565078294e-05,
+      "loss": 0.2393,
+      "step": 2959
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.286445164826013e-05,
+      "loss": 0.3051,
+      "step": 2960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2855368530775895e-05,
+      "loss": 0.3308,
+      "step": 2961
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.284628621602538e-05,
+      "loss": 0.3231,
+      "step": 2962
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2837204707408096e-05,
+      "loss": 0.3027,
+      "step": 2963
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2828124008323248e-05,
+      "loss": 0.3076,
+      "step": 2964
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2819044122169728e-05,
+      "loss": 0.2268,
+      "step": 2965
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2809965052346123e-05,
+      "loss": 0.264,
+      "step": 2966
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2800886802250732e-05,
+      "loss": 0.2594,
+      "step": 2967
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2791809375281535e-05,
+      "loss": 0.2996,
+      "step": 2968
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2782732774836202e-05,
+      "loss": 0.2935,
+      "step": 2969
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2773657004312096e-05,
+      "loss": 0.2839,
+      "step": 2970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2764582067106267e-05,
+      "loss": 0.2942,
+      "step": 2971
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2755507966615469e-05,
+      "loss": 0.2744,
+      "step": 2972
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.274643470623612e-05,
+      "loss": 0.3533,
+      "step": 2973
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2737362289364331e-05,
+      "loss": 0.2673,
+      "step": 2974
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2728290719395903e-05,
+      "loss": 0.3182,
+      "step": 2975
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2719219999726323e-05,
+      "loss": 0.2673,
+      "step": 2976
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2710150133750752e-05,
+      "loss": 0.318,
+      "step": 2977
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2701081124864023e-05,
+      "loss": 0.2093,
+      "step": 2978
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.269201297646067e-05,
+      "loss": 0.2415,
+      "step": 2979
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2682945691934882e-05,
+      "loss": 0.3037,
+      "step": 2980
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2673879274680549e-05,
+      "loss": 0.287,
+      "step": 2981
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2664813728091214e-05,
+      "loss": 0.2475,
+      "step": 2982
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2655749055560107e-05,
+      "loss": 0.2421,
+      "step": 2983
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.264668526048012e-05,
+      "loss": 0.2347,
+      "step": 2984
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2637622346243837e-05,
+      "loss": 0.2815,
+      "step": 2985
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2628560316243488e-05,
+      "loss": 0.2305,
+      "step": 2986
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.261949917387099e-05,
+      "loss": 0.2226,
+      "step": 2987
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2610438922517916e-05,
+      "loss": 0.265,
+      "step": 2988
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.260137956557551e-05,
+      "loss": 0.3068,
+      "step": 2989
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2592321106434684e-05,
+      "loss": 0.2869,
+      "step": 2990
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2583263548486016e-05,
+      "loss": 0.2363,
+      "step": 2991
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2574206895119735e-05,
+      "loss": 0.3085,
+      "step": 2992
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2565151149725737e-05,
+      "loss": 0.246,
+      "step": 2993
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2556096315693586e-05,
+      "loss": 0.2917,
+      "step": 2994
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.25470423964125e-05,
+      "loss": 0.2742,
+      "step": 2995
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2537989395271346e-05,
+      "loss": 0.2878,
+      "step": 2996
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2528937315658662e-05,
+      "loss": 0.3152,
+      "step": 2997
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2519886160962623e-05,
+      "loss": 0.2758,
+      "step": 2998
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2510835934571081e-05,
+      "loss": 0.2893,
+      "step": 2999
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.250178663987152e-05,
+      "loss": 0.2838,
+      "step": 3000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.249273828025109e-05,
+      "loss": 0.3179,
+      "step": 3001
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2483690859096571e-05,
+      "loss": 0.2784,
+      "step": 3002
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2474644379794421e-05,
+      "loss": 0.2716,
+      "step": 3003
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2465598845730721e-05,
+      "loss": 0.249,
+      "step": 3004
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2456554260291212e-05,
+      "loss": 0.2679,
+      "step": 3005
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2447510626861267e-05,
+      "loss": 0.2139,
+      "step": 3006
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2438467948825912e-05,
+      "loss": 0.2477,
+      "step": 3007
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2429426229569816e-05,
+      "loss": 0.3231,
+      "step": 3008
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.242038547247729e-05,
+      "loss": 0.3115,
+      "step": 3009
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2411345680932277e-05,
+      "loss": 0.256,
+      "step": 3010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2402306858318361e-05,
+      "loss": 0.2699,
+      "step": 3011
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2393269008018768e-05,
+      "loss": 0.2457,
+      "step": 3012
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2384232133416364e-05,
+      "loss": 0.2736,
+      "step": 3013
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2375196237893637e-05,
+      "loss": 0.2794,
+      "step": 3014
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2366161324832716e-05,
+      "loss": 0.2365,
+      "step": 3015
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2357127397615356e-05,
+      "loss": 0.2603,
+      "step": 3016
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2348094459622959e-05,
+      "loss": 0.2687,
+      "step": 3017
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2339062514236541e-05,
+      "loss": 0.2699,
+      "step": 3018
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2330031564836749e-05,
+      "loss": 0.236,
+      "step": 3019
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2321001614803863e-05,
+      "loss": 0.2697,
+      "step": 3020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2311972667517779e-05,
+      "loss": 0.2114,
+      "step": 3021
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2302944726358031e-05,
+      "loss": 0.243,
+      "step": 3022
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2293917794703771e-05,
+      "loss": 0.3052,
+      "step": 3023
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2284891875933773e-05,
+      "loss": 0.297,
+      "step": 3024
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2275866973426415e-05,
+      "loss": 0.2487,
+      "step": 3025
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2266843090559728e-05,
+      "loss": 0.2559,
+      "step": 3026
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.225782023071134e-05,
+      "loss": 0.2675,
+      "step": 3027
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2248798397258498e-05,
+      "loss": 0.3206,
+      "step": 3028
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2239777593578061e-05,
+      "loss": 0.2552,
+      "step": 3029
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2230757823046511e-05,
+      "loss": 0.303,
+      "step": 3030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2221739089039948e-05,
+      "loss": 0.2551,
+      "step": 3031
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.221272139493407e-05,
+      "loss": 0.2827,
+      "step": 3032
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2203704744104191e-05,
+      "loss": 0.2453,
+      "step": 3033
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2194689139925235e-05,
+      "loss": 0.2565,
+      "step": 3034
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2185674585771742e-05,
+      "loss": 0.2628,
+      "step": 3035
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2176661085017847e-05,
+      "loss": 0.2343,
+      "step": 3036
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2167648641037296e-05,
+      "loss": 0.2802,
+      "step": 3037
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.215863725720344e-05,
+      "loss": 0.2333,
+      "step": 3038
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2149626936889225e-05,
+      "loss": 0.2128,
+      "step": 3039
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2140617683467221e-05,
+      "loss": 0.2153,
+      "step": 3040
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.213160950030957e-05,
+      "loss": 0.2642,
+      "step": 3041
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2122602390788039e-05,
+      "loss": 0.246,
+      "step": 3042
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2113596358273967e-05,
+      "loss": 0.2917,
+      "step": 3043
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.210459140613832e-05,
+      "loss": 0.2706,
+      "step": 3044
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.209558753775163e-05,
+      "loss": 0.2804,
+      "step": 3045
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.208658475648405e-05,
+      "loss": 0.2686,
+      "step": 3046
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2077583065705304e-05,
+      "loss": 0.2563,
+      "step": 3047
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2068582468784719e-05,
+      "loss": 0.2152,
+      "step": 3048
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.205958296909121e-05,
+      "loss": 0.2729,
+      "step": 3049
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2050584569993289e-05,
+      "loss": 0.3092,
+      "step": 3050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2041587274859042e-05,
+      "loss": 0.3313,
+      "step": 3051
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2032591087056148e-05,
+      "loss": 0.2301,
+      "step": 3052
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2023596009951877e-05,
+      "loss": 0.3036,
+      "step": 3053
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.201460204691308e-05,
+      "loss": 0.2712,
+      "step": 3054
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2005609201306183e-05,
+      "loss": 0.2891,
+      "step": 3055
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1996617476497208e-05,
+      "loss": 0.2564,
+      "step": 3056
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1987626875851739e-05,
+      "loss": 0.3049,
+      "step": 3057
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1978637402734963e-05,
+      "loss": 0.2743,
+      "step": 3058
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1969649060511624e-05,
+      "loss": 0.3022,
+      "step": 3059
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1960661852546058e-05,
+      "loss": 0.2736,
+      "step": 3060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1951675782202156e-05,
+      "loss": 0.2455,
+      "step": 3061
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.194269085284341e-05,
+      "loss": 0.2384,
+      "step": 3062
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1933707067832865e-05,
+      "loss": 0.3085,
+      "step": 3063
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1924724430533147e-05,
+      "loss": 0.2797,
+      "step": 3064
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.191574294430645e-05,
+      "loss": 0.247,
+      "step": 3065
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1906762612514525e-05,
+      "loss": 0.3185,
+      "step": 3066
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1897783438518716e-05,
+      "loss": 0.297,
+      "step": 3067
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1888805425679919e-05,
+      "loss": 0.2814,
+      "step": 3068
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1879828577358595e-05,
+      "loss": 0.303,
+      "step": 3069
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.187085289691476e-05,
+      "loss": 0.2952,
+      "step": 3070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1861878387708019e-05,
+      "loss": 0.2865,
+      "step": 3071
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1852905053097519e-05,
+      "loss": 0.1815,
+      "step": 3072
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1843932896441968e-05,
+      "loss": 0.2748,
+      "step": 3073
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1834961921099637e-05,
+      "loss": 0.2238,
+      "step": 3074
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1825992130428349e-05,
+      "loss": 0.2327,
+      "step": 3075
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1817023527785504e-05,
+      "loss": 0.2936,
+      "step": 3076
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1808056116528033e-05,
+      "loss": 0.3041,
+      "step": 3077
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1799089900012427e-05,
+      "loss": 0.2485,
+      "step": 3078
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1790124881594736e-05,
+      "loss": 0.2462,
+      "step": 3079
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1781161064630564e-05,
+      "loss": 0.2709,
+      "step": 3080
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1772198452475056e-05,
+      "loss": 0.2887,
+      "step": 3081
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.176323704848291e-05,
+      "loss": 0.2284,
+      "step": 3082
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1754276856008374e-05,
+      "loss": 0.2534,
+      "step": 3083
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1745317878405229e-05,
+      "loss": 0.2944,
+      "step": 3084
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1736360119026827e-05,
+      "loss": 0.2654,
+      "step": 3085
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1727403581226048e-05,
+      "loss": 0.2678,
+      "step": 3086
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1718448268355312e-05,
+      "loss": 0.2694,
+      "step": 3087
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.170949418376658e-05,
+      "loss": 0.2448,
+      "step": 3088
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1700541330811367e-05,
+      "loss": 0.2151,
+      "step": 3089
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1691589712840719e-05,
+      "loss": 0.2748,
+      "step": 3090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1682639333205217e-05,
+      "loss": 0.2786,
+      "step": 3091
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1673690195254975e-05,
+      "loss": 0.3553,
+      "step": 3092
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1664742302339646e-05,
+      "loss": 0.2653,
+      "step": 3093
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1655795657808432e-05,
+      "loss": 0.2423,
+      "step": 3094
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1646850265010046e-05,
+      "loss": 0.275,
+      "step": 3095
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.163790612729274e-05,
+      "loss": 0.2734,
+      "step": 3096
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1628963248004293e-05,
+      "loss": 0.3275,
+      "step": 3097
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.162002163049203e-05,
+      "loss": 0.2321,
+      "step": 3098
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1611081278102785e-05,
+      "loss": 0.2451,
+      "step": 3099
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1602142194182918e-05,
+      "loss": 0.2776,
+      "step": 3100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.159320438207833e-05,
+      "loss": 0.26,
+      "step": 3101
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1584267845134423e-05,
+      "loss": 0.3186,
+      "step": 3102
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1575332586696151e-05,
+      "loss": 0.2123,
+      "step": 3103
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1566398610107963e-05,
+      "loss": 0.2269,
+      "step": 3104
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1557465918713847e-05,
+      "loss": 0.2511,
+      "step": 3105
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.154853451585729e-05,
+      "loss": 0.2781,
+      "step": 3106
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.153960440488132e-05,
+      "loss": 0.3166,
+      "step": 3107
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1530675589128463e-05,
+      "loss": 0.2543,
+      "step": 3108
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1521748071940773e-05,
+      "loss": 0.2936,
+      "step": 3109
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1512821856659805e-05,
+      "loss": 0.21,
+      "step": 3110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1503896946626638e-05,
+      "loss": 0.2693,
+      "step": 3111
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1494973345181857e-05,
+      "loss": 0.3451,
+      "step": 3112
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1486051055665561e-05,
+      "loss": 0.2847,
+      "step": 3113
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1477130081417352e-05,
+      "loss": 0.2684,
+      "step": 3114
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.146821042577634e-05,
+      "loss": 0.2546,
+      "step": 3115
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1459292092081155e-05,
+      "loss": 0.2697,
+      "step": 3116
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1450375083669916e-05,
+      "loss": 0.2474,
+      "step": 3117
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1441459403880246e-05,
+      "loss": 0.2675,
+      "step": 3118
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1432545056049286e-05,
+      "loss": 0.3303,
+      "step": 3119
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1423632043513656e-05,
+      "loss": 0.22,
+      "step": 3120
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1414720369609505e-05,
+      "loss": 0.2755,
+      "step": 3121
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1405810037672455e-05,
+      "loss": 0.3145,
+      "step": 3122
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1396901051037639e-05,
+      "loss": 0.2106,
+      "step": 3123
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1387993413039676e-05,
+      "loss": 0.2778,
+      "step": 3124
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.13790871270127e-05,
+      "loss": 0.265,
+      "step": 3125
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1370182196290315e-05,
+      "loss": 0.2163,
+      "step": 3126
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1361278624205637e-05,
+      "loss": 0.275,
+      "step": 3127
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1352376414091263e-05,
+      "loss": 0.2324,
+      "step": 3128
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1343475569279272e-05,
+      "loss": 0.2032,
+      "step": 3129
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1334576093101258e-05,
+      "loss": 0.2573,
+      "step": 3130
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1325677988888283e-05,
+      "loss": 0.2228,
+      "step": 3131
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1316781259970898e-05,
+      "loss": 0.268,
+      "step": 3132
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1307885909679134e-05,
+      "loss": 0.3149,
+      "step": 3133
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1298991941342526e-05,
+      "loss": 0.2966,
+      "step": 3134
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1290099358290074e-05,
+      "loss": 0.4143,
+      "step": 3135
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.128120816385026e-05,
+      "loss": 0.2255,
+      "step": 3136
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1272318361351058e-05,
+      "loss": 0.2747,
+      "step": 3137
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1263429954119901e-05,
+      "loss": 0.225,
+      "step": 3138
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1254542945483728e-05,
+      "loss": 0.2231,
+      "step": 3139
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1245657338768927e-05,
+      "loss": 0.278,
+      "step": 3140
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1236773137301381e-05,
+      "loss": 0.2736,
+      "step": 3141
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1227890344406428e-05,
+      "loss": 0.2673,
+      "step": 3142
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1219008963408904e-05,
+      "loss": 0.2224,
+      "step": 3143
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1210128997633092e-05,
+      "loss": 0.2115,
+      "step": 3144
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1201250450402762e-05,
+      "loss": 0.2063,
+      "step": 3145
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1192373325041142e-05,
+      "loss": 0.231,
+      "step": 3146
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1183497624870929e-05,
+      "loss": 0.2858,
+      "step": 3147
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1174623353214299e-05,
+      "loss": 0.3096,
+      "step": 3148
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.116575051339288e-05,
+      "loss": 0.3457,
+      "step": 3149
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.115687910872777e-05,
+      "loss": 0.2382,
+      "step": 3150
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1148009142539526e-05,
+      "loss": 0.2549,
+      "step": 3151
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1139140618148164e-05,
+      "loss": 0.3005,
+      "step": 3152
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.113027353887318e-05,
+      "loss": 0.2618,
+      "step": 3153
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.11214079080335e-05,
+      "loss": 0.2178,
+      "step": 3154
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1112543728947525e-05,
+      "loss": 0.3038,
+      "step": 3155
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1103681004933111e-05,
+      "loss": 0.2344,
+      "step": 3156
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1094819739307573e-05,
+      "loss": 0.2439,
+      "step": 3157
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.108595993538767e-05,
+      "loss": 0.3069,
+      "step": 3158
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1077101596489618e-05,
+      "loss": 0.3042,
+      "step": 3159
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.106824472592909e-05,
+      "loss": 0.2881,
+      "step": 3160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1059389327021197e-05,
+      "loss": 0.3735,
+      "step": 3161
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1050535403080517e-05,
+      "loss": 0.2501,
+      "step": 3162
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.104168295742106e-05,
+      "loss": 0.3158,
+      "step": 3163
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1032831993356293e-05,
+      "loss": 0.2881,
+      "step": 3164
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1023982514199119e-05,
+      "loss": 0.23,
+      "step": 3165
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1015134523261896e-05,
+      "loss": 0.2526,
+      "step": 3166
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1006288023856416e-05,
+      "loss": 0.2626,
+      "step": 3167
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.099744301929392e-05,
+      "loss": 0.2358,
+      "step": 3168
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0988599512885077e-05,
+      "loss": 0.2151,
+      "step": 3169
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0979757507940012e-05,
+      "loss": 0.2733,
+      "step": 3170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0970917007768282e-05,
+      "loss": 0.3217,
+      "step": 3171
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.096207801567887e-05,
+      "loss": 0.2275,
+      "step": 3172
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0953240534980209e-05,
+      "loss": 0.2332,
+      "step": 3173
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0944404568980155e-05,
+      "loss": 0.2875,
+      "step": 3174
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0935570120986009e-05,
+      "loss": 0.2843,
+      "step": 3175
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.09267371943045e-05,
+      "loss": 0.2692,
+      "step": 3176
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0917905792241772e-05,
+      "loss": 0.2419,
+      "step": 3177
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0909075918103425e-05,
+      "loss": 0.2919,
+      "step": 3178
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0900247575194462e-05,
+      "loss": 0.2616,
+      "step": 3179
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0891420766819334e-05,
+      "loss": 0.2529,
+      "step": 3180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0882595496281899e-05,
+      "loss": 0.3879,
+      "step": 3181
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0873771766885458e-05,
+      "loss": 0.2082,
+      "step": 3182
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0864949581932713e-05,
+      "loss": 0.2551,
+      "step": 3183
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0856128944725813e-05,
+      "loss": 0.2582,
+      "step": 3184
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0847309858566307e-05,
+      "loss": 0.3345,
+      "step": 3185
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0838492326755176e-05,
+      "loss": 0.2492,
+      "step": 3186
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0829676352592813e-05,
+      "loss": 0.263,
+      "step": 3187
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0820861939379022e-05,
+      "loss": 0.3114,
+      "step": 3188
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.081204909041304e-05,
+      "loss": 0.3201,
+      "step": 3189
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.080323780899351e-05,
+      "loss": 0.2704,
+      "step": 3190
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0794428098418482e-05,
+      "loss": 0.2421,
+      "step": 3191
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0785619961985419e-05,
+      "loss": 0.2092,
+      "step": 3192
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0776813402991206e-05,
+      "loss": 0.241,
+      "step": 3193
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0768008424732133e-05,
+      "loss": 0.3357,
+      "step": 3194
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.075920503050389e-05,
+      "loss": 0.3063,
+      "step": 3195
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.075040322360158e-05,
+      "loss": 0.286,
+      "step": 3196
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.074160300731971e-05,
+      "loss": 0.2424,
+      "step": 3197
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0732804384952205e-05,
+      "loss": 0.2683,
+      "step": 3198
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0724007359792367e-05,
+      "loss": 0.2544,
+      "step": 3199
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0715211935132928e-05,
+      "loss": 0.2453,
+      "step": 3200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0706418114265994e-05,
+      "loss": 0.2553,
+      "step": 3201
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0697625900483097e-05,
+      "loss": 0.2043,
+      "step": 3202
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0688835297075147e-05,
+      "loss": 0.2651,
+      "step": 3203
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0680046307332468e-05,
+      "loss": 0.4464,
+      "step": 3204
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0671258934544763e-05,
+      "loss": 0.2974,
+      "step": 3205
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0662473182001137e-05,
+      "loss": 0.3275,
+      "step": 3206
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0653689052990097e-05,
+      "loss": 0.2348,
+      "step": 3207
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0644906550799535e-05,
+      "loss": 0.2711,
+      "step": 3208
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.063612567871673e-05,
+      "loss": 0.282,
+      "step": 3209
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.062734644002835e-05,
+      "loss": 0.2717,
+      "step": 3210
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0618568838020464e-05,
+      "loss": 0.3147,
+      "step": 3211
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0609792875978524e-05,
+      "loss": 0.2573,
+      "step": 3212
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.060101855718736e-05,
+      "loss": 0.2736,
+      "step": 3213
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0592245884931188e-05,
+      "loss": 0.3229,
+      "step": 3214
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0583474862493614e-05,
+      "loss": 0.2361,
+      "step": 3215
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0574705493157635e-05,
+      "loss": 0.296,
+      "step": 3216
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0565937780205606e-05,
+      "loss": 0.2345,
+      "step": 3217
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0557171726919276e-05,
+      "loss": 0.2826,
+      "step": 3218
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0548407336579771e-05,
+      "loss": 0.2491,
+      "step": 3219
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0539644612467603e-05,
+      "loss": 0.2754,
+      "step": 3220
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0530883557862648e-05,
+      "loss": 0.3126,
+      "step": 3221
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0522124176044154e-05,
+      "loss": 0.2016,
+      "step": 3222
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.051336647029076e-05,
+      "loss": 0.307,
+      "step": 3223
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0504610443880457e-05,
+      "loss": 0.2582,
+      "step": 3224
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0495856100090631e-05,
+      "loss": 0.2618,
+      "step": 3225
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0487103442198019e-05,
+      "loss": 0.1992,
+      "step": 3226
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0478352473478732e-05,
+      "loss": 0.2822,
+      "step": 3227
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0469603197208249e-05,
+      "loss": 0.2546,
+      "step": 3228
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0460855616661423e-05,
+      "loss": 0.2626,
+      "step": 3229
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0452109735112462e-05,
+      "loss": 0.2302,
+      "step": 3230
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0443365555834943e-05,
+      "loss": 0.2354,
+      "step": 3231
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0434623082101802e-05,
+      "loss": 0.2462,
+      "step": 3232
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0425882317185338e-05,
+      "loss": 0.2894,
+      "step": 3233
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0417143264357223e-05,
+      "loss": 0.2554,
+      "step": 3234
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0408405926888468e-05,
+      "loss": 0.2542,
+      "step": 3235
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.039967030804945e-05,
+      "loss": 0.2664,
+      "step": 3236
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0390936411109902e-05,
+      "loss": 0.2224,
+      "step": 3237
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0382204239338923e-05,
+      "loss": 0.2666,
+      "step": 3238
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0373473796004956e-05,
+      "loss": 0.2691,
+      "step": 3239
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.036474508437579e-05,
+      "loss": 0.2531,
+      "step": 3240
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0356018107718584e-05,
+      "loss": 0.2368,
+      "step": 3241
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0347292869299825e-05,
+      "loss": 0.2507,
+      "step": 3242
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0338569372385377e-05,
+      "loss": 0.3396,
+      "step": 3243
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0329847620240429e-05,
+      "loss": 0.3094,
+      "step": 3244
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.032112761612953e-05,
+      "loss": 0.2661,
+      "step": 3245
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0312409363316562e-05,
+      "loss": 0.2797,
+      "step": 3246
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.030369286506477e-05,
+      "loss": 0.2667,
+      "step": 3247
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0294978124636724e-05,
+      "loss": 0.2821,
+      "step": 3248
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0286265145294351e-05,
+      "loss": 0.2434,
+      "step": 3249
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0277553930298908e-05,
+      "loss": 0.3042,
+      "step": 3250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0268844482910987e-05,
+      "loss": 0.2795,
+      "step": 3251
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0260136806390538e-05,
+      "loss": 0.2941,
+      "step": 3252
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0251430903996838e-05,
+      "loss": 0.327,
+      "step": 3253
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0242726778988489e-05,
+      "loss": 0.2622,
+      "step": 3254
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0234024434623443e-05,
+      "loss": 0.2134,
+      "step": 3255
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0225323874158982e-05,
+      "loss": 0.2463,
+      "step": 3256
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0216625100851718e-05,
+      "loss": 0.2637,
+      "step": 3257
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.020792811795759e-05,
+      "loss": 0.2528,
+      "step": 3258
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0199232928731879e-05,
+      "loss": 0.2709,
+      "step": 3259
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0190539536429172e-05,
+      "loss": 0.2573,
+      "step": 3260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0181847944303417e-05,
+      "loss": 0.2808,
+      "step": 3261
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0173158155607856e-05,
+      "loss": 0.2257,
+      "step": 3262
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0164470173595078e-05,
+      "loss": 0.2166,
+      "step": 3263
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0155784001516978e-05,
+      "loss": 0.2391,
+      "step": 3264
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0147099642624794e-05,
+      "loss": 0.2668,
+      "step": 3265
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0138417100169064e-05,
+      "loss": 0.2638,
+      "step": 3266
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0129736377399664e-05,
+      "loss": 0.2024,
+      "step": 3267
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0121057477565778e-05,
+      "loss": 0.3252,
+      "step": 3268
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0112380403915904e-05,
+      "loss": 0.2469,
+      "step": 3269
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0103705159697874e-05,
+      "loss": 0.2684,
+      "step": 3270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0095031748158822e-05,
+      "loss": 0.3079,
+      "step": 3271
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0086360172545197e-05,
+      "loss": 0.2754,
+      "step": 3272
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0077690436102756e-05,
+      "loss": 0.3103,
+      "step": 3273
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0069022542076585e-05,
+      "loss": 0.3212,
+      "step": 3274
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0060356493711065e-05,
+      "loss": 0.2563,
+      "step": 3275
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.005169229424989e-05,
+      "loss": 0.3414,
+      "step": 3276
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0043029946936061e-05,
+      "loss": 0.2423,
+      "step": 3277
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0034369455011883e-05,
+      "loss": 0.2665,
+      "step": 3278
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0025710821718983e-05,
+      "loss": 0.2823,
+      "step": 3279
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0017054050298275e-05,
+      "loss": 0.3127,
+      "step": 3280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0008399143989975e-05,
+      "loss": 0.3087,
+      "step": 3281
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.999746106033615e-06,
+      "loss": 0.2676,
+      "step": 3282
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.99109493966801e-06,
+      "loss": 0.1913,
+      "step": 3283
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.982445648131298e-06,
+      "loss": 0.277,
+      "step": 3284
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.973798234660888e-06,
+      "loss": 0.2404,
+      "step": 3285
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.965152702493508e-06,
+      "loss": 0.2279,
+      "step": 3286
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.956509054865167e-06,
+      "loss": 0.2275,
+      "step": 3287
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.947867295011179e-06,
+      "loss": 0.3187,
+      "step": 3288
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.939227426166154e-06,
+      "loss": 0.288,
+      "step": 3289
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.930589451563977e-06,
+      "loss": 0.229,
+      "step": 3290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.921953374437835e-06,
+      "loss": 0.2855,
+      "step": 3291
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.913319198020205e-06,
+      "loss": 0.2588,
+      "step": 3292
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.90468692554286e-06,
+      "loss": 0.2153,
+      "step": 3293
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.896056560236847e-06,
+      "loss": 0.2483,
+      "step": 3294
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.887428105332499e-06,
+      "loss": 0.2639,
+      "step": 3295
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.878801564059442e-06,
+      "loss": 0.2189,
+      "step": 3296
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.87017693964659e-06,
+      "loss": 0.3324,
+      "step": 3297
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.861554235322129e-06,
+      "loss": 0.2786,
+      "step": 3298
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.852933454313523e-06,
+      "loss": 0.2659,
+      "step": 3299
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.844314599847535e-06,
+      "loss": 0.2373,
+      "step": 3300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.835697675150179e-06,
+      "loss": 0.3136,
+      "step": 3301
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.82708268344678e-06,
+      "loss": 0.2556,
+      "step": 3302
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.818469627961907e-06,
+      "loss": 0.2338,
+      "step": 3303
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.809858511919435e-06,
+      "loss": 0.2365,
+      "step": 3304
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.801249338542477e-06,
+      "loss": 0.2699,
+      "step": 3305
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.79264211105346e-06,
+      "loss": 0.3162,
+      "step": 3306
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.784036832674046e-06,
+      "loss": 0.2611,
+      "step": 3307
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.775433506625197e-06,
+      "loss": 0.2426,
+      "step": 3308
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.766832136127123e-06,
+      "loss": 0.2103,
+      "step": 3309
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.758232724399302e-06,
+      "loss": 0.2759,
+      "step": 3310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749635274660499e-06,
+      "loss": 0.2382,
+      "step": 3311
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.741039790128729e-06,
+      "loss": 0.2264,
+      "step": 3312
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.732446274021272e-06,
+      "loss": 0.2677,
+      "step": 3313
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.723854729554676e-06,
+      "loss": 0.2587,
+      "step": 3314
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.715265159944744e-06,
+      "loss": 0.2238,
+      "step": 3315
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.706677568406555e-06,
+      "loss": 0.2584,
+      "step": 3316
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.698091958154425e-06,
+      "loss": 0.2693,
+      "step": 3317
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.68950833240195e-06,
+      "loss": 0.313,
+      "step": 3318
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.680926694361966e-06,
+      "loss": 0.2529,
+      "step": 3319
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.67234704724658e-06,
+      "loss": 0.2837,
+      "step": 3320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.663769394267141e-06,
+      "loss": 0.2283,
+      "step": 3321
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.655193738634261e-06,
+      "loss": 0.27,
+      "step": 3322
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.646620083557794e-06,
+      "loss": 0.3143,
+      "step": 3323
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.638048432246864e-06,
+      "loss": 0.2893,
+      "step": 3324
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.629478787909822e-06,
+      "loss": 0.2798,
+      "step": 3325
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.620911153754286e-06,
+      "loss": 0.2631,
+      "step": 3326
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.61234553298711e-06,
+      "loss": 0.3016,
+      "step": 3327
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.603781928814389e-06,
+      "loss": 0.236,
+      "step": 3328
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.595220344441488e-06,
+      "loss": 0.2977,
+      "step": 3329
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.586660783073e-06,
+      "loss": 0.2975,
+      "step": 3330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.578103247912755e-06,
+      "loss": 0.2336,
+      "step": 3331
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.569547742163827e-06,
+      "loss": 0.2645,
+      "step": 3332
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.560994269028542e-06,
+      "loss": 0.3057,
+      "step": 3333
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.55244283170846e-06,
+      "loss": 0.2313,
+      "step": 3334
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.543893433404373e-06,
+      "loss": 0.2537,
+      "step": 3335
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.535346077316311e-06,
+      "loss": 0.2561,
+      "step": 3336
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.52680076664354e-06,
+      "loss": 0.2789,
+      "step": 3337
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.518257504584573e-06,
+      "loss": 0.2598,
+      "step": 3338
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.50971629433714e-06,
+      "loss": 0.2333,
+      "step": 3339
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.501177139098204e-06,
+      "loss": 0.2732,
+      "step": 3340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.492640042063965e-06,
+      "loss": 0.2411,
+      "step": 3341
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.48410500642986e-06,
+      "loss": 0.24,
+      "step": 3342
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.475572035390537e-06,
+      "loss": 0.194,
+      "step": 3343
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.467041132139884e-06,
+      "loss": 0.2524,
+      "step": 3344
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.458512299871009e-06,
+      "loss": 0.2911,
+      "step": 3345
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.44998554177624e-06,
+      "loss": 0.2621,
+      "step": 3346
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.441460861047148e-06,
+      "loss": 0.2655,
+      "step": 3347
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.432938260874508e-06,
+      "loss": 0.2816,
+      "step": 3348
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.424417744448322e-06,
+      "loss": 0.2524,
+      "step": 3349
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.415899314957803e-06,
+      "loss": 0.2479,
+      "step": 3350
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.407382975591407e-06,
+      "loss": 0.2767,
+      "step": 3351
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.398868729536787e-06,
+      "loss": 0.3063,
+      "step": 3352
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.390356579980817e-06,
+      "loss": 0.2686,
+      "step": 3353
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.381846530109583e-06,
+      "loss": 0.2222,
+      "step": 3354
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.373338583108394e-06,
+      "loss": 0.2627,
+      "step": 3355
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.36483274216177e-06,
+      "loss": 0.2833,
+      "step": 3356
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.356329010453437e-06,
+      "loss": 0.2611,
+      "step": 3357
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.347827391166333e-06,
+      "loss": 0.2635,
+      "step": 3358
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.339327887482605e-06,
+      "loss": 0.231,
+      "step": 3359
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.330830502583619e-06,
+      "loss": 0.3269,
+      "step": 3360
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.322335239649934e-06,
+      "loss": 0.2787,
+      "step": 3361
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.313842101861313e-06,
+      "loss": 0.3802,
+      "step": 3362
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.305351092396743e-06,
+      "loss": 0.2588,
+      "step": 3363
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.296862214434385e-06,
+      "loss": 0.2573,
+      "step": 3364
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.288375471151637e-06,
+      "loss": 0.2878,
+      "step": 3365
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.279890865725067e-06,
+      "loss": 0.2758,
+      "step": 3366
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.271408401330462e-06,
+      "loss": 0.2125,
+      "step": 3367
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.262928081142794e-06,
+      "loss": 0.257,
+      "step": 3368
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.25444990833625e-06,
+      "loss": 0.2562,
+      "step": 3369
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.245973886084196e-06,
+      "loss": 0.2678,
+      "step": 3370
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.237500017559202e-06,
+      "loss": 0.234,
+      "step": 3371
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.229028305933028e-06,
+      "loss": 0.257,
+      "step": 3372
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.220558754376628e-06,
+      "loss": 0.2551,
+      "step": 3373
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.212091366060148e-06,
+      "loss": 0.2317,
+      "step": 3374
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.203626144152933e-06,
+      "loss": 0.1911,
+      "step": 3375
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.195163091823493e-06,
+      "loss": 0.2805,
+      "step": 3376
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.186702212239548e-06,
+      "loss": 0.2139,
+      "step": 3377
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.178243508568005e-06,
+      "loss": 0.3219,
+      "step": 3378
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.169786983974946e-06,
+      "loss": 0.2067,
+      "step": 3379
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.161332641625634e-06,
+      "loss": 0.2545,
+      "step": 3380
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.152880484684526e-06,
+      "loss": 0.2521,
+      "step": 3381
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.144430516315256e-06,
+      "loss": 0.3048,
+      "step": 3382
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.13598273968065e-06,
+      "loss": 0.2911,
+      "step": 3383
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.127537157942686e-06,
+      "loss": 0.2451,
+      "step": 3384
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.119093774262556e-06,
+      "loss": 0.226,
+      "step": 3385
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.110652591800595e-06,
+      "loss": 0.2405,
+      "step": 3386
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.102213613716343e-06,
+      "loss": 0.2422,
+      "step": 3387
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.093776843168494e-06,
+      "loss": 0.2174,
+      "step": 3388
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.085342283314932e-06,
+      "loss": 0.2827,
+      "step": 3389
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.0769099373127e-06,
+      "loss": 0.2738,
+      "step": 3390
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.068479808318014e-06,
+      "loss": 0.319,
+      "step": 3391
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.06005189948627e-06,
+      "loss": 0.3042,
+      "step": 3392
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.05162621397203e-06,
+      "loss": 0.2548,
+      "step": 3393
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.043202754929016e-06,
+      "loss": 0.2587,
+      "step": 3394
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.034781525510118e-06,
+      "loss": 0.2166,
+      "step": 3395
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.0263625288674e-06,
+      "loss": 0.22,
+      "step": 3396
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.017945768152093e-06,
+      "loss": 0.236,
+      "step": 3397
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.009531246514569e-06,
+      "loss": 0.2424,
+      "step": 3398
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.001118967104382e-06,
+      "loss": 0.2645,
+      "step": 3399
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.992708933070239e-06,
+      "loss": 0.2482,
+      "step": 3400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.984301147560016e-06,
+      "loss": 0.2249,
+      "step": 3401
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.97589561372073e-06,
+      "loss": 0.2614,
+      "step": 3402
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.967492334698576e-06,
+      "loss": 0.2577,
+      "step": 3403
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.95909131363888e-06,
+      "loss": 0.2878,
+      "step": 3404
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.950692553686151e-06,
+      "loss": 0.2939,
+      "step": 3405
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.94229605798403e-06,
+      "loss": 0.2075,
+      "step": 3406
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.933901829675321e-06,
+      "loss": 0.2543,
+      "step": 3407
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.925509871901976e-06,
+      "loss": 0.2509,
+      "step": 3408
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.917120187805091e-06,
+      "loss": 0.2426,
+      "step": 3409
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.908732780524927e-06,
+      "loss": 0.2623,
+      "step": 3410
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.900347653200882e-06,
+      "loss": 0.2535,
+      "step": 3411
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.891964808971502e-06,
+      "loss": 0.2432,
+      "step": 3412
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.88358425097447e-06,
+      "loss": 0.2988,
+      "step": 3413
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.87520598234663e-06,
+      "loss": 0.2461,
+      "step": 3414
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.866830006223962e-06,
+      "loss": 0.2808,
+      "step": 3415
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.858456325741585e-06,
+      "loss": 0.2902,
+      "step": 3416
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.850084944033756e-06,
+      "loss": 0.2031,
+      "step": 3417
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.841715864233874e-06,
+      "loss": 0.2358,
+      "step": 3418
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.83334908947449e-06,
+      "loss": 0.2141,
+      "step": 3419
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.824984622887274e-06,
+      "loss": 0.296,
+      "step": 3420
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.816622467603037e-06,
+      "loss": 0.285,
+      "step": 3421
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.80826262675173e-06,
+      "loss": 0.2518,
+      "step": 3422
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.799905103462429e-06,
+      "loss": 0.2587,
+      "step": 3423
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.791549900863356e-06,
+      "loss": 0.2466,
+      "step": 3424
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.783197022081848e-06,
+      "loss": 0.3073,
+      "step": 3425
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.774846470244388e-06,
+      "loss": 0.2646,
+      "step": 3426
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.76649824847657e-06,
+      "loss": 0.2176,
+      "step": 3427
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.75815235990314e-06,
+      "loss": 0.2574,
+      "step": 3428
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.749808807647945e-06,
+      "loss": 0.2576,
+      "step": 3429
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.741467594833977e-06,
+      "loss": 0.2422,
+      "step": 3430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.733128724583345e-06,
+      "loss": 0.2688,
+      "step": 3431
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.724792200017267e-06,
+      "loss": 0.3536,
+      "step": 3432
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.716458024256122e-06,
+      "loss": 0.2474,
+      "step": 3433
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.708126200419364e-06,
+      "loss": 0.3226,
+      "step": 3434
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.699796731625593e-06,
+      "loss": 0.3124,
+      "step": 3435
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.691469620992525e-06,
+      "loss": 0.2302,
+      "step": 3436
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.68314487163699e-06,
+      "loss": 0.2744,
+      "step": 3437
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.674822486674932e-06,
+      "loss": 0.2725,
+      "step": 3438
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.666502469221419e-06,
+      "loss": 0.3136,
+      "step": 3439
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.658184822390618e-06,
+      "loss": 0.2256,
+      "step": 3440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.64986954929582e-06,
+      "loss": 0.2427,
+      "step": 3441
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.641556653049424e-06,
+      "loss": 0.2661,
+      "step": 3442
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.633246136762944e-06,
+      "loss": 0.2881,
+      "step": 3443
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.624938003546993e-06,
+      "loss": 0.2327,
+      "step": 3444
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.616632256511303e-06,
+      "loss": 0.3063,
+      "step": 3445
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.608328898764709e-06,
+      "loss": 0.2478,
+      "step": 3446
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.60002793341515e-06,
+      "loss": 0.27,
+      "step": 3447
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.591729363569665e-06,
+      "loss": 0.2795,
+      "step": 3448
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.583433192334404e-06,
+      "loss": 0.2526,
+      "step": 3449
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.575139422814619e-06,
+      "loss": 0.3077,
+      "step": 3450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.56684805811466e-06,
+      "loss": 0.2561,
+      "step": 3451
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.558559101337977e-06,
+      "loss": 0.2625,
+      "step": 3452
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.550272555587128e-06,
+      "loss": 0.3099,
+      "step": 3453
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.541988423963735e-06,
+      "loss": 0.2834,
+      "step": 3454
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.533706709568576e-06,
+      "loss": 0.2742,
+      "step": 3455
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.525427415501462e-06,
+      "loss": 0.2542,
+      "step": 3456
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.517150544861336e-06,
+      "loss": 0.2769,
+      "step": 3457
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.508876100746233e-06,
+      "loss": 0.2478,
+      "step": 3458
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.500604086253244e-06,
+      "loss": 0.2056,
+      "step": 3459
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.4923345044786e-06,
+      "loss": 0.2841,
+      "step": 3460
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.484067358517602e-06,
+      "loss": 0.2554,
+      "step": 3461
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.475802651464615e-06,
+      "loss": 0.2863,
+      "step": 3462
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.467540386413126e-06,
+      "loss": 0.2772,
+      "step": 3463
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.45928056645569e-06,
+      "loss": 0.2949,
+      "step": 3464
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.451023194683949e-06,
+      "loss": 0.2256,
+      "step": 3465
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.442768274188637e-06,
+      "loss": 0.2933,
+      "step": 3466
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.434515808059555e-06,
+      "loss": 0.3135,
+      "step": 3467
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.42626579938559e-06,
+      "loss": 0.2448,
+      "step": 3468
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.418018251254735e-06,
+      "loss": 0.3107,
+      "step": 3469
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.409773166754021e-06,
+      "loss": 0.2581,
+      "step": 3470
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.401530548969576e-06,
+      "loss": 0.2471,
+      "step": 3471
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.393290400986613e-06,
+      "loss": 0.3065,
+      "step": 3472
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.38505272588941e-06,
+      "loss": 0.2953,
+      "step": 3473
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.376817526761327e-06,
+      "loss": 0.2078,
+      "step": 3474
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.36858480668478e-06,
+      "loss": 0.2151,
+      "step": 3475
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.360354568741273e-06,
+      "loss": 0.2755,
+      "step": 3476
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.352126816011382e-06,
+      "loss": 0.2253,
+      "step": 3477
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.343901551574744e-06,
+      "loss": 0.2683,
+      "step": 3478
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.33567877851007e-06,
+      "loss": 0.3127,
+      "step": 3479
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.327458499895142e-06,
+      "loss": 0.2669,
+      "step": 3480
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.319240718806785e-06,
+      "loss": 0.2271,
+      "step": 3481
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.311025438320928e-06,
+      "loss": 0.2831,
+      "step": 3482
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.30281266151254e-06,
+      "loss": 0.261,
+      "step": 3483
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.294602391455644e-06,
+      "loss": 0.2413,
+      "step": 3484
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.286394631223346e-06,
+      "loss": 0.327,
+      "step": 3485
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.278189383887806e-06,
+      "loss": 0.2595,
+      "step": 3486
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.269986652520234e-06,
+      "loss": 0.273,
+      "step": 3487
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.261786440190916e-06,
+      "loss": 0.2184,
+      "step": 3488
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.253588749969174e-06,
+      "loss": 0.2783,
+      "step": 3489
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.24539358492339e-06,
+      "loss": 0.2335,
+      "step": 3490
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.237200948121032e-06,
+      "loss": 0.3525,
+      "step": 3491
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.229010842628575e-06,
+      "loss": 0.2903,
+      "step": 3492
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.220823271511577e-06,
+      "loss": 0.2599,
+      "step": 3493
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.212638237834643e-06,
+      "loss": 0.2371,
+      "step": 3494
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.204455744661405e-06,
+      "loss": 0.3029,
+      "step": 3495
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.19627579505459e-06,
+      "loss": 0.38,
+      "step": 3496
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.188098392075929e-06,
+      "loss": 0.3239,
+      "step": 3497
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.179923538786218e-06,
+      "loss": 0.2382,
+      "step": 3498
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.171751238245302e-06,
+      "loss": 0.2749,
+      "step": 3499
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.163581493512065e-06,
+      "loss": 0.2664,
+      "step": 3500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.155414307644435e-06,
+      "loss": 0.2125,
+      "step": 3501
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.147249683699389e-06,
+      "loss": 0.2456,
+      "step": 3502
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.139087624732925e-06,
+      "loss": 0.3147,
+      "step": 3503
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.130928133800105e-06,
+      "loss": 0.2224,
+      "step": 3504
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.122771213955015e-06,
+      "loss": 0.2198,
+      "step": 3505
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.114616868250786e-06,
+      "loss": 0.2745,
+      "step": 3506
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.10646509973958e-06,
+      "loss": 0.1834,
+      "step": 3507
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.0983159114726e-06,
+      "loss": 0.2719,
+      "step": 3508
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.090169306500078e-06,
+      "loss": 0.2084,
+      "step": 3509
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.082025287871288e-06,
+      "loss": 0.2108,
+      "step": 3510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.073883858634518e-06,
+      "loss": 0.2139,
+      "step": 3511
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.065745021837102e-06,
+      "loss": 0.262,
+      "step": 3512
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.057608780525403e-06,
+      "loss": 0.2363,
+      "step": 3513
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.049475137744804e-06,
+      "loss": 0.2937,
+      "step": 3514
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.041344096539727e-06,
+      "loss": 0.2531,
+      "step": 3515
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.033215659953616e-06,
+      "loss": 0.3101,
+      "step": 3516
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.025089831028917e-06,
+      "loss": 0.3096,
+      "step": 3517
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.016966612807153e-06,
+      "loss": 0.2545,
+      "step": 3518
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.008846008328813e-06,
+      "loss": 0.2903,
+      "step": 3519
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.000728020633437e-06,
+      "loss": 0.2222,
+      "step": 3520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.992612652759593e-06,
+      "loss": 0.2675,
+      "step": 3521
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.984499907744835e-06,
+      "loss": 0.2146,
+      "step": 3522
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.976389788625774e-06,
+      "loss": 0.3246,
+      "step": 3523
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.96828229843802e-06,
+      "loss": 0.2657,
+      "step": 3524
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.960177440216193e-06,
+      "loss": 0.2535,
+      "step": 3525
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.952075216993933e-06,
+      "loss": 0.2675,
+      "step": 3526
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.943975631803902e-06,
+      "loss": 0.2692,
+      "step": 3527
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.935878687677764e-06,
+      "loss": 0.2864,
+      "step": 3528
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.927784387646202e-06,
+      "loss": 0.2739,
+      "step": 3529
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.919692734738899e-06,
+      "loss": 0.2457,
+      "step": 3530
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.91160373198455e-06,
+      "loss": 0.3098,
+      "step": 3531
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.90351738241088e-06,
+      "loss": 0.3019,
+      "step": 3532
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.89543368904459e-06,
+      "loss": 0.3196,
+      "step": 3533
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.887352654911398e-06,
+      "loss": 0.2267,
+      "step": 3534
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.879274283036032e-06,
+      "loss": 0.2738,
+      "step": 3535
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.871198576442219e-06,
+      "loss": 0.2201,
+      "step": 3536
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.86312553815269e-06,
+      "loss": 0.255,
+      "step": 3537
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.855055171189181e-06,
+      "loss": 0.2435,
+      "step": 3538
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.846987478572411e-06,
+      "loss": 0.2628,
+      "step": 3539
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.83892246332212e-06,
+      "loss": 0.2078,
+      "step": 3540
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.830860128457033e-06,
+      "loss": 0.2256,
+      "step": 3541
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.822800476994875e-06,
+      "loss": 0.2668,
+      "step": 3542
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.814743511952374e-06,
+      "loss": 0.2701,
+      "step": 3543
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.806689236345227e-06,
+      "loss": 0.2355,
+      "step": 3544
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.798637653188161e-06,
+      "loss": 0.2305,
+      "step": 3545
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.790588765494874e-06,
+      "loss": 0.2731,
+      "step": 3546
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.78254257627805e-06,
+      "loss": 0.2792,
+      "step": 3547
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.774499088549372e-06,
+      "loss": 0.2949,
+      "step": 3548
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.766458305319512e-06,
+      "loss": 0.2901,
+      "step": 3549
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.758420229598133e-06,
+      "loss": 0.2508,
+      "step": 3550
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.750384864393881e-06,
+      "loss": 0.2847,
+      "step": 3551
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.742352212714374e-06,
+      "loss": 0.2708,
+      "step": 3552
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.734322277566229e-06,
+      "loss": 0.2617,
+      "step": 3553
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.726295061955062e-06,
+      "loss": 0.3197,
+      "step": 3554
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.718270568885433e-06,
+      "loss": 0.2769,
+      "step": 3555
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.71024880136091e-06,
+      "loss": 0.2825,
+      "step": 3556
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.702229762384039e-06,
+      "loss": 0.2489,
+      "step": 3557
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.694213454956323e-06,
+      "loss": 0.2717,
+      "step": 3558
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.68619988207828e-06,
+      "loss": 0.2406,
+      "step": 3559
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.678189046749368e-06,
+      "loss": 0.2271,
+      "step": 3560
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.670180951968044e-06,
+      "loss": 0.199,
+      "step": 3561
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.662175600731727e-06,
+      "loss": 0.292,
+      "step": 3562
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.654172996036814e-06,
+      "loss": 0.2994,
+      "step": 3563
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.646173140878675e-06,
+      "loss": 0.197,
+      "step": 3564
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.63817603825165e-06,
+      "loss": 0.2487,
+      "step": 3565
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.630181691149044e-06,
+      "loss": 0.3457,
+      "step": 3566
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.622190102563128e-06,
+      "loss": 0.2334,
+      "step": 3567
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.614201275485165e-06,
+      "loss": 0.2734,
+      "step": 3568
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.60621521290535e-06,
+      "loss": 0.2125,
+      "step": 3569
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.598231917812865e-06,
+      "loss": 0.2209,
+      "step": 3570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.590251393195857e-06,
+      "loss": 0.311,
+      "step": 3571
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.582273642041411e-06,
+      "loss": 0.2302,
+      "step": 3572
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.574298667335615e-06,
+      "loss": 0.2537,
+      "step": 3573
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.566326472063479e-06,
+      "loss": 0.2708,
+      "step": 3574
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.558357059208997e-06,
+      "loss": 0.2188,
+      "step": 3575
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.550390431755107e-06,
+      "loss": 0.2695,
+      "step": 3576
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.542426592683718e-06,
+      "loss": 0.2775,
+      "step": 3577
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.534465544975683e-06,
+      "loss": 0.2662,
+      "step": 3578
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.526507291610826e-06,
+      "loss": 0.2404,
+      "step": 3579
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.518551835567896e-06,
+      "loss": 0.2993,
+      "step": 3580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.510599179824626e-06,
+      "loss": 0.2968,
+      "step": 3581
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.502649327357684e-06,
+      "loss": 0.2841,
+      "step": 3582
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.494702281142695e-06,
+      "loss": 0.2831,
+      "step": 3583
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.486758044154236e-06,
+      "loss": 0.2231,
+      "step": 3584
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.478816619365813e-06,
+      "loss": 0.235,
+      "step": 3585
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.4708780097499105e-06,
+      "loss": 0.3457,
+      "step": 3586
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.462942218277941e-06,
+      "loss": 0.2222,
+      "step": 3587
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.455009247920256e-06,
+      "loss": 0.2081,
+      "step": 3588
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.4470791016461656e-06,
+      "loss": 0.2474,
+      "step": 3589
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.439151782423915e-06,
+      "loss": 0.2742,
+      "step": 3590
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.431227293220693e-06,
+      "loss": 0.3256,
+      "step": 3591
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.423305637002631e-06,
+      "loss": 0.2405,
+      "step": 3592
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.415386816734801e-06,
+      "loss": 0.2286,
+      "step": 3593
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.407470835381195e-06,
+      "loss": 0.2623,
+      "step": 3594
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3995576959047835e-06,
+      "loss": 0.2146,
+      "step": 3595
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.391647401267426e-06,
+      "loss": 0.2844,
+      "step": 3596
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.383739954429947e-06,
+      "loss": 0.2708,
+      "step": 3597
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3758353583521034e-06,
+      "loss": 0.234,
+      "step": 3598
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.367933615992562e-06,
+      "loss": 0.2574,
+      "step": 3599
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.360034730308955e-06,
+      "loss": 0.2819,
+      "step": 3600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.352138704257826e-06,
+      "loss": 0.2565,
+      "step": 3601
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.344245540794641e-06,
+      "loss": 0.2247,
+      "step": 3602
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3363552428738126e-06,
+      "loss": 0.2469,
+      "step": 3603
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3284678134486685e-06,
+      "loss": 0.2565,
+      "step": 3604
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3205832554714685e-06,
+      "loss": 0.2571,
+      "step": 3605
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.312701571893402e-06,
+      "loss": 0.2816,
+      "step": 3606
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.304822765664565e-06,
+      "loss": 0.1879,
+      "step": 3607
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.2969468397339885e-06,
+      "loss": 0.2918,
+      "step": 3608
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.289073797049641e-06,
+      "loss": 0.2678,
+      "step": 3609
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.281203640558379e-06,
+      "loss": 0.2661,
+      "step": 3610
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.273336373205999e-06,
+      "loss": 0.1863,
+      "step": 3611
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.265471997937216e-06,
+      "loss": 0.2203,
+      "step": 3612
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.257610517695657e-06,
+      "loss": 0.2469,
+      "step": 3613
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.2497519354238725e-06,
+      "loss": 0.257,
+      "step": 3614
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.241896254063312e-06,
+      "loss": 0.2258,
+      "step": 3615
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.234043476554358e-06,
+      "loss": 0.29,
+      "step": 3616
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.226193605836297e-06,
+      "loss": 0.2618,
+      "step": 3617
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.218346644847326e-06,
+      "loss": 0.2121,
+      "step": 3618
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.210502596524561e-06,
+      "loss": 0.2357,
+      "step": 3619
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.202661463804025e-06,
+      "loss": 0.1683,
+      "step": 3620
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.194823249620633e-06,
+      "loss": 0.2855,
+      "step": 3621
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.186987956908241e-06,
+      "loss": 0.2541,
+      "step": 3622
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.1791555885995766e-06,
+      "loss": 0.2853,
+      "step": 3623
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.171326147626296e-06,
+      "loss": 0.2517,
+      "step": 3624
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.16349963691895e-06,
+      "loss": 0.2657,
+      "step": 3625
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.155676059406995e-06,
+      "loss": 0.2427,
+      "step": 3626
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.147855418018792e-06,
+      "loss": 0.2146,
+      "step": 3627
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.140037715681602e-06,
+      "loss": 0.2982,
+      "step": 3628
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.132222955321577e-06,
+      "loss": 0.2111,
+      "step": 3629
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.124411139863774e-06,
+      "loss": 0.2841,
+      "step": 3630
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.116602272232164e-06,
+      "loss": 0.2677,
+      "step": 3631
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.108796355349584e-06,
+      "loss": 0.2513,
+      "step": 3632
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.100993392137787e-06,
+      "loss": 0.2325,
+      "step": 3633
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.093193385517421e-06,
+      "loss": 0.2438,
+      "step": 3634
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.085396338408007e-06,
+      "loss": 0.2192,
+      "step": 3635
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.077602253727994e-06,
+      "loss": 0.2675,
+      "step": 3636
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.069811134394682e-06,
+      "loss": 0.2286,
+      "step": 3637
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.06202298332429e-06,
+      "loss": 0.2769,
+      "step": 3638
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.054237803431916e-06,
+      "loss": 0.2607,
+      "step": 3639
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.046455597631542e-06,
+      "loss": 0.2665,
+      "step": 3640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.038676368836045e-06,
+      "loss": 0.2466,
+      "step": 3641
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.030900119957187e-06,
+      "loss": 0.2355,
+      "step": 3642
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.023126853905602e-06,
+      "loss": 0.3223,
+      "step": 3643
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.015356573590821e-06,
+      "loss": 0.2174,
+      "step": 3644
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.007589281921256e-06,
+      "loss": 0.2424,
+      "step": 3645
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.999824981804193e-06,
+      "loss": 0.2402,
+      "step": 3646
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.992063676145812e-06,
+      "loss": 0.2449,
+      "step": 3647
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.984305367851144e-06,
+      "loss": 0.2183,
+      "step": 3648
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.976550059824137e-06,
+      "loss": 0.1997,
+      "step": 3649
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.968797754967594e-06,
+      "loss": 0.2332,
+      "step": 3650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.961048456183186e-06,
+      "loss": 0.3116,
+      "step": 3651
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.953302166371473e-06,
+      "loss": 0.2138,
+      "step": 3652
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.945558888431885e-06,
+      "loss": 0.2572,
+      "step": 3653
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.937818625262724e-06,
+      "loss": 0.2715,
+      "step": 3654
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.930081379761166e-06,
+      "loss": 0.3174,
+      "step": 3655
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.9223471548232605e-06,
+      "loss": 0.2913,
+      "step": 3656
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.9146159533439026e-06,
+      "loss": 0.2175,
+      "step": 3657
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.906887778216897e-06,
+      "loss": 0.2958,
+      "step": 3658
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.899162632334879e-06,
+      "loss": 0.2285,
+      "step": 3659
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.891440518589367e-06,
+      "loss": 0.2871,
+      "step": 3660
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.883721439870748e-06,
+      "loss": 0.3344,
+      "step": 3661
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.876005399068249e-06,
+      "loss": 0.2263,
+      "step": 3662
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.868292399069996e-06,
+      "loss": 0.2582,
+      "step": 3663
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.860582442762957e-06,
+      "loss": 0.2782,
+      "step": 3664
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.852875533032951e-06,
+      "loss": 0.2847,
+      "step": 3665
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.845171672764675e-06,
+      "loss": 0.2008,
+      "step": 3666
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.837470864841676e-06,
+      "loss": 0.2599,
+      "step": 3667
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.829773112146357e-06,
+      "loss": 0.313,
+      "step": 3668
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.822078417559991e-06,
+      "loss": 0.276,
+      "step": 3669
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.81438678396268e-06,
+      "loss": 0.2281,
+      "step": 3670
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.806698214233398e-06,
+      "loss": 0.269,
+      "step": 3671
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.799012711249989e-06,
+      "loss": 0.2734,
+      "step": 3672
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7913302778891094e-06,
+      "loss": 0.2183,
+      "step": 3673
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.783650917026295e-06,
+      "loss": 0.2606,
+      "step": 3674
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.775974631535924e-06,
+      "loss": 0.2434,
+      "step": 3675
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.768301424291223e-06,
+      "loss": 0.2574,
+      "step": 3676
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.760631298164273e-06,
+      "loss": 0.3306,
+      "step": 3677
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.752964256025984e-06,
+      "loss": 0.3181,
+      "step": 3678
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.74530030074613e-06,
+      "loss": 0.2233,
+      "step": 3679
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7376394351933205e-06,
+      "loss": 0.2643,
+      "step": 3680
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.729981662235013e-06,
+      "loss": 0.279,
+      "step": 3681
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.722326984737507e-06,
+      "loss": 0.2142,
+      "step": 3682
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.714675405565944e-06,
+      "loss": 0.3029,
+      "step": 3683
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.707026927584287e-06,
+      "loss": 0.3068,
+      "step": 3684
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.699381553655374e-06,
+      "loss": 0.267,
+      "step": 3685
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.6917392866408594e-06,
+      "loss": 0.2359,
+      "step": 3686
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.684100129401227e-06,
+      "loss": 0.2674,
+      "step": 3687
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.676464084795814e-06,
+      "loss": 0.2643,
+      "step": 3688
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.668831155682782e-06,
+      "loss": 0.2585,
+      "step": 3689
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.661201344919134e-06,
+      "loss": 0.2628,
+      "step": 3690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.653574655360703e-06,
+      "loss": 0.2373,
+      "step": 3691
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.645951089862144e-06,
+      "loss": 0.2617,
+      "step": 3692
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.63833065127695e-06,
+      "loss": 0.2957,
+      "step": 3693
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.630713342457463e-06,
+      "loss": 0.2394,
+      "step": 3694
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.623099166254816e-06,
+      "loss": 0.2687,
+      "step": 3695
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.615488125518996e-06,
+      "loss": 0.2115,
+      "step": 3696
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.607880223098813e-06,
+      "loss": 0.2539,
+      "step": 3697
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.600275461841885e-06,
+      "loss": 0.2526,
+      "step": 3698
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.592673844594685e-06,
+      "loss": 0.278,
+      "step": 3699
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.585075374202477e-06,
+      "loss": 0.2992,
+      "step": 3700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.577480053509371e-06,
+      "loss": 0.2567,
+      "step": 3701
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.56988788535829e-06,
+      "loss": 0.2393,
+      "step": 3702
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.562298872590961e-06,
+      "loss": 0.2284,
+      "step": 3703
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.554713018047964e-06,
+      "loss": 0.3187,
+      "step": 3704
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.547130324568674e-06,
+      "loss": 0.27,
+      "step": 3705
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.539550794991278e-06,
+      "loss": 0.2166,
+      "step": 3706
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.531974432152791e-06,
+      "loss": 0.3015,
+      "step": 3707
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.524401238889042e-06,
+      "loss": 0.2869,
+      "step": 3708
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5168312180346694e-06,
+      "loss": 0.2725,
+      "step": 3709
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.509264372423125e-06,
+      "loss": 0.2865,
+      "step": 3710
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5017007048866774e-06,
+      "loss": 0.3127,
+      "step": 3711
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.494140218256387e-06,
+      "loss": 0.2344,
+      "step": 3712
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.486582915362158e-06,
+      "loss": 0.2144,
+      "step": 3713
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.479028799032664e-06,
+      "loss": 0.2548,
+      "step": 3714
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.471477872095415e-06,
+      "loss": 0.2289,
+      "step": 3715
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4639301373767116e-06,
+      "loss": 0.2599,
+      "step": 3716
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.456385597701666e-06,
+      "loss": 0.2197,
+      "step": 3717
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4488442558941925e-06,
+      "loss": 0.2653,
+      "step": 3718
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.441306114777015e-06,
+      "loss": 0.3059,
+      "step": 3719
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.433771177171642e-06,
+      "loss": 0.3506,
+      "step": 3720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.426239445898402e-06,
+      "loss": 0.3104,
+      "step": 3721
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.418710923776412e-06,
+      "loss": 0.2853,
+      "step": 3722
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.411185613623593e-06,
+      "loss": 0.2678,
+      "step": 3723
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.403663518256668e-06,
+      "loss": 0.3153,
+      "step": 3724
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.396144640491133e-06,
+      "loss": 0.2587,
+      "step": 3725
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.388628983141317e-06,
+      "loss": 0.2351,
+      "step": 3726
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.38111654902032e-06,
+      "loss": 0.256,
+      "step": 3727
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.373607340940031e-06,
+      "loss": 0.254,
+      "step": 3728
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.366101361711146e-06,
+      "loss": 0.2961,
+      "step": 3729
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3585986141431476e-06,
+      "loss": 0.2784,
+      "step": 3730
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.351099101044304e-06,
+      "loss": 0.2545,
+      "step": 3731
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.343602825221685e-06,
+      "loss": 0.2275,
+      "step": 3732
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.336109789481128e-06,
+      "loss": 0.3048,
+      "step": 3733
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.328619996627272e-06,
+      "loss": 0.2673,
+      "step": 3734
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.321133449463553e-06,
+      "loss": 0.2316,
+      "step": 3735
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3136501507921656e-06,
+      "loss": 0.2455,
+      "step": 3736
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.306170103414108e-06,
+      "loss": 0.2053,
+      "step": 3737
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.298693310129158e-06,
+      "loss": 0.2484,
+      "step": 3738
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.291219773735859e-06,
+      "loss": 0.275,
+      "step": 3739
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.2837494970315655e-06,
+      "loss": 0.239,
+      "step": 3740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.276282482812394e-06,
+      "loss": 0.2264,
+      "step": 3741
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.268818733873233e-06,
+      "loss": 0.2338,
+      "step": 3742
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.26135825300776e-06,
+      "loss": 0.2751,
+      "step": 3743
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.253901043008428e-06,
+      "loss": 0.3174,
+      "step": 3744
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.2464471066664654e-06,
+      "loss": 0.3352,
+      "step": 3745
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.238996446771879e-06,
+      "loss": 0.2548,
+      "step": 3746
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.231549066113434e-06,
+      "loss": 0.2396,
+      "step": 3747
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.224104967478676e-06,
+      "loss": 0.2263,
+      "step": 3748
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.216664153653945e-06,
+      "loss": 0.2574,
+      "step": 3749
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.209226627424311e-06,
+      "loss": 0.3136,
+      "step": 3750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.2017923915736426e-06,
+      "loss": 0.2692,
+      "step": 3751
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.194361448884565e-06,
+      "loss": 0.2794,
+      "step": 3752
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.186933802138478e-06,
+      "loss": 0.2473,
+      "step": 3753
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1795094541155426e-06,
+      "loss": 0.2097,
+      "step": 3754
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1720884075946795e-06,
+      "loss": 0.2225,
+      "step": 3755
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.164670665353585e-06,
+      "loss": 0.28,
+      "step": 3756
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.157256230168714e-06,
+      "loss": 0.2338,
+      "step": 3757
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.14984510481528e-06,
+      "loss": 0.2677,
+      "step": 3758
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.142437292067265e-06,
+      "loss": 0.2638,
+      "step": 3759
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.135032794697411e-06,
+      "loss": 0.2124,
+      "step": 3760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1276316154771985e-06,
+      "loss": 0.2534,
+      "step": 3761
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.120233757176903e-06,
+      "loss": 0.2064,
+      "step": 3762
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.112839222565522e-06,
+      "loss": 0.2825,
+      "step": 3763
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.105448014410829e-06,
+      "loss": 0.2308,
+      "step": 3764
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.098060135479346e-06,
+      "loss": 0.2176,
+      "step": 3765
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.09067558853635e-06,
+      "loss": 0.2686,
+      "step": 3766
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.08329437634587e-06,
+      "loss": 0.2449,
+      "step": 3767
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.075916501670693e-06,
+      "loss": 0.2535,
+      "step": 3768
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.068541967272342e-06,
+      "loss": 0.2479,
+      "step": 3769
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.061170775911104e-06,
+      "loss": 0.2281,
+      "step": 3770
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.053802930346008e-06,
+      "loss": 0.2316,
+      "step": 3771
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.046438433334832e-06,
+      "loss": 0.2345,
+      "step": 3772
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.039077287634104e-06,
+      "loss": 0.2896,
+      "step": 3773
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.031719495999097e-06,
+      "loss": 0.2148,
+      "step": 3774
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.024365061183812e-06,
+      "loss": 0.2505,
+      "step": 3775
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.017013985941027e-06,
+      "loss": 0.2639,
+      "step": 3776
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.009666273022229e-06,
+      "loss": 0.2836,
+      "step": 3777
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.002321925177666e-06,
+      "loss": 0.3167,
+      "step": 3778
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.994980945156318e-06,
+      "loss": 0.2739,
+      "step": 3779
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9876433357059116e-06,
+      "loss": 0.2461,
+      "step": 3780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.980309099572906e-06,
+      "loss": 0.2761,
+      "step": 3781
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.972978239502505e-06,
+      "loss": 0.2402,
+      "step": 3782
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.965650758238632e-06,
+      "loss": 0.2457,
+      "step": 3783
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.958326658523964e-06,
+      "loss": 0.2629,
+      "step": 3784
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.951005943099903e-06,
+      "loss": 0.2576,
+      "step": 3785
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.943688614706588e-06,
+      "loss": 0.2542,
+      "step": 3786
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.936374676082894e-06,
+      "loss": 0.2441,
+      "step": 3787
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.929064129966406e-06,
+      "loss": 0.2964,
+      "step": 3788
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.921756979093471e-06,
+      "loss": 0.2272,
+      "step": 3789
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.914453226199149e-06,
+      "loss": 0.2563,
+      "step": 3790
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9071528740172175e-06,
+      "loss": 0.2771,
+      "step": 3791
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.899855925280199e-06,
+      "loss": 0.2755,
+      "step": 3792
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.892562382719333e-06,
+      "loss": 0.3081,
+      "step": 3793
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.885272249064587e-06,
+      "loss": 0.2406,
+      "step": 3794
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.877985527044658e-06,
+      "loss": 0.2142,
+      "step": 3795
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.87070221938695e-06,
+      "loss": 0.2385,
+      "step": 3796
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.863422328817596e-06,
+      "loss": 0.2816,
+      "step": 3797
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8561458580614716e-06,
+      "loss": 0.2515,
+      "step": 3798
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8488728098421345e-06,
+      "loss": 0.2708,
+      "step": 3799
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.841603186881889e-06,
+      "loss": 0.2623,
+      "step": 3800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.834336991901752e-06,
+      "loss": 0.2368,
+      "step": 3801
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.82707422762144e-06,
+      "loss": 0.2294,
+      "step": 3802
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.819814896759414e-06,
+      "loss": 0.2271,
+      "step": 3803
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.812559002032834e-06,
+      "loss": 0.277,
+      "step": 3804
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.805306546157568e-06,
+      "loss": 0.3137,
+      "step": 3805
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.798057531848207e-06,
+      "loss": 0.2488,
+      "step": 3806
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.79081196181805e-06,
+      "loss": 0.2485,
+      "step": 3807
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.783569838779111e-06,
+      "loss": 0.2061,
+      "step": 3808
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.776331165442113e-06,
+      "loss": 0.2721,
+      "step": 3809
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.769095944516474e-06,
+      "loss": 0.282,
+      "step": 3810
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.761864178710335e-06,
+      "loss": 0.2734,
+      "step": 3811
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7546358707305505e-06,
+      "loss": 0.2733,
+      "step": 3812
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.747411023282659e-06,
+      "loss": 0.2737,
+      "step": 3813
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.740189639070916e-06,
+      "loss": 0.2241,
+      "step": 3814
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.732971720798283e-06,
+      "loss": 0.3057,
+      "step": 3815
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7257572711664166e-06,
+      "loss": 0.2757,
+      "step": 3816
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.718546292875687e-06,
+      "loss": 0.2306,
+      "step": 3817
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.711338788625148e-06,
+      "loss": 0.2615,
+      "step": 3818
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.704134761112565e-06,
+      "loss": 0.3035,
+      "step": 3819
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6969342130344005e-06,
+      "loss": 0.3009,
+      "step": 3820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.689737147085813e-06,
+      "loss": 0.2628,
+      "step": 3821
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.682543565960658e-06,
+      "loss": 0.2816,
+      "step": 3822
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.675353472351493e-06,
+      "loss": 0.2217,
+      "step": 3823
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.668166868949547e-06,
+      "loss": 0.2454,
+      "step": 3824
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.660983758444781e-06,
+      "loss": 0.2559,
+      "step": 3825
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.653804143525811e-06,
+      "loss": 0.3018,
+      "step": 3826
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.6466280268799665e-06,
+      "loss": 0.2345,
+      "step": 3827
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.639455411193263e-06,
+      "loss": 0.2571,
+      "step": 3828
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.6322862991504015e-06,
+      "loss": 0.2755,
+      "step": 3829
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.625120693434776e-06,
+      "loss": 0.2314,
+      "step": 3830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.617958596728475e-06,
+      "loss": 0.1986,
+      "step": 3831
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.610800011712248e-06,
+      "loss": 0.2058,
+      "step": 3832
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.603644941065552e-06,
+      "loss": 0.2715,
+      "step": 3833
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.596493387466539e-06,
+      "loss": 0.2402,
+      "step": 3834
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.589345353592013e-06,
+      "loss": 0.3137,
+      "step": 3835
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5822008421174835e-06,
+      "loss": 0.2274,
+      "step": 3836
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.575059855717139e-06,
+      "loss": 0.3195,
+      "step": 3837
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.567922397063828e-06,
+      "loss": 0.2488,
+      "step": 3838
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.560788468829116e-06,
+      "loss": 0.3125,
+      "step": 3839
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.553658073683212e-06,
+      "loss": 0.2444,
+      "step": 3840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.546531214295024e-06,
+      "loss": 0.2057,
+      "step": 3841
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.539407893332132e-06,
+      "loss": 0.3071,
+      "step": 3842
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.532288113460773e-06,
+      "loss": 0.2993,
+      "step": 3843
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5251718773458906e-06,
+      "loss": 0.2601,
+      "step": 3844
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5180591876510854e-06,
+      "loss": 0.2721,
+      "step": 3845
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.510950047038622e-06,
+      "loss": 0.195,
+      "step": 3846
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.503844458169451e-06,
+      "loss": 0.3062,
+      "step": 3847
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.496742423703188e-06,
+      "loss": 0.2174,
+      "step": 3848
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.4896439462981165e-06,
+      "loss": 0.2112,
+      "step": 3849
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.482549028611197e-06,
+      "loss": 0.2742,
+      "step": 3850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.475457673298042e-06,
+      "loss": 0.2573,
+      "step": 3851
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.468369883012938e-06,
+      "loss": 0.2487,
+      "step": 3852
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.461285660408856e-06,
+      "loss": 0.2999,
+      "step": 3853
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.454205008137397e-06,
+      "loss": 0.2433,
+      "step": 3854
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.447127928848849e-06,
+      "loss": 0.2875,
+      "step": 3855
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.440054425192157e-06,
+      "loss": 0.2196,
+      "step": 3856
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.432984499814927e-06,
+      "loss": 0.3107,
+      "step": 3857
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.425918155363427e-06,
+      "loss": 0.2148,
+      "step": 3858
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.418855394482589e-06,
+      "loss": 0.24,
+      "step": 3859
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.411796219815986e-06,
+      "loss": 0.2474,
+      "step": 3860
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.404740634005869e-06,
+      "loss": 0.2419,
+      "step": 3861
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.397688639693136e-06,
+      "loss": 0.2551,
+      "step": 3862
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.390640239517341e-06,
+      "loss": 0.1815,
+      "step": 3863
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.383595436116703e-06,
+      "loss": 0.3175,
+      "step": 3864
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3765542321280675e-06,
+      "loss": 0.2079,
+      "step": 3865
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.369516630186968e-06,
+      "loss": 0.2509,
+      "step": 3866
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.362482632927571e-06,
+      "loss": 0.2341,
+      "step": 3867
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.355452242982688e-06,
+      "loss": 0.2551,
+      "step": 3868
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.348425462983791e-06,
+      "loss": 0.2731,
+      "step": 3869
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.341402295561001e-06,
+      "loss": 0.241,
+      "step": 3870
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.334382743343078e-06,
+      "loss": 0.2294,
+      "step": 3871
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3273668089574444e-06,
+      "loss": 0.2217,
+      "step": 3872
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.320354495030146e-06,
+      "loss": 0.2941,
+      "step": 3873
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3133458041858875e-06,
+      "loss": 0.2446,
+      "step": 3874
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.306340739048029e-06,
+      "loss": 0.2434,
+      "step": 3875
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.299339302238546e-06,
+      "loss": 0.233,
+      "step": 3876
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.292341496378077e-06,
+      "loss": 0.2423,
+      "step": 3877
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.285347324085898e-06,
+      "loss": 0.2389,
+      "step": 3878
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.278356787979907e-06,
+      "loss": 0.2479,
+      "step": 3879
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.271369890676676e-06,
+      "loss": 0.2488,
+      "step": 3880
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.264386634791379e-06,
+      "loss": 0.197,
+      "step": 3881
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.257407022937849e-06,
+      "loss": 0.2127,
+      "step": 3882
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.250431057728548e-06,
+      "loss": 0.2692,
+      "step": 3883
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.243458741774573e-06,
+      "loss": 0.3021,
+      "step": 3884
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.236490077685658e-06,
+      "loss": 0.2479,
+      "step": 3885
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.22952506807017e-06,
+      "loss": 0.2819,
+      "step": 3886
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.222563715535099e-06,
+      "loss": 0.2877,
+      "step": 3887
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.215606022686078e-06,
+      "loss": 0.2029,
+      "step": 3888
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.208651992127363e-06,
+      "loss": 0.2583,
+      "step": 3889
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.201701626461843e-06,
+      "loss": 0.3234,
+      "step": 3890
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.194754928291036e-06,
+      "loss": 0.2679,
+      "step": 3891
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.187811900215084e-06,
+      "loss": 0.2859,
+      "step": 3892
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.180872544832754e-06,
+      "loss": 0.2827,
+      "step": 3893
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.173936864741449e-06,
+      "loss": 0.2315,
+      "step": 3894
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1670048625371755e-06,
+      "loss": 0.2155,
+      "step": 3895
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.160076540814585e-06,
+      "loss": 0.2367,
+      "step": 3896
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.153151902166937e-06,
+      "loss": 0.2555,
+      "step": 3897
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.146230949186121e-06,
+      "loss": 0.2838,
+      "step": 3898
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.139313684462645e-06,
+      "loss": 0.2692,
+      "step": 3899
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.132400110585638e-06,
+      "loss": 0.2344,
+      "step": 3900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.12549023014283e-06,
+      "loss": 0.275,
+      "step": 3901
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.118584045720606e-06,
+      "loss": 0.2117,
+      "step": 3902
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.111681559903926e-06,
+      "loss": 0.3057,
+      "step": 3903
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.104782775276391e-06,
+      "loss": 0.2809,
+      "step": 3904
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.097887694420216e-06,
+      "loss": 0.2864,
+      "step": 3905
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.090996319916208e-06,
+      "loss": 0.2791,
+      "step": 3906
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.084108654343819e-06,
+      "loss": 0.2361,
+      "step": 3907
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.077224700281096e-06,
+      "loss": 0.2443,
+      "step": 3908
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.070344460304685e-06,
+      "loss": 0.2886,
+      "step": 3909
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.06346793698986e-06,
+      "loss": 0.2076,
+      "step": 3910
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.056595132910498e-06,
+      "loss": 0.2571,
+      "step": 3911
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.049726050639086e-06,
+      "loss": 0.2332,
+      "step": 3912
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.042860692746716e-06,
+      "loss": 0.257,
+      "step": 3913
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.03599906180308e-06,
+      "loss": 0.2847,
+      "step": 3914
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.029141160376478e-06,
+      "loss": 0.2664,
+      "step": 3915
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.022286991033832e-06,
+      "loss": 0.2341,
+      "step": 3916
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.015436556340638e-06,
+      "loss": 0.3025,
+      "step": 3917
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.008589858861012e-06,
+      "loss": 0.2786,
+      "step": 3918
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.001746901157666e-06,
+      "loss": 0.2588,
+      "step": 3919
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.994907685791915e-06,
+      "loss": 0.2462,
+      "step": 3920
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.988072215323669e-06,
+      "loss": 0.2896,
+      "step": 3921
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.981240492311448e-06,
+      "loss": 0.2906,
+      "step": 3922
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.974412519312346e-06,
+      "loss": 0.243,
+      "step": 3923
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.967588298882074e-06,
+      "loss": 0.298,
+      "step": 3924
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.960767833574932e-06,
+      "loss": 0.2197,
+      "step": 3925
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.953951125943815e-06,
+      "loss": 0.2524,
+      "step": 3926
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.947138178540213e-06,
+      "loss": 0.2561,
+      "step": 3927
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9403289939141954e-06,
+      "loss": 0.256,
+      "step": 3928
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.933523574614447e-06,
+      "loss": 0.1923,
+      "step": 3929
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.92672192318823e-06,
+      "loss": 0.2423,
+      "step": 3930
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.919924042181388e-06,
+      "loss": 0.3036,
+      "step": 3931
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.913129934138365e-06,
+      "loss": 0.2418,
+      "step": 3932
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.906339601602192e-06,
+      "loss": 0.2228,
+      "step": 3933
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.899553047114484e-06,
+      "loss": 0.1682,
+      "step": 3934
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.892770273215447e-06,
+      "loss": 0.2524,
+      "step": 3935
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8859912824438595e-06,
+      "loss": 0.1973,
+      "step": 3936
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.879216077337089e-06,
+      "loss": 0.2417,
+      "step": 3937
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.872444660431106e-06,
+      "loss": 0.2765,
+      "step": 3938
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.86567703426043e-06,
+      "loss": 0.2559,
+      "step": 3939
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.858913201358184e-06,
+      "loss": 0.2759,
+      "step": 3940
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.852153164256069e-06,
+      "loss": 0.222,
+      "step": 3941
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.845396925484345e-06,
+      "loss": 0.3082,
+      "step": 3942
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.838644487571887e-06,
+      "loss": 0.25,
+      "step": 3943
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.831895853046113e-06,
+      "loss": 0.2255,
+      "step": 3944
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.825151024433036e-06,
+      "loss": 0.2371,
+      "step": 3945
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8184100042572394e-06,
+      "loss": 0.2403,
+      "step": 3946
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8116727950418796e-06,
+      "loss": 0.3058,
+      "step": 3947
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.804939399308691e-06,
+      "loss": 0.2649,
+      "step": 3948
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.79820981957798e-06,
+      "loss": 0.2618,
+      "step": 3949
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.791484058368614e-06,
+      "loss": 0.2281,
+      "step": 3950
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.784762118198041e-06,
+      "loss": 0.2304,
+      "step": 3951
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.77804400158229e-06,
+      "loss": 0.2759,
+      "step": 3952
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.771329711035933e-06,
+      "loss": 0.2759,
+      "step": 3953
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7646192490721265e-06,
+      "loss": 0.2732,
+      "step": 3954
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.757912618202593e-06,
+      "loss": 0.2531,
+      "step": 3955
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751209820937616e-06,
+      "loss": 0.2861,
+      "step": 3956
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.744510859786054e-06,
+      "loss": 0.3188,
+      "step": 3957
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.737815737255312e-06,
+      "loss": 0.2766,
+      "step": 3958
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.731124455851375e-06,
+      "loss": 0.1878,
+      "step": 3959
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.724437018078781e-06,
+      "loss": 0.1425,
+      "step": 3960
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.717753426440633e-06,
+      "loss": 0.1831,
+      "step": 3961
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7110736834385945e-06,
+      "loss": 0.2267,
+      "step": 3962
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.704397791572891e-06,
+      "loss": 0.2299,
+      "step": 3963
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6977257533422924e-06,
+      "loss": 0.2413,
+      "step": 3964
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6910575712441515e-06,
+      "loss": 0.2332,
+      "step": 3965
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.684393247774351e-06,
+      "loss": 0.2697,
+      "step": 3966
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6777327854273475e-06,
+      "loss": 0.2009,
+      "step": 3967
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.671076186696149e-06,
+      "loss": 0.2737,
+      "step": 3968
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6644234540723e-06,
+      "loss": 0.22,
+      "step": 3969
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.657774590045928e-06,
+      "loss": 0.2826,
+      "step": 3970
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.651129597105698e-06,
+      "loss": 0.283,
+      "step": 3971
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.644488477738815e-06,
+      "loss": 0.2141,
+      "step": 3972
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.637851234431048e-06,
+      "loss": 0.2188,
+      "step": 3973
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.631217869666714e-06,
+      "loss": 0.1872,
+      "step": 3974
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.624588385928673e-06,
+      "loss": 0.2046,
+      "step": 3975
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.617962785698338e-06,
+      "loss": 0.2599,
+      "step": 3976
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.611341071455665e-06,
+      "loss": 0.2323,
+      "step": 3977
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.604723245679149e-06,
+      "loss": 0.2022,
+      "step": 3978
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.59810931084585e-06,
+      "loss": 0.2736,
+      "step": 3979
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.591499269431345e-06,
+      "loss": 0.254,
+      "step": 3980
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.584893123909773e-06,
+      "loss": 0.3219,
+      "step": 3981
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.578290876753812e-06,
+      "loss": 0.2554,
+      "step": 3982
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.571692530434666e-06,
+      "loss": 0.2491,
+      "step": 3983
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.565098087422101e-06,
+      "loss": 0.2617,
+      "step": 3984
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.558507550184414e-06,
+      "loss": 0.2476,
+      "step": 3985
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5519209211884246e-06,
+      "loss": 0.2487,
+      "step": 3986
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5453382028995115e-06,
+      "loss": 0.2594,
+      "step": 3987
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.538759397781578e-06,
+      "loss": 0.2856,
+      "step": 3988
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.532184508297065e-06,
+      "loss": 0.2711,
+      "step": 3989
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.525613536906954e-06,
+      "loss": 0.2612,
+      "step": 3990
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.519046486070745e-06,
+      "loss": 0.2975,
+      "step": 3991
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.512483358246478e-06,
+      "loss": 0.2742,
+      "step": 3992
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.505924155890739e-06,
+      "loss": 0.2712,
+      "step": 3993
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4993688814586215e-06,
+      "loss": 0.2139,
+      "step": 3994
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.492817537403761e-06,
+      "loss": 0.2119,
+      "step": 3995
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.486270126178322e-06,
+      "loss": 0.2555,
+      "step": 3996
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.479726650232991e-06,
+      "loss": 0.2018,
+      "step": 3997
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4731871120169945e-06,
+      "loss": 0.2037,
+      "step": 3998
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.466651513978065e-06,
+      "loss": 0.2328,
+      "step": 3999
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.460119858562477e-06,
+      "loss": 0.265,
+      "step": 4000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.453592148215021e-06,
+      "loss": 0.2546,
+      "step": 4001
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.447068385379016e-06,
+      "loss": 0.2444,
+      "step": 4002
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.440548572496299e-06,
+      "loss": 0.2698,
+      "step": 4003
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.434032712007236e-06,
+      "loss": 0.2771,
+      "step": 4004
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.427520806350694e-06,
+      "loss": 0.2327,
+      "step": 4005
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4210128579640896e-06,
+      "loss": 0.2282,
+      "step": 4006
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.414508869283338e-06,
+      "loss": 0.2316,
+      "step": 4007
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.40800884274287e-06,
+      "loss": 0.2189,
+      "step": 4008
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4015127807756445e-06,
+      "loss": 0.2197,
+      "step": 4009
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.395020685813133e-06,
+      "loss": 0.2661,
+      "step": 4010
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.388532560285318e-06,
+      "loss": 0.1787,
+      "step": 4011
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.382048406620706e-06,
+      "loss": 0.226,
+      "step": 4012
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.375568227246302e-06,
+      "loss": 0.3043,
+      "step": 4013
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.369092024587628e-06,
+      "loss": 0.1794,
+      "step": 4014
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.36261980106874e-06,
+      "loss": 0.2161,
+      "step": 4015
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.356151559112167e-06,
+      "loss": 0.2424,
+      "step": 4016
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.349687301138974e-06,
+      "loss": 0.2083,
+      "step": 4017
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3432270295687325e-06,
+      "loss": 0.2176,
+      "step": 4018
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.336770746819499e-06,
+      "loss": 0.267,
+      "step": 4019
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3303184553078766e-06,
+      "loss": 0.2527,
+      "step": 4020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.323870157448938e-06,
+      "loss": 0.2454,
+      "step": 4021
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.317425855656279e-06,
+      "loss": 0.2315,
+      "step": 4022
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.310985552341998e-06,
+      "loss": 0.2299,
+      "step": 4023
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.304549249916692e-06,
+      "loss": 0.2653,
+      "step": 4024
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.298116950789467e-06,
+      "loss": 0.2314,
+      "step": 4025
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.29168865736793e-06,
+      "loss": 0.198,
+      "step": 4026
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.285264372058176e-06,
+      "loss": 0.2795,
+      "step": 4027
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.278844097264817e-06,
+      "loss": 0.2571,
+      "step": 4028
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.272427835390951e-06,
+      "loss": 0.2611,
+      "step": 4029
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.266015588838185e-06,
+      "loss": 0.197,
+      "step": 4030
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.259607360006614e-06,
+      "loss": 0.2218,
+      "step": 4031
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.253203151294833e-06,
+      "loss": 0.2667,
+      "step": 4032
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.246802965099933e-06,
+      "loss": 0.3,
+      "step": 4033
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2404068038175015e-06,
+      "loss": 0.269,
+      "step": 4034
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.234014669841608e-06,
+      "loss": 0.2108,
+      "step": 4035
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2276265655648275e-06,
+      "loss": 0.1919,
+      "step": 4036
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.221242493378221e-06,
+      "loss": 0.2524,
+      "step": 4037
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.214862455671342e-06,
+      "loss": 0.1859,
+      "step": 4038
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2084864548322325e-06,
+      "loss": 0.2354,
+      "step": 4039
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.20211449324743e-06,
+      "loss": 0.2638,
+      "step": 4040
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.195746573301939e-06,
+      "loss": 0.3311,
+      "step": 4041
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.189382697379286e-06,
+      "loss": 0.2133,
+      "step": 4042
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.18302286786145e-06,
+      "loss": 0.2365,
+      "step": 4043
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.176667087128916e-06,
+      "loss": 0.22,
+      "step": 4044
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.170315357560647e-06,
+      "loss": 0.2731,
+      "step": 4045
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1639676815340825e-06,
+      "loss": 0.2494,
+      "step": 4046
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.157624061425161e-06,
+      "loss": 0.2927,
+      "step": 4047
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.151284499608298e-06,
+      "loss": 0.2694,
+      "step": 4048
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1449489984563746e-06,
+      "loss": 0.2589,
+      "step": 4049
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.138617560340768e-06,
+      "loss": 0.2221,
+      "step": 4050
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.13229018763133e-06,
+      "loss": 0.219,
+      "step": 4051
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.125966882696392e-06,
+      "loss": 0.292,
+      "step": 4052
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.119647647902767e-06,
+      "loss": 0.2797,
+      "step": 4053
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.113332485615726e-06,
+      "loss": 0.2059,
+      "step": 4054
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.107021398199034e-06,
+      "loss": 0.2716,
+      "step": 4055
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.100714388014934e-06,
+      "loss": 0.2723,
+      "step": 4056
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.094411457424126e-06,
+      "loss": 0.2346,
+      "step": 4057
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.088112608785792e-06,
+      "loss": 0.2125,
+      "step": 4058
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.081817844457589e-06,
+      "loss": 0.2991,
+      "step": 4059
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.075527166795639e-06,
+      "loss": 0.1964,
+      "step": 4060
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.069240578154544e-06,
+      "loss": 0.2401,
+      "step": 4061
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.062958080887358e-06,
+      "loss": 0.2382,
+      "step": 4062
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.05667967734562e-06,
+      "loss": 0.2387,
+      "step": 4063
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.050405369879331e-06,
+      "loss": 0.1945,
+      "step": 4064
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.044135160836959e-06,
+      "loss": 0.1846,
+      "step": 4065
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.037869052565436e-06,
+      "loss": 0.2416,
+      "step": 4066
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.031607047410168e-06,
+      "loss": 0.2283,
+      "step": 4067
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.025349147715004e-06,
+      "loss": 0.2678,
+      "step": 4068
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.019095355822284e-06,
+      "loss": 0.286,
+      "step": 4069
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.012845674072797e-06,
+      "loss": 0.2922,
+      "step": 4070
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0066001048057836e-06,
+      "loss": 0.2179,
+      "step": 4071
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0003586503589586e-06,
+      "loss": 0.2633,
+      "step": 4072
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.994121313068497e-06,
+      "loss": 0.2645,
+      "step": 4073
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.987888095269025e-06,
+      "loss": 0.2421,
+      "step": 4074
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.981658999293638e-06,
+      "loss": 0.2661,
+      "step": 4075
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.97543402747387e-06,
+      "loss": 0.2245,
+      "step": 4076
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.969213182139724e-06,
+      "loss": 0.1749,
+      "step": 4077
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.96299646561967e-06,
+      "loss": 0.271,
+      "step": 4078
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.956783880240607e-06,
+      "loss": 0.2712,
+      "step": 4079
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9505754283279035e-06,
+      "loss": 0.2802,
+      "step": 4080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.944371112205384e-06,
+      "loss": 0.2273,
+      "step": 4081
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.938170934195304e-06,
+      "loss": 0.2396,
+      "step": 4082
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.931974896618401e-06,
+      "loss": 0.2275,
+      "step": 4083
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.925783001793839e-06,
+      "loss": 0.3379,
+      "step": 4084
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.919595252039238e-06,
+      "loss": 0.2781,
+      "step": 4085
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.913411649670668e-06,
+      "loss": 0.215,
+      "step": 4086
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.907232197002646e-06,
+      "loss": 0.2132,
+      "step": 4087
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.90105689634814e-06,
+      "loss": 0.1985,
+      "step": 4088
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.89488575001856e-06,
+      "loss": 0.2195,
+      "step": 4089
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.888718760323752e-06,
+      "loss": 0.2798,
+      "step": 4090
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.882555929572023e-06,
+      "loss": 0.244,
+      "step": 4091
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.876397260070111e-06,
+      "loss": 0.3013,
+      "step": 4092
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.870242754123203e-06,
+      "loss": 0.2267,
+      "step": 4093
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8640924140349255e-06,
+      "loss": 0.2645,
+      "step": 4094
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.857946242107344e-06,
+      "loss": 0.2841,
+      "step": 4095
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.851804240640966e-06,
+      "loss": 0.2108,
+      "step": 4096
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.845666411934744e-06,
+      "loss": 0.3298,
+      "step": 4097
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.839532758286049e-06,
+      "loss": 0.2454,
+      "step": 4098
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.833403281990711e-06,
+      "loss": 0.2455,
+      "step": 4099
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.827277985342985e-06,
+      "loss": 0.2164,
+      "step": 4100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.821156870635564e-06,
+      "loss": 0.27,
+      "step": 4101
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.815039940159579e-06,
+      "loss": 0.2241,
+      "step": 4102
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.808927196204593e-06,
+      "loss": 0.249,
+      "step": 4103
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8028186410585897e-06,
+      "loss": 0.2362,
+      "step": 4104
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.796714277008014e-06,
+      "loss": 0.2322,
+      "step": 4105
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7906141063377077e-06,
+      "loss": 0.2583,
+      "step": 4106
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7845181313309677e-06,
+      "loss": 0.2752,
+      "step": 4107
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7784263542695147e-06,
+      "loss": 0.2179,
+      "step": 4108
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.772338777433482e-06,
+      "loss": 0.2291,
+      "step": 4109
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7662554031014597e-06,
+      "loss": 0.2169,
+      "step": 4110
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7601762335504487e-06,
+      "loss": 0.1952,
+      "step": 4111
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.754101271055866e-06,
+      "loss": 0.2535,
+      "step": 4112
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7480305178915725e-06,
+      "loss": 0.2719,
+      "step": 4113
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.741963976329844e-06,
+      "loss": 0.2294,
+      "step": 4114
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.735901648641383e-06,
+      "loss": 0.2791,
+      "step": 4115
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7298435370953167e-06,
+      "loss": 0.2063,
+      "step": 4116
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.723789643959182e-06,
+      "loss": 0.2867,
+      "step": 4117
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7177399714989476e-06,
+      "loss": 0.3149,
+      "step": 4118
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7116945219790137e-06,
+      "loss": 0.25,
+      "step": 4119
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7056532976621725e-06,
+      "loss": 0.2404,
+      "step": 4120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6996163008096533e-06,
+      "loss": 0.1702,
+      "step": 4121
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.693583533681105e-06,
+      "loss": 0.2169,
+      "step": 4122
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6875549985345736e-06,
+      "loss": 0.2738,
+      "step": 4123
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6815306976265466e-06,
+      "loss": 0.2177,
+      "step": 4124
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.675510633211916e-06,
+      "loss": 0.2224,
+      "step": 4125
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6694948075439774e-06,
+      "loss": 0.2817,
+      "step": 4126
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6634832228744523e-06,
+      "loss": 0.2959,
+      "step": 4127
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6574758814534737e-06,
+      "loss": 0.2391,
+      "step": 4128
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6514727855295836e-06,
+      "loss": 0.2318,
+      "step": 4129
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.645473937349742e-06,
+      "loss": 0.2694,
+      "step": 4130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.639479339159299e-06,
+      "loss": 0.2501,
+      "step": 4131
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6334889932020333e-06,
+      "loss": 0.2337,
+      "step": 4132
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6275029017201345e-06,
+      "loss": 0.2418,
+      "step": 4133
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.621521066954182e-06,
+      "loss": 0.2662,
+      "step": 4134
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6155434911431734e-06,
+      "loss": 0.2058,
+      "step": 4135
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.609570176524512e-06,
+      "loss": 0.2471,
+      "step": 4136
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6036011253340027e-06,
+      "loss": 0.2286,
+      "step": 4137
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.597636339805862e-06,
+      "loss": 0.2167,
+      "step": 4138
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5916758221726952e-06,
+      "loss": 0.2172,
+      "step": 4139
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.585719574665523e-06,
+      "loss": 0.2179,
+      "step": 4140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5797675995137635e-06,
+      "loss": 0.2535,
+      "step": 4141
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.573819898945236e-06,
+      "loss": 0.2247,
+      "step": 4142
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.567876475186161e-06,
+      "loss": 0.197,
+      "step": 4143
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5619373304611593e-06,
+      "loss": 0.2786,
+      "step": 4144
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5560024669932344e-06,
+      "loss": 0.176,
+      "step": 4145
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.55007188700382e-06,
+      "loss": 0.2433,
+      "step": 4146
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5441455927127137e-06,
+      "loss": 0.2375,
+      "step": 4147
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.538223586338126e-06,
+      "loss": 0.2571,
+      "step": 4148
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5323058700966583e-06,
+      "loss": 0.2908,
+      "step": 4149
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5263924462033075e-06,
+      "loss": 0.2959,
+      "step": 4150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5204833168714635e-06,
+      "loss": 0.2305,
+      "step": 4151
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5145784843129114e-06,
+      "loss": 0.2946,
+      "step": 4152
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5086779507378184e-06,
+      "loss": 0.1971,
+      "step": 4153
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5027817183547485e-06,
+      "loss": 0.2363,
+      "step": 4154
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4968897893706684e-06,
+      "loss": 0.2361,
+      "step": 4155
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4910021659909085e-06,
+      "loss": 0.2753,
+      "step": 4156
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.485118850419207e-06,
+      "loss": 0.2871,
+      "step": 4157
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4792398448576897e-06,
+      "loss": 0.2262,
+      "step": 4158
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.473365151506847e-06,
+      "loss": 0.2085,
+      "step": 4159
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.467494772565591e-06,
+      "loss": 0.184,
+      "step": 4160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.461628710231185e-06,
+      "loss": 0.3079,
+      "step": 4161
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.455766966699298e-06,
+      "loss": 0.2288,
+      "step": 4162
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4499095441639734e-06,
+      "loss": 0.2517,
+      "step": 4163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4440564448176385e-06,
+      "loss": 0.1875,
+      "step": 4164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.438207670851107e-06,
+      "loss": 0.2595,
+      "step": 4165
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.432363224453573e-06,
+      "loss": 0.2335,
+      "step": 4166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4265231078125958e-06,
+      "loss": 0.2225,
+      "step": 4167
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.420687323114135e-06,
+      "loss": 0.2789,
+      "step": 4168
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4148558725425184e-06,
+      "loss": 0.2371,
+      "step": 4169
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.40902875828045e-06,
+      "loss": 0.2456,
+      "step": 4170
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.403205982509023e-06,
+      "loss": 0.2655,
+      "step": 4171
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.397387547407681e-06,
+      "loss": 0.3003,
+      "step": 4172
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.391573455154272e-06,
+      "loss": 0.2507,
+      "step": 4173
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.385763707925007e-06,
+      "loss": 0.2072,
+      "step": 4174
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.379958307894463e-06,
+      "loss": 0.2516,
+      "step": 4175
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.374157257235597e-06,
+      "loss": 0.2419,
+      "step": 4176
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.368360558119738e-06,
+      "loss": 0.2275,
+      "step": 4177
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.362568212716586e-06,
+      "loss": 0.2497,
+      "step": 4178
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3567802231942125e-06,
+      "loss": 0.2031,
+      "step": 4179
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3509965917190583e-06,
+      "loss": 0.2118,
+      "step": 4180
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3452173204559205e-06,
+      "loss": 0.2532,
+      "step": 4181
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.339442411567994e-06,
+      "loss": 0.2004,
+      "step": 4182
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3336718672168058e-06,
+      "loss": 0.2861,
+      "step": 4183
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3279056895622737e-06,
+      "loss": 0.2408,
+      "step": 4184
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3221438807626747e-06,
+      "loss": 0.2387,
+      "step": 4185
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3163864429746383e-06,
+      "loss": 0.246,
+      "step": 4186
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3106333783531787e-06,
+      "loss": 0.2864,
+      "step": 4187
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3048846890516658e-06,
+      "loss": 0.2574,
+      "step": 4188
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.29914037722182e-06,
+      "loss": 0.209,
+      "step": 4189
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.293400445013734e-06,
+      "loss": 0.2446,
+      "step": 4190
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.287664894575863e-06,
+      "loss": 0.2404,
+      "step": 4191
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2819337280550167e-06,
+      "loss": 0.2981,
+      "step": 4192
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.27620694759637e-06,
+      "loss": 0.2443,
+      "step": 4193
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.270484555343444e-06,
+      "loss": 0.2799,
+      "step": 4194
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2647665534381254e-06,
+      "loss": 0.2494,
+      "step": 4195
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2590529440206683e-06,
+      "loss": 0.2121,
+      "step": 4196
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2533437292296624e-06,
+      "loss": 0.2191,
+      "step": 4197
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2476389112020634e-06,
+      "loss": 0.2854,
+      "step": 4198
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2419384920731825e-06,
+      "loss": 0.2346,
+      "step": 4199
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.236242473976679e-06,
+      "loss": 0.1909,
+      "step": 4200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2305508590445715e-06,
+      "loss": 0.2412,
+      "step": 4201
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2248636494072212e-06,
+      "loss": 0.1865,
+      "step": 4202
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2191808471933477e-06,
+      "loss": 0.2881,
+      "step": 4203
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2135024545300203e-06,
+      "loss": 0.2843,
+      "step": 4204
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2078284735426553e-06,
+      "loss": 0.2317,
+      "step": 4205
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.20215890635502e-06,
+      "loss": 0.2014,
+      "step": 4206
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1964937550892325e-06,
+      "loss": 0.2631,
+      "step": 4207
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1908330218657404e-06,
+      "loss": 0.2696,
+      "step": 4208
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1851767088033707e-06,
+      "loss": 0.235,
+      "step": 4209
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.179524818019261e-06,
+      "loss": 0.1956,
+      "step": 4210
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1738773516289142e-06,
+      "loss": 0.2364,
+      "step": 4211
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.168234311746173e-06,
+      "loss": 0.2198,
+      "step": 4212
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1625957004832214e-06,
+      "loss": 0.2717,
+      "step": 4213
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1569615199505885e-06,
+      "loss": 0.2808,
+      "step": 4214
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1513317722571468e-06,
+      "loss": 0.2758,
+      "step": 4215
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1457064595100947e-06,
+      "loss": 0.3334,
+      "step": 4216
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1400855838149873e-06,
+      "loss": 0.2535,
+      "step": 4217
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.134469147275721e-06,
+      "loss": 0.1981,
+      "step": 4218
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.128857151994514e-06,
+      "loss": 0.2097,
+      "step": 4219
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.123249600071935e-06,
+      "loss": 0.257,
+      "step": 4220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.117646493606888e-06,
+      "loss": 0.2542,
+      "step": 4221
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.112047834696601e-06,
+      "loss": 0.2736,
+      "step": 4222
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1064536254366633e-06,
+      "loss": 0.275,
+      "step": 4223
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.100863867920969e-06,
+      "loss": 0.2343,
+      "step": 4224
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.095278564241764e-06,
+      "loss": 0.2368,
+      "step": 4225
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0896977164896236e-06,
+      "loss": 0.2828,
+      "step": 4226
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.084121326753454e-06,
+      "loss": 0.2468,
+      "step": 4227
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.078549397120493e-06,
+      "loss": 0.2368,
+      "step": 4228
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.072981929676314e-06,
+      "loss": 0.2399,
+      "step": 4229
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.067418926504805e-06,
+      "loss": 0.196,
+      "step": 4230
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0618603896881995e-06,
+      "loss": 0.2686,
+      "step": 4231
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.056306321307053e-06,
+      "loss": 0.2689,
+      "step": 4232
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0507567234402465e-06,
+      "loss": 0.3055,
+      "step": 4233
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0452115981649957e-06,
+      "loss": 0.2325,
+      "step": 4234
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0396709475568234e-06,
+      "loss": 0.2737,
+      "step": 4235
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0341347736896023e-06,
+      "loss": 0.238,
+      "step": 4236
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.02860307863552e-06,
+      "loss": 0.2831,
+      "step": 4237
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0230758644650736e-06,
+      "loss": 0.2171,
+      "step": 4238
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0175531332470997e-06,
+      "loss": 0.2559,
+      "step": 4239
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.012034887048753e-06,
+      "loss": 0.2179,
+      "step": 4240
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.006521127935508e-06,
+      "loss": 0.2749,
+      "step": 4241
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.001011857971159e-06,
+      "loss": 0.2169,
+      "step": 4242
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9955070792178267e-06,
+      "loss": 0.2938,
+      "step": 4243
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.990006793735933e-06,
+      "loss": 0.2296,
+      "step": 4244
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9845110035842433e-06,
+      "loss": 0.2231,
+      "step": 4245
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.979019710819818e-06,
+      "loss": 0.2743,
+      "step": 4246
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9735329174980463e-06,
+      "loss": 0.2311,
+      "step": 4247
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9680506256726335e-06,
+      "loss": 0.2593,
+      "step": 4248
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9625728373955867e-06,
+      "loss": 0.2561,
+      "step": 4249
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9570995547172475e-06,
+      "loss": 0.1946,
+      "step": 4250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9516307796862633e-06,
+      "loss": 0.242,
+      "step": 4251
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.946166514349581e-06,
+      "loss": 0.2556,
+      "step": 4252
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9407067607524753e-06,
+      "loss": 0.23,
+      "step": 4253
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.935251520938528e-06,
+      "loss": 0.2067,
+      "step": 4254
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9298007969496305e-06,
+      "loss": 0.2298,
+      "step": 4255
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9243545908259894e-06,
+      "loss": 0.2006,
+      "step": 4256
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9189129046061045e-06,
+      "loss": 0.2649,
+      "step": 4257
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9134757403267953e-06,
+      "loss": 0.37,
+      "step": 4258
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9080431000232e-06,
+      "loss": 0.265,
+      "step": 4259
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9026149857287406e-06,
+      "loss": 0.2269,
+      "step": 4260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8971913994751565e-06,
+      "loss": 0.2761,
+      "step": 4261
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8917723432924976e-06,
+      "loss": 0.2622,
+      "step": 4262
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8863578192090994e-06,
+      "loss": 0.3191,
+      "step": 4263
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.880947829251631e-06,
+      "loss": 0.229,
+      "step": 4264
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8755423754450327e-06,
+      "loss": 0.3513,
+      "step": 4265
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.870141459812566e-06,
+      "loss": 0.204,
+      "step": 4266
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.86474508437579e-06,
+      "loss": 0.2765,
+      "step": 4267
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.859353251154563e-06,
+      "loss": 0.2887,
+      "step": 4268
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8539659621670418e-06,
+      "loss": 0.2419,
+      "step": 4269
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.84858321942969e-06,
+      "loss": 0.2759,
+      "step": 4270
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8432050249572567e-06,
+      "loss": 0.2662,
+      "step": 4271
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8378313807627926e-06,
+      "loss": 0.274,
+      "step": 4272
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8324622888576623e-06,
+      "loss": 0.2021,
+      "step": 4273
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8270977512514985e-06,
+      "loss": 0.217,
+      "step": 4274
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8217377699522494e-06,
+      "loss": 0.2804,
+      "step": 4275
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8163823469661495e-06,
+      "loss": 0.2479,
+      "step": 4276
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8110314842977293e-06,
+      "loss": 0.2595,
+      "step": 4277
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8056851839498172e-06,
+      "loss": 0.2768,
+      "step": 4278
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.800343447923519e-06,
+      "loss": 0.2605,
+      "step": 4279
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7950062782182474e-06,
+      "loss": 0.2678,
+      "step": 4280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7896736768317003e-06,
+      "loss": 0.2,
+      "step": 4281
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.784345645759867e-06,
+      "loss": 0.2174,
+      "step": 4282
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7790221869970215e-06,
+      "loss": 0.3085,
+      "step": 4283
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.773703302535737e-06,
+      "loss": 0.2151,
+      "step": 4284
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7683889943668543e-06,
+      "loss": 0.255,
+      "step": 4285
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7630792644795304e-06,
+      "loss": 0.2136,
+      "step": 4286
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.757774114861182e-06,
+      "loss": 0.239,
+      "step": 4287
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.752473547497523e-06,
+      "loss": 0.2319,
+      "step": 4288
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7471775643725578e-06,
+      "loss": 0.2568,
+      "step": 4289
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.741886167468557e-06,
+      "loss": 0.275,
+      "step": 4290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.736599358766095e-06,
+      "loss": 0.288,
+      "step": 4291
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7313171402440216e-06,
+      "loss": 0.2045,
+      "step": 4292
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.726039513879458e-06,
+      "loss": 0.3036,
+      "step": 4293
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7207664816478202e-06,
+      "loss": 0.2126,
+      "step": 4294
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.715498045522799e-06,
+      "loss": 0.2953,
+      "step": 4295
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7102342074763675e-06,
+      "loss": 0.219,
+      "step": 4296
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.704974969478771e-06,
+      "loss": 0.2422,
+      "step": 4297
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.699720333498547e-06,
+      "loss": 0.1906,
+      "step": 4298
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.694470301502487e-06,
+      "loss": 0.2694,
+      "step": 4299
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6892248754556886e-06,
+      "loss": 0.1913,
+      "step": 4300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6839840573214987e-06,
+      "loss": 0.2478,
+      "step": 4301
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6787478490615557e-06,
+      "loss": 0.2553,
+      "step": 4302
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6735162526357694e-06,
+      "loss": 0.2507,
+      "step": 4303
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6682892700023194e-06,
+      "loss": 0.2028,
+      "step": 4304
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.663066903117661e-06,
+      "loss": 0.217,
+      "step": 4305
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.657849153936525e-06,
+      "loss": 0.2255,
+      "step": 4306
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.652636024411907e-06,
+      "loss": 0.1818,
+      "step": 4307
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6474275164950752e-06,
+      "loss": 0.2858,
+      "step": 4308
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6422236321355736e-06,
+      "loss": 0.2433,
+      "step": 4309
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6370243732812096e-06,
+      "loss": 0.2257,
+      "step": 4310
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.631829741878066e-06,
+      "loss": 0.2153,
+      "step": 4311
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6266397398704774e-06,
+      "loss": 0.2643,
+      "step": 4312
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6214543692010684e-06,
+      "loss": 0.2255,
+      "step": 4313
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6162736318107185e-06,
+      "loss": 0.2334,
+      "step": 4314
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.611097529638566e-06,
+      "loss": 0.2618,
+      "step": 4315
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6059260646220246e-06,
+      "loss": 0.2792,
+      "step": 4316
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.600759238696771e-06,
+      "loss": 0.2693,
+      "step": 4317
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5955970537967433e-06,
+      "loss": 0.2202,
+      "step": 4318
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.590439511854144e-06,
+      "loss": 0.2982,
+      "step": 4319
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5852866147994326e-06,
+      "loss": 0.1788,
+      "step": 4320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5801383645613325e-06,
+      "loss": 0.2073,
+      "step": 4321
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5749947630668407e-06,
+      "loss": 0.2874,
+      "step": 4322
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.569855812241191e-06,
+      "loss": 0.2284,
+      "step": 4323
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.564721514007894e-06,
+      "loss": 0.2079,
+      "step": 4324
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5595918702887145e-06,
+      "loss": 0.2458,
+      "step": 4325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.554466883003663e-06,
+      "loss": 0.2883,
+      "step": 4326
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5493465540710326e-06,
+      "loss": 0.2306,
+      "step": 4327
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.544230885407353e-06,
+      "loss": 0.2051,
+      "step": 4328
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.539119878927408e-06,
+      "loss": 0.2043,
+      "step": 4329
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5340135365442515e-06,
+      "loss": 0.2833,
+      "step": 4330
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5289118601691756e-06,
+      "loss": 0.2373,
+      "step": 4331
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.523814851711741e-06,
+      "loss": 0.2202,
+      "step": 4332
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.518722513079751e-06,
+      "loss": 0.2773,
+      "step": 4333
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.513634846179261e-06,
+      "loss": 0.254,
+      "step": 4334
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5085518529145773e-06,
+      "loss": 0.2816,
+      "step": 4335
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.503473535188272e-06,
+      "loss": 0.1979,
+      "step": 4336
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4983998949011426e-06,
+      "loss": 0.2534,
+      "step": 4337
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4933309339522554e-06,
+      "loss": 0.2434,
+      "step": 4338
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4882666542389153e-06,
+      "loss": 0.252,
+      "step": 4339
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.48320705765668e-06,
+      "loss": 0.2936,
+      "step": 4340
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.478152146099353e-06,
+      "loss": 0.2653,
+      "step": 4341
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4731019214589785e-06,
+      "loss": 0.2236,
+      "step": 4342
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.468056385625853e-06,
+      "loss": 0.2371,
+      "step": 4343
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4630155404885195e-06,
+      "loss": 0.1981,
+      "step": 4344
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.457979387933759e-06,
+      "loss": 0.2675,
+      "step": 4345
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.452947929846599e-06,
+      "loss": 0.2648,
+      "step": 4346
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.447921168110314e-06,
+      "loss": 0.3018,
+      "step": 4347
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4428991046064075e-06,
+      "loss": 0.2496,
+      "step": 4348
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.437881741214647e-06,
+      "loss": 0.3026,
+      "step": 4349
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4328690798130155e-06,
+      "loss": 0.2627,
+      "step": 4350
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4278611222777534e-06,
+      "loss": 0.2483,
+      "step": 4351
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.422857870483334e-06,
+      "loss": 0.2388,
+      "step": 4352
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4178593263024693e-06,
+      "loss": 0.2444,
+      "step": 4353
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.412865491606111e-06,
+      "loss": 0.2529,
+      "step": 4354
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.40787636826345e-06,
+      "loss": 0.2656,
+      "step": 4355
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4028919581419053e-06,
+      "loss": 0.2714,
+      "step": 4356
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3979122631071383e-06,
+      "loss": 0.2496,
+      "step": 4357
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.392937285023048e-06,
+      "loss": 0.2272,
+      "step": 4358
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3879670257517616e-06,
+      "loss": 0.2551,
+      "step": 4359
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.383001487153646e-06,
+      "loss": 0.2554,
+      "step": 4360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3780406710872975e-06,
+      "loss": 0.2261,
+      "step": 4361
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3730845794095377e-06,
+      "loss": 0.2531,
+      "step": 4362
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3681332139754414e-06,
+      "loss": 0.239,
+      "step": 4363
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3631865766382873e-06,
+      "loss": 0.2059,
+      "step": 4364
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.358244669249604e-06,
+      "loss": 0.2203,
+      "step": 4365
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3533074936591424e-06,
+      "loss": 0.2477,
+      "step": 4366
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3483750517148827e-06,
+      "loss": 0.2363,
+      "step": 4367
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3434473452630346e-06,
+      "loss": 0.2256,
+      "step": 4368
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3385243761480356e-06,
+      "loss": 0.2183,
+      "step": 4369
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.333606146212544e-06,
+      "loss": 0.216,
+      "step": 4370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.328692657297452e-06,
+      "loss": 0.3109,
+      "step": 4371
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.323783911241875e-06,
+      "loss": 0.1697,
+      "step": 4372
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3188799098831502e-06,
+      "loss": 0.3146,
+      "step": 4373
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3139806550568478e-06,
+      "loss": 0.2439,
+      "step": 4374
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3090861485967417e-06,
+      "loss": 0.2207,
+      "step": 4375
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.304196392334854e-06,
+      "loss": 0.3047,
+      "step": 4376
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.299311388101414e-06,
+      "loss": 0.2671,
+      "step": 4377
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2944311377248685e-06,
+      "loss": 0.2717,
+      "step": 4378
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2895556430318965e-06,
+      "loss": 0.3075,
+      "step": 4379
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2846849058473884e-06,
+      "loss": 0.2656,
+      "step": 4380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2798189279944577e-06,
+      "loss": 0.3009,
+      "step": 4381
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.274957711294442e-06,
+      "loss": 0.2454,
+      "step": 4382
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2701012575668796e-06,
+      "loss": 0.2542,
+      "step": 4383
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.265249568629539e-06,
+      "loss": 0.1787,
+      "step": 4384
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.260402646298414e-06,
+      "loss": 0.2001,
+      "step": 4385
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2555604923876936e-06,
+      "loss": 0.3213,
+      "step": 4386
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.250723108709794e-06,
+      "loss": 0.2603,
+      "step": 4387
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2458904970753462e-06,
+      "loss": 0.2958,
+      "step": 4388
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2410626592931854e-06,
+      "loss": 0.2797,
+      "step": 4389
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.236239597170374e-06,
+      "loss": 0.2584,
+      "step": 4390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2314213125121836e-06,
+      "loss": 0.2125,
+      "step": 4391
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.226607807122084e-06,
+      "loss": 0.2415,
+      "step": 4392
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2217990828017716e-06,
+      "loss": 0.2227,
+      "step": 4393
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2169951413511457e-06,
+      "loss": 0.2335,
+      "step": 4394
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2121959845683206e-06,
+      "loss": 0.2798,
+      "step": 4395
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2074016142496168e-06,
+      "loss": 0.2491,
+      "step": 4396
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.202612032189559e-06,
+      "loss": 0.2913,
+      "step": 4397
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1978272401808806e-06,
+      "loss": 0.2404,
+      "step": 4398
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.19304724001454e-06,
+      "loss": 0.285,
+      "step": 4399
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1882720334796713e-06,
+      "loss": 0.2219,
+      "step": 4400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1835016223636377e-06,
+      "loss": 0.2061,
+      "step": 4401
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.178736008452003e-06,
+      "loss": 0.3207,
+      "step": 4402
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.17397519352852e-06,
+      "loss": 0.2401,
+      "step": 4403
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1692191793751735e-06,
+      "loss": 0.2296,
+      "step": 4404
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1644679677721245e-06,
+      "loss": 0.2759,
+      "step": 4405
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.159721560497749e-06,
+      "loss": 0.2563,
+      "step": 4406
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1549799593286267e-06,
+      "loss": 0.2509,
+      "step": 4407
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.150243166039532e-06,
+      "loss": 0.2239,
+      "step": 4408
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.145511182403445e-06,
+      "loss": 0.2802,
+      "step": 4409
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1407840101915456e-06,
+      "loss": 0.2336,
+      "step": 4410
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1360616511732027e-06,
+      "loss": 0.3306,
+      "step": 4411
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1313441071159934e-06,
+      "loss": 0.2158,
+      "step": 4412
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1266313797856947e-06,
+      "loss": 0.2144,
+      "step": 4413
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.121923470946272e-06,
+      "loss": 0.2493,
+      "step": 4414
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1172203823598944e-06,
+      "loss": 0.2011,
+      "step": 4415
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1125221157869224e-06,
+      "loss": 0.287,
+      "step": 4416
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1078286729859135e-06,
+      "loss": 0.2661,
+      "step": 4417
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.103140055713625e-06,
+      "loss": 0.2907,
+      "step": 4418
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0984562657249907e-06,
+      "loss": 0.2239,
+      "step": 4419
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.093777304773157e-06,
+      "loss": 0.2642,
+      "step": 4420
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0891031746094524e-06,
+      "loss": 0.245,
+      "step": 4421
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0844338769834014e-06,
+      "loss": 0.2687,
+      "step": 4422
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0797694136427175e-06,
+      "loss": 0.2296,
+      "step": 4423
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.075109786333309e-06,
+      "loss": 0.2844,
+      "step": 4424
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.070454996799261e-06,
+      "loss": 0.2667,
+      "step": 4425
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0658050467828677e-06,
+      "loss": 0.2329,
+      "step": 4426
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0611599380245936e-06,
+      "loss": 0.2012,
+      "step": 4427
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0565196722631036e-06,
+      "loss": 0.3251,
+      "step": 4428
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.051884251235246e-06,
+      "loss": 0.2618,
+      "step": 4429
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.047253676676046e-06,
+      "loss": 0.2667,
+      "step": 4430
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0426279503187335e-06,
+      "loss": 0.2789,
+      "step": 4431
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0380070738947153e-06,
+      "loss": 0.2108,
+      "step": 4432
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0333910491335723e-06,
+      "loss": 0.2397,
+      "step": 4433
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.028779877763083e-06,
+      "loss": 0.2118,
+      "step": 4434
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0241735615092067e-06,
+      "loss": 0.2325,
+      "step": 4435
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.019572102096081e-06,
+      "loss": 0.2477,
+      "step": 4436
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0149755012460323e-06,
+      "loss": 0.2521,
+      "step": 4437
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0103837606795595e-06,
+      "loss": 0.207,
+      "step": 4438
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0057968821153437e-06,
+      "loss": 0.2058,
+      "step": 4439
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.001214867270264e-06,
+      "loss": 0.209,
+      "step": 4440
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9966377178593513e-06,
+      "loss": 0.3163,
+      "step": 4441
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9920654355958322e-06,
+      "loss": 0.2167,
+      "step": 4442
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.987498022191111e-06,
+      "loss": 0.2368,
+      "step": 4443
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9829354793547634e-06,
+      "loss": 0.1982,
+      "step": 4444
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9783778087945465e-06,
+      "loss": 0.2083,
+      "step": 4445
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.973825012216394e-06,
+      "loss": 0.2246,
+      "step": 4446
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.969277091324412e-06,
+      "loss": 0.2307,
+      "step": 4447
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9647340478208808e-06,
+      "loss": 0.239,
+      "step": 4448
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9601958834062627e-06,
+      "loss": 0.1947,
+      "step": 4449
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9556625997791853e-06,
+      "loss": 0.2521,
+      "step": 4450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9511341986364584e-06,
+      "loss": 0.297,
+      "step": 4451
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.946610681673047e-06,
+      "loss": 0.2626,
+      "step": 4452
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.942092050582111e-06,
+      "loss": 0.231,
+      "step": 4453
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9375783070549687e-06,
+      "loss": 0.2468,
+      "step": 4454
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.933069452781105e-06,
+      "loss": 0.2886,
+      "step": 4455
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9285654894481818e-06,
+      "loss": 0.2278,
+      "step": 4456
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.924066418742031e-06,
+      "loss": 0.2589,
+      "step": 4457
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.91957224234665e-06,
+      "loss": 0.2765,
+      "step": 4458
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9150829619442084e-06,
+      "loss": 0.2678,
+      "step": 4459
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9105985792150337e-06,
+      "loss": 0.2734,
+      "step": 4460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9061190958376268e-06,
+      "loss": 0.1876,
+      "step": 4461
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9016445134886628e-06,
+      "loss": 0.2213,
+      "step": 4462
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8971748338429656e-06,
+      "loss": 0.2875,
+      "step": 4463
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8927100585735353e-06,
+      "loss": 0.2504,
+      "step": 4464
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.888250189351538e-06,
+      "loss": 0.2589,
+      "step": 4465
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8837952278462855e-06,
+      "loss": 0.2296,
+      "step": 4466
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8793451757252838e-06,
+      "loss": 0.2298,
+      "step": 4467
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8749000346541707e-06,
+      "loss": 0.2457,
+      "step": 4468
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.870459806296763e-06,
+      "loss": 0.2011,
+      "step": 4469
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8660244923150354e-06,
+      "loss": 0.2855,
+      "step": 4470
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8615940943691194e-06,
+      "loss": 0.2352,
+      "step": 4471
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8571686141173122e-06,
+      "loss": 0.2305,
+      "step": 4472
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8527480532160669e-06,
+      "loss": 0.2509,
+      "step": 4473
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.848332413319993e-06,
+      "loss": 0.2648,
+      "step": 4474
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8439216960818623e-06,
+      "loss": 0.1801,
+      "step": 4475
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.839515903152602e-06,
+      "loss": 0.2028,
+      "step": 4476
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8351150361812984e-06,
+      "loss": 0.2585,
+      "step": 4477
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8307190968151904e-06,
+      "loss": 0.2032,
+      "step": 4478
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.826328086699675e-06,
+      "loss": 0.2196,
+      "step": 4479
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8219420074783027e-06,
+      "loss": 0.2176,
+      "step": 4480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8175608607927852e-06,
+      "loss": 0.2403,
+      "step": 4481
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8131846482829722e-06,
+      "loss": 0.2336,
+      "step": 4482
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8088133715868804e-06,
+      "loss": 0.2272,
+      "step": 4483
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8044470323406758e-06,
+      "loss": 0.1968,
+      "step": 4484
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8000856321786762e-06,
+      "loss": 0.2195,
+      "step": 4485
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7957291727333475e-06,
+      "loss": 0.2626,
+      "step": 4486
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.791377655635315e-06,
+      "loss": 0.2059,
+      "step": 4487
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7870310825133347e-06,
+      "loss": 0.2734,
+      "step": 4488
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7826894549943413e-06,
+      "loss": 0.2385,
+      "step": 4489
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7783527747033901e-06,
+      "loss": 0.2577,
+      "step": 4490
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7740210432637011e-06,
+      "loss": 0.2178,
+      "step": 4491
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7696942622966423e-06,
+      "loss": 0.261,
+      "step": 4492
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7653724334217114e-06,
+      "loss": 0.2353,
+      "step": 4493
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.761055558256578e-06,
+      "loss": 0.2942,
+      "step": 4494
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7567436384170428e-06,
+      "loss": 0.3295,
+      "step": 4495
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7524366755170479e-06,
+      "loss": 0.245,
+      "step": 4496
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.748134671168689e-06,
+      "loss": 0.2479,
+      "step": 4497
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7438376269822015e-06,
+      "loss": 0.3057,
+      "step": 4498
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7395455445659686e-06,
+      "loss": 0.2651,
+      "step": 4499
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7352584255265136e-06,
+      "loss": 0.1962,
+      "step": 4500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7309762714684946e-06,
+      "loss": 0.2124,
+      "step": 4501
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.726699083994721e-06,
+      "loss": 0.2336,
+      "step": 4502
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7224268647061475e-06,
+      "loss": 0.2699,
+      "step": 4503
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.718159615201853e-06,
+      "loss": 0.2423,
+      "step": 4504
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7138973370790684e-06,
+      "loss": 0.2589,
+      "step": 4505
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7096400319331624e-06,
+      "loss": 0.2618,
+      "step": 4506
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7053877013576402e-06,
+      "loss": 0.2958,
+      "step": 4507
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.701140346944144e-06,
+      "loss": 0.2202,
+      "step": 4508
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6968979702824582e-06,
+      "loss": 0.2416,
+      "step": 4509
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6926605729604966e-06,
+      "loss": 0.1815,
+      "step": 4510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6884281565643133e-06,
+      "loss": 0.249,
+      "step": 4511
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6842007226781015e-06,
+      "loss": 0.1997,
+      "step": 4512
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.679978272884185e-06,
+      "loss": 0.2275,
+      "step": 4513
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6757608087630249e-06,
+      "loss": 0.1987,
+      "step": 4514
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.671548331893205e-06,
+      "loss": 0.184,
+      "step": 4515
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6673408438514641e-06,
+      "loss": 0.2211,
+      "step": 4516
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6631383462126593e-06,
+      "loss": 0.1958,
+      "step": 4517
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6589408405497752e-06,
+      "loss": 0.293,
+      "step": 4518
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.654748328433937e-06,
+      "loss": 0.2406,
+      "step": 4519
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6505608114344012e-06,
+      "loss": 0.2344,
+      "step": 4520
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6463782911185505e-06,
+      "loss": 0.2205,
+      "step": 4521
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.642200769051901e-06,
+      "loss": 0.2288,
+      "step": 4522
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6380282467980901e-06,
+      "loss": 0.2394,
+      "step": 4523
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6338607259188904e-06,
+      "loss": 0.2255,
+      "step": 4524
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.629698207974209e-06,
+      "loss": 0.2144,
+      "step": 4525
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.625540694522063e-06,
+      "loss": 0.2583,
+      "step": 4526
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.621388187118611e-06,
+      "loss": 0.1821,
+      "step": 4527
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6172406873181344e-06,
+      "loss": 0.2458,
+      "step": 4528
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.613098196673032e-06,
+      "loss": 0.2548,
+      "step": 4529
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.608960716733846e-06,
+      "loss": 0.2823,
+      "step": 4530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.60482824904922e-06,
+      "loss": 0.2652,
+      "step": 4531
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.600700795165938e-06,
+      "loss": 0.1791,
+      "step": 4532
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5965783566289043e-06,
+      "loss": 0.2248,
+      "step": 4533
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5924609349811398e-06,
+      "loss": 0.1705,
+      "step": 4534
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5883485317637964e-06,
+      "loss": 0.2609,
+      "step": 4535
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.584241148516143e-06,
+      "loss": 0.1625,
+      "step": 4536
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5801387867755645e-06,
+      "loss": 0.2207,
+      "step": 4537
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.576041448077573e-06,
+      "loss": 0.2537,
+      "step": 4538
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.571949133955805e-06,
+      "loss": 0.2595,
+      "step": 4539
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5678618459420036e-06,
+      "loss": 0.2611,
+      "step": 4540
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5637795855660382e-06,
+      "loss": 0.2028,
+      "step": 4541
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5597023543558997e-06,
+      "loss": 0.231,
+      "step": 4542
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5556301538376844e-06,
+      "loss": 0.2137,
+      "step": 4543
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5515629855356227e-06,
+      "loss": 0.2415,
+      "step": 4544
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.547500850972044e-06,
+      "loss": 0.2328,
+      "step": 4545
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5434437516674066e-06,
+      "loss": 0.2201,
+      "step": 4546
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5393916891402776e-06,
+      "loss": 0.2471,
+      "step": 4547
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.535344664907342e-06,
+      "loss": 0.2615,
+      "step": 4548
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5313026804833952e-06,
+      "loss": 0.2178,
+      "step": 4549
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5272657373813532e-06,
+      "loss": 0.2764,
+      "step": 4550
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5232338371122339e-06,
+      "loss": 0.2332,
+      "step": 4551
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5192069811851788e-06,
+      "loss": 0.2921,
+      "step": 4552
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5151851711074339e-06,
+      "loss": 0.2524,
+      "step": 4553
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5111684083843596e-06,
+      "loss": 0.2487,
+      "step": 4554
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5071566945194321e-06,
+      "loss": 0.2321,
+      "step": 4555
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5031500310142216e-06,
+      "loss": 0.2557,
+      "step": 4556
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4991484193684286e-06,
+      "loss": 0.2341,
+      "step": 4557
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4951518610798516e-06,
+      "loss": 0.2412,
+      "step": 4558
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4911603576443938e-06,
+      "loss": 0.2719,
+      "step": 4559
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4871739105560756e-06,
+      "loss": 0.2324,
+      "step": 4560
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.483192521307018e-06,
+      "loss": 0.2565,
+      "step": 4561
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4792161913874542e-06,
+      "loss": 0.2526,
+      "step": 4562
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4752449222857218e-06,
+      "loss": 0.207,
+      "step": 4563
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4712787154882634e-06,
+      "loss": 0.2549,
+      "step": 4564
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.46731757247962e-06,
+      "loss": 0.1987,
+      "step": 4565
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.463361494742454e-06,
+      "loss": 0.2827,
+      "step": 4566
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4594104837575157e-06,
+      "loss": 0.2289,
+      "step": 4567
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4554645410036654e-06,
+      "loss": 0.2437,
+      "step": 4568
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4515236679578718e-06,
+      "loss": 0.2664,
+      "step": 4569
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4475878660951887e-06,
+      "loss": 0.2349,
+      "step": 4570
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4436571368887936e-06,
+      "loss": 0.2809,
+      "step": 4571
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4397314818099565e-06,
+      "loss": 0.2343,
+      "step": 4572
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4358109023280375e-06,
+      "loss": 0.2708,
+      "step": 4573
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4318953999105123e-06,
+      "loss": 0.2401,
+      "step": 4574
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4279849760229485e-06,
+      "loss": 0.2009,
+      "step": 4575
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.424079632129014e-06,
+      "loss": 0.2332,
+      "step": 4576
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4201793696904807e-06,
+      "loss": 0.2772,
+      "step": 4577
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4162841901672047e-06,
+      "loss": 0.2325,
+      "step": 4578
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.412394095017151e-06,
+      "loss": 0.2362,
+      "step": 4579
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4085090856963851e-06,
+      "loss": 0.2646,
+      "step": 4580
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.404629163659057e-06,
+      "loss": 0.2408,
+      "step": 4581
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4007543303574184e-06,
+      "loss": 0.236,
+      "step": 4582
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.396884587241819e-06,
+      "loss": 0.1982,
+      "step": 4583
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.393019935760697e-06,
+      "loss": 0.2147,
+      "step": 4584
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.389160377360592e-06,
+      "loss": 0.2385,
+      "step": 4585
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.38530591348613e-06,
+      "loss": 0.2261,
+      "step": 4586
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3814565455800338e-06,
+      "loss": 0.2215,
+      "step": 4587
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3776122750831188e-06,
+      "loss": 0.2328,
+      "step": 4588
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3737731034342927e-06,
+      "loss": 0.2203,
+      "step": 4589
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3699390320705557e-06,
+      "loss": 0.2493,
+      "step": 4590
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3661100624269967e-06,
+      "loss": 0.2737,
+      "step": 4591
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.362286195936791e-06,
+      "loss": 0.2427,
+      "step": 4592
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3584674340312153e-06,
+      "loss": 0.3371,
+      "step": 4593
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3546537781396278e-06,
+      "loss": 0.2366,
+      "step": 4594
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3508452296894735e-06,
+      "loss": 0.1893,
+      "step": 4595
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3470417901062886e-06,
+      "loss": 0.2367,
+      "step": 4596
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3432434608137011e-06,
+      "loss": 0.2396,
+      "step": 4597
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3394502432334189e-06,
+      "loss": 0.2333,
+      "step": 4598
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3356621387852458e-06,
+      "loss": 0.2148,
+      "step": 4599
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.331879148887058e-06,
+      "loss": 0.2611,
+      "step": 4600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.328101274954826e-06,
+      "loss": 0.2153,
+      "step": 4601
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.324328518402614e-06,
+      "loss": 0.2428,
+      "step": 4602
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3205608806425533e-06,
+      "loss": 0.2257,
+      "step": 4603
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3167983630848684e-06,
+      "loss": 0.2338,
+      "step": 4604
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3130409671378724e-06,
+      "loss": 0.25,
+      "step": 4605
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3092886942079436e-06,
+      "loss": 0.2607,
+      "step": 4606
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3055415456995667e-06,
+      "loss": 0.2729,
+      "step": 4607
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.30179952301529e-06,
+      "loss": 0.2906,
+      "step": 4608
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.29806262755575e-06,
+      "loss": 0.2742,
+      "step": 4609
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2943308607196641e-06,
+      "loss": 0.1961,
+      "step": 4610
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.29060422390383e-06,
+      "loss": 0.2345,
+      "step": 4611
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2868827185031224e-06,
+      "loss": 0.2538,
+      "step": 4612
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.283166345910503e-06,
+      "loss": 0.2686,
+      "step": 4613
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2794551075170013e-06,
+      "loss": 0.2103,
+      "step": 4614
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2757490047117309e-06,
+      "loss": 0.2294,
+      "step": 4615
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2720480388818856e-06,
+      "loss": 0.2402,
+      "step": 4616
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2683522114127322e-06,
+      "loss": 0.1943,
+      "step": 4617
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2646615236876159e-06,
+      "loss": 0.2882,
+      "step": 4618
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.260975977087958e-06,
+      "loss": 0.2266,
+      "step": 4619
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2572955729932574e-06,
+      "loss": 0.2256,
+      "step": 4620
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2536203127810863e-06,
+      "loss": 0.2407,
+      "step": 4621
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2499501978270866e-06,
+      "loss": 0.1867,
+      "step": 4622
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2462852295049825e-06,
+      "loss": 0.2495,
+      "step": 4623
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2426254091865685e-06,
+      "loss": 0.3235,
+      "step": 4624
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.238970738241712e-06,
+      "loss": 0.2646,
+      "step": 4625
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2353212180383521e-06,
+      "loss": 0.2065,
+      "step": 4626
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2316768499425062e-06,
+      "loss": 0.2629,
+      "step": 4627
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2280376353182483e-06,
+      "loss": 0.2167,
+      "step": 4628
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.224403575527745e-06,
+      "loss": 0.2462,
+      "step": 4629
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2207746719312136e-06,
+      "loss": 0.2363,
+      "step": 4630
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2171509258869523e-06,
+      "loss": 0.2412,
+      "step": 4631
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2135323387513264e-06,
+      "loss": 0.223,
+      "step": 4632
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2099189118787657e-06,
+      "loss": 0.2549,
+      "step": 4633
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2063106466217799e-06,
+      "loss": 0.241,
+      "step": 4634
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.202707544330937e-06,
+      "loss": 0.2012,
+      "step": 4635
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1991096063548733e-06,
+      "loss": 0.2223,
+      "step": 4636
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.195516834040295e-06,
+      "loss": 0.2295,
+      "step": 4637
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1919292287319727e-06,
+      "loss": 0.243,
+      "step": 4638
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1883467917727443e-06,
+      "loss": 0.2067,
+      "step": 4639
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1847695245035167e-06,
+      "loss": 0.2385,
+      "step": 4640
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1811974282632525e-06,
+      "loss": 0.2638,
+      "step": 4641
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1776305043889858e-06,
+      "loss": 0.2065,
+      "step": 4642
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1740687542158162e-06,
+      "loss": 0.2653,
+      "step": 4643
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1705121790769014e-06,
+      "loss": 0.2161,
+      "step": 4644
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1669607803034627e-06,
+      "loss": 0.2836,
+      "step": 4645
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1634145592247885e-06,
+      "loss": 0.2452,
+      "step": 4646
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.159873517168224e-06,
+      "loss": 0.2966,
+      "step": 4647
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1563376554591837e-06,
+      "loss": 0.2759,
+      "step": 4648
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1528069754211306e-06,
+      "loss": 0.2787,
+      "step": 4649
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1492814783755978e-06,
+      "loss": 0.2445,
+      "step": 4650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1457611656421746e-06,
+      "loss": 0.2624,
+      "step": 4651
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1422460385385137e-06,
+      "loss": 0.2231,
+      "step": 4652
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.138736098380324e-06,
+      "loss": 0.2747,
+      "step": 4653
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1352313464813734e-06,
+      "loss": 0.2657,
+      "step": 4654
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1317317841534824e-06,
+      "loss": 0.2081,
+      "step": 4655
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1282374127065399e-06,
+      "loss": 0.236,
+      "step": 4656
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.124748233448486e-06,
+      "loss": 0.1901,
+      "step": 4657
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1212642476853146e-06,
+      "loss": 0.251,
+      "step": 4658
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1177854567210778e-06,
+      "loss": 0.2284,
+      "step": 4659
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1143118618578885e-06,
+      "loss": 0.2836,
+      "step": 4660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.110843464395907e-06,
+      "loss": 0.2346,
+      "step": 4661
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1073802656333548e-06,
+      "loss": 0.2123,
+      "step": 4662
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1039222668664984e-06,
+      "loss": 0.2288,
+      "step": 4663
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1004694693896654e-06,
+      "loss": 0.2003,
+      "step": 4664
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0970218744952426e-06,
+      "loss": 0.2496,
+      "step": 4665
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0935794834736524e-06,
+      "loss": 0.1645,
+      "step": 4666
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0901422976133823e-06,
+      "loss": 0.2046,
+      "step": 4667
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.086710318200972e-06,
+      "loss": 0.2407,
+      "step": 4668
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0832835465209984e-06,
+      "loss": 0.2533,
+      "step": 4669
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0798619838561109e-06,
+      "loss": 0.3126,
+      "step": 4670
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0764456314869897e-06,
+      "loss": 0.2075,
+      "step": 4671
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.073034490692374e-06,
+      "loss": 0.2226,
+      "step": 4672
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0696285627490538e-06,
+      "loss": 0.2596,
+      "step": 4673
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0662278489318593e-06,
+      "loss": 0.2242,
+      "step": 4674
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.062832350513681e-06,
+      "loss": 0.2377,
+      "step": 4675
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0594420687654504e-06,
+      "loss": 0.2473,
+      "step": 4676
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0560570049561425e-06,
+      "loss": 0.2355,
+      "step": 4677
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0526771603527867e-06,
+      "loss": 0.2218,
+      "step": 4678
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.049302536220455e-06,
+      "loss": 0.2278,
+      "step": 4679
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0459331338222683e-06,
+      "loss": 0.2351,
+      "step": 4680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0425689544193895e-06,
+      "loss": 0.2296,
+      "step": 4681
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0392099992710297e-06,
+      "loss": 0.2766,
+      "step": 4682
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.035856269634438e-06,
+      "loss": 0.3286,
+      "step": 4683
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0325077667649208e-06,
+      "loss": 0.2444,
+      "step": 4684
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0291644919158117e-06,
+      "loss": 0.29,
+      "step": 4685
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0258264463384998e-06,
+      "loss": 0.2388,
+      "step": 4686
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0224936312824123e-06,
+      "loss": 0.2299,
+      "step": 4687
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0191660479950199e-06,
+      "loss": 0.2252,
+      "step": 4688
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.015843697721831e-06,
+      "loss": 0.2063,
+      "step": 4689
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0125265817064028e-06,
+      "loss": 0.2587,
+      "step": 4690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0092147011903252e-06,
+      "loss": 0.2639,
+      "step": 4691
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0059080574132334e-06,
+      "loss": 0.2122,
+      "step": 4692
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0026066516128023e-06,
+      "loss": 0.2667,
+      "step": 4693
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.993104850247454e-07,
+      "loss": 0.2308,
+      "step": 4694
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.960195588828153e-07,
+      "loss": 0.2472,
+      "step": 4695
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.92733874418797e-07,
+      "loss": 0.295,
+      "step": 4696
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.894534328625265e-07,
+      "loss": 0.2079,
+      "step": 4697
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.861782354418703e-07,
+      "loss": 0.2534,
+      "step": 4698
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.829082833827291e-07,
+      "loss": 0.2716,
+      "step": 4699
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.796435779090424e-07,
+      "loss": 0.3025,
+      "step": 4700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.763841202427875e-07,
+      "loss": 0.1898,
+      "step": 4701
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.73129911603977e-07,
+      "loss": 0.2177,
+      "step": 4702
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.698809532106618e-07,
+      "loss": 0.2134,
+      "step": 4703
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.666372462789186e-07,
+      "loss": 0.3065,
+      "step": 4704
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.633987920228648e-07,
+      "loss": 0.2045,
+      "step": 4705
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.601655916546558e-07,
+      "loss": 0.2676,
+      "step": 4706
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.56937646384472e-07,
+      "loss": 0.2783,
+      "step": 4707
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.5371495742053e-07,
+      "loss": 0.2736,
+      "step": 4708
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.504975259690835e-07,
+      "loss": 0.197,
+      "step": 4709
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.472853532344072e-07,
+      "loss": 0.2374,
+      "step": 4710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.44078440418823e-07,
+      "loss": 0.2111,
+      "step": 4711
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.408767887226721e-07,
+      "loss": 0.2231,
+      "step": 4712
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.376803993443278e-07,
+      "loss": 0.2081,
+      "step": 4713
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.34489273480198e-07,
+      "loss": 0.2612,
+      "step": 4714
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.313034123247171e-07,
+      "loss": 0.219,
+      "step": 4715
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.281228170703515e-07,
+      "loss": 0.2232,
+      "step": 4716
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.249474889075987e-07,
+      "loss": 0.2252,
+      "step": 4717
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.21777429024973e-07,
+      "loss": 0.2051,
+      "step": 4718
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.186126386090287e-07,
+      "loss": 0.2668,
+      "step": 4719
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.154531188443499e-07,
+      "loss": 0.2509,
+      "step": 4720
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.122988709135355e-07,
+      "loss": 0.2689,
+      "step": 4721
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.091498959972195e-07,
+      "loss": 0.2002,
+      "step": 4722
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.060061952740606e-07,
+      "loss": 0.2241,
+      "step": 4723
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.028677699207428e-07,
+      "loss": 0.2395,
+      "step": 4724
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.997346211119795e-07,
+      "loss": 0.1849,
+      "step": 4725
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.966067500204994e-07,
+      "loss": 0.2469,
+      "step": 4726
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.934841578170677e-07,
+      "loss": 0.2323,
+      "step": 4727
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.903668456704645e-07,
+      "loss": 0.192,
+      "step": 4728
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.872548147474979e-07,
+      "loss": 0.2427,
+      "step": 4729
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.841480662129981e-07,
+      "loss": 0.2634,
+      "step": 4730
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.81046601229823e-07,
+      "loss": 0.1984,
+      "step": 4731
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.77950420958839e-07,
+      "loss": 0.2278,
+      "step": 4732
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.748595265589526e-07,
+      "loss": 0.2424,
+      "step": 4733
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.717739191870782e-07,
+      "loss": 0.1969,
+      "step": 4734
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.686935999981588e-07,
+      "loss": 0.218,
+      "step": 4735
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.656185701451518e-07,
+      "loss": 0.2643,
+      "step": 4736
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.625488307790402e-07,
+      "loss": 0.2765,
+      "step": 4737
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.594843830488247e-07,
+      "loss": 0.2462,
+      "step": 4738
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.564252281015278e-07,
+      "loss": 0.3029,
+      "step": 4739
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.533713670821852e-07,
+      "loss": 0.2062,
+      "step": 4740
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.503228011338526e-07,
+      "loss": 0.1943,
+      "step": 4741
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.472795313976123e-07,
+      "loss": 0.2607,
+      "step": 4742
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.442415590125529e-07,
+      "loss": 0.2079,
+      "step": 4743
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.412088851157851e-07,
+      "loss": 0.1982,
+      "step": 4744
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.38181510842439e-07,
+      "loss": 0.1882,
+      "step": 4745
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.351594373256515e-07,
+      "loss": 0.1777,
+      "step": 4746
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.321426656965913e-07,
+      "loss": 0.2823,
+      "step": 4747
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.291311970844268e-07,
+      "loss": 0.223,
+      "step": 4748
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.261250326163511e-07,
+      "loss": 0.264,
+      "step": 4749
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.231241734175676e-07,
+      "loss": 0.2545,
+      "step": 4750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.201286206112945e-07,
+      "loss": 0.2493,
+      "step": 4751
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.171383753187667e-07,
+      "loss": 0.3053,
+      "step": 4752
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.141534386592336e-07,
+      "loss": 0.1964,
+      "step": 4753
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.111738117499468e-07,
+      "loss": 0.2888,
+      "step": 4754
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.081994957061822e-07,
+      "loss": 0.2227,
+      "step": 4755
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.052304916412229e-07,
+      "loss": 0.2433,
+      "step": 4756
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.022668006663647e-07,
+      "loss": 0.2462,
+      "step": 4757
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.993084238909188e-07,
+      "loss": 0.2528,
+      "step": 4758
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.963553624221942e-07,
+      "loss": 0.2776,
+      "step": 4759
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.934076173655269e-07,
+      "loss": 0.2544,
+      "step": 4760
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.904651898242544e-07,
+      "loss": 0.2039,
+      "step": 4761
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.875280808997221e-07,
+      "loss": 0.2117,
+      "step": 4762
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.845962916912875e-07,
+      "loss": 0.2328,
+      "step": 4763
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.816698232963176e-07,
+      "loss": 0.2151,
+      "step": 4764
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.787486768101881e-07,
+      "loss": 0.2169,
+      "step": 4765
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.758328533262809e-07,
+      "loss": 0.181,
+      "step": 4766
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.729223539359881e-07,
+      "loss": 0.229,
+      "step": 4767
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.700171797287015e-07,
+      "loss": 0.198,
+      "step": 4768
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.671173317918312e-07,
+      "loss": 0.2585,
+      "step": 4769
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.642228112107857e-07,
+      "loss": 0.2654,
+      "step": 4770
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.613336190689801e-07,
+      "loss": 0.2172,
+      "step": 4771
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.58449756447841e-07,
+      "loss": 0.2272,
+      "step": 4772
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.555712244267882e-07,
+      "loss": 0.2495,
+      "step": 4773
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.526980240832598e-07,
+      "loss": 0.2384,
+      "step": 4774
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.498301564926924e-07,
+      "loss": 0.2612,
+      "step": 4775
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.469676227285238e-07,
+      "loss": 0.1993,
+      "step": 4776
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.441104238621988e-07,
+      "loss": 0.2417,
+      "step": 4777
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.412585609631633e-07,
+      "loss": 0.2776,
+      "step": 4778
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.384120350988688e-07,
+      "loss": 0.2296,
+      "step": 4779
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.355708473347695e-07,
+      "loss": 0.2063,
+      "step": 4780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.327349987343146e-07,
+      "loss": 0.2085,
+      "step": 4781
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.299044903589603e-07,
+      "loss": 0.249,
+      "step": 4782
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.270793232681689e-07,
+      "loss": 0.2456,
+      "step": 4783
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.242594985193945e-07,
+      "loss": 0.1811,
+      "step": 4784
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.214450171680931e-07,
+      "loss": 0.2421,
+      "step": 4785
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.186358802677251e-07,
+      "loss": 0.1801,
+      "step": 4786
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.158320888697494e-07,
+      "loss": 0.3153,
+      "step": 4787
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.130336440236218e-07,
+      "loss": 0.2874,
+      "step": 4788
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.102405467767941e-07,
+      "loss": 0.2539,
+      "step": 4789
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.074527981747253e-07,
+      "loss": 0.2347,
+      "step": 4790
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.046703992608639e-07,
+      "loss": 0.2965,
+      "step": 4791
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.018933510766601e-07,
+      "loss": 0.2822,
+      "step": 4792
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.991216546615625e-07,
+      "loss": 0.267,
+      "step": 4793
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.963553110530146e-07,
+      "loss": 0.235,
+      "step": 4794
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.935943212864515e-07,
+      "loss": 0.1899,
+      "step": 4795
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.908386863953165e-07,
+      "loss": 0.2161,
+      "step": 4796
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.880884074110378e-07,
+      "loss": 0.3376,
+      "step": 4797
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.853434853630419e-07,
+      "loss": 0.23,
+      "step": 4798
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.826039212787505e-07,
+      "loss": 0.2518,
+      "step": 4799
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.798697161835815e-07,
+      "loss": 0.2524,
+      "step": 4800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.771408711009464e-07,
+      "loss": 0.2178,
+      "step": 4801
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.744173870522497e-07,
+      "loss": 0.219,
+      "step": 4802
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.716992650568859e-07,
+      "loss": 0.2242,
+      "step": 4803
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.689865061322448e-07,
+      "loss": 0.254,
+      "step": 4804
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.662791112937173e-07,
+      "loss": 0.3132,
+      "step": 4805
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.635770815546727e-07,
+      "loss": 0.2637,
+      "step": 4806
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.608804179264805e-07,
+      "loss": 0.2276,
+      "step": 4807
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.581891214184998e-07,
+      "loss": 0.2117,
+      "step": 4808
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.555031930380767e-07,
+      "loss": 0.1886,
+      "step": 4809
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.528226337905585e-07,
+      "loss": 0.2377,
+      "step": 4810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.501474446792711e-07,
+      "loss": 0.2435,
+      "step": 4811
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.474776267055366e-07,
+      "loss": 0.2756,
+      "step": 4812
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.448131808686691e-07,
+      "loss": 0.2188,
+      "step": 4813
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.421541081659627e-07,
+      "loss": 0.2183,
+      "step": 4814
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.395004095927093e-07,
+      "loss": 0.2855,
+      "step": 4815
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.368520861421862e-07,
+      "loss": 0.2423,
+      "step": 4816
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.342091388056587e-07,
+      "loss": 0.2661,
+      "step": 4817
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.315715685723772e-07,
+      "loss": 0.2669,
+      "step": 4818
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.289393764295853e-07,
+      "loss": 0.1887,
+      "step": 4819
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.263125633625067e-07,
+      "loss": 0.2794,
+      "step": 4820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.236911303543613e-07,
+      "loss": 0.1975,
+      "step": 4821
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.210750783863411e-07,
+      "loss": 0.248,
+      "step": 4822
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.184644084376361e-07,
+      "loss": 0.2186,
+      "step": 4823
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.158591214854214e-07,
+      "loss": 0.2375,
+      "step": 4824
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.132592185048486e-07,
+      "loss": 0.1727,
+      "step": 4825
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.106647004690608e-07,
+      "loss": 0.218,
+      "step": 4826
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.080755683491845e-07,
+      "loss": 0.2352,
+      "step": 4827
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.05491823114328e-07,
+      "loss": 0.2152,
+      "step": 4828
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.029134657315877e-07,
+      "loss": 0.2941,
+      "step": 4829
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.003404971660398e-07,
+      "loss": 0.2711,
+      "step": 4830
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.977729183807407e-07,
+      "loss": 0.2194,
+      "step": 4831
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.952107303367366e-07,
+      "loss": 0.204,
+      "step": 4832
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.926539339930503e-07,
+      "loss": 0.2622,
+      "step": 4833
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.901025303066898e-07,
+      "loss": 0.2173,
+      "step": 4834
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.875565202326461e-07,
+      "loss": 0.2467,
+      "step": 4835
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.850159047238801e-07,
+      "loss": 0.2437,
+      "step": 4836
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.824806847313496e-07,
+      "loss": 0.2689,
+      "step": 4837
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.799508612039855e-07,
+      "loss": 0.3071,
+      "step": 4838
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.774264350886937e-07,
+      "loss": 0.3187,
+      "step": 4839
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.749074073303684e-07,
+      "loss": 0.2397,
+      "step": 4840
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.72393778871877e-07,
+      "loss": 0.2245,
+      "step": 4841
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.698855506540701e-07,
+      "loss": 0.2489,
+      "step": 4842
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.673827236157769e-07,
+      "loss": 0.2026,
+      "step": 4843
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.648852986937997e-07,
+      "loss": 0.2331,
+      "step": 4844
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.623932768229239e-07,
+      "loss": 0.2147,
+      "step": 4845
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.599066589359136e-07,
+      "loss": 0.1948,
+      "step": 4846
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.574254459635058e-07,
+      "loss": 0.2156,
+      "step": 4847
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.549496388344144e-07,
+      "loss": 0.1989,
+      "step": 4848
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.52479238475338e-07,
+      "loss": 0.2927,
+      "step": 4849
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.500142458109353e-07,
+      "loss": 0.2367,
+      "step": 4850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.475546617638616e-07,
+      "loss": 0.2621,
+      "step": 4851
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.451004872547322e-07,
+      "loss": 0.2648,
+      "step": 4852
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.426517232021405e-07,
+      "loss": 0.271,
+      "step": 4853
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.402083705226601e-07,
+      "loss": 0.2949,
+      "step": 4854
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.377704301308345e-07,
+      "loss": 0.2258,
+      "step": 4855
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.353379029391836e-07,
+      "loss": 0.2618,
+      "step": 4856
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.329107898582008e-07,
+      "loss": 0.2701,
+      "step": 4857
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.304890917963495e-07,
+      "loss": 0.2454,
+      "step": 4858
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.280728096600679e-07,
+      "loss": 0.2681,
+      "step": 4859
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.256619443537774e-07,
+      "loss": 0.1992,
+      "step": 4860
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.232564967798531e-07,
+      "loss": 0.2106,
+      "step": 4861
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.208564678386579e-07,
+      "loss": 0.3129,
+      "step": 4862
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.184618584285166e-07,
+      "loss": 0.2272,
+      "step": 4863
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.160726694457324e-07,
+      "loss": 0.2617,
+      "step": 4864
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.136889017845764e-07,
+      "loss": 0.2766,
+      "step": 4865
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.113105563372883e-07,
+      "loss": 0.3251,
+      "step": 4866
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.089376339940827e-07,
+      "loss": 0.2538,
+      "step": 4867
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.065701356431424e-07,
+      "loss": 0.2524,
+      "step": 4868
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.042080621706186e-07,
+      "loss": 0.1976,
+      "step": 4869
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.018514144606356e-07,
+      "loss": 0.277,
+      "step": 4870
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.995001933952865e-07,
+      "loss": 0.2869,
+      "step": 4871
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.971543998546235e-07,
+      "loss": 0.2268,
+      "step": 4872
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.948140347166829e-07,
+      "loss": 0.2128,
+      "step": 4873
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.924790988574573e-07,
+      "loss": 0.2372,
+      "step": 4874
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.901495931509126e-07,
+      "loss": 0.2469,
+      "step": 4875
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.878255184689828e-07,
+      "loss": 0.2654,
+      "step": 4876
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.855068756815606e-07,
+      "loss": 0.202,
+      "step": 4877
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.831936656565184e-07,
+      "loss": 0.1761,
+      "step": 4878
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.808858892596884e-07,
+      "loss": 0.2203,
+      "step": 4879
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.78583547354865e-07,
+      "loss": 0.2472,
+      "step": 4880
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.7628664080381523e-07,
+      "loss": 0.2449,
+      "step": 4881
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.739951704662698e-07,
+      "loss": 0.278,
+      "step": 4882
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.7170913719992437e-07,
+      "loss": 0.204,
+      "step": 4883
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.694285418604377e-07,
+      "loss": 0.1926,
+      "step": 4884
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.671533853014387e-07,
+      "loss": 0.2211,
+      "step": 4885
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.648836683745078e-07,
+      "loss": 0.2263,
+      "step": 4886
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6261939192920876e-07,
+      "loss": 0.2296,
+      "step": 4887
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6036055681305034e-07,
+      "loss": 0.244,
+      "step": 4888
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.5810716387151617e-07,
+      "loss": 0.2864,
+      "step": 4889
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.558592139480483e-07,
+      "loss": 0.2162,
+      "step": 4890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.536167078840536e-07,
+      "loss": 0.213,
+      "step": 4891
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.513796465188974e-07,
+      "loss": 0.2899,
+      "step": 4892
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.491480306899148e-07,
+      "loss": 0.2524,
+      "step": 4893
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4692186123239276e-07,
+      "loss": 0.2294,
+      "step": 4894
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.447011389795863e-07,
+      "loss": 0.2537,
+      "step": 4895
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4248586476271057e-07,
+      "loss": 0.2404,
+      "step": 4896
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4027603941094063e-07,
+      "loss": 0.2469,
+      "step": 4897
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3807166375141315e-07,
+      "loss": 0.2898,
+      "step": 4898
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3587273860921985e-07,
+      "loss": 0.2198,
+      "step": 4899
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.336792648074206e-07,
+      "loss": 0.2141,
+      "step": 4900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3149124316703036e-07,
+      "loss": 0.1883,
+      "step": 4901
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2930867450702405e-07,
+      "loss": 0.286,
+      "step": 4902
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.271315596443315e-07,
+      "loss": 0.2816,
+      "step": 4903
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2495989939384916e-07,
+      "loss": 0.22,
+      "step": 4904
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2279369456842687e-07,
+      "loss": 0.2656,
+      "step": 4905
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.206329459788744e-07,
+      "loss": 0.25,
+      "step": 4906
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.18477654433953e-07,
+      "loss": 0.2009,
+      "step": 4907
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.163278207403892e-07,
+      "loss": 0.1882,
+      "step": 4908
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.141834457028676e-07,
+      "loss": 0.2917,
+      "step": 4909
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.1204453012401944e-07,
+      "loss": 0.2693,
+      "step": 4910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.099110748044443e-07,
+      "loss": 0.2529,
+      "step": 4911
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0778308054269186e-07,
+      "loss": 0.2276,
+      "step": 4912
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.056605481352649e-07,
+      "loss": 0.265,
+      "step": 4913
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0354347837662784e-07,
+      "loss": 0.2259,
+      "step": 4914
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.014318720592003e-07,
+      "loss": 0.215,
+      "step": 4915
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.993257299733516e-07,
+      "loss": 0.3075,
+      "step": 4916
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9722505290741116e-07,
+      "loss": 0.2734,
+      "step": 4917
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.951298416476584e-07,
+      "loss": 0.2476,
+      "step": 4918
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.930400969783293e-07,
+      "loss": 0.2264,
+      "step": 4919
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9095581968161655e-07,
+      "loss": 0.2354,
+      "step": 4920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8887701053765933e-07,
+      "loss": 0.2145,
+      "step": 4921
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.868036703245553e-07,
+      "loss": 0.2289,
+      "step": 4922
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8473579981835707e-07,
+      "loss": 0.2512,
+      "step": 4923
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.826733997930637e-07,
+      "loss": 0.2388,
+      "step": 4924
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.806164710206295e-07,
+      "loss": 0.203,
+      "step": 4925
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.785650142709618e-07,
+      "loss": 0.2317,
+      "step": 4926
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.765190303119198e-07,
+      "loss": 0.2607,
+      "step": 4927
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7447851990931425e-07,
+      "loss": 0.2639,
+      "step": 4928
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.724434838269025e-07,
+      "loss": 0.2904,
+      "step": 4929
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7041392282640196e-07,
+      "loss": 0.2642,
+      "step": 4930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6838983766747157e-07,
+      "loss": 0.2411,
+      "step": 4931
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.66371229107727e-07,
+      "loss": 0.2365,
+      "step": 4932
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6435809790273224e-07,
+      "loss": 0.2883,
+      "step": 4933
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.623504448059994e-07,
+      "loss": 0.2219,
+      "step": 4934
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6034827056898924e-07,
+      "loss": 0.2289,
+      "step": 4935
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.583515759411171e-07,
+      "loss": 0.2266,
+      "step": 4936
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5636036166974364e-07,
+      "loss": 0.3035,
+      "step": 4937
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.543746285001759e-07,
+      "loss": 0.2415,
+      "step": 4938
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.52394377175676e-07,
+      "loss": 0.2172,
+      "step": 4939
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5041960843744783e-07,
+      "loss": 0.2458,
+      "step": 4940
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.48450323024645e-07,
+      "loss": 0.2755,
+      "step": 4941
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.464865216743729e-07,
+      "loss": 0.2407,
+      "step": 4942
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4452820512167695e-07,
+      "loss": 0.3445,
+      "step": 4943
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4257537409955574e-07,
+      "loss": 0.2521,
+      "step": 4944
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4062802933894965e-07,
+      "loss": 0.3049,
+      "step": 4945
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3868617156875224e-07,
+      "loss": 0.2234,
+      "step": 4946
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.367498015157955e-07,
+      "loss": 0.245,
+      "step": 4947
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3481891990486293e-07,
+      "loss": 0.2611,
+      "step": 4948
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.328935274586797e-07,
+      "loss": 0.2508,
+      "step": 4949
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.309736248979228e-07,
+      "loss": 0.249,
+      "step": 4950
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2905921294120566e-07,
+      "loss": 0.2565,
+      "step": 4951
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.271502923050934e-07,
+      "loss": 0.2814,
+      "step": 4952
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2524686370409453e-07,
+      "loss": 0.1848,
+      "step": 4953
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.233489278506557e-07,
+      "loss": 0.2901,
+      "step": 4954
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2145648545517716e-07,
+      "loss": 0.2329,
+      "step": 4955
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.195695372259988e-07,
+      "loss": 0.3265,
+      "step": 4956
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.176880838694024e-07,
+      "loss": 0.2131,
+      "step": 4957
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1581212608961297e-07,
+      "loss": 0.2284,
+      "step": 4958
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1394166458880216e-07,
+      "loss": 0.2734,
+      "step": 4959
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.120767000670799e-07,
+      "loss": 0.2181,
+      "step": 4960
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.102172332225045e-07,
+      "loss": 0.2018,
+      "step": 4961
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.083632647510692e-07,
+      "loss": 0.2697,
+      "step": 4962
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0651479534671233e-07,
+      "loss": 0.2231,
+      "step": 4963
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.046718257013187e-07,
+      "loss": 0.2034,
+      "step": 4964
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.028343565047065e-07,
+      "loss": 0.3384,
+      "step": 4965
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.010023884446389e-07,
+      "loss": 0.2009,
+      "step": 4966
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9917592220682246e-07,
+      "loss": 0.1962,
+      "step": 4967
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9735495847490023e-07,
+      "loss": 0.2206,
+      "step": 4968
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9553949793045874e-07,
+      "loss": 0.2173,
+      "step": 4969
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.937295412530211e-07,
+      "loss": 0.2279,
+      "step": 4970
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9192508912005365e-07,
+      "loss": 0.1826,
+      "step": 4971
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9012614220696286e-07,
+      "loss": 0.2773,
+      "step": 4972
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.883327011870901e-07,
+      "loss": 0.2446,
+      "step": 4973
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8654476673172005e-07,
+      "loss": 0.1812,
+      "step": 4974
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8476233951007734e-07,
+      "loss": 0.2341,
+      "step": 4975
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.829854201893184e-07,
+      "loss": 0.2432,
+      "step": 4976
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.812140094345461e-07,
+      "loss": 0.3038,
+      "step": 4977
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.794481079087985e-07,
+      "loss": 0.2346,
+      "step": 4978
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.7768771627304845e-07,
+      "loss": 0.2213,
+      "step": 4979
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.759328351862106e-07,
+      "loss": 0.2402,
+      "step": 4980
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.741834653051328e-07,
+      "loss": 0.1978,
+      "step": 4981
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.724396072846064e-07,
+      "loss": 0.2465,
+      "step": 4982
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.707012617773558e-07,
+      "loss": 0.2425,
+      "step": 4983
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6896842943403887e-07,
+      "loss": 0.1965,
+      "step": 4984
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.672411109032552e-07,
+      "loss": 0.2127,
+      "step": 4985
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.655193068315409e-07,
+      "loss": 0.2478,
+      "step": 4986
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6380301786336047e-07,
+      "loss": 0.2448,
+      "step": 4987
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6209224464112503e-07,
+      "loss": 0.2162,
+      "step": 4988
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6038698780517236e-07,
+      "loss": 0.2267,
+      "step": 4989
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.586872479937785e-07,
+      "loss": 0.2383,
+      "step": 4990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5699302584315633e-07,
+      "loss": 0.2803,
+      "step": 4991
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.553043219874501e-07,
+      "loss": 0.2075,
+      "step": 4992
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.53621137058741e-07,
+      "loss": 0.205,
+      "step": 4993
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5194347168704336e-07,
+      "loss": 0.2123,
+      "step": 4994
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.502713265003065e-07,
+      "loss": 0.2803,
+      "step": 4995
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.486047021244131e-07,
+      "loss": 0.238,
+      "step": 4996
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4694359918317923e-07,
+      "loss": 0.2087,
+      "step": 4997
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4528801829835243e-07,
+      "loss": 0.2896,
+      "step": 4998
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4363796008961713e-07,
+      "loss": 0.2469,
+      "step": 4999
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.419934251745892e-07,
+      "loss": 0.1943,
+      "step": 5000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4035441416881465e-07,
+      "loss": 0.1938,
+      "step": 5001
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.387209276857744e-07,
+      "loss": 0.2373,
+      "step": 5002
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3709296633688094e-07,
+      "loss": 0.2279,
+      "step": 5003
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3547053073148028e-07,
+      "loss": 0.2098,
+      "step": 5004
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3385362147684996e-07,
+      "loss": 0.2781,
+      "step": 5005
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.322422391781942e-07,
+      "loss": 0.2006,
+      "step": 5006
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.306363844386522e-07,
+      "loss": 0.2261,
+      "step": 5007
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2903605785929648e-07,
+      "loss": 0.2314,
+      "step": 5008
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2744126003912458e-07,
+      "loss": 0.2817,
+      "step": 5009
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.258519915750723e-07,
+      "loss": 0.2316,
+      "step": 5010
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2426825306199884e-07,
+      "loss": 0.2288,
+      "step": 5011
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2269004509269497e-07,
+      "loss": 0.2906,
+      "step": 5012
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2111736825788653e-07,
+      "loss": 0.2122,
+      "step": 5013
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.19550223146221e-07,
+      "loss": 0.1894,
+      "step": 5014
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.179886103442824e-07,
+      "loss": 0.2412,
+      "step": 5015
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1643253043657995e-07,
+      "loss": 0.2263,
+      "step": 5016
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1488198400555103e-07,
+      "loss": 0.2775,
+      "step": 5017
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.133369716315664e-07,
+      "loss": 0.2393,
+      "step": 5018
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1179749389292513e-07,
+      "loss": 0.2405,
+      "step": 5019
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1026355136584628e-07,
+      "loss": 0.2141,
+      "step": 5020
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0873514462448728e-07,
+      "loss": 0.2529,
+      "step": 5021
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.072122742409288e-07,
+      "loss": 0.248,
+      "step": 5022
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.056949407851799e-07,
+      "loss": 0.2678,
+      "step": 5023
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0418314482517798e-07,
+      "loss": 0.1959,
+      "step": 5024
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0267688692678366e-07,
+      "loss": 0.2212,
+      "step": 5025
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0117616765379098e-07,
+      "loss": 0.2278,
+      "step": 5026
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9968098756791897e-07,
+      "loss": 0.2635,
+      "step": 5027
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9819134722880993e-07,
+      "loss": 0.2411,
+      "step": 5028
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9670724719403453e-07,
+      "loss": 0.2151,
+      "step": 5029
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9522868801909177e-07,
+      "loss": 0.2314,
+      "step": 5030
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9375567025740558e-07,
+      "loss": 0.2522,
+      "step": 5031
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.922881944603233e-07,
+      "loss": 0.2017,
+      "step": 5032
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.908262611771222e-07,
+      "loss": 0.2347,
+      "step": 5033
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.893698709549996e-07,
+      "loss": 0.2897,
+      "step": 5034
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.879190243390827e-07,
+      "loss": 0.3201,
+      "step": 5035
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8647372187242218e-07,
+      "loss": 0.2358,
+      "step": 5036
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8503396409599527e-07,
+      "loss": 0.2681,
+      "step": 5037
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8359975154869923e-07,
+      "loss": 0.2391,
+      "step": 5038
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.82171084767358e-07,
+      "loss": 0.2712,
+      "step": 5039
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.807479642867238e-07,
+      "loss": 0.2277,
+      "step": 5040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.793303906394689e-07,
+      "loss": 0.2239,
+      "step": 5041
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7791836435618714e-07,
+      "loss": 0.2687,
+      "step": 5042
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.765118859654008e-07,
+      "loss": 0.1977,
+      "step": 5043
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7511095599355208e-07,
+      "loss": 0.2626,
+      "step": 5044
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.737155749650099e-07,
+      "loss": 0.2512,
+      "step": 5045
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7232574340206309e-07,
+      "loss": 0.2078,
+      "step": 5046
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7094146182492388e-07,
+      "loss": 0.309,
+      "step": 5047
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6956273075172779e-07,
+      "loss": 0.212,
+      "step": 5048
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.681895506985337e-07,
+      "loss": 0.3623,
+      "step": 5049
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6682192217932213e-07,
+      "loss": 0.2255,
+      "step": 5050
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.654598457059936e-07,
+      "loss": 0.2511,
+      "step": 5051
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.641033217883736e-07,
+      "loss": 0.2188,
+      "step": 5052
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6275235093420436e-07,
+      "loss": 0.2191,
+      "step": 5053
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6140693364915804e-07,
+      "loss": 0.2462,
+      "step": 5054
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.600670704368218e-07,
+      "loss": 0.1824,
+      "step": 5055
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5873276179870279e-07,
+      "loss": 0.2454,
+      "step": 5056
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5740400823423317e-07,
+      "loss": 0.2819,
+      "step": 5057
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5608081024076515e-07,
+      "loss": 0.2733,
+      "step": 5058
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5476316831357085e-07,
+      "loss": 0.2093,
+      "step": 5059
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5345108294584077e-07,
+      "loss": 0.2405,
+      "step": 5060
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.521445546286887e-07,
+      "loss": 0.2105,
+      "step": 5061
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5084358385114515e-07,
+      "loss": 0.2174,
+      "step": 5062
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4954817110016394e-07,
+      "loss": 0.2236,
+      "step": 5063
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.482583168606172e-07,
+      "loss": 0.2557,
+      "step": 5064
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4697402161529537e-07,
+      "loss": 0.252,
+      "step": 5065
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4569528584491066e-07,
+      "loss": 0.2162,
+      "step": 5066
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4442211002809015e-07,
+      "loss": 0.2108,
+      "step": 5067
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4315449464138598e-07,
+      "loss": 0.1973,
+      "step": 5068
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4189244015926196e-07,
+      "loss": 0.2648,
+      "step": 5069
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4063594705410522e-07,
+      "loss": 0.1911,
+      "step": 5070
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3938501579622122e-07,
+      "loss": 0.2344,
+      "step": 5071
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.381396468538304e-07,
+      "loss": 0.2266,
+      "step": 5072
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3689984069307493e-07,
+      "loss": 0.193,
+      "step": 5073
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3566559777801357e-07,
+      "loss": 0.1899,
+      "step": 5074
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3443691857062012e-07,
+      "loss": 0.2321,
+      "step": 5075
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.332138035307917e-07,
+      "loss": 0.1768,
+      "step": 5076
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3199625311633545e-07,
+      "loss": 0.2579,
+      "step": 5077
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.307842677829818e-07,
+      "loss": 0.2943,
+      "step": 5078
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2957784798437621e-07,
+      "loss": 0.2242,
+      "step": 5079
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2837699417207749e-07,
+      "loss": 0.2629,
+      "step": 5080
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2718170679556606e-07,
+      "loss": 0.2534,
+      "step": 5081
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2599198630223907e-07,
+      "loss": 0.2239,
+      "step": 5082
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2480783313740528e-07,
+      "loss": 0.2411,
+      "step": 5083
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2362924774429185e-07,
+      "loss": 0.3212,
+      "step": 5084
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.224562305640442e-07,
+      "loss": 0.2135,
+      "step": 5085
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2128878203571946e-07,
+      "loss": 0.2698,
+      "step": 5086
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2012690259629644e-07,
+      "loss": 0.22,
+      "step": 5087
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1897059268066058e-07,
+      "loss": 0.2748,
+      "step": 5088
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1781985272162065e-07,
+      "loss": 0.2615,
+      "step": 5089
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1667468314989882e-07,
+      "loss": 0.1978,
+      "step": 5090
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1553508439412886e-07,
+      "loss": 0.1881,
+      "step": 5091
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1440105688086122e-07,
+      "loss": 0.2214,
+      "step": 5092
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1327260103456305e-07,
+      "loss": 0.244,
+      "step": 5093
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1214971727761314e-07,
+      "loss": 0.2963,
+      "step": 5094
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1103240603030696e-07,
+      "loss": 0.2483,
+      "step": 5095
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.099206677108533e-07,
+      "loss": 0.198,
+      "step": 5096
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.088145027353743e-07,
+      "loss": 0.225,
+      "step": 5097
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0771391151790543e-07,
+      "loss": 0.2267,
+      "step": 5098
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0661889447039886e-07,
+      "loss": 0.2506,
+      "step": 5099
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0552945200271836e-07,
+      "loss": 0.2866,
+      "step": 5100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0444558452264108e-07,
+      "loss": 0.2112,
+      "step": 5101
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0336729243585585e-07,
+      "loss": 0.1881,
+      "step": 5102
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0229457614596816e-07,
+      "loss": 0.2293,
+      "step": 5103
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.012274360544968e-07,
+      "loss": 0.2421,
+      "step": 5104
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0016587256086728e-07,
+      "loss": 0.2121,
+      "step": 5105
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.91098860624251e-08,
+      "loss": 0.3038,
+      "step": 5106
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.805947695442242e-08,
+      "loss": 0.2,
+      "step": 5107
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.70146456300297e-08,
+      "loss": 0.2959,
+      "step": 5108
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.597539248032417e-08,
+      "loss": 0.3037,
+      "step": 5109
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.494171789429795e-08,
+      "loss": 0.286,
+      "step": 5110
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.391362225885492e-08,
+      "loss": 0.2693,
+      "step": 5111
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.28911059588089e-08,
+      "loss": 0.2533,
+      "step": 5112
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.187416937688875e-08,
+      "loss": 0.2174,
+      "step": 5113
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.086281289373166e-08,
+      "loss": 0.2623,
+      "step": 5114
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.985703688788816e-08,
+      "loss": 0.2224,
+      "step": 5115
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.885684173581876e-08,
+      "loss": 0.2007,
+      "step": 5116
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.786222781189734e-08,
+      "loss": 0.2476,
+      "step": 5117
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.687319548840611e-08,
+      "loss": 0.2438,
+      "step": 5118
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.588974513553727e-08,
+      "loss": 0.2021,
+      "step": 5119
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.491187712139804e-08,
+      "loss": 0.2266,
+      "step": 5120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.393959181200061e-08,
+      "loss": 0.27,
+      "step": 5121
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.297288957127391e-08,
+      "loss": 0.2745,
+      "step": 5122
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.201177076105182e-08,
+      "loss": 0.329,
+      "step": 5123
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.105623574108157e-08,
+      "loss": 0.2811,
+      "step": 5124
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.010628486901705e-08,
+      "loss": 0.2139,
+      "step": 5125
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.916191850042553e-08,
+      "loss": 0.2009,
+      "step": 5126
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.822313698878259e-08,
+      "loss": 0.2297,
+      "step": 5127
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.728994068547213e-08,
+      "loss": 0.238,
+      "step": 5128
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.636232993979141e-08,
+      "loss": 0.2432,
+      "step": 5129
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.544030509894106e-08,
+      "loss": 0.2688,
+      "step": 5130
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.4523866508035e-08,
+      "loss": 0.2099,
+      "step": 5131
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.361301451009716e-08,
+      "loss": 0.2196,
+      "step": 5132
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.270774944605819e-08,
+      "loss": 0.2347,
+      "step": 5133
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.180807165475534e-08,
+      "loss": 0.1948,
+      "step": 5134
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.091398147294093e-08,
+      "loss": 0.2232,
+      "step": 5135
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.002547923527058e-08,
+      "loss": 0.2606,
+      "step": 5136
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.914256527430995e-08,
+      "loss": 0.1974,
+      "step": 5137
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.826523992053302e-08,
+      "loss": 0.2592,
+      "step": 5138
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.739350350232043e-08,
+      "loss": 0.2162,
+      "step": 5139
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.652735634596452e-08,
+      "loss": 0.2137,
+      "step": 5140
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.566679877566262e-08,
+      "loss": 0.2393,
+      "step": 5141
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.481183111352206e-08,
+      "loss": 0.2408,
+      "step": 5142
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.396245367955189e-08,
+      "loss": 0.2107,
+      "step": 5143
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.311866679167944e-08,
+      "loss": 0.2247,
+      "step": 5144
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.228047076573041e-08,
+      "loss": 0.2369,
+      "step": 5145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.144786591544049e-08,
+      "loss": 0.2058,
+      "step": 5146
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.062085255245375e-08,
+      "loss": 0.181,
+      "step": 5147
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.979943098631923e-08,
+      "loss": 0.2126,
+      "step": 5148
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8983601524497666e-08,
+      "loss": 0.2732,
+      "step": 5149
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.817336447234978e-08,
+      "loss": 0.2697,
+      "step": 5150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.736872013314798e-08,
+      "loss": 0.2189,
+      "step": 5151
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.6569668808069686e-08,
+      "loss": 0.2919,
+      "step": 5152
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.577621079620066e-08,
+      "loss": 0.2509,
+      "step": 5153
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.498834639452832e-08,
+      "loss": 0.2646,
+      "step": 5154
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.4206075897953454e-08,
+      "loss": 0.2167,
+      "step": 5155
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.342939959927684e-08,
+      "loss": 0.2089,
+      "step": 5156
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.26583177892076e-08,
+      "loss": 0.287,
+      "step": 5157
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.189283075636153e-08,
+      "loss": 0.2433,
+      "step": 5158
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.1132938787261107e-08,
+      "loss": 0.2459,
+      "step": 5159
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.037864216633215e-08,
+      "loss": 0.2054,
+      "step": 5160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.9629941175907137e-08,
+      "loss": 0.2275,
+      "step": 5161
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.888683609622357e-08,
+      "loss": 0.2528,
+      "step": 5162
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.8149327205425645e-08,
+      "loss": 0.263,
+      "step": 5163
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.741741477956252e-08,
+      "loss": 0.2478,
+      "step": 5164
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6691099092588397e-08,
+      "loss": 0.1817,
+      "step": 5165
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5970380416362476e-08,
+      "loss": 0.1835,
+      "step": 5166
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.525525902064897e-08,
+      "loss": 0.2151,
+      "step": 5167
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.454573517311711e-08,
+      "loss": 0.3114,
+      "step": 5168
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.384180913934111e-08,
+      "loss": 0.2783,
+      "step": 5169
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.314348118280187e-08,
+      "loss": 0.2876,
+      "step": 5170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.2450751564878653e-08,
+      "loss": 0.3027,
+      "step": 5171
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.176362054486404e-08,
+      "loss": 0.219,
+      "step": 5172
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.108208837994898e-08,
+      "loss": 0.2784,
+      "step": 5173
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.04061553252294e-08,
+      "loss": 0.2237,
+      "step": 5174
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.973582163370626e-08,
+      "loss": 0.2378,
+      "step": 5175
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.9071087556285524e-08,
+      "loss": 0.1715,
+      "step": 5176
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.841195334177816e-08,
+      "loss": 0.2302,
+      "step": 5177
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.775841923689516e-08,
+      "loss": 0.203,
+      "step": 5178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.711048548625584e-08,
+      "loss": 0.1987,
+      "step": 5179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.646815233237788e-08,
+      "loss": 0.2802,
+      "step": 5180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.583142001569062e-08,
+      "loss": 0.2053,
+      "step": 5181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.520028877452008e-08,
+      "loss": 0.2488,
+      "step": 5182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.457475884509731e-08,
+      "loss": 0.2502,
+      "step": 5183
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.3954830461560006e-08,
+      "loss": 0.2686,
+      "step": 5184
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.334050385594423e-08,
+      "loss": 0.2314,
+      "step": 5185
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2731779258192706e-08,
+      "loss": 0.2436,
+      "step": 5186
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.212865689615152e-08,
+      "loss": 0.2931,
+      "step": 5187
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.1531136995566756e-08,
+      "loss": 0.2148,
+      "step": 5188
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0939219780092844e-08,
+      "loss": 0.174,
+      "step": 5189
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0352905471280887e-08,
+      "loss": 0.2295,
+      "step": 5190
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.9772194288590348e-08,
+      "loss": 0.3203,
+      "step": 5191
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.9197086449379018e-08,
+      "loss": 0.2605,
+      "step": 5192
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8627582168909704e-08,
+      "loss": 0.2762,
+      "step": 5193
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.806368166034856e-08,
+      "loss": 0.2473,
+      "step": 5194
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7505385134760087e-08,
+      "loss": 0.2502,
+      "step": 5195
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.695269280111712e-08,
+      "loss": 0.2466,
+      "step": 5196
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6405604866290845e-08,
+      "loss": 0.2596,
+      "step": 5197
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5864121535055795e-08,
+      "loss": 0.2773,
+      "step": 5198
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5328243010089848e-08,
+      "loss": 0.2532,
+      "step": 5199
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4797969491970885e-08,
+      "loss": 0.248,
+      "step": 5200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4273301179178476e-08,
+      "loss": 0.2747,
+      "step": 5201
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3754238268097196e-08,
+      "loss": 0.2552,
+      "step": 5202
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3240780953013295e-08,
+      "loss": 0.2517,
+      "step": 5203
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2732929426109718e-08,
+      "loss": 0.2153,
+      "step": 5204
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2230683877476066e-08,
+      "loss": 0.235,
+      "step": 5205
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.1734044495105298e-08,
+      "loss": 0.2396,
+      "step": 5206
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.1243011464885388e-08,
+      "loss": 0.2712,
+      "step": 5207
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.075758497061264e-08,
+      "loss": 0.2043,
+      "step": 5208
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.0277765193978392e-08,
+      "loss": 0.1967,
+      "step": 5209
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.980355231458231e-08,
+      "loss": 0.2714,
+      "step": 5210
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9334946509919072e-08,
+      "loss": 0.2526,
+      "step": 5211
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8871947955390045e-08,
+      "loss": 0.2302,
+      "step": 5212
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8414556824291605e-08,
+      "loss": 0.2628,
+      "step": 5213
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.796277328782847e-08,
+      "loss": 0.2565,
+      "step": 5214
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7516597515100374e-08,
+      "loss": 0.2231,
+      "step": 5215
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7076029673113723e-08,
+      "loss": 0.1898,
+      "step": 5216
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6641069926768283e-08,
+      "loss": 0.27,
+      "step": 5217
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6211718438872148e-08,
+      "loss": 0.2239,
+      "step": 5218
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5787975370130102e-08,
+      "loss": 0.2119,
+      "step": 5219
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5369840879148612e-08,
+      "loss": 0.251,
+      "step": 5220
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.495731512243581e-08,
+      "loss": 0.2113,
+      "step": 5221
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4550398254399855e-08,
+      "loss": 0.2458,
+      "step": 5222
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.414909042734891e-08,
+      "loss": 0.2676,
+      "step": 5223
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3753391791491154e-08,
+      "loss": 0.2267,
+      "step": 5224
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3363302494938112e-08,
+      "loss": 0.2175,
+      "step": 5225
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2978822683699653e-08,
+      "loss": 0.3129,
+      "step": 5226
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2599952501683998e-08,
+      "loss": 0.2213,
+      "step": 5227
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2226692090706037e-08,
+      "loss": 0.2161,
+      "step": 5228
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1859041590472352e-08,
+      "loss": 0.2264,
+      "step": 5229
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1497001138596197e-08,
+      "loss": 0.3199,
+      "step": 5230
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1140570870590838e-08,
+      "loss": 0.2567,
+      "step": 5231
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.078975091986456e-08,
+      "loss": 0.2504,
+      "step": 5232
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0444541417730657e-08,
+      "loss": 0.2348,
+      "step": 5233
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0104942493400771e-08,
+      "loss": 0.2222,
+      "step": 5234
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.770954273986555e-09,
+      "loss": 0.2415,
+      "step": 5235
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.442576884498011e-09,
+      "loss": 0.2538,
+      "step": 5236
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.119810447848487e-09,
+      "loss": 0.2194,
+      "step": 5237
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.80265508484801e-09,
+      "loss": 0.2248,
+      "step": 5238
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.491110914208288e-09,
+      "loss": 0.241,
+      "step": 5239
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.185178052539377e-09,
+      "loss": 0.1877,
+      "step": 5240
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.88485661435301e-09,
+      "loss": 0.244,
+      "step": 5241
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.5901467120576e-09,
+      "loss": 0.2305,
+      "step": 5242
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.301048455963244e-09,
+      "loss": 0.282,
+      "step": 5243
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.01756195428005e-09,
+      "loss": 0.2722,
+      "step": 5244
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.739687313114806e-09,
+      "loss": 0.1991,
+      "step": 5245
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.467424636477648e-09,
+      "loss": 0.2319,
+      "step": 5246
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.200774026277056e-09,
+      "loss": 0.2329,
+      "step": 5247
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.93973558231653e-09,
+      "loss": 0.1908,
+      "step": 5248
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.6843094023062424e-09,
+      "loss": 0.2219,
+      "step": 5249
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4344955818513844e-09,
+      "loss": 0.2865,
+      "step": 5250
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.190294214453828e-09,
+      "loss": 0.2117,
+      "step": 5251
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.951705391522121e-09,
+      "loss": 0.2533,
+      "step": 5252
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.718729202356497e-09,
+      "loss": 0.2217,
+      "step": 5253
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4913657341622006e-09,
+      "loss": 0.2631,
+      "step": 5254
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2696150720411556e-09,
+      "loss": 0.2689,
+      "step": 5255
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.05347729899197e-09,
+      "loss": 0.223,
+      "step": 5256
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.842952495916596e-09,
+      "loss": 0.2642,
+      "step": 5257
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.6380407416136685e-09,
+      "loss": 0.2836,
+      "step": 5258
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4387421127834993e-09,
+      "loss": 0.2034,
+      "step": 5259
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.2450566840214193e-09,
+      "loss": 0.2732,
+      "step": 5260
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0569845278227705e-09,
+      "loss": 0.2572,
+      "step": 5261
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.874525714584575e-09,
+      "loss": 0.1858,
+      "step": 5262
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.697680312602202e-09,
+      "loss": 0.2758,
+      "step": 5263
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5264483880660382e-09,
+      "loss": 0.174,
+      "step": 5264
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3608300050681484e-09,
+      "loss": 0.2126,
+      "step": 5265
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.200825225600611e-09,
+      "loss": 0.2662,
+      "step": 5266
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0464341095538517e-09,
+      "loss": 0.234,
+      "step": 5267
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8976567147149794e-09,
+      "loss": 0.2806,
+      "step": 5268
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7544930967711147e-09,
+      "loss": 0.2705,
+      "step": 5269
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6169433093077279e-09,
+      "loss": 0.2158,
+      "step": 5270
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4850074038119665e-09,
+      "loss": 0.2217,
+      "step": 5271
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3586854296659955e-09,
+      "loss": 0.2183,
+      "step": 5272
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2379774341519934e-09,
+      "loss": 0.2361,
+      "step": 5273
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1228834624488205e-09,
+      "loss": 0.2333,
+      "step": 5274
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0134035576403467e-09,
+      "loss": 0.2407,
+      "step": 5275
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.095377607021283e-10,
+      "loss": 0.2346,
+      "step": 5276
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.112861105097347e-10,
+      "loss": 0.2786,
+      "step": 5277
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.186486438420791e-10,
+      "loss": 0.2246,
+      "step": 5278
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.316253953714268e-10,
+      "loss": 0.2606,
+      "step": 5279
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.502163976700558e-10,
+      "loss": 0.2543,
+      "step": 5280
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.744216812102575e-10,
+      "loss": 0.2875,
+      "step": 5281
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0424127436100575e-10,
+      "loss": 0.2272,
+      "step": 5282
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3967520339128757e-10,
+      "loss": 0.2269,
+      "step": 5283
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.8072349246843767e-10,
+      "loss": 0.2906,
+      "step": 5284
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.273861636581387e-10,
+      "loss": 0.2178,
+      "step": 5285
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.7966323692275576e-10,
+      "loss": 0.2526,
+      "step": 5286
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.375547301279978e-10,
+      "loss": 0.2219,
+      "step": 5287
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0106065903126016e-10,
+      "loss": 0.2517,
+      "step": 5288
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.018103729661273e-11,
+      "loss": 0.276,
+      "step": 5289
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.491587647814655e-11,
+      "loss": 0.1996,
+      "step": 5290
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5265186034961752e-11,
+      "loss": 0.2118,
+      "step": 5291
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1228973322841008e-11,
+      "loss": 0.2815,
+      "step": 5292
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.807243592584108e-12,
+      "loss": 0.295,
+      "step": 5293
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.1407,
+      "step": 5294
+    },
+    {
+      "epoch": 2.0,
+      "step": 5294,
+      "total_flos": 4.0812787897435423e+18,
+      "train_loss": 0.4312453462710955,
+      "train_runtime": 38589.1017,
+      "train_samples_per_second": 17.555,
+      "train_steps_per_second": 0.137
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5294,
+  "num_train_epochs": 2,
+  "save_steps": 50000,
+  "total_flos": 4.0812787897435423e+18,
+  "trial_name": null,
+  "trial_params": null
+}